Mercurial > hgrepos > Python2 > PyMuPDF
annotate mupdf-source/thirdparty/tesseract/src/ccmain/paragraphs.cpp @ 2:b50eed0cc0ef upstream
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
The directory name has changed: no version number in the expanded directory now.
| author | Franz Glasner <fzglas.hg@dom66.de> |
|---|---|
| date | Mon, 15 Sep 2025 11:43:07 +0200 |
| parents | |
| children |
| rev | line source |
|---|---|
|
2
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1 /********************************************************************** |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2 * File: paragraphs.cpp |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
3 * Description: Paragraph detection for tesseract. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
4 * Author: David Eger |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
5 * |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
6 * (C) Copyright 2011, Google Inc. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
7 ** Licensed under the Apache License, Version 2.0 (the "License"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
8 ** you may not use this file except in compliance with the License. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
9 ** You may obtain a copy of the License at |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
10 ** http://www.apache.org/licenses/LICENSE-2.0 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
11 ** Unless required by applicable law or agreed to in writing, software |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
12 ** distributed under the License is distributed on an "AS IS" BASIS, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
13 ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
14 ** See the License for the specific language governing permissions and |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
15 ** limitations under the License. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
16 * |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
17 **********************************************************************/ |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
18 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
19 #include "paragraphs.h" |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
20 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
21 #include "helpers.h" // for UpdateRange, ClipToRange |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
22 #include "host.h" // for NearlyEqual |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
23 #include "mutableiterator.h" // for MutableIterator |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
24 #include "ocrblock.h" // for BLOCK |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
25 #include "ocrpara.h" // for ParagraphModel, PARA, PARA_IT, PARA... |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
26 #include "ocrrow.h" // for ROW |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
27 #include "pageres.h" // for PAGE_RES_IT, WERD_RES, ROW_RES, BLO... |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
28 #include "paragraphs_internal.h" // for RowScratchRegisters, SetOfModels |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
29 #include "pdblock.h" // for PDBLK |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
30 #include "polyblk.h" // for POLY_BLOCK |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
31 #include "ratngs.h" // for WERD_CHOICE |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
32 #include "rect.h" // for TBOX |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
33 #include "statistc.h" // for STATS |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
34 #include "tesserrstream.h" // for tesserr |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
35 #include "tprintf.h" // for tprintf |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
36 #include "unicharset.h" // for UNICHARSET |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
37 #include "werd.h" // for WERD, W_REP_CHAR |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
38 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
39 #include <tesseract/pageiterator.h> // for PageIterator |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
40 #include <tesseract/publictypes.h> // for JUSTIFICATION_LEFT, JUSTIFICATION_R... |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
41 #include <tesseract/unichar.h> // for UNICHAR, UNICHAR_ID |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
42 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
43 #include <algorithm> // for max |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
44 #include <cctype> // for isspace |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
45 #include <cmath> // for abs |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
46 #include <cstdio> // for snprintf |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
47 #include <cstdlib> // for abs |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
48 #include <cstring> // for strchr, strlen |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
49 #include <memory> // for unique_ptr |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
50 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
51 static const char *const kRLE = "\u202A"; // Right-to-Left Embedding |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
52 static const char *const kPDF = "\u202C"; // Pop Directional Formatting |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
53 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
54 namespace tesseract { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
55 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
56 // Special "weak" ParagraphModels. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
57 const ParagraphModel *kCrownLeft = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
58 reinterpret_cast<ParagraphModel *>(static_cast<uintptr_t>(0xDEAD111F)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
59 const ParagraphModel *kCrownRight = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
60 reinterpret_cast<ParagraphModel *>(static_cast<uintptr_t>(0xDEAD888F)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
61 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
62 // Do the text and geometry of two rows support a paragraph break between them? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
63 static bool LikelyParagraphStart(const RowScratchRegisters &before, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
64 const RowScratchRegisters &after, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
65 tesseract::ParagraphJustification j); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
66 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
67 // Given the width of a typical space between words, what is the threshold |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
68 // by which by which we think left and right alignments for paragraphs |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
69 // can vary and still be aligned. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
70 static int Epsilon(int space_pix) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
71 return space_pix * 4 / 5; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
72 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
73 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
74 static bool AcceptableRowArgs(int debug_level, int min_num_rows, const char *function_name, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
75 const std::vector<RowScratchRegisters> *rows, int row_start, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
76 int row_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
77 if (row_start < 0 || static_cast<size_t>(row_end) > rows->size() || row_start > row_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
78 tesserr << "Invalid arguments rows[" << row_start << ", " << row_end |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
79 << ") while rows is of size " << rows->size() << ".\n"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
80 return false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
81 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
82 if (row_end - row_start < min_num_rows) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
83 if (debug_level > 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
84 tprintf("# Too few rows[%d, %d) for %s.\n", row_start, row_end, function_name); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
85 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
86 return false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
87 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
88 return true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
89 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
90 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
91 // =============================== Debug Code ================================ |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
92 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
93 // Given a row-major matrix of unicode text and a column separator, print |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
94 // a formatted table. For ASCII, we get good column alignment. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
95 static void PrintTable(const std::vector<std::vector<std::string>> &rows, const char *colsep) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
96 std::vector<int> max_col_widths; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
97 for (const auto &row : rows) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
98 auto num_columns = row.size(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
99 for (size_t c = 0; c < num_columns; c++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
100 int num_unicodes = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
101 for (char i : row[c]) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
102 if ((i & 0xC0) != 0x80) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
103 num_unicodes++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
104 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
105 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
106 if (c >= max_col_widths.size()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
107 max_col_widths.push_back(num_unicodes); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
108 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
109 if (num_unicodes > max_col_widths[c]) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
110 max_col_widths[c] = num_unicodes; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
111 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
112 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
113 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
114 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
115 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
116 std::vector<std::string> col_width_patterns; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
117 col_width_patterns.reserve(max_col_widths.size()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
118 for (int max_col_width : max_col_widths) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
119 col_width_patterns.push_back(std::string("%-") + std::to_string(max_col_width) + "s"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
120 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
121 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
122 for (const auto &row : rows) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
123 for (unsigned c = 0; c < row.size(); c++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
124 if (c > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
125 tprintf("%s", colsep); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
126 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
127 tprintf(col_width_patterns[c].c_str(), row[c].c_str()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
128 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
129 tprintf("\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
130 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
131 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
132 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
133 static std::string RtlEmbed(const std::string &word, bool rtlify) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
134 if (rtlify) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
135 return std::string(kRLE) + word + std::string(kPDF); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
136 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
137 return word; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
138 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
139 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
140 // Print the current thoughts of the paragraph detector. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
141 static void PrintDetectorState(const ParagraphTheory &theory, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
142 const std::vector<RowScratchRegisters> &rows) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
143 std::vector<std::vector<std::string>> output; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
144 output.emplace_back(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
145 output.back().push_back("#row"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
146 output.back().push_back("space"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
147 output.back().push_back(".."); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
148 output.back().push_back("lword[widthSEL]"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
149 output.back().push_back("rword[widthSEL]"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
150 RowScratchRegisters::AppendDebugHeaderFields(output.back()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
151 output.back().push_back("text"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
152 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
153 for (unsigned i = 0; i < rows.size(); i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
154 output.emplace_back(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
155 std::vector<std::string> &row = output.back(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
156 const RowInfo &ri = *rows[i].ri_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
157 row.push_back(std::to_string(i)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
158 row.push_back(std::to_string(ri.average_interword_space)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
159 row.emplace_back(ri.has_leaders ? ".." : " "); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
160 row.push_back(RtlEmbed(ri.lword_text, !ri.ltr) + "[" + std::to_string(ri.lword_box.width()) + |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
161 (ri.lword_likely_starts_idea ? "S" : "s") + |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
162 (ri.lword_likely_ends_idea ? "E" : "e") + |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
163 (ri.lword_indicates_list_item ? "L" : "l") + "]"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
164 row.push_back(RtlEmbed(ri.rword_text, !ri.ltr) + "[" + std::to_string(ri.rword_box.width()) + |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
165 (ri.rword_likely_starts_idea ? "S" : "s") + |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
166 (ri.rword_likely_ends_idea ? "E" : "e") + |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
167 (ri.rword_indicates_list_item ? "L" : "l") + "]"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
168 rows[i].AppendDebugInfo(theory, row); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
169 row.push_back(RtlEmbed(ri.text, !ri.ltr)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
170 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
171 PrintTable(output, " "); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
172 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
173 tprintf("Active Paragraph Models:\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
174 unsigned m = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
175 for (const auto &model : theory.models()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
176 tprintf(" %d: %s\n", ++m, model->ToString().c_str()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
177 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
178 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
179 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
180 static void DebugDump(bool should_print, const char *phase, const ParagraphTheory &theory, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
181 const std::vector<RowScratchRegisters> &rows) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
182 if (!should_print) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
183 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
184 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
185 tprintf("# %s\n", phase); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
186 PrintDetectorState(theory, rows); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
187 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
188 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
189 // Print out the text for rows[row_start, row_end) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
190 static void PrintRowRange(const std::vector<RowScratchRegisters> &rows, int row_start, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
191 int row_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
192 tprintf("======================================\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
193 for (int row = row_start; row < row_end; row++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
194 tprintf("%s\n", rows[row].ri_->text.c_str()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
195 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
196 tprintf("======================================\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
197 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
198 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
199 // ============= Brain Dead Language Model (ASCII Version) =================== |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
200 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
201 static bool IsLatinLetter(int ch) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
202 return (ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z'); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
203 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
204 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
205 static bool IsDigitLike(int ch) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
206 return ch == 'o' || ch == 'O' || ch == 'l' || ch == 'I'; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
207 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
208 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
209 static bool IsOpeningPunct(int ch) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
210 return strchr("'\"({[", ch) != nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
211 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
212 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
213 static bool IsTerminalPunct(int ch) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
214 return strchr(":'\".?!]})", ch) != nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
215 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
216 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
217 // Return a pointer after consuming as much text as qualifies as roman numeral. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
218 static const char *SkipChars(const char *str, const char *toskip) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
219 while (*str != '\0' && strchr(toskip, *str)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
220 str++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
221 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
222 return str; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
223 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
224 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
225 static const char *SkipChars(const char *str, bool (*skip)(int)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
226 while (*str != '\0' && skip(*str)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
227 str++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
228 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
229 return str; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
230 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
231 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
232 static const char *SkipOne(const char *str, const char *toskip) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
233 if (*str != '\0' && strchr(toskip, *str)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
234 return str + 1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
235 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
236 return str; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
237 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
238 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
239 // Return whether it is very likely that this is a numeral marker that could |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
240 // start a list item. Some examples include: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
241 // A I iii. VI (2) 3.5. [C-4] |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
242 static bool LikelyListNumeral(const std::string &word) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
243 const char *kRomans = "ivxlmdIVXLMD"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
244 const char *kDigits = "012345789"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
245 const char *kOpen = "[{("; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
246 const char *kSep = ":;-.,"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
247 const char *kClose = "]})"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
248 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
249 int num_segments = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
250 const char *pos = word.c_str(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
251 while (*pos != '\0' && num_segments < 3) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
252 // skip up to two open parens. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
253 const char *numeral_start = SkipOne(SkipOne(pos, kOpen), kOpen); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
254 const char *numeral_end = SkipChars(numeral_start, kRomans); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
255 if (numeral_end != numeral_start) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
256 // Got Roman Numeral. Great. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
257 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
258 numeral_end = SkipChars(numeral_start, kDigits); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
259 if (numeral_end == numeral_start) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
260 // If there's a single latin letter, we can use that. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
261 numeral_end = SkipChars(numeral_start, IsLatinLetter); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
262 if (numeral_end - numeral_start != 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
263 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
264 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
265 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
266 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
267 // We got some sort of numeral. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
268 num_segments++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
269 // Skip any trailing parens or punctuation. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
270 pos = SkipChars(SkipChars(numeral_end, kClose), kSep); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
271 if (pos == numeral_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
272 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
273 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
274 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
275 return *pos == '\0'; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
276 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
277 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
278 static bool LikelyListMark(const std::string &word) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
279 const char *kListMarks = "0Oo*.,+."; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
280 return word.size() == 1 && strchr(kListMarks, word[0]) != nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
281 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
282 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
283 bool AsciiLikelyListItem(const std::string &word) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
284 return LikelyListMark(word) || LikelyListNumeral(word); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
285 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
286 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
287 // ========== Brain Dead Language Model (Tesseract Version) ================ |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
288 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
289 // Return the first Unicode Codepoint from werd[pos]. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
290 static int UnicodeFor(const UNICHARSET *u, const WERD_CHOICE *werd, unsigned pos) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
291 if (!u || !werd || pos > werd->length()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
292 return 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
293 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
294 return UNICHAR(u->id_to_unichar(werd->unichar_id(pos)), -1).first_uni(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
295 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
296 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
297 // A useful helper class for finding the first j >= i so that word[j] |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
298 // does not have given character type. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
299 class UnicodeSpanSkipper { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
300 public: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
301 UnicodeSpanSkipper(const UNICHARSET *unicharset, const WERD_CHOICE *word) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
302 : u_(unicharset), word_(word), wordlen_(word->length()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
303 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
304 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
305 // Given an input position, return the first position >= pos not punc. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
306 unsigned SkipPunc(unsigned pos); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
307 // Given an input position, return the first position >= pos not digit. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
308 unsigned SkipDigits(unsigned pos); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
309 // Given an input position, return the first position >= pos not roman. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
310 unsigned SkipRomans(unsigned pos); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
311 // Given an input position, return the first position >= pos not alpha. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
312 unsigned SkipAlpha(unsigned pos); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
313 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
314 private: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
315 const UNICHARSET *u_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
316 const WERD_CHOICE *word_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
317 unsigned wordlen_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
318 }; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
319 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
320 unsigned UnicodeSpanSkipper::SkipPunc(unsigned pos) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
321 while (pos < wordlen_ && u_->get_ispunctuation(word_->unichar_id(pos))) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
322 pos++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
323 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
324 return pos; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
325 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
326 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
327 unsigned UnicodeSpanSkipper::SkipDigits(unsigned pos) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
328 while (pos < wordlen_ && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
329 (u_->get_isdigit(word_->unichar_id(pos)) || IsDigitLike(UnicodeFor(u_, word_, pos)))) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
330 pos++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
331 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
332 return pos; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
333 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
334 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
335 unsigned UnicodeSpanSkipper::SkipRomans(unsigned pos) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
336 const char *kRomans = "ivxlmdIVXLMD"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
337 while (pos < wordlen_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
338 int ch = UnicodeFor(u_, word_, pos); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
339 if (ch >= 0xF0 || strchr(kRomans, ch) == nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
340 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
341 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
342 pos++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
343 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
344 return pos; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
345 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
346 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
347 unsigned UnicodeSpanSkipper::SkipAlpha(unsigned pos) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
348 while (pos < wordlen_ && u_->get_isalpha(word_->unichar_id(pos))) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
349 pos++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
350 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
351 return pos; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
352 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
353 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
354 static bool LikelyListMarkUnicode(int ch) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
355 if (ch < 0x80) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
356 std::string single_ch; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
357 single_ch += ch; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
358 return LikelyListMark(single_ch); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
359 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
360 switch (ch) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
361 // TODO(eger) expand this list of unicodes as needed. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
362 case 0x00B0: // degree sign |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
363 case 0x2022: // bullet |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
364 case 0x25E6: // white bullet |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
365 case 0x00B7: // middle dot |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
366 case 0x25A1: // white square |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
367 case 0x25A0: // black square |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
368 case 0x25AA: // black small square |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
369 case 0x2B1D: // black very small square |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
370 case 0x25BA: // black right-pointing pointer |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
371 case 0x25CF: // black circle |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
372 case 0x25CB: // white circle |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
373 return true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
374 default: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
375 break; // fall through |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
376 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
377 return false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
378 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
379 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
380 // Return whether it is very likely that this is a numeral marker that could |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
381 // start a list item. Some examples include: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
382 // A I iii. VI (2) 3.5. [C-4] |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
383 static bool UniLikelyListItem(const UNICHARSET *u, const WERD_CHOICE *werd) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
384 if (werd->length() == 1 && LikelyListMarkUnicode(UnicodeFor(u, werd, 0))) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
385 return true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
386 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
387 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
388 UnicodeSpanSkipper m(u, werd); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
389 int num_segments = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
390 unsigned pos = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
391 while (pos < werd->length() && num_segments < 3) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
392 auto numeral_start = m.SkipPunc(pos); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
393 if (numeral_start > pos + 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
394 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
395 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
396 auto numeral_end = m.SkipRomans(numeral_start); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
397 if (numeral_end == numeral_start) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
398 numeral_end = m.SkipDigits(numeral_start); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
399 if (numeral_end == numeral_start) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
400 // If there's a single latin letter, we can use that. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
401 numeral_end = m.SkipAlpha(numeral_start); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
402 if (numeral_end - numeral_start != 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
403 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
404 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
405 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
406 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
407 // We got some sort of numeral. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
408 num_segments++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
409 // Skip any trailing punctuation. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
410 pos = m.SkipPunc(numeral_end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
411 if (pos == numeral_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
412 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
413 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
414 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
415 return pos == werd->length(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
416 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
417 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
418 template<class T> |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
419 void push_back_new(std::vector<T> &vector, const T &data) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
420 if (std::find(vector.begin(), vector.end(), data) == vector.end()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
421 vector.push_back(data); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
422 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
423 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
424 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
425 // ========= Brain Dead Language Model (combined entry points) ================ |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
426 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
427 // Given the leftmost word of a line either as a Tesseract unicharset + werd |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
428 // or a utf8 string, set the following attributes for it: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
429 // is_list - this word might be a list number or bullet. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
430 // starts_idea - this word is likely to start a sentence. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
431 // ends_idea - this word is likely to end a sentence. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
432 void LeftWordAttributes(const UNICHARSET *unicharset, const WERD_CHOICE *werd, const std::string &utf8, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
433 bool *is_list, bool *starts_idea, bool *ends_idea) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
434 *is_list = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
435 *starts_idea = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
436 *ends_idea = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
437 if (utf8.empty() || (werd != nullptr && werd->empty())) { // Empty |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
438 *ends_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
439 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
440 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
441 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
442 if (unicharset && werd) { // We have a proper werd and unicharset so use it. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
443 if (UniLikelyListItem(unicharset, werd)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
444 *is_list = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
445 *starts_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
446 *ends_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
447 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
448 if (unicharset->get_isupper(werd->unichar_id(0))) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
449 *starts_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
450 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
451 if (unicharset->get_ispunctuation(werd->unichar_id(0))) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
452 *starts_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
453 *ends_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
454 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
455 } else { // Assume utf8 is mostly ASCII |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
456 if (AsciiLikelyListItem(utf8)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
457 *is_list = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
458 *starts_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
459 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
460 int start_letter = utf8[0]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
461 if (IsOpeningPunct(start_letter)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
462 *starts_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
463 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
464 if (IsTerminalPunct(start_letter)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
465 *ends_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
466 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
467 if (start_letter >= 'A' && start_letter <= 'Z') { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
468 *starts_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
469 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
470 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
471 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
472 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
473 // Given the rightmost word of a line either as a Tesseract unicharset + werd |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
474 // or a utf8 string, set the following attributes for it: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
475 // is_list - this word might be a list number or bullet. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
476 // starts_idea - this word is likely to start a sentence. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
477 // ends_idea - this word is likely to end a sentence. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
478 void RightWordAttributes(const UNICHARSET *unicharset, const WERD_CHOICE *werd, const std::string &utf8, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
479 bool *is_list, bool *starts_idea, bool *ends_idea) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
480 *is_list = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
481 *starts_idea = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
482 *ends_idea = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
483 if (utf8.empty() || (werd != nullptr && werd->empty())) { // Empty |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
484 *ends_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
485 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
486 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
487 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
488 if (unicharset && werd) { // We have a proper werd and unicharset so use it. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
489 if (UniLikelyListItem(unicharset, werd)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
490 *is_list = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
491 *starts_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
492 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
493 UNICHAR_ID last_letter = werd->unichar_id(werd->length() - 1); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
494 if (unicharset->get_ispunctuation(last_letter)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
495 *ends_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
496 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
497 } else { // Assume utf8 is mostly ASCII |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
498 if (AsciiLikelyListItem(utf8)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
499 *is_list = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
500 *starts_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
501 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
502 int last_letter = utf8[utf8.size() - 1]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
503 if (IsOpeningPunct(last_letter) || IsTerminalPunct(last_letter)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
504 *ends_idea = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
505 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
506 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
507 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
508 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
509 // =============== Implementation of RowScratchRegisters ===================== |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
510 /* static */ |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
511 void RowScratchRegisters::AppendDebugHeaderFields(std::vector<std::string> &header) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
512 header.emplace_back("[lmarg,lind;rind,rmarg]"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
513 header.emplace_back("model"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
514 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
515 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
516 void RowScratchRegisters::AppendDebugInfo(const ParagraphTheory &theory, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
517 std::vector<std::string> &dbg) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
518 char s[60]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
519 // The largest (positive and negative) numbers are reported for lindent & rindent. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
520 // While the column header has widths 5,4,4,5, it is therefore opportune to slightly |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
521 // offset the widths in the format string here to allow ample space for lindent & rindent |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
522 // while keeping the final table output nicely readable: 4,5,5,4. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
523 snprintf(s, sizeof(s), "[%4d,%5d;%5d,%4d]", lmargin_, lindent_, rindent_, rmargin_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
524 dbg.emplace_back(s); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
525 std::string model_string; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
526 model_string += static_cast<char>(GetLineType()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
527 model_string += ":"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
528 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
529 int model_numbers = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
530 for (const auto &hypothese : hypotheses_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
531 if (hypothese.model == nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
532 continue; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
533 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
534 if (model_numbers > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
535 model_string += ","; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
536 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
537 if (StrongModel(hypothese.model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
538 model_string += std::to_string(1 + theory.IndexOf(hypothese.model)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
539 } else if (hypothese.model == kCrownLeft) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
540 model_string += "CrL"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
541 } else if (hypothese.model == kCrownRight) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
542 model_string += "CrR"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
543 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
544 model_numbers++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
545 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
546 if (model_numbers == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
547 model_string += "0"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
548 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
549 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
550 dbg.push_back(model_string); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
551 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
552 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
553 void RowScratchRegisters::Init(const RowInfo &row) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
554 ri_ = &row; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
555 lmargin_ = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
556 lindent_ = row.pix_ldistance; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
557 rmargin_ = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
558 rindent_ = row.pix_rdistance; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
559 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
560 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
561 LineType RowScratchRegisters::GetLineType() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
562 if (hypotheses_.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
563 return LT_UNKNOWN; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
564 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
565 bool has_start = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
566 bool has_body = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
567 for (const auto &hypothese : hypotheses_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
568 switch (hypothese.ty) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
569 case LT_START: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
570 has_start = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
571 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
572 case LT_BODY: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
573 has_body = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
574 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
575 default: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
576 tprintf("Encountered bad value in hypothesis list: %c\n", hypothese.ty); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
577 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
578 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
579 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
580 if (has_start && has_body) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
581 return LT_MULTIPLE; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
582 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
583 return has_start ? LT_START : LT_BODY; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
584 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
585 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
586 LineType RowScratchRegisters::GetLineType(const ParagraphModel *model) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
587 if (hypotheses_.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
588 return LT_UNKNOWN; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
589 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
590 bool has_start = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
591 bool has_body = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
592 for (const auto &hypothese : hypotheses_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
593 if (hypothese.model != model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
594 continue; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
595 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
596 switch (hypothese.ty) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
597 case LT_START: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
598 has_start = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
599 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
600 case LT_BODY: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
601 has_body = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
602 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
603 default: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
604 tprintf("Encountered bad value in hypothesis list: %c\n", hypothese.ty); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
605 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
606 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
607 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
608 if (has_start && has_body) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
609 return LT_MULTIPLE; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
610 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
611 return has_start ? LT_START : LT_BODY; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
612 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
613 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
614 void RowScratchRegisters::SetStartLine() { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
615 LineType current_lt = GetLineType(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
616 if (current_lt != LT_UNKNOWN && current_lt != LT_START) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
617 tprintf("Trying to set a line to be START when it's already BODY.\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
618 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
619 if (current_lt == LT_UNKNOWN || current_lt == LT_BODY) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
620 push_back_new(hypotheses_, LineHypothesis(LT_START, nullptr)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
621 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
622 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
623 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
624 void RowScratchRegisters::SetBodyLine() { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
625 LineType current_lt = GetLineType(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
626 if (current_lt != LT_UNKNOWN && current_lt != LT_BODY) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
627 tprintf("Trying to set a line to be BODY when it's already START.\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
628 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
629 if (current_lt == LT_UNKNOWN || current_lt == LT_START) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
630 push_back_new(hypotheses_, LineHypothesis(LT_BODY, nullptr)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
631 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
632 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
633 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
634 void RowScratchRegisters::AddStartLine(const ParagraphModel *model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
635 push_back_new(hypotheses_, LineHypothesis(LT_START, model)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
636 auto found = std::find(hypotheses_.begin(), hypotheses_.end(), LineHypothesis(LT_START, nullptr)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
637 if (found != hypotheses_.end()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
638 hypotheses_.erase(found); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
639 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
640 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
641 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
642 void RowScratchRegisters::AddBodyLine(const ParagraphModel *model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
643 push_back_new(hypotheses_, LineHypothesis(LT_BODY, model)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
644 auto found = std::find(hypotheses_.begin(), hypotheses_.end(), LineHypothesis(LT_BODY, nullptr)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
645 if (found != hypotheses_.end()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
646 hypotheses_.erase(found); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
647 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
648 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
649 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
650 void RowScratchRegisters::StartHypotheses(SetOfModels *models) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
651 for (const auto &hypothese : hypotheses_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
652 if (hypothese.ty == LT_START && StrongModel(hypothese.model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
653 push_back_new(*models, hypothese.model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
654 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
655 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
656 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
657 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
658 void RowScratchRegisters::StrongHypotheses(SetOfModels *models) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
659 for (const auto &hypothese : hypotheses_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
660 if (StrongModel(hypothese.model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
661 push_back_new(*models, hypothese.model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
662 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
663 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
664 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
665 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
666 void RowScratchRegisters::NonNullHypotheses(SetOfModels *models) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
667 for (const auto &hypothese : hypotheses_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
668 if (hypothese.model != nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
669 push_back_new(*models, hypothese.model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
670 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
671 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
672 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
673 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
674 const ParagraphModel *RowScratchRegisters::UniqueStartHypothesis() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
675 if (hypotheses_.size() != 1 || hypotheses_[0].ty != LT_START) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
676 return nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
677 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
678 return hypotheses_[0].model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
679 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
680 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
681 const ParagraphModel *RowScratchRegisters::UniqueBodyHypothesis() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
682 if (hypotheses_.size() != 1 || hypotheses_[0].ty != LT_BODY) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
683 return nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
684 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
685 return hypotheses_[0].model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
686 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
687 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
688 // Discard any hypotheses whose model is not in the given list. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
689 void RowScratchRegisters::DiscardNonMatchingHypotheses(const SetOfModels &models) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
690 if (models.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
691 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
692 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
693 for (int h = hypotheses_.size() - 1; h >= 0; h--) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
694 if (!contains(models, hypotheses_[h].model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
695 hypotheses_.erase(hypotheses_.begin() + h); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
696 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
697 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
698 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
699 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
700 // ============ Geometry based Paragraph Detection Algorithm ================= |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
701 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
702 struct Cluster { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
703 Cluster() : center(0), count(0) {} |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
704 Cluster(int cen, int num) : center(cen), count(num) {} |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
705 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
706 int center; // The center of the cluster. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
707 int count; // The number of entries within the cluster. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
708 }; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
709 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
710 class SimpleClusterer { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
711 public: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
712 explicit SimpleClusterer(int max_cluster_width) : max_cluster_width_(max_cluster_width) {} |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
713 void Add(int value) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
714 values_.push_back(value); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
715 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
716 size_t size() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
717 return values_.size(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
718 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
719 void GetClusters(std::vector<Cluster> *clusters); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
720 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
721 private: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
722 int max_cluster_width_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
723 std::vector<int> values_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
724 }; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
725 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
726 // Return the index of the cluster closest to value. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
727 static int ClosestCluster(const std::vector<Cluster> &clusters, int value) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
728 unsigned best_index = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
729 for (unsigned i = 0; i < clusters.size(); i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
730 if (abs(value - clusters[i].center) < abs(value - clusters[best_index].center)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
731 best_index = i; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
732 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
733 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
734 return best_index; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
735 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
736 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
737 void SimpleClusterer::GetClusters(std::vector<Cluster> *clusters) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
738 clusters->clear(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
739 std::sort(values_.begin(), values_.end()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
740 for (unsigned i = 0; i < values_.size();) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
741 int orig_i = i; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
742 int lo = values_[i]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
743 int hi = lo; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
744 while (++i < values_.size() && values_[i] <= lo + max_cluster_width_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
745 hi = values_[i]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
746 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
747 clusters->push_back(Cluster((hi + lo) / 2, i - orig_i)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
748 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
749 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
750 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
751 // Calculate left- and right-indent tab stop values seen in |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
752 // rows[row_start, row_end) given a tolerance of tolerance. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
753 static void CalculateTabStops(std::vector<RowScratchRegisters> *rows, int row_start, int row_end, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
754 int tolerance, std::vector<Cluster> *left_tabs, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
755 std::vector<Cluster> *right_tabs) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
756 if (!AcceptableRowArgs(0, 1, __func__, rows, row_start, row_end)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
757 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
758 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
759 // First pass: toss all left and right indents into clusterers. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
760 SimpleClusterer initial_lefts(tolerance); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
761 SimpleClusterer initial_rights(tolerance); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
762 std::vector<Cluster> initial_left_tabs; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
763 std::vector<Cluster> initial_right_tabs; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
764 for (int i = row_start; i < row_end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
765 initial_lefts.Add((*rows)[i].lindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
766 initial_rights.Add((*rows)[i].rindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
767 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
768 initial_lefts.GetClusters(&initial_left_tabs); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
769 initial_rights.GetClusters(&initial_right_tabs); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
770 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
771 // Second pass: cluster only lines that are not "stray" |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
772 // An example of a stray line is a page number -- a line whose start |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
773 // and end tab-stops are far outside the typical start and end tab-stops |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
774 // for the block. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
775 // Put another way, we only cluster data from lines whose start or end |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
776 // tab stop is frequent. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
777 SimpleClusterer lefts(tolerance); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
778 SimpleClusterer rights(tolerance); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
779 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
780 // Outlier elimination. We might want to switch this to test outlier-ness |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
781 // based on how strange a position an outlier is in instead of or in addition |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
782 // to how rare it is. These outliers get re-added if we end up having too |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
783 // few tab stops, to work with, however. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
784 int infrequent_enough_to_ignore = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
785 if (row_end - row_start >= 8) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
786 infrequent_enough_to_ignore = 1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
787 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
788 if (row_end - row_start >= 20) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
789 infrequent_enough_to_ignore = 2; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
790 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
791 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
792 for (int i = row_start; i < row_end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
793 int lidx = ClosestCluster(initial_left_tabs, (*rows)[i].lindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
794 int ridx = ClosestCluster(initial_right_tabs, (*rows)[i].rindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
795 if (initial_left_tabs[lidx].count > infrequent_enough_to_ignore || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
796 initial_right_tabs[ridx].count > infrequent_enough_to_ignore) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
797 lefts.Add((*rows)[i].lindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
798 rights.Add((*rows)[i].rindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
799 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
800 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
801 lefts.GetClusters(left_tabs); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
802 rights.GetClusters(right_tabs); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
803 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
804 if ((left_tabs->size() == 1 && right_tabs->size() >= 4) || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
805 (right_tabs->size() == 1 && left_tabs->size() >= 4)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
806 // One side is really ragged, and the other only has one tab stop, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
807 // so those "insignificant outliers" are probably important, actually. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
808 // This often happens on a page of an index. Add back in the ones |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
809 // we omitted in the first pass. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
810 for (int i = row_start; i < row_end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
811 int lidx = ClosestCluster(initial_left_tabs, (*rows)[i].lindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
812 int ridx = ClosestCluster(initial_right_tabs, (*rows)[i].rindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
813 if (!(initial_left_tabs[lidx].count > infrequent_enough_to_ignore || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
814 initial_right_tabs[ridx].count > infrequent_enough_to_ignore)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
815 lefts.Add((*rows)[i].lindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
816 rights.Add((*rows)[i].rindent_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
817 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
818 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
819 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
820 lefts.GetClusters(left_tabs); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
821 rights.GetClusters(right_tabs); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
822 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
823 // If one side is almost a two-indent aligned side, and the other clearly |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
824 // isn't, try to prune out the least frequent tab stop from that side. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
825 if (left_tabs->size() == 3 && right_tabs->size() >= 4) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
826 int to_prune = -1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
827 for (int i = left_tabs->size() - 1; i >= 0; i--) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
828 if (to_prune < 0 || (*left_tabs)[i].count < (*left_tabs)[to_prune].count) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
829 to_prune = i; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
830 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
831 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
832 if (to_prune >= 0 && (*left_tabs)[to_prune].count <= infrequent_enough_to_ignore) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
833 left_tabs->erase(left_tabs->begin() + to_prune); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
834 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
835 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
836 if (right_tabs->size() == 3 && left_tabs->size() >= 4) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
837 int to_prune = -1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
838 for (int i = right_tabs->size() - 1; i >= 0; i--) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
839 if (to_prune < 0 || (*right_tabs)[i].count < (*right_tabs)[to_prune].count) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
840 to_prune = i; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
841 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
842 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
843 if (to_prune >= 0 && (*right_tabs)[to_prune].count <= infrequent_enough_to_ignore) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
844 right_tabs->erase(right_tabs->begin() + to_prune); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
845 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
846 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
847 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
848 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
849 // Given a paragraph model mark rows[row_start, row_end) as said model |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
850 // start or body lines. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
851 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
852 // Case 1: model->first_indent_ != model->body_indent_ |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
853 // Differentiating the paragraph start lines from the paragraph body lines in |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
854 // this case is easy, we just see how far each line is indented. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
855 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
856 // Case 2: model->first_indent_ == model->body_indent_ |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
857 // Here, we find end-of-paragraph lines by looking for "short lines." |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
858 // What constitutes a "short line" changes depending on whether the text |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
859 // ragged-right[left] or fully justified (aligned left and right). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
860 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
861 // Case 2a: Ragged Right (or Left) text. (eop_threshold == 0) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
862 // We have a new paragraph it the first word would have at the end |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
863 // of the previous line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
864 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
865 // Case 2b: Fully Justified. (eop_threshold > 0) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
866 // We mark a line as short (end of paragraph) if the offside indent |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
867 // is greater than eop_threshold. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
868 static void MarkRowsWithModel(std::vector<RowScratchRegisters> *rows, int row_start, int row_end, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
869 const ParagraphModel *model, bool ltr, int eop_threshold) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
870 if (!AcceptableRowArgs(0, 0, __func__, rows, row_start, row_end)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
871 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
872 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
873 for (int row = row_start; row < row_end; row++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
874 bool valid_first = ValidFirstLine(rows, row, model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
875 bool valid_body = ValidBodyLine(rows, row, model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
876 if (valid_first && !valid_body) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
877 (*rows)[row].AddStartLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
878 } else if (valid_body && !valid_first) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
879 (*rows)[row].AddBodyLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
880 } else if (valid_body && valid_first) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
881 bool after_eop = (row == row_start); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
882 if (row > row_start) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
883 if (eop_threshold > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
884 if (model->justification() == JUSTIFICATION_LEFT) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
885 after_eop = (*rows)[row - 1].rindent_ > eop_threshold; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
886 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
887 after_eop = (*rows)[row - 1].lindent_ > eop_threshold; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
888 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
889 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
890 after_eop = FirstWordWouldHaveFit((*rows)[row - 1], (*rows)[row], model->justification()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
891 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
892 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
893 if (after_eop) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
894 (*rows)[row].AddStartLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
895 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
896 (*rows)[row].AddBodyLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
897 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
898 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
899 // Do nothing. Stray row. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
900 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
901 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
902 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
903 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
904 // GeometricClassifierState holds all of the information we'll use while |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
905 // trying to determine a paragraph model for the text lines in a block of |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
906 // text: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
907 // + the rows under consideration [row_start, row_end) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
908 // + the common left- and right-indent tab stops |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
909 // + does the block start out left-to-right or right-to-left |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
910 // Further, this struct holds the data we amass for the (single) ParagraphModel |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
911 // we'll assign to the text lines (assuming we get that far). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
912 struct GeometricClassifierState { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
913 GeometricClassifierState(int dbg_level, std::vector<RowScratchRegisters> *r, int r_start, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
914 int r_end) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
915 : debug_level(dbg_level), rows(r), row_start(r_start), row_end(r_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
916 tolerance = InterwordSpace(*r, r_start, r_end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
917 CalculateTabStops(r, r_start, r_end, tolerance, &left_tabs, &right_tabs); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
918 if (debug_level >= 3) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
919 tesserr << "Geometry: TabStop cluster tolerance = " << tolerance << "; " |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
920 << left_tabs.size() << " left tabs; " |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
921 << right_tabs.size() << " right tabs\n"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
922 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
923 ltr = (*r)[r_start].ri_->ltr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
924 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
925 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
926 void AssumeLeftJustification() { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
927 just = tesseract::JUSTIFICATION_LEFT; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
928 margin = (*rows)[row_start].lmargin_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
929 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
930 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
931 void AssumeRightJustification() { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
932 just = tesseract::JUSTIFICATION_RIGHT; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
933 margin = (*rows)[row_start].rmargin_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
934 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
935 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
936 // Align tabs are the tab stops the text is aligned to. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
937 const std::vector<Cluster> &AlignTabs() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
938 if (just == tesseract::JUSTIFICATION_RIGHT) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
939 return right_tabs; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
940 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
941 return left_tabs; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
942 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
943 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
944 // Offside tabs are the tab stops opposite the tabs used to align the text. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
945 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
946 // Note that for a left-to-right text which is aligned to the right such as |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
947 // this function comment, the offside tabs are the horizontal tab stops |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
948 // marking the beginning of ("Note", "this" and "marking"). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
949 const std::vector<Cluster> &OffsideTabs() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
950 if (just == tesseract::JUSTIFICATION_RIGHT) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
951 return left_tabs; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
952 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
953 return right_tabs; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
954 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
955 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
956 // Return whether the i'th row extends from the leftmost left tab stop |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
957 // to the right most right tab stop. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
958 bool IsFullRow(int i) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
959 return ClosestCluster(left_tabs, (*rows)[i].lindent_) == 0 && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
960 ClosestCluster(right_tabs, (*rows)[i].rindent_) == 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
961 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
962 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
963 int AlignsideTabIndex(int row_idx) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
964 return ClosestCluster(AlignTabs(), (*rows)[row_idx].AlignsideIndent(just)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
965 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
966 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
967 // Given what we know about the paragraph justification (just), would the |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
968 // first word of row_b have fit at the end of row_a? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
969 bool FirstWordWouldHaveFit(int row_a, int row_b) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
970 return ::tesseract::FirstWordWouldHaveFit((*rows)[row_a], (*rows)[row_b], just); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
971 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
972 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
973 void PrintRows() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
974 PrintRowRange(*rows, row_start, row_end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
975 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
976 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
977 void Fail(int min_debug_level, const char *why) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
978 if (debug_level < min_debug_level) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
979 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
980 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
981 tprintf("# %s\n", why); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
982 PrintRows(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
983 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
984 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
985 ParagraphModel Model() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
986 return ParagraphModel(just, margin, first_indent, body_indent, tolerance); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
987 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
988 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
989 // We print out messages with a debug level at least as great as debug_level. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
990 int debug_level = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
991 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
992 // The Geometric Classifier was asked to find a single paragraph model |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
993 // to fit the text rows (*rows)[row_start, row_end) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
994 std::vector<RowScratchRegisters> *rows; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
995 int row_start = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
996 int row_end = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
997 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
998 // The amount by which we expect the text edge can vary and still be aligned. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
999 int tolerance = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1000 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1001 // Is the script in this text block left-to-right? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1002 // HORRIBLE ROUGH APPROXIMATION. TODO(eger): Improve |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1003 bool ltr = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1004 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1005 // These left and right tab stops were determined to be the common tab |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1006 // stops for the given text. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1007 std::vector<Cluster> left_tabs; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1008 std::vector<Cluster> right_tabs; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1009 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1010 // These are parameters we must determine to create a ParagraphModel. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1011 tesseract::ParagraphJustification just = JUSTIFICATION_UNKNOWN; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1012 int margin = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1013 int first_indent = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1014 int body_indent = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1015 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1016 // eop_threshold > 0 if the text is fully justified. See MarkRowsWithModel() |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1017 int eop_threshold = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1018 }; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1019 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1020 // Given a section of text where strong textual clues did not help identifying |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1021 // paragraph breaks, and for which the left and right indents have exactly |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1022 // three tab stops between them, attempt to find the paragraph breaks based |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1023 // solely on the outline of the text and whether the script is left-to-right. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1024 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1025 // Algorithm Detail: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1026 // The selected rows are in the form of a rectangle except |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1027 // for some number of "short lines" of the same length: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1028 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1029 // (A1) xxxxxxxxxxxxx (B1) xxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1030 // xxxxxxxxxxx xxxxxxxxxx # A "short" line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1031 // xxxxxxxxxxxxx xxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1032 // xxxxxxxxxxxxx xxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1033 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1034 // We have a slightly different situation if the only short |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1035 // line is at the end of the excerpt. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1036 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1037 // (A2) xxxxxxxxxxxxx (B2) xxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1038 // xxxxxxxxxxxxx xxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1039 // xxxxxxxxxxxxx xxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1040 // xxxxxxxxxxx xxxxxxxxxx # A "short" line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1041 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1042 // We'll interpret these as follows based on the reasoning in the comment for |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1043 // GeometricClassify(): |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1044 // [script direction: first indent, body indent] |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1045 // (A1) LtR: 2,0 RtL: 0,0 (B1) LtR: 0,0 RtL: 2,0 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1046 // (A2) LtR: 2,0 RtL: CrR (B2) LtR: CrL RtL: 2,0 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1047 static void GeometricClassifyThreeTabStopTextBlock(int debug_level, GeometricClassifierState &s, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1048 ParagraphTheory *theory) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1049 int num_rows = s.row_end - s.row_start; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1050 int num_full_rows = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1051 int last_row_full = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1052 for (int i = s.row_start; i < s.row_end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1053 if (s.IsFullRow(i)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1054 num_full_rows++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1055 if (i == s.row_end - 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1056 last_row_full++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1057 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1058 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1059 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1060 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1061 if (num_full_rows < 0.7 * num_rows) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1062 s.Fail(1, "Not enough full lines to know which lines start paras."); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1063 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1064 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1065 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1066 // eop_threshold gets set if we're fully justified; see MarkRowsWithModel() |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1067 s.eop_threshold = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1068 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1069 if (s.ltr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1070 s.AssumeLeftJustification(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1071 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1072 s.AssumeRightJustification(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1073 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1074 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1075 if (debug_level > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1076 tprintf( |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1077 "# Not enough variety for clear outline classification. " |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1078 "Guessing these are %s aligned based on script.\n", |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1079 s.ltr ? "left" : "right"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1080 s.PrintRows(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1081 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1082 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1083 if (s.AlignTabs().size() == 2) { // case A1 or A2 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1084 s.first_indent = s.AlignTabs()[1].center; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1085 s.body_indent = s.AlignTabs()[0].center; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1086 } else { // case B1 or B2 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1087 if (num_rows - 1 == num_full_rows - last_row_full) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1088 // case B2 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1089 const ParagraphModel *model = s.ltr ? kCrownLeft : kCrownRight; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1090 (*s.rows)[s.row_start].AddStartLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1091 for (int i = s.row_start + 1; i < s.row_end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1092 (*s.rows)[i].AddBodyLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1093 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1094 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1095 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1096 // case B1 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1097 s.first_indent = s.body_indent = s.AlignTabs()[0].center; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1098 s.eop_threshold = (s.OffsideTabs()[0].center + s.OffsideTabs()[1].center) / 2; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1099 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1100 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1101 const ParagraphModel *model = theory->AddModel(s.Model()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1102 MarkRowsWithModel(s.rows, s.row_start, s.row_end, model, s.ltr, s.eop_threshold); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1103 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1104 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1105 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1106 // This function is called if strong textual clues were not available, but |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1107 // the caller hopes that the paragraph breaks will be super obvious just |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1108 // by the outline of the text. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1109 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1110 // The particularly difficult case is figuring out what's going on if you |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1111 // don't have enough short paragraph end lines to tell us what's going on. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1112 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1113 // For instance, let's say you have the following outline: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1114 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1115 // (A1) xxxxxxxxxxxxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1116 // xxxxxxxxxxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1117 // xxxxxxxxxxxxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1118 // xxxxxxxxxxxxxxxxxxxxxx |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1119 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1120 // Even if we know that the text is left-to-right and so will probably be |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1121 // left-aligned, both of the following are possible texts: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1122 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1123 // (A1a) 1. Here our list item |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1124 // with two full lines. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1125 // 2. Here a second item. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1126 // 3. Here our third one. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1127 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1128 // (A1b) so ends paragraph one. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1129 // Here starts another |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1130 // paragraph we want to |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1131 // read. This continues |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1132 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1133 // These examples are obvious from the text and should have been caught |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1134 // by the StrongEvidenceClassify pass. However, for languages where we don't |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1135 // have capital letters to go on (e.g. Hebrew, Arabic, Hindi, Chinese), |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1136 // it's worth guessing that (A1b) is the correct interpretation if there are |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1137 // far more "full" lines than "short" lines. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1138 static void GeometricClassify(int debug_level, std::vector<RowScratchRegisters> *rows, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1139 int row_start, int row_end, ParagraphTheory *theory) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1140 if (!AcceptableRowArgs(debug_level, 4, __func__, rows, row_start, row_end)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1141 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1142 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1143 if (debug_level > 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1144 tprintf("###############################################\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1145 tprintf("##### GeometricClassify( rows[%d:%d) ) ####\n", row_start, row_end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1146 tprintf("###############################################\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1147 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1148 RecomputeMarginsAndClearHypotheses(rows, row_start, row_end, 10); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1149 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1150 GeometricClassifierState s(debug_level, rows, row_start, row_end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1151 if (s.left_tabs.size() > 2 && s.right_tabs.size() > 2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1152 s.Fail(2, "Too much variety for simple outline classification."); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1153 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1154 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1155 if (s.left_tabs.size() <= 1 && s.right_tabs.size() <= 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1156 s.Fail(1, "Not enough variety for simple outline classification."); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1157 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1158 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1159 if (s.left_tabs.size() + s.right_tabs.size() == 3) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1160 GeometricClassifyThreeTabStopTextBlock(debug_level, s, theory); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1161 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1162 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1163 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1164 // At this point, we know that one side has at least two tab stops, and the |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1165 // other side has one or two tab stops. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1166 // Left to determine: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1167 // (1) Which is the body indent and which is the first line indent? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1168 // (2) Is the text fully justified? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1169 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1170 // If one side happens to have three or more tab stops, assume that side |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1171 // is opposite of the aligned side. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1172 if (s.right_tabs.size() > 2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1173 s.AssumeLeftJustification(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1174 } else if (s.left_tabs.size() > 2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1175 s.AssumeRightJustification(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1176 } else if (s.ltr) { // guess based on script direction |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1177 s.AssumeLeftJustification(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1178 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1179 s.AssumeRightJustification(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1180 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1181 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1182 if (s.AlignTabs().size() == 2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1183 // For each tab stop on the aligned side, how many of them appear |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1184 // to be paragraph start lines? [first lines] |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1185 int firsts[2] = {0, 0}; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1186 // Count the first line as a likely paragraph start line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1187 firsts[s.AlignsideTabIndex(s.row_start)]++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1188 // For each line, if the first word would have fit on the previous |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1189 // line count it as a likely paragraph start line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1190 bool jam_packed = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1191 for (int i = s.row_start + 1; i < s.row_end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1192 if (s.FirstWordWouldHaveFit(i - 1, i)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1193 firsts[s.AlignsideTabIndex(i)]++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1194 jam_packed = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1195 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1196 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1197 // Make an extra accounting for the last line of the paragraph just |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1198 // in case it's the only short line in the block. That is, take its |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1199 // first word as typical and see if this looks like the *last* line |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1200 // of a paragraph. If so, mark the *other* indent as probably a first. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1201 if (jam_packed && s.FirstWordWouldHaveFit(s.row_end - 1, s.row_end - 1)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1202 firsts[1 - s.AlignsideTabIndex(s.row_end - 1)]++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1203 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1204 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1205 int percent0firsts, percent1firsts; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1206 percent0firsts = (100 * firsts[0]) / s.AlignTabs()[0].count; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1207 percent1firsts = (100 * firsts[1]) / s.AlignTabs()[1].count; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1208 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1209 // TODO(eger): Tune these constants if necessary. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1210 if ((percent0firsts < 20 && 30 < percent1firsts) || percent0firsts + 30 < percent1firsts) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1211 s.first_indent = s.AlignTabs()[1].center; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1212 s.body_indent = s.AlignTabs()[0].center; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1213 } else if ((percent1firsts < 20 && 30 < percent0firsts) || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1214 percent1firsts + 30 < percent0firsts) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1215 s.first_indent = s.AlignTabs()[0].center; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1216 s.body_indent = s.AlignTabs()[1].center; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1217 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1218 // Ambiguous! Probably lineated (poetry) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1219 if (debug_level > 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1220 tprintf("# Cannot determine %s indent likely to start paragraphs.\n", |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1221 s.just == tesseract::JUSTIFICATION_LEFT ? "left" : "right"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1222 tprintf("# Indent of %d looks like a first line %d%% of the time.\n", |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1223 s.AlignTabs()[0].center, percent0firsts); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1224 tprintf("# Indent of %d looks like a first line %d%% of the time.\n", |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1225 s.AlignTabs()[1].center, percent1firsts); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1226 s.PrintRows(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1227 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1228 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1229 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1230 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1231 // There's only one tab stop for the "aligned to" side. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1232 s.first_indent = s.body_indent = s.AlignTabs()[0].center; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1233 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1234 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1235 // At this point, we have our model. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1236 const ParagraphModel *model = theory->AddModel(s.Model()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1237 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1238 // Now all we have to do is figure out if the text is fully justified or not. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1239 // eop_threshold: default to fully justified unless we see evidence below. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1240 // See description on MarkRowsWithModel() |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1241 s.eop_threshold = (s.OffsideTabs()[0].center + s.OffsideTabs()[1].center) / 2; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1242 // If the text is not fully justified, re-set the eop_threshold to 0. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1243 if (s.AlignTabs().size() == 2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1244 // Paragraphs with a paragraph-start indent. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1245 for (int i = s.row_start; i < s.row_end - 1; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1246 if (ValidFirstLine(s.rows, i + 1, model) && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1247 !NearlyEqual(s.OffsideTabs()[0].center, (*s.rows)[i].OffsideIndent(s.just), |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1248 s.tolerance)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1249 // We found a non-end-of-paragraph short line: not fully justified. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1250 s.eop_threshold = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1251 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1252 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1253 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1254 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1255 // Paragraphs with no paragraph-start indent. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1256 for (int i = s.row_start; i < s.row_end - 1; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1257 if (!s.FirstWordWouldHaveFit(i, i + 1) && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1258 !NearlyEqual(s.OffsideTabs()[0].center, (*s.rows)[i].OffsideIndent(s.just), |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1259 s.tolerance)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1260 // We found a non-end-of-paragraph short line: not fully justified. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1261 s.eop_threshold = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1262 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1263 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1264 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1265 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1266 MarkRowsWithModel(rows, row_start, row_end, model, s.ltr, s.eop_threshold); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1267 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1268 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1269 // =============== Implementation of ParagraphTheory ===================== |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1270 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1271 const ParagraphModel *ParagraphTheory::AddModel(const ParagraphModel &model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1272 for (const auto &m : *models_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1273 if (m->Comparable(model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1274 return m; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1275 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1276 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1277 auto *m = new ParagraphModel(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1278 models_->push_back(m); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1279 push_back_new(models_we_added_, m); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1280 return m; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1281 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1282 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1283 void ParagraphTheory::DiscardUnusedModels(const SetOfModels &used_models) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1284 size_t w = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1285 for (size_t r = 0; r < models_->size(); r++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1286 ParagraphModel *m = (*models_)[r]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1287 if (!contains(used_models, static_cast<const ParagraphModel *>(m)) && contains(models_we_added_, m)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1288 delete m; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1289 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1290 if (r > w) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1291 (*models_)[w] = m; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1292 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1293 w++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1294 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1295 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1296 models_->resize(w); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1297 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1298 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1299 // Examine rows[start, end) and try to determine if an existing non-centered |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1300 // paragraph model would fit them perfectly. If so, return a pointer to it. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1301 // If not, return nullptr. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1302 const ParagraphModel *ParagraphTheory::Fits(const std::vector<RowScratchRegisters> *rows, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1303 int start, int end) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1304 for (const auto *model : *models_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1305 if (model->justification() != JUSTIFICATION_CENTER && RowsFitModel(rows, start, end, model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1306 return model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1307 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1308 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1309 return nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1310 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1311 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1312 void ParagraphTheory::NonCenteredModels(SetOfModels *models) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1313 for (const auto *model : *models_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1314 if (model->justification() != JUSTIFICATION_CENTER) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1315 push_back_new(*models, model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1316 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1317 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1318 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1319 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1320 int ParagraphTheory::IndexOf(const ParagraphModel *model) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1321 int i = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1322 for (const auto *m : *models_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1323 if (m == model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1324 return i; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1325 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1326 i++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1327 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1328 return -1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1329 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1330 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1331 bool ValidFirstLine(const std::vector<RowScratchRegisters> *rows, int row, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1332 const ParagraphModel *model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1333 if (!StrongModel(model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1334 tprintf("ValidFirstLine() should only be called with strong models!\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1335 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1336 return StrongModel(model) && model->ValidFirstLine((*rows)[row].lmargin_, (*rows)[row].lindent_, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1337 (*rows)[row].rindent_, (*rows)[row].rmargin_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1338 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1339 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1340 bool ValidBodyLine(const std::vector<RowScratchRegisters> *rows, int row, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1341 const ParagraphModel *model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1342 if (!StrongModel(model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1343 tprintf("ValidBodyLine() should only be called with strong models!\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1344 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1345 return StrongModel(model) && model->ValidBodyLine((*rows)[row].lmargin_, (*rows)[row].lindent_, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1346 (*rows)[row].rindent_, (*rows)[row].rmargin_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1347 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1348 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1349 bool CrownCompatible(const std::vector<RowScratchRegisters> *rows, int a, int b, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1350 const ParagraphModel *model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1351 if (model != kCrownRight && model != kCrownLeft) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1352 tprintf("CrownCompatible() should only be called with crown models!\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1353 return false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1354 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1355 auto &row_a = (*rows)[a]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1356 auto &row_b = (*rows)[b]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1357 if (model == kCrownRight) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1358 return NearlyEqual(row_a.rindent_ + row_a.rmargin_, row_b.rindent_ + row_b.rmargin_, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1359 Epsilon(row_a.ri_->average_interword_space)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1360 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1361 return NearlyEqual(row_a.lindent_ + row_a.lmargin_, row_b.lindent_ + row_b.lmargin_, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1362 Epsilon(row_a.ri_->average_interword_space)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1363 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1364 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1365 // =============== Implementation of ParagraphModelSmearer ==================== |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1366 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1367 ParagraphModelSmearer::ParagraphModelSmearer(std::vector<RowScratchRegisters> *rows, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1368 int row_start, int row_end, ParagraphTheory *theory) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1369 : theory_(theory), rows_(rows), row_start_(row_start), row_end_(row_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1370 if (!AcceptableRowArgs(0, 0, __func__, rows, row_start, row_end)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1371 row_start_ = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1372 row_end_ = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1373 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1374 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1375 open_models_.resize(open_models_.size() + row_end - row_start + 2); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1376 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1377 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1378 // see paragraphs_internal.h |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1379 void ParagraphModelSmearer::CalculateOpenModels(int row_start, int row_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1380 SetOfModels no_models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1381 if (row_start < row_start_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1382 row_start = row_start_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1383 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1384 if (row_end > row_end_) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1385 row_end = row_end_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1386 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1387 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1388 for (int row = (row_start > 0) ? row_start - 1 : row_start; row < row_end; row++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1389 if ((*rows_)[row].ri_->num_words == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1390 OpenModels(row + 1) = no_models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1391 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1392 SetOfModels &opened = OpenModels(row); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1393 (*rows_)[row].StartHypotheses(&opened); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1394 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1395 // Which models survive the transition from row to row + 1? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1396 SetOfModels still_open; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1397 for (auto &m : opened) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1398 if (ValidFirstLine(rows_, row, m) || ValidBodyLine(rows_, row, m)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1399 // This is basic filtering; we check likely paragraph starty-ness down |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1400 // below in Smear() -- you know, whether the first word would have fit |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1401 // and such. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1402 push_back_new(still_open, m); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1403 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1404 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1405 OpenModels(row + 1) = std::move(still_open); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1406 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1407 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1408 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1409 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1410 // see paragraphs_internal.h |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1411 void ParagraphModelSmearer::Smear() { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1412 CalculateOpenModels(row_start_, row_end_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1413 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1414 // For each row which we're unsure about (that is, it is LT_UNKNOWN or |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1415 // we have multiple LT_START hypotheses), see if there's a model that |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1416 // was recently used (an "open" model) which might model it well. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1417 for (int i = row_start_; i < row_end_; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1418 RowScratchRegisters &row = (*rows_)[i]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1419 if (row.ri_->num_words == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1420 continue; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1421 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1422 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1423 // Step One: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1424 // Figure out if there are "open" models which are left-alined or |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1425 // right-aligned. This is important for determining whether the |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1426 // "first" word in a row would fit at the "end" of the previous row. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1427 bool left_align_open = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1428 bool right_align_open = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1429 for (auto &m : OpenModels(i)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1430 switch (m->justification()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1431 case JUSTIFICATION_LEFT: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1432 left_align_open = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1433 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1434 case JUSTIFICATION_RIGHT: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1435 right_align_open = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1436 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1437 default: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1438 left_align_open = right_align_open = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1439 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1440 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1441 // Step Two: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1442 // Use that knowledge to figure out if this row is likely to |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1443 // start a paragraph. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1444 bool likely_start; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1445 if (i == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1446 likely_start = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1447 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1448 if ((left_align_open && right_align_open) || (!left_align_open && !right_align_open)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1449 likely_start = LikelyParagraphStart((*rows_)[i - 1], row, JUSTIFICATION_LEFT) || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1450 LikelyParagraphStart((*rows_)[i - 1], row, JUSTIFICATION_RIGHT); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1451 } else if (left_align_open) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1452 likely_start = LikelyParagraphStart((*rows_)[i - 1], row, JUSTIFICATION_LEFT); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1453 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1454 likely_start = LikelyParagraphStart((*rows_)[i - 1], row, JUSTIFICATION_RIGHT); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1455 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1456 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1457 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1458 // Step Three: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1459 // If this text line seems like an obvious first line of an |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1460 // open model, or an obvious continuation of an existing |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1461 // modelled paragraph, mark it up. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1462 if (likely_start) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1463 // Add Start Hypotheses for all Open models that fit. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1464 for (unsigned m = 0; m < OpenModels(i).size(); m++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1465 if (ValidFirstLine(rows_, i, OpenModels(i)[m])) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1466 row.AddStartLine(OpenModels(i)[m]); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1467 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1468 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1469 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1470 // Add relevant body line hypotheses. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1471 SetOfModels last_line_models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1472 if (i > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1473 (*rows_)[i - 1].StrongHypotheses(&last_line_models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1474 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1475 theory_->NonCenteredModels(&last_line_models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1476 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1477 for (auto model : last_line_models) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1478 if (ValidBodyLine(rows_, i, model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1479 row.AddBodyLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1480 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1481 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1482 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1483 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1484 // Step Four: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1485 // If we're still quite unsure about this line, go through all |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1486 // models in our theory and see if this row could be the start |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1487 // of any of our models. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1488 if (row.GetLineType() == LT_UNKNOWN || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1489 (row.GetLineType() == LT_START && !row.UniqueStartHypothesis())) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1490 SetOfModels all_models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1491 theory_->NonCenteredModels(&all_models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1492 for (auto &all_model : all_models) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1493 if (ValidFirstLine(rows_, i, all_model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1494 row.AddStartLine(all_model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1495 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1496 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1497 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1498 // Step Five: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1499 // Since we may have updated the hypotheses about this row, we need |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1500 // to recalculate the Open models for the rest of rows[i + 1, row_end) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1501 if (row.GetLineType() != LT_UNKNOWN) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1502 CalculateOpenModels(i + 1, row_end_); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1503 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1504 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1505 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1506 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1507 // ================ Main Paragraph Detection Algorithm ======================= |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1508 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1509 // Find out what ParagraphModels are actually used, and discard any |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1510 // that are not. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1511 static void DiscardUnusedModels(const std::vector<RowScratchRegisters> &rows, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1512 ParagraphTheory *theory) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1513 SetOfModels used_models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1514 for (const auto &row : rows) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1515 row.StrongHypotheses(&used_models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1516 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1517 theory->DiscardUnusedModels(used_models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1518 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1519 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1520 // DowngradeWeakestToCrowns: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1521 // Forget any flush-{left, right} models unless we see two or more |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1522 // of them in sequence. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1523 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1524 // In pass 3, we start to classify even flush-left paragraphs (paragraphs |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1525 // where the first line and body indent are the same) as having proper Models. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1526 // This is generally dangerous, since if you start imagining that flush-left |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1527 // is a typical paragraph model when it is not, it will lead you to chop normal |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1528 // indented paragraphs in the middle whenever a sentence happens to start on a |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1529 // new line (see "This" above). What to do? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1530 // What we do is to take any paragraph which is flush left and is not |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1531 // preceded by another paragraph of the same model and convert it to a "Crown" |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1532 // paragraph. This is a weak pseudo-ParagraphModel which is a placeholder |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1533 // for later. It means that the paragraph is flush, but it would be desirable |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1534 // to mark it as the same model as following text if it fits. This downgrade |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1535 // FlushLeft -> CrownLeft -> Model of following paragraph. Means that we |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1536 // avoid making flush left Paragraph Models whenever we see a top-of-the-page |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1537 // half-of-a-paragraph. and instead we mark it the same as normal body text. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1538 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1539 // Implementation: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1540 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1541 // Comb backwards through the row scratch registers, and turn any |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1542 // sequences of body lines of equivalent type abutted against the beginning |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1543 // or a body or start line of a different type into a crown paragraph. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1544 static void DowngradeWeakestToCrowns(int debug_level, ParagraphTheory *theory, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1545 std::vector<RowScratchRegisters> *rows) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1546 int start; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1547 for (int end = rows->size(); end > 0; end = start) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1548 // Search back for a body line of a unique type. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1549 const ParagraphModel *model = nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1550 while (end > 0 && (model = (*rows)[end - 1].UniqueBodyHypothesis()) == nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1551 end--; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1552 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1553 if (end == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1554 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1555 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1556 start = end - 1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1557 while (start >= 0 && (*rows)[start].UniqueBodyHypothesis() == model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1558 start--; // walk back to the first line that is not the same body type. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1559 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1560 if (start >= 0 && (*rows)[start].UniqueStartHypothesis() == model && StrongModel(model) && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1561 NearlyEqual(model->first_indent(), model->body_indent(), model->tolerance())) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1562 start--; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1563 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1564 start++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1565 // Now rows[start, end) is a sequence of unique body hypotheses of model. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1566 if (StrongModel(model) && model->justification() == JUSTIFICATION_CENTER) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1567 continue; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1568 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1569 if (!StrongModel(model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1570 while (start > 0 && CrownCompatible(rows, start - 1, start, model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1571 start--; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1572 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1573 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1574 if (start == 0 || (!StrongModel(model)) || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1575 (StrongModel(model) && !ValidFirstLine(rows, start - 1, model))) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1576 // crownify rows[start, end) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1577 const ParagraphModel *crown_model = model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1578 if (StrongModel(model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1579 if (model->justification() == JUSTIFICATION_LEFT) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1580 crown_model = kCrownLeft; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1581 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1582 crown_model = kCrownRight; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1583 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1584 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1585 (*rows)[start].SetUnknown(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1586 (*rows)[start].AddStartLine(crown_model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1587 for (int row = start + 1; row < end; row++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1588 (*rows)[row].SetUnknown(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1589 (*rows)[row].AddBodyLine(crown_model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1590 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1591 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1592 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1593 DiscardUnusedModels(*rows, theory); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1594 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1595 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1596 // Clear all hypotheses about lines [start, end) and reset margins. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1597 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1598 // The empty space between the left of a row and the block boundary (and |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1599 // similarly for the right) is split into two pieces: margin and indent. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1600 // In initial processing, we assume the block is tight and the margin for |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1601 // all lines is set to zero. However, if our first pass does not yield |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1602 // models for everything, it may be due to an inset paragraph like a |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1603 // block-quote. In that case, we make a second pass over that unmarked |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1604 // section of the page and reset the "margin" portion of the empty space |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1605 // to the common amount of space at the ends of the lines under consid- |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1606 // eration. This would be equivalent to percentile set to 0. However, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1607 // sometimes we have a single character sticking out in the right margin |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1608 // of a text block (like the 'r' in 'for' on line 3 above), and we can |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1609 // really just ignore it as an outlier. To express this, we allow the |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1610 // user to specify the percentile (0..100) of indent values to use as |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1611 // the common margin for each row in the run of rows[start, end). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1612 void RecomputeMarginsAndClearHypotheses(std::vector<RowScratchRegisters> *rows, int start, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1613 int end, int percentile) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1614 if (!AcceptableRowArgs(0, 0, __func__, rows, start, end)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1615 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1616 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1617 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1618 int lmin, lmax, rmin, rmax; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1619 lmin = lmax = (*rows)[start].lmargin_ + (*rows)[start].lindent_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1620 rmin = rmax = (*rows)[start].rmargin_ + (*rows)[start].rindent_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1621 for (int i = start; i < end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1622 RowScratchRegisters &sr = (*rows)[i]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1623 sr.SetUnknown(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1624 if (sr.ri_->num_words == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1625 continue; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1626 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1627 UpdateRange(sr.lmargin_ + sr.lindent_, &lmin, &lmax); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1628 UpdateRange(sr.rmargin_ + sr.rindent_, &rmin, &rmax); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1629 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1630 STATS lefts(lmin, lmax); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1631 STATS rights(rmin, rmax); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1632 for (int i = start; i < end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1633 RowScratchRegisters &sr = (*rows)[i]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1634 if (sr.ri_->num_words == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1635 continue; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1636 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1637 lefts.add(sr.lmargin_ + sr.lindent_, 1); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1638 rights.add(sr.rmargin_ + sr.rindent_, 1); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1639 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1640 int ignorable_left = lefts.ile(ClipToRange(percentile, 0, 100) / 100.0); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1641 int ignorable_right = rights.ile(ClipToRange(percentile, 0, 100) / 100.0); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1642 for (int i = start; i < end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1643 RowScratchRegisters &sr = (*rows)[i]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1644 int ldelta = ignorable_left - sr.lmargin_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1645 sr.lmargin_ += ldelta; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1646 sr.lindent_ -= ldelta; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1647 int rdelta = ignorable_right - sr.rmargin_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1648 sr.rmargin_ += rdelta; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1649 sr.rindent_ -= rdelta; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1650 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1651 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1652 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1653 // Return the median inter-word space in rows[row_start, row_end). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1654 int InterwordSpace(const std::vector<RowScratchRegisters> &rows, int row_start, int row_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1655 if (row_end < row_start + 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1656 return 1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1657 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1658 int word_height = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1659 (rows[row_start].ri_->lword_box.height() + rows[row_end - 1].ri_->lword_box.height()) / 2; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1660 int word_width = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1661 (rows[row_start].ri_->lword_box.width() + rows[row_end - 1].ri_->lword_box.width()) / 2; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1662 STATS spacing_widths(0, 4 + word_width); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1663 for (int i = row_start; i < row_end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1664 if (rows[i].ri_->num_words > 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1665 spacing_widths.add(rows[i].ri_->average_interword_space, 1); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1666 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1667 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1668 int minimum_reasonable_space = word_height / 3; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1669 if (minimum_reasonable_space < 2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1670 minimum_reasonable_space = 2; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1671 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1672 int median = spacing_widths.median(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1673 return (median > minimum_reasonable_space) ? median : minimum_reasonable_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1674 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1675 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1676 // Return whether the first word on the after line can fit in the space at |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1677 // the end of the before line (knowing which way the text is aligned and read). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1678 bool FirstWordWouldHaveFit(const RowScratchRegisters &before, const RowScratchRegisters &after, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1679 tesseract::ParagraphJustification justification) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1680 if (before.ri_->num_words == 0 || after.ri_->num_words == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1681 return true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1682 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1683 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1684 if (justification == JUSTIFICATION_UNKNOWN) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1685 tprintf("Don't call FirstWordWouldHaveFit(r, s, JUSTIFICATION_UNKNOWN).\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1686 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1687 int available_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1688 if (justification == JUSTIFICATION_CENTER) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1689 available_space = before.lindent_ + before.rindent_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1690 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1691 available_space = before.OffsideIndent(justification); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1692 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1693 available_space -= before.ri_->average_interword_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1694 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1695 if (before.ri_->ltr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1696 return after.ri_->lword_box.width() < available_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1697 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1698 return after.ri_->rword_box.width() < available_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1699 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1700 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1701 // Return whether the first word on the after line can fit in the space at |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1702 // the end of the before line (not knowing which way the text goes) in a left |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1703 // or right alignment. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1704 bool FirstWordWouldHaveFit(const RowScratchRegisters &before, const RowScratchRegisters &after) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1705 if (before.ri_->num_words == 0 || after.ri_->num_words == 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1706 return true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1707 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1708 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1709 int available_space = before.lindent_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1710 if (before.rindent_ > available_space) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1711 available_space = before.rindent_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1712 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1713 available_space -= before.ri_->average_interword_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1714 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1715 if (before.ri_->ltr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1716 return after.ri_->lword_box.width() < available_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1717 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1718 return after.ri_->rword_box.width() < available_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1719 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1720 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1721 static bool TextSupportsBreak(const RowScratchRegisters &before, const RowScratchRegisters &after) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1722 if (before.ri_->ltr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1723 return before.ri_->rword_likely_ends_idea && after.ri_->lword_likely_starts_idea; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1724 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1725 return before.ri_->lword_likely_ends_idea && after.ri_->rword_likely_starts_idea; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1726 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1727 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1728 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1729 static bool LikelyParagraphStart(const RowScratchRegisters &before, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1730 const RowScratchRegisters &after, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1731 tesseract::ParagraphJustification j) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1732 return before.ri_->num_words == 0 || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1733 (FirstWordWouldHaveFit(before, after, j) && TextSupportsBreak(before, after)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1734 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1735 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1736 // Examine rows[start, end) and try to determine what sort of ParagraphModel |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1737 // would fit them as a single paragraph. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1738 // If we can't produce a unique model justification_ = JUSTIFICATION_UNKNOWN. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1739 // If the rows given could be a consistent start to a paragraph, set *consistent |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1740 // true. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1741 static ParagraphModel InternalParagraphModelByOutline( |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1742 const std::vector<RowScratchRegisters> *rows, int start, int end, int tolerance, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1743 bool *consistent) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1744 int ltr_line_count = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1745 for (int i = start; i < end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1746 ltr_line_count += static_cast<int>((*rows)[i].ri_->ltr); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1747 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1748 bool ltr = (ltr_line_count >= (end - start) / 2); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1749 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1750 *consistent = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1751 if (!AcceptableRowArgs(0, 2, __func__, rows, start, end)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1752 return ParagraphModel(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1753 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1754 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1755 // Ensure the caller only passed us a region with a common rmargin and |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1756 // lmargin. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1757 int lmargin = (*rows)[start].lmargin_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1758 int rmargin = (*rows)[start].rmargin_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1759 int lmin, lmax, rmin, rmax, cmin, cmax; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1760 lmin = lmax = (*rows)[start + 1].lindent_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1761 rmin = rmax = (*rows)[start + 1].rindent_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1762 cmin = cmax = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1763 for (int i = start + 1; i < end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1764 if ((*rows)[i].lmargin_ != lmargin || (*rows)[i].rmargin_ != rmargin) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1765 tprintf("Margins don't match! Software error.\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1766 *consistent = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1767 return ParagraphModel(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1768 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1769 UpdateRange((*rows)[i].lindent_, &lmin, &lmax); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1770 UpdateRange((*rows)[i].rindent_, &rmin, &rmax); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1771 UpdateRange((*rows)[i].rindent_ - (*rows)[i].lindent_, &cmin, &cmax); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1772 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1773 int ldiff = lmax - lmin; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1774 int rdiff = rmax - rmin; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1775 int cdiff = cmax - cmin; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1776 if (rdiff > tolerance && ldiff > tolerance) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1777 if (cdiff < tolerance * 2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1778 if (end - start < 3) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1779 return ParagraphModel(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1780 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1781 return ParagraphModel(JUSTIFICATION_CENTER, 0, 0, 0, tolerance); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1782 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1783 *consistent = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1784 return ParagraphModel(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1785 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1786 if (end - start < 3) { // Don't return a model for two line paras. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1787 return ParagraphModel(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1788 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1789 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1790 // These booleans keep us from saying something is aligned left when the body |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1791 // left variance is too large. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1792 bool body_admits_left_alignment = ldiff < tolerance; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1793 bool body_admits_right_alignment = rdiff < tolerance; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1794 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1795 ParagraphModel left_model = ParagraphModel(JUSTIFICATION_LEFT, lmargin, (*rows)[start].lindent_, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1796 (lmin + lmax) / 2, tolerance); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1797 ParagraphModel right_model = ParagraphModel(JUSTIFICATION_RIGHT, rmargin, (*rows)[start].rindent_, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1798 (rmin + rmax) / 2, tolerance); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1799 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1800 // These booleans keep us from having an indent on the "wrong side" for the |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1801 // first line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1802 bool text_admits_left_alignment = ltr || left_model.is_flush(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1803 bool text_admits_right_alignment = !ltr || right_model.is_flush(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1804 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1805 // At least one of the edges is less than tolerance in variance. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1806 // If the other is obviously ragged, it can't be the one aligned to. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1807 // [Note the last line is included in this raggedness.] |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1808 if (tolerance < rdiff) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1809 if (body_admits_left_alignment && text_admits_left_alignment) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1810 return left_model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1811 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1812 *consistent = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1813 return ParagraphModel(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1814 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1815 if (tolerance < ldiff) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1816 if (body_admits_right_alignment && text_admits_right_alignment) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1817 return right_model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1818 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1819 *consistent = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1820 return ParagraphModel(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1821 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1822 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1823 // At this point, we know the body text doesn't vary much on either side. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1824 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1825 // If the first line juts out oddly in one direction or the other, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1826 // that likely indicates the side aligned to. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1827 int first_left = (*rows)[start].lindent_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1828 int first_right = (*rows)[start].rindent_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1829 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1830 if (ltr && body_admits_left_alignment && (first_left < lmin || first_left > lmax)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1831 return left_model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1832 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1833 if (!ltr && body_admits_right_alignment && (first_right < rmin || first_right > rmax)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1834 return right_model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1835 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1836 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1837 *consistent = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1838 return ParagraphModel(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1839 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1840 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1841 // Examine rows[start, end) and try to determine what sort of ParagraphModel |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1842 // would fit them as a single paragraph. If nothing fits, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1843 // justification_ = JUSTIFICATION_UNKNOWN and print the paragraph to debug |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1844 // output if we're debugging. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1845 static ParagraphModel ParagraphModelByOutline(int debug_level, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1846 const std::vector<RowScratchRegisters> *rows, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1847 int start, int end, int tolerance) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1848 bool unused_consistent; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1849 ParagraphModel retval = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1850 InternalParagraphModelByOutline(rows, start, end, tolerance, &unused_consistent); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1851 if (debug_level >= 2 && retval.justification() == JUSTIFICATION_UNKNOWN) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1852 tprintf("Could not determine a model for this paragraph:\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1853 PrintRowRange(*rows, start, end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1854 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1855 return retval; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1856 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1857 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1858 // Do rows[start, end) form a single instance of the given paragraph model? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1859 bool RowsFitModel(const std::vector<RowScratchRegisters> *rows, int start, int end, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1860 const ParagraphModel *model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1861 if (!AcceptableRowArgs(0, 1, __func__, rows, start, end)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1862 return false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1863 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1864 if (!ValidFirstLine(rows, start, model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1865 return false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1866 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1867 for (int i = start + 1; i < end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1868 if (!ValidBodyLine(rows, i, model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1869 return false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1870 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1871 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1872 return true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1873 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1874 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1875 // Examine rows[row_start, row_end) as an independent section of text, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1876 // and mark rows that are exceptionally clear as start-of-paragraph |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1877 // and paragraph-body lines. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1878 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1879 // We presume that any lines surrounding rows[row_start, row_end) may |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1880 // have wildly different paragraph models, so we don't key any data off |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1881 // of those lines. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1882 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1883 // We only take the very strongest signals, as we don't want to get |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1884 // confused and marking up centered text, poetry, or source code as |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1885 // clearly part of a typical paragraph. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1886 static void MarkStrongEvidence(std::vector<RowScratchRegisters> *rows, int row_start, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1887 int row_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1888 // Record patently obvious body text. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1889 for (int i = row_start + 1; i < row_end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1890 const RowScratchRegisters &prev = (*rows)[i - 1]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1891 RowScratchRegisters &curr = (*rows)[i]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1892 tesseract::ParagraphJustification typical_justification = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1893 prev.ri_->ltr ? JUSTIFICATION_LEFT : JUSTIFICATION_RIGHT; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1894 if (!curr.ri_->rword_likely_starts_idea && !curr.ri_->lword_likely_starts_idea && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1895 !FirstWordWouldHaveFit(prev, curr, typical_justification)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1896 curr.SetBodyLine(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1897 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1898 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1899 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1900 // Record patently obvious start paragraph lines. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1901 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1902 // It's an extremely good signal of the start of a paragraph that |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1903 // the first word would have fit on the end of the previous line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1904 // However, applying just that signal would have us mark random |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1905 // start lines of lineated text (poetry and source code) and some |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1906 // centered headings as paragraph start lines. Therefore, we use |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1907 // a second qualification for a paragraph start: Not only should |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1908 // the first word of this line have fit on the previous line, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1909 // but also, this line should go full to the right of the block, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1910 // disallowing a subsequent word from having fit on this line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1911 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1912 // First row: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1913 { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1914 RowScratchRegisters &curr = (*rows)[row_start]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1915 RowScratchRegisters &next = (*rows)[row_start + 1]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1916 tesseract::ParagraphJustification j = curr.ri_->ltr ? JUSTIFICATION_LEFT : JUSTIFICATION_RIGHT; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1917 if (curr.GetLineType() == LT_UNKNOWN && !FirstWordWouldHaveFit(curr, next, j) && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1918 (curr.ri_->lword_likely_starts_idea || curr.ri_->rword_likely_starts_idea)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1919 curr.SetStartLine(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1920 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1921 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1922 // Middle rows |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1923 for (int i = row_start + 1; i < row_end - 1; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1924 RowScratchRegisters &prev = (*rows)[i - 1]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1925 RowScratchRegisters &curr = (*rows)[i]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1926 RowScratchRegisters &next = (*rows)[i + 1]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1927 tesseract::ParagraphJustification j = curr.ri_->ltr ? JUSTIFICATION_LEFT : JUSTIFICATION_RIGHT; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1928 if (curr.GetLineType() == LT_UNKNOWN && !FirstWordWouldHaveFit(curr, next, j) && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1929 LikelyParagraphStart(prev, curr, j)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1930 curr.SetStartLine(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1931 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1932 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1933 // Last row |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1934 { // the short circuit at the top means we have at least two lines. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1935 RowScratchRegisters &prev = (*rows)[row_end - 2]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1936 RowScratchRegisters &curr = (*rows)[row_end - 1]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1937 tesseract::ParagraphJustification j = curr.ri_->ltr ? JUSTIFICATION_LEFT : JUSTIFICATION_RIGHT; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1938 if (curr.GetLineType() == LT_UNKNOWN && !FirstWordWouldHaveFit(curr, curr, j) && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1939 LikelyParagraphStart(prev, curr, j)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1940 curr.SetStartLine(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1941 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1942 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1943 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1944 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1945 // Look for sequences of a start line followed by some body lines in |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1946 // rows[row_start, row_end) and create ParagraphModels for them if |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1947 // they seem coherent. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1948 static void ModelStrongEvidence(int debug_level, std::vector<RowScratchRegisters> *rows, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1949 int row_start, int row_end, bool allow_flush_models, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1950 ParagraphTheory *theory) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1951 if (!AcceptableRowArgs(debug_level, 2, __func__, rows, row_start, row_end)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1952 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1953 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1954 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1955 int start = row_start; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1956 while (start < row_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1957 while (start < row_end && (*rows)[start].GetLineType() != LT_START) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1958 start++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1959 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1960 if (start >= row_end - 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1961 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1962 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1963 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1964 int tolerance = Epsilon((*rows)[start + 1].ri_->average_interword_space); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1965 int end = start; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1966 ParagraphModel last_model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1967 bool next_consistent; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1968 do { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1969 ++end; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1970 // rows[row, end) was consistent. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1971 // If rows[row, end + 1) is not consistent, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1972 // just model rows[row, end) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1973 if (end < row_end - 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1974 RowScratchRegisters &next = (*rows)[end]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1975 LineType lt = next.GetLineType(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1976 next_consistent = lt == LT_BODY || (lt == LT_UNKNOWN && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1977 !FirstWordWouldHaveFit((*rows)[end - 1], (*rows)[end])); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1978 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1979 next_consistent = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1980 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1981 if (next_consistent) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1982 ParagraphModel next_model = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1983 InternalParagraphModelByOutline(rows, start, end + 1, tolerance, &next_consistent); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1984 if (((*rows)[start].ri_->ltr && last_model.justification() == JUSTIFICATION_LEFT && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1985 next_model.justification() != JUSTIFICATION_LEFT) || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1986 (!(*rows)[start].ri_->ltr && last_model.justification() == JUSTIFICATION_RIGHT && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1987 next_model.justification() != JUSTIFICATION_RIGHT)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1988 next_consistent = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1989 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1990 last_model = next_model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1991 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1992 next_consistent = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1993 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1994 } while (next_consistent && end < row_end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1995 // At this point, rows[start, end) looked like it could have been a |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1996 // single paragraph. If we can make a good ParagraphModel for it, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1997 // do so and mark this sequence with that model. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1998 if (end > start + 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1999 // emit a new paragraph if we have more than one line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2000 const ParagraphModel *model = nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2001 ParagraphModel new_model = ParagraphModelByOutline( |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2002 debug_level, rows, start, end, Epsilon(InterwordSpace(*rows, start, end))); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2003 if (new_model.justification() == JUSTIFICATION_UNKNOWN) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2004 // couldn't create a good model, oh well. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2005 } else if (new_model.is_flush()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2006 if (end == start + 2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2007 // It's very likely we just got two paragraph starts in a row. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2008 end = start + 1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2009 } else if (start == row_start) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2010 // Mark this as a Crown. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2011 if (new_model.justification() == JUSTIFICATION_LEFT) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2012 model = kCrownLeft; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2013 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2014 model = kCrownRight; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2015 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2016 } else if (allow_flush_models) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2017 model = theory->AddModel(new_model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2018 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2019 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2020 model = theory->AddModel(new_model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2021 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2022 if (model) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2023 (*rows)[start].AddStartLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2024 for (int i = start + 1; i < end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2025 (*rows)[i].AddBodyLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2026 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2027 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2028 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2029 start = end; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2030 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2031 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2032 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2033 // We examine rows[row_start, row_end) and do the following: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2034 // (1) Clear all existing hypotheses for the rows being considered. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2035 // (2) Mark up any rows as exceptionally likely to be paragraph starts |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2036 // or paragraph body lines as such using both geometric and textual |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2037 // clues. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2038 // (3) Form models for any sequence of start + continuation lines. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2039 // (4) Smear the paragraph models to cover surrounding text. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2040 static void StrongEvidenceClassify(int debug_level, std::vector<RowScratchRegisters> *rows, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2041 int row_start, int row_end, ParagraphTheory *theory) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2042 if (!AcceptableRowArgs(debug_level, 2, __func__, rows, row_start, row_end)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2043 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2044 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2045 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2046 if (debug_level > 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2047 tprintf("#############################################\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2048 tprintf("# StrongEvidenceClassify( rows[%d:%d) )\n", row_start, row_end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2049 tprintf("#############################################\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2050 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2051 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2052 RecomputeMarginsAndClearHypotheses(rows, row_start, row_end, 10); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2053 MarkStrongEvidence(rows, row_start, row_end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2054 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2055 DebugDump(debug_level > 2, "Initial strong signals.", *theory, *rows); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2056 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2057 // Create paragraph models. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2058 ModelStrongEvidence(debug_level, rows, row_start, row_end, false, theory); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2059 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2060 DebugDump(debug_level > 2, "Unsmeared hypotheses.s.", *theory, *rows); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2061 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2062 // At this point, some rows are marked up as paragraphs with model numbers, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2063 // and some rows are marked up as either LT_START or LT_BODY. Now let's |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2064 // smear any good paragraph hypotheses forward and backward. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2065 ParagraphModelSmearer smearer(rows, row_start, row_end, theory); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2066 smearer.Smear(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2067 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2068 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2069 static void SeparateSimpleLeaderLines(std::vector<RowScratchRegisters> *rows, int row_start, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2070 int row_end, ParagraphTheory *theory) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2071 for (int i = row_start + 1; i < row_end - 1; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2072 if ((*rows)[i - 1].ri_->has_leaders && (*rows)[i].ri_->has_leaders && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2073 (*rows)[i + 1].ri_->has_leaders) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2074 const ParagraphModel *model = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2075 theory->AddModel(ParagraphModel(JUSTIFICATION_UNKNOWN, 0, 0, 0, 0)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2076 (*rows)[i].AddStartLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2077 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2078 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2079 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2080 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2081 // Collect sequences of unique hypotheses in row registers and create proper |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2082 // paragraphs for them, referencing the paragraphs in row_owners. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2083 static void ConvertHypothesizedModelRunsToParagraphs(int debug_level, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2084 std::vector<RowScratchRegisters> &rows, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2085 std::vector<PARA *> *row_owners, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2086 ParagraphTheory *theory) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2087 int end = rows.size(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2088 int start; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2089 for (; end > 0; end = start) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2090 start = end - 1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2091 const ParagraphModel *model = nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2092 // TODO(eger): Be smarter about dealing with multiple hypotheses. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2093 bool single_line_paragraph = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2094 SetOfModels models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2095 rows[start].NonNullHypotheses(&models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2096 if (!models.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2097 model = models[0]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2098 if (rows[start].GetLineType(model) != LT_BODY) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2099 single_line_paragraph = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2100 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2101 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2102 if (model && !single_line_paragraph) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2103 // walk back looking for more body lines and then a start line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2104 while (--start > 0 && rows[start].GetLineType(model) == LT_BODY) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2105 // do nothing |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2106 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2107 if (start < 0 || rows[start].GetLineType(model) != LT_START) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2108 model = nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2109 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2110 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2111 if (model == nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2112 continue; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2113 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2114 // rows[start, end) should be a paragraph. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2115 PARA *p = new PARA(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2116 if (model == kCrownLeft || model == kCrownRight) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2117 p->is_very_first_or_continuation = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2118 // Crown paragraph. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2119 // If we can find an existing ParagraphModel that fits, use it, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2120 // else create a new one. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2121 for (unsigned row = end; row < rows.size(); row++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2122 if ((*row_owners)[row] && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2123 (ValidBodyLine(&rows, start, (*row_owners)[row]->model) && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2124 (start == 0 || ValidFirstLine(&rows, start, (*row_owners)[row]->model)))) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2125 model = (*row_owners)[row]->model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2126 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2127 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2128 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2129 if (model == kCrownLeft) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2130 // No subsequent model fits, so cons one up. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2131 model = theory->AddModel(ParagraphModel(JUSTIFICATION_LEFT, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2132 rows[start].lmargin_ + rows[start].lindent_, 0, 0, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2133 Epsilon(rows[start].ri_->average_interword_space))); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2134 } else if (model == kCrownRight) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2135 // No subsequent model fits, so cons one up. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2136 model = theory->AddModel(ParagraphModel(JUSTIFICATION_RIGHT, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2137 rows[start].rmargin_ + rows[start].rmargin_, 0, 0, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2138 Epsilon(rows[start].ri_->average_interword_space))); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2139 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2140 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2141 rows[start].SetUnknown(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2142 rows[start].AddStartLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2143 for (int i = start + 1; i < end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2144 rows[i].SetUnknown(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2145 rows[i].AddBodyLine(model); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2146 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2147 p->model = model; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2148 p->has_drop_cap = rows[start].ri_->has_drop_cap; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2149 p->is_list_item = model->justification() == JUSTIFICATION_RIGHT |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2150 ? rows[start].ri_->rword_indicates_list_item |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2151 : rows[start].ri_->lword_indicates_list_item; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2152 for (int row = start; row < end; row++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2153 if ((*row_owners)[row] != nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2154 tprintf( |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2155 "Memory leak! ConvertHypothesizeModelRunsToParagraphs() called " |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2156 "more than once!\n"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2157 delete (*row_owners)[row]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2158 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2159 (*row_owners)[row] = p; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2160 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2161 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2162 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2163 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2164 struct Interval { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2165 Interval() : begin(0), end(0) {} |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2166 Interval(int b, int e) : begin(b), end(e) {} |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2167 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2168 int begin; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2169 int end; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2170 }; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2171 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2172 // Return whether rows[row] appears to be stranded, meaning that the evidence |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2173 // for this row is very weak due to context. For instance, two lines of source |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2174 // code may happen to be indented at the same tab vector as body text starts, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2175 // leading us to think they are two start-of-paragraph lines. This is not |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2176 // optimal. However, we also don't want to mark a sequence of short dialog |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2177 // as "weak," so our heuristic is: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2178 // (1) If a line is surrounded by lines of unknown type, it's weak. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2179 // (2) If two lines in a row are start lines for a given paragraph type, but |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2180 // after that the same paragraph type does not continue, they're weak. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2181 static bool RowIsStranded(const std::vector<RowScratchRegisters> &rows, int row) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2182 SetOfModels row_models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2183 rows[row].StrongHypotheses(&row_models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2184 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2185 for (auto &row_model : row_models) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2186 bool all_starts = rows[row].GetLineType(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2187 int run_length = 1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2188 bool continues = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2189 for (int i = row - 1; i >= 0 && continues; i--) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2190 SetOfModels models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2191 rows[i].NonNullHypotheses(&models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2192 switch (rows[i].GetLineType(row_model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2193 case LT_START: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2194 run_length++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2195 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2196 case LT_MULTIPLE: // explicit fall-through |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2197 case LT_BODY: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2198 run_length++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2199 all_starts = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2200 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2201 case LT_UNKNOWN: // explicit fall-through |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2202 default: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2203 continues = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2204 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2205 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2206 continues = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2207 for (unsigned i = row + 1; i < rows.size() && continues; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2208 SetOfModels models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2209 rows[i].NonNullHypotheses(&models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2210 switch (rows[i].GetLineType(row_model)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2211 case LT_START: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2212 run_length++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2213 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2214 case LT_MULTIPLE: // explicit fall-through |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2215 case LT_BODY: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2216 run_length++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2217 all_starts = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2218 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2219 case LT_UNKNOWN: // explicit fall-through |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2220 default: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2221 continues = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2222 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2223 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2224 if (run_length > 2 || (!all_starts && run_length > 1)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2225 return false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2226 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2227 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2228 return true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2229 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2230 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2231 // Go through rows[row_start, row_end) and gather up sequences that need better |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2232 // classification. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2233 // + Sequences of non-empty rows without hypotheses. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2234 // + Crown paragraphs not immediately followed by a strongly modeled line. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2235 // + Single line paragraphs surrounded by text that doesn't match the |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2236 // model. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2237 static void LeftoverSegments(const std::vector<RowScratchRegisters> &rows, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2238 std::vector<Interval> *to_fix, int row_start, int row_end) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2239 to_fix->clear(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2240 for (int i = row_start; i < row_end; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2241 bool needs_fixing = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2242 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2243 SetOfModels models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2244 SetOfModels models_w_crowns; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2245 rows[i].StrongHypotheses(&models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2246 rows[i].NonNullHypotheses(&models_w_crowns); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2247 if (models.empty() && !models_w_crowns.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2248 // Crown paragraph. Is it followed by a modeled line? |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2249 for (unsigned end = i + 1; end < rows.size(); end++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2250 SetOfModels end_models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2251 SetOfModels strong_end_models; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2252 rows[end].NonNullHypotheses(&end_models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2253 rows[end].StrongHypotheses(&strong_end_models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2254 if (end_models.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2255 needs_fixing = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2256 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2257 } else if (!strong_end_models.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2258 needs_fixing = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2259 break; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2260 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2261 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2262 } else if (models.empty() && rows[i].ri_->num_words > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2263 // No models at all. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2264 needs_fixing = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2265 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2266 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2267 if (!needs_fixing && !models.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2268 needs_fixing = RowIsStranded(rows, i); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2269 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2270 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2271 if (needs_fixing) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2272 if (!to_fix->empty() && to_fix->back().end == i - 1) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2273 to_fix->back().end = i; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2274 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2275 to_fix->push_back(Interval(i, i)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2276 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2277 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2278 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2279 // Convert inclusive intervals to half-open intervals. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2280 for (auto &i : *to_fix) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2281 i.end = i.end + 1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2282 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2283 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2284 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2285 // Given a set of row_owners pointing to PARAs or nullptr (no paragraph known), |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2286 // normalize each row_owner to point to an actual PARA, and output the |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2287 // paragraphs in order onto paragraphs. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2288 void CanonicalizeDetectionResults(std::vector<PARA *> *row_owners, PARA_LIST *paragraphs) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2289 std::vector<PARA *> &rows = *row_owners; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2290 paragraphs->clear(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2291 PARA_IT out(paragraphs); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2292 PARA *formerly_null = nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2293 for (unsigned i = 0; i < rows.size(); i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2294 if (rows[i] == nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2295 if (i == 0 || rows[i - 1] != formerly_null) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2296 rows[i] = formerly_null = new PARA(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2297 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2298 rows[i] = formerly_null; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2299 continue; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2300 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2301 } else if (i > 0 && rows[i - 1] == rows[i]) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2302 continue; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2303 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2304 out.add_after_then_move(rows[i]); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2305 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2306 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2307 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2308 // Main entry point for Paragraph Detection Algorithm. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2309 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2310 // Given a set of equally spaced textlines (described by row_infos), |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2311 // Split them into paragraphs. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2312 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2313 // Output: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2314 // row_owners - one pointer for each row, to the paragraph it belongs to. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2315 // paragraphs - this is the actual list of PARA objects. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2316 // models - the list of paragraph models referenced by the PARA objects. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2317 // caller is responsible for deleting the models. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2318 void DetectParagraphs(int debug_level, std::vector<RowInfo> *row_infos, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2319 std::vector<PARA *> *row_owners, PARA_LIST *paragraphs, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2320 std::vector<ParagraphModel *> *models) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2321 ParagraphTheory theory(models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2322 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2323 // Initialize row_owners to be a bunch of nullptr pointers. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2324 row_owners->clear(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2325 row_owners->resize(row_infos->size()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2326 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2327 // Set up row scratch registers for the main algorithm. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2328 std::vector<RowScratchRegisters> rows(row_infos->size()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2329 for (unsigned i = 0; i < row_infos->size(); i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2330 rows[i].Init((*row_infos)[i]); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2331 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2332 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2333 // Pass 1: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2334 // Detect sequences of lines that all contain leader dots (.....) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2335 // These are likely Tables of Contents. If there are three text lines in |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2336 // a row with leader dots, it's pretty safe to say the middle one should |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2337 // be a paragraph of its own. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2338 SeparateSimpleLeaderLines(&rows, 0, rows.size(), &theory); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2339 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2340 DebugDump(debug_level > 1, "End of Pass 1", theory, rows); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2341 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2342 std::vector<Interval> leftovers; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2343 LeftoverSegments(rows, &leftovers, 0, rows.size()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2344 for (auto &leftover : leftovers) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2345 // Pass 2a: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2346 // Find any strongly evidenced start-of-paragraph lines. If they're |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2347 // followed by two lines that look like body lines, make a paragraph |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2348 // model for that and see if that model applies throughout the text |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2349 // (that is, "smear" it). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2350 StrongEvidenceClassify(debug_level, &rows, leftover.begin, leftover.end, &theory); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2351 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2352 // Pass 2b: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2353 // If we had any luck in pass 2a, we got part of the page and didn't |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2354 // know how to classify a few runs of rows. Take the segments that |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2355 // didn't find a model and reprocess them individually. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2356 std::vector<Interval> leftovers2; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2357 LeftoverSegments(rows, &leftovers2, leftover.begin, leftover.end); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2358 bool pass2a_was_useful = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2359 leftovers2.size() > 1 || |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2360 (leftovers2.size() == 1 && (leftovers2[0].begin != 0 || static_cast<size_t>(leftovers2[0].end) != rows.size())); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2361 if (pass2a_was_useful) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2362 for (auto &leftover2 : leftovers2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2363 StrongEvidenceClassify(debug_level, &rows, leftover2.begin, leftover2.end, &theory); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2364 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2365 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2366 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2367 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2368 DebugDump(debug_level > 1, "End of Pass 2", theory, rows); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2369 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2370 // Pass 3: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2371 // These are the dregs for which we didn't have enough strong textual |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2372 // and geometric clues to form matching models for. Let's see if |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2373 // the geometric clues are simple enough that we could just use those. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2374 LeftoverSegments(rows, &leftovers, 0, rows.size()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2375 for (auto &leftover : leftovers) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2376 GeometricClassify(debug_level, &rows, leftover.begin, leftover.end, &theory); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2377 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2378 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2379 // Undo any flush models for which there's little evidence. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2380 DowngradeWeakestToCrowns(debug_level, &theory, &rows); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2381 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2382 DebugDump(debug_level > 1, "End of Pass 3", theory, rows); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2383 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2384 // Pass 4: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2385 // Take everything that's still not marked up well and clear all markings. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2386 LeftoverSegments(rows, &leftovers, 0, rows.size()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2387 for (auto &leftover : leftovers) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2388 for (int j = leftover.begin; j < leftover.end; j++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2389 rows[j].SetUnknown(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2390 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2391 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2392 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2393 DebugDump(debug_level > 1, "End of Pass 4", theory, rows); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2394 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2395 // Convert all of the unique hypothesis runs to PARAs. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2396 ConvertHypothesizedModelRunsToParagraphs(debug_level, rows, row_owners, &theory); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2397 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2398 DebugDump(debug_level > 0, "Final Paragraph Segmentation", theory, rows); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2399 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2400 // Finally, clean up any dangling nullptr row paragraph parents. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2401 CanonicalizeDetectionResults(row_owners, paragraphs); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2402 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2403 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2404 // ============ Code interfacing with the rest of Tesseract ================== |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2405 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2406 static void InitializeTextAndBoxesPreRecognition(const MutableIterator &it, RowInfo *info) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2407 // Set up text, lword_text, and rword_text (mostly for debug printing). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2408 std::string fake_text; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2409 PageIterator pit(static_cast<const PageIterator &>(it)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2410 if (!pit.Empty(RIL_WORD)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2411 bool first_word = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2412 do { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2413 fake_text += "x"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2414 if (first_word) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2415 info->lword_text += "x"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2416 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2417 info->rword_text += "x"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2418 if (pit.IsAtFinalElement(RIL_WORD, RIL_SYMBOL) && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2419 !pit.IsAtFinalElement(RIL_TEXTLINE, RIL_SYMBOL)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2420 fake_text += " "; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2421 info->rword_text = ""; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2422 first_word = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2423 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2424 } while (!pit.IsAtFinalElement(RIL_TEXTLINE, RIL_SYMBOL) && pit.Next(RIL_SYMBOL)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2425 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2426 if (fake_text.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2427 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2428 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2429 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2430 int lspaces = info->pix_ldistance / info->average_interword_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2431 for (int i = 0; i < lspaces; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2432 info->text += ' '; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2433 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2434 info->text += fake_text; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2435 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2436 // Set up lword_box, rword_box, and num_words. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2437 PAGE_RES_IT page_res_it = *it.PageResIt(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2438 WERD_RES *word_res = page_res_it.restart_row(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2439 ROW_RES *this_row = page_res_it.row(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2440 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2441 WERD_RES *lword = nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2442 WERD_RES *rword = nullptr; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2443 info->num_words = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2444 do { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2445 if (word_res) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2446 if (!lword) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2447 lword = word_res; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2448 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2449 if (rword != word_res) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2450 info->num_words++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2451 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2452 rword = word_res; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2453 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2454 word_res = page_res_it.forward(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2455 } while (page_res_it.row() == this_row); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2456 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2457 if (lword) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2458 info->lword_box = lword->word->bounding_box(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2459 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2460 if (rword) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2461 info->rword_box = rword->word->bounding_box(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2462 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2463 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2464 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2465 // Given a Tesseract Iterator pointing to a text line, fill in the paragraph |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2466 // detector RowInfo with all relevant information from the row. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2467 static void InitializeRowInfo(bool after_recognition, const MutableIterator &it, RowInfo *info) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2468 if (it.PageResIt()->row() != nullptr) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2469 ROW *row = it.PageResIt()->row()->row; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2470 info->pix_ldistance = row->lmargin(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2471 info->pix_rdistance = row->rmargin(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2472 info->average_interword_space = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2473 row->space() > 0 ? row->space() : std::max(static_cast<int>(row->x_height()), 1); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2474 info->pix_xheight = row->x_height(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2475 info->has_leaders = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2476 info->has_drop_cap = row->has_drop_cap(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2477 info->ltr = true; // set below depending on word scripts |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2478 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2479 info->pix_ldistance = info->pix_rdistance = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2480 info->average_interword_space = 1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2481 info->pix_xheight = 1.0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2482 info->has_leaders = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2483 info->has_drop_cap = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2484 info->ltr = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2485 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2486 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2487 info->num_words = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2488 info->lword_indicates_list_item = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2489 info->lword_likely_starts_idea = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2490 info->lword_likely_ends_idea = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2491 info->rword_indicates_list_item = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2492 info->rword_likely_starts_idea = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2493 info->rword_likely_ends_idea = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2494 info->has_leaders = false; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2495 info->ltr = true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2496 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2497 if (!after_recognition) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2498 InitializeTextAndBoxesPreRecognition(it, info); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2499 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2500 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2501 info->text = ""; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2502 const std::unique_ptr<const char[]> text(it.GetUTF8Text(RIL_TEXTLINE)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2503 int trailing_ws_idx = strlen(text.get()); // strip trailing space |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2504 while (trailing_ws_idx > 0 && |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2505 // isspace() only takes ASCII |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2506 isascii(text[trailing_ws_idx - 1]) && isspace(text[trailing_ws_idx - 1])) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2507 trailing_ws_idx--; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2508 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2509 if (trailing_ws_idx > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2510 int lspaces = info->pix_ldistance / info->average_interword_space; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2511 for (int i = 0; i < lspaces; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2512 info->text += ' '; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2513 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2514 for (int i = 0; i < trailing_ws_idx; i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2515 info->text += text[i]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2516 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2517 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2518 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2519 if (info->text.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2520 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2521 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2522 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2523 PAGE_RES_IT page_res_it = *it.PageResIt(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2524 std::vector<WERD_RES *> werds; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2525 WERD_RES *word_res = page_res_it.restart_row(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2526 ROW_RES *this_row = page_res_it.row(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2527 int num_leaders = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2528 int ltr = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2529 int rtl = 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2530 do { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2531 if (word_res && word_res->best_choice->unichar_string().length() > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2532 werds.push_back(word_res); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2533 ltr += word_res->AnyLtrCharsInWord() ? 1 : 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2534 rtl += word_res->AnyRtlCharsInWord() ? 1 : 0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2535 if (word_res->word->flag(W_REP_CHAR)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2536 num_leaders++; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2537 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2538 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2539 word_res = page_res_it.forward(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2540 } while (page_res_it.row() == this_row); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2541 info->ltr = ltr >= rtl; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2542 info->has_leaders = num_leaders > 3; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2543 info->num_words = werds.size(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2544 if (!werds.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2545 WERD_RES *lword = werds[0], *rword = werds[werds.size() - 1]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2546 info->lword_text = lword->best_choice->unichar_string().c_str(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2547 info->rword_text = rword->best_choice->unichar_string().c_str(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2548 info->lword_box = lword->word->bounding_box(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2549 info->rword_box = rword->word->bounding_box(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2550 LeftWordAttributes(lword->uch_set, lword->best_choice, info->lword_text, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2551 &info->lword_indicates_list_item, &info->lword_likely_starts_idea, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2552 &info->lword_likely_ends_idea); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2553 RightWordAttributes(rword->uch_set, rword->best_choice, info->rword_text, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2554 &info->rword_indicates_list_item, &info->rword_likely_starts_idea, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2555 &info->rword_likely_ends_idea); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2556 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2557 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2558 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2559 // This is called after rows have been identified and words are recognized. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2560 // Much of this could be implemented before word recognition, but text helps |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2561 // to identify bulleted lists and gives good signals for sentence boundaries. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2562 void DetectParagraphs(int debug_level, bool after_text_recognition, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2563 const MutableIterator *block_start, std::vector<ParagraphModel *> *models) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2564 // Clear out any preconceived notions. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2565 if (block_start->Empty(RIL_TEXTLINE)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2566 return; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2567 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2568 BLOCK *block = block_start->PageResIt()->block()->block; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2569 block->para_list()->clear(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2570 bool is_image_block = block->pdblk.poly_block() && !block->pdblk.poly_block()->IsText(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2571 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2572 // Convert the Tesseract structures to RowInfos |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2573 // for the paragraph detection algorithm. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2574 MutableIterator row(*block_start); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2575 if (row.Empty(RIL_TEXTLINE)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2576 return; // end of input already. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2577 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2578 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2579 std::vector<RowInfo> row_infos; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2580 do { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2581 if (!row.PageResIt()->row()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2582 continue; // empty row. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2583 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2584 row.PageResIt()->row()->row->set_para(nullptr); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2585 row_infos.emplace_back(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2586 RowInfo &ri = row_infos.back(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2587 InitializeRowInfo(after_text_recognition, row, &ri); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2588 } while (!row.IsAtFinalElement(RIL_BLOCK, RIL_TEXTLINE) && row.Next(RIL_TEXTLINE)); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2589 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2590 // If we're called before text recognition, we might not have |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2591 // tight block bounding boxes, so trim by the minimum on each side. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2592 if (!row_infos.empty()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2593 int min_lmargin = row_infos[0].pix_ldistance; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2594 int min_rmargin = row_infos[0].pix_rdistance; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2595 for (unsigned i = 1; i < row_infos.size(); i++) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2596 if (row_infos[i].pix_ldistance < min_lmargin) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2597 min_lmargin = row_infos[i].pix_ldistance; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2598 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2599 if (row_infos[i].pix_rdistance < min_rmargin) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2600 min_rmargin = row_infos[i].pix_rdistance; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2601 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2602 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2603 if (min_lmargin > 0 || min_rmargin > 0) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2604 for (auto &row_info : row_infos) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2605 row_info.pix_ldistance -= min_lmargin; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2606 row_info.pix_rdistance -= min_rmargin; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2607 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2608 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2609 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2610 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2611 // Run the paragraph detection algorithm. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2612 std::vector<PARA *> row_owners; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2613 if (!is_image_block) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2614 DetectParagraphs(debug_level, &row_infos, &row_owners, block->para_list(), models); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2615 } else { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2616 row_owners.resize(row_infos.size()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2617 CanonicalizeDetectionResults(&row_owners, block->para_list()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2618 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2619 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2620 // Now stitch in the row_owners into the rows. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2621 row = *block_start; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2622 for (auto &row_owner : row_owners) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2623 while (!row.PageResIt()->row()) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2624 row.Next(RIL_TEXTLINE); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2625 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2626 row.PageResIt()->row()->row->set_para(row_owner); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2627 row.Next(RIL_TEXTLINE); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2628 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2629 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2630 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2631 } // namespace tesseract |
