Mercurial > hgrepos > Python2 > PyMuPDF

annotate mupdf-source/thirdparty/tesseract/src/training/unicharset/validator.h @ 21:2f43e400f144

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Provide an "all" target to build both the sdist and the wheel
author Franz Glasner <fzglas.hg@dom66.de>
date Fri, 19 Sep 2025 10:28:53 +0200
parents b50eed0cc0ef
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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: validator.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
3 * Description: Base class for various text validators. Intended mainly 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
4 * scripts that use a virama character.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
5 * Author: Ray Smith
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
6 *
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
7 * (C) Copyright 2017, 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
8 * 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
9 * 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
10 * 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
11 * 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
12 * 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
13 * 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
14 * 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
15 * 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
16 * 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
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
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
20 #ifndef TESSERACT_TRAINING_VALIDATOR_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
21 #define TESSERACT_TRAINING_VALIDATOR_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
22
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 "export.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
24
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 <tesseract/unichar.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
26
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 <memory>
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 <vector>
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
29
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
30 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
31
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
32 // Different kinds of grapheme normalization - not just for Indic!
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
33 // A grapheme is a syllable unit in Indic and can be several 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
34 // In other scripts, a grapheme is a base character and accent/diacritic
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
35 // combination, as not all accented characters have a single composed form.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
36 enum class GraphemeNormMode {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
37 // Validation result is a single string, even if input is multi-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
38 kSingleString,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
39 // Standard unicode graphemes are validated and output as grapheme units.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
40 kCombined,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
41 // Graphemes are validated and sub-divided. For virama-using scripts, units
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
42 // that correspond to repeatable glyphs are generated. (Mostly single 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
43 // but viramas and joiners are paired with the most sensible neighbor.)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
44 // For non-virama scripts, this means that base/accent pairs are separated,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
45 // ie the output is individual 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
46 kGlyphSplit,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
47 // The output is always single unicodes, regardless of the script.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
48 kIndividualUnicodes,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
49 };
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 // An enum representing the scripts that use a virama character. It 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
52 // guaranteed that the value of any element, (except kNonVirama) can be cast
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
53 // to a unicode (char32) value that represents the start of the unicode range
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
54 // of the corresponding script.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
55 enum class ViramaScript : char32 {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
56 kNonVirama = 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
57 kDevanagari = 0x900,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
58 kBengali = 0x980,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
59 kGurmukhi = 0xa00,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
60 kGujarati = 0xa80,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
61 kOriya = 0xb00,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
62 kTamil = 0xb80,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
63 kTelugu = 0xc00,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
64 kKannada = 0xc80,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
65 kMalayalam = 0xd00,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
66 kSinhala = 0xd80,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
67 kMyanmar = 0x1000,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
68 kKhmer = 0x1780,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
69 kJavanese = 0xa980,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
70 };
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
71
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
72 // Base class offers a validation API and protected methods to allow subclasses
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
73 // to easily build the validated/segmented 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
74 class TESS_UNICHARSET_TRAINING_API Validator {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
75 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
76 // Validates and cleans the src vector of unicodes to the *dest, according 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
77 // g_mode. In the case of kSingleString, a single vector containing the whole
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
78 // result is added to *dest. With kCombined, multiple vectors are added 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
79 // *dest with one grapheme in each. With kGlyphSplit, multiple vectors 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
80 // added to *dest with a smaller unit representing a glyph in each.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
81 // In case of validation error, returns false and as much as possible of 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
82 // input, without discarding invalid 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
83 static bool ValidateCleanAndSegment(GraphemeNormMode g_mode, bool report_errors,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
84 const std::vector<char32> &src,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
85 std::vector<std::vector<char32>> *dest);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
86
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
87 // Returns true if the unicode ch is a non-printing zero-width mark of no
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
88 // significance to OCR training or evaluation.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
89 static bool IsZeroWidthMark(char32 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
90 return ch == kZeroWidthSpace || ch == kLeftToRightMark || ch == kRightToLeftMark ||
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
91 ch == kInvalid;
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 virtual ~Validator();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
94
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
95 // Some specific but universally useful 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
96 static const char32 kZeroWidthSpace;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
97 static const char32 kZeroWidthNonJoiner;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
98 static const char32 kZeroWidthJoiner;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
99 static const char32 kLeftToRightMark;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
100 static const char32 kRightToLeftMark;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
101 static const char32 kInvalid;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
102
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
103 protected:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
104 // These are more or less the character class identifiers in the ISCII
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
105 // standard, section 8. They have been augmented with the Unicode meta
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
106 // characters Zero Width Joiner and Zero Width Non Joiner, 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
107 // Unicode Vedic Marks.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
108 // The best sources of information on Unicode and Indic scripts 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
109 // http://varamozhi.sourceforge.net/iscii91.pdf
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
110 // http://www.unicode.org/versions/Unicode9.0.0/ch12.pdf
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
111 // http://unicode.org/faq/indic.html
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
112 // http://www.microsoft.com/typography/otfntdev/teluguot/shaping.aspx
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
113 enum class CharClass {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
114 // NOTE: The values of the enum members are meaningless and arbitrary, ie
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
115 // they are not used for sorting, or any other risky application.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
116 // The reason they are what they are is they are a single character
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
117 // abbreviation that can be used in a regexp/BNF definition of a grammar,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
118 // IN A COMMENT, and still not relied upon in the 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
119 kConsonant = '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
120 kVowel = 'V',
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
121 kVirama = 'H', // (aka Halant)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
122 kMatra = 'M', // (aka Dependent Vowel)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
123 kMatraPiece = 'P', // unicode provides pieces of Matras.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
124 kVowelModifier = 'D', // (candrabindu, anusvara, visarga, other marks)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
125 kZeroWidthNonJoiner = 'z', // Unicode Zero Width Non-Joiner U+200C
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
126 kZeroWidthJoiner = 'Z', // Unicode Zero Width Joiner U+200D
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
127 kVedicMark = 'v', // Modifiers can come modify any indic syllable.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
128 kNukta = 'N', // Occurs only immediately after consonants.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
129 kRobat = 'R', // Khmer only.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
130 kOther = 'O', // (digits, measures, non-Indic, etc)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
131 // Additional classes used only by ValidateGrapheme.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
132 kWhitespace = ' ',
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
133 kCombiner = 'c', // Combiners other than virama.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
134 };
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
135 using IndicPair = std::pair<CharClass, char32>;
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 Validator(ViramaScript script, bool report_errors)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
138 : script_(script), codes_used_(0), output_used_(0), report_errors_(report_errors) {}
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 // Factory method that understands how to map script to the right subclass.
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 std::unique_ptr<Validator> ScriptValidator(ViramaScript script, bool report_errors);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
142
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
143 // Internal version of the public static ValidateCleanAndSegment.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
144 // Validates and cleans the src vector of unicodes to the *dest, according 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
145 // its type and the given g_mode.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
146 // In case of validation error, returns false and returns as much as possible
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
147 // of the input, without discarding invalid 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
148 bool ValidateCleanAndSegmentInternal(GraphemeNormMode g_mode, const std::vector<char32> &src,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
149 std::vector<std::vector<char32>> *dest);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
150 // Moves the results from parts_ or output_ to dest according to g_mode.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
151 void MoveResultsToDest(GraphemeNormMode g_mode, std::vector<std::vector<char32>> *dest);
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 // Computes and returns the ViramaScript corresponding to the most 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
154 // virama-using script in the input, or kNonVirama if none are present.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
155 static ViramaScript MostFrequentViramaScript(const std::vector<char32> &utf32);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
156 // Returns true if the given UTF-32 unicode is a "virama" character.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
157 static bool IsVirama(char32 unicode);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
158 // Returns true if the given UTF-32 unicode is a vedic accent.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
159 static bool IsVedicAccent(char32 unicode);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
160 // Returns true if the script is one that uses subscripts for conjuncts.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
161 bool IsSubscriptScript() 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
162
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
163 // Helper function appends the next element of codes_ only to 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
164 // without touching parts_
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
165 // Returns true at the end of codes_.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
166 bool CodeOnlyToOutput() {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
167 output_.push_back(codes_[codes_used_].second);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
168 return ++codes_used_ == codes_.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
169 }
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 // Helper function adds a length-element vector to parts_ from the last 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
172 // elements of output_. If there are more than length unused elements 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
173 // output_, adds unicodes as single-element vectors to parts_ to catch
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
174 // output_used_ up to output->size() - length before adding the length-element
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
175 // vector.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
176 void MultiCodePart(unsigned 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
177 while (output_used_ + length < output_.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
178 parts_.emplace_back(std::initializer_list<char32>{output_[output_used_++]});
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 parts_.emplace_back(std::initializer_list<char32>{output_[output_used_]});
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
181 while (++output_used_ < output_.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
182 parts_.back().push_back(output_[output_used_]);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
183 }
b50eed0cc0ef ADD: MuPDF v1.26.7: 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
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
186 // Helper function appends the next element of codes_ to output_, and then
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
187 // calls MultiCodePart to add the appropriate components to parts_.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
188 // Returns true at the end of codes_.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
189 bool UseMultiCode(unsigned 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
190 output_.push_back(codes_[codes_used_].second);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
191 MultiCodePart(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
192 return ++codes_used_ == codes_.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
193 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
194
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
195 // Consumes the next Grapheme in codes_[codes_used_++...] and copies it 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
196 // parts_ and output_. Returns true if a valid Grapheme was consumed,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
197 // otherwise does not increment codes_used_.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
198 virtual bool ConsumeGraphemeIfValid() = 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
199 // Sets codes_ to the class codes for the given unicode 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
200 void ComputeClassCodes(const std::vector<char32> &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
201 // Returns the CharClass corresponding to the given Unicode 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 virtual CharClass UnicodeToCharClass(char32 ch) const = 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
203 // Resets to the initial state.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
204 void 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
205
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
206 // Number of unicodes in each Indic codepage.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
207 static const int kIndicCodePageSize = 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
208 // Lowest unicode value of any Indic script. (Devanagari).
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 const char32 kMinIndicUnicode = 0x900;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
210 // Highest unicode value of any consistent (ISCII-based) Indic script.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
211 static const char32 kMaxSinhalaUnicode = 0xdff;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
212 // Highest unicode value of any virama-using script. (Khmer).
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 const char32 kMaxViramaScriptUnicode = 0x17ff;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
214 // Some special 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
215 static const char32 kSinhalaVirama = 0xdca;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
216 static const char32 kMyanmarVirama = 0x1039;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
217 static const char32 kKhmerVirama = 0x17d2;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
218 // Javanese Script - aksarajawa
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
219 static const char32 kJavaneseVirama = 0xa9c0;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
220 static const char32 kMaxJavaneseUnicode = 0xa9df;
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 // Script we are operating 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
223 ViramaScript script_;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
224 // Input unicodes with assigned CharClass is the data to be validated.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
225 std::vector<IndicPair> codes_;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
226 // Glyph-like components of the input.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
227 std::vector<std::vector<char32>> parts_;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
228 // Copied validated unicodes from codes_ that are OK to 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
229 std::vector<char32> 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
230 // The number of elements of codes_ that have been processed so 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
231 unsigned codes_used_;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
232 // The number of elements of output_ that have already been added to parts_.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
233 unsigned output_used_;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
234 // Log error messages for reasons why text is invalid.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
235 bool report_errors_;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
236 };
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 } // 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
239
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
240 #endif // TESSERACT_TRAINING_VALIDATOR_H_