Mercurial > hgrepos > Python2 > PyMuPDF
annotate mupdf-source/thirdparty/tesseract/src/training/unicharset_extractor.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: unicharset_extractor.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: Unicode character/ligature set extractor. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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: Thomas Kielbus |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 2006, 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 // Given a list of box files or text files on the command line, this program |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
20 // normalizes the text according to command-line options and generates |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
21 // a 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
|
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 <cstdlib> |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 <filesystem> |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 "boxread.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 #include "commandlineflags.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
|
27 #include "commontraining.h" // CheckSharedLibraryVersion |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 "lang_model_helpers.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
|
29 #include "normstrngs.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
|
30 #include "unicharset.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
|
31 #include "unicharset_training_utils.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
|
32 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
33 using 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
|
34 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
35 static STRING_PARAM_FLAG(output_unicharset, "unicharset", "Output file path"); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
36 static INT_PARAM_FLAG(norm_mode, 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
|
37 "Normalization mode: 1=Combine graphemes, " |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
38 "2=Split graphemes, 3=Pure 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
|
39 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
40 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
|
41 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
42 // Helper normalizes and segments the given strings according to norm_mode, 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
|
43 // adds the segmented parts to 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
|
44 static void AddStringsToUnicharset(const std::vector<std::string> &strings, int norm_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
|
45 UNICHARSET *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
|
46 for (const auto &string : strings) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
47 std::vector<std::string> normalized; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
48 if (NormalizeCleanAndSegmentUTF8(UnicodeNormMode::kNFC, OCRNorm::kNone, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
49 static_cast<GraphemeNormMode>(norm_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
|
50 /*report_errors*/ true, string.c_str(), &normalized)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
51 for (const std::string &normed : normalized) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
52 // normed is a UTF-8 encoded 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
|
53 if (normed.empty() || IsUTF8Whitespace(normed.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
|
54 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
|
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 unicharset->unichar_insert(normed.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
|
57 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
58 } 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
|
59 tprintf("Normalization failed for string '%s'\n", 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
|
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
|
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 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
63 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
64 static int Main(int argc, char **argv) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
65 UNICHARSET 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
|
66 // Load input files |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
67 for (int arg = 1; arg < argc; ++arg) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
68 std::filesystem::path filePath = argv[arg]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
69 std::string file_data = tesseract::ReadFile(argv[arg]); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
70 if (file_data.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
|
71 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
|
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 std::vector<std::string> 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
|
74 if (filePath.extension() == ".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
|
75 tprintf("Extracting unicharset from box file %s\n", argv[arg]); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
76 bool res = ReadMemBoxes(-1, /*skip_blanks*/ true, &file_data[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
|
77 /*continue_on_failure*/ false, /*boxes*/ nullptr, &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
|
78 /*box_texts*/ nullptr, /*pages*/ 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
|
79 if (!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
|
80 tprintf("Cannot read box data from '%s'\n", argv[arg]); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
81 return EXIT_FAILURE; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
82 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
83 } 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
|
84 tprintf("Extracting unicharset from plain text file %s\n", argv[arg]); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
85 texts.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
|
86 texts = split(file_data, '\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
|
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 AddStringsToUnicharset(texts, FLAGS_norm_mode, &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
|
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 SetupBasicProperties(/*report_errors*/ true, /*decompose*/ false, &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
|
91 // Write unicharset file. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
92 if (unicharset.save_to_file(FLAGS_output_unicharset.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
|
93 tprintf("Wrote unicharset file %s\n", FLAGS_output_unicharset.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
|
94 } 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
|
95 tprintf("Cannot save unicharset file %s\n", FLAGS_output_unicharset.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
|
96 return EXIT_FAILURE; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
97 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
98 return EXIT_SUCCESS; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
99 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
100 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
101 } // 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
|
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 int main(int argc, char **argv) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
104 tesseract::CheckSharedLibraryVersion(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
105 if (argc > 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
|
106 tesseract::ParseCommandLineFlags(argv[0], &argc, &argv, 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
|
107 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
108 if (argc < 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
|
109 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
|
110 "Usage: %s [--output_unicharset filename] [--norm_mode 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
|
111 " box_or_text_file [...]\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
|
112 argv[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
|
113 tprintf("Where mode means:\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
|
114 tprintf(" 1=combine graphemes (use for Latin and other simple scripts)\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
|
115 tprintf(" 2=split graphemes (use for Indic/Khmer/Myanmar)\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
|
116 tprintf(" 3=pure unicode (use for Arabic/Hebrew/Thai/Tibetan)\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
|
117 tprintf("Reads box or plain text files to extract the unicharset.\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
|
118 return EXIT_FAILURE; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
119 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
120 return tesseract::Main(argc, argv); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
121 } |
