Mercurial > hgrepos > Python2 > PyMuPDF
annotate mupdf-source/thirdparty/tesseract/src/training/unicharset/lang_model_helpers.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 // Copyright 2017 Google Inc. All Rights Reserved. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2 // Author: rays@google.com (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
|
3 // Purpose: Collection of convenience functions to simplify creation 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
|
4 // unicharset, recoder, and dawgs for an LSTM 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
|
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 // 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
|
7 // 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
|
8 // 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
|
9 // 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
|
10 // 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
|
11 // 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
|
12 // 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
|
13 // 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
|
14 // 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
|
15 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
16 #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
|
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 #include "dawg.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
|
19 #include "fileio.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 #include "tessdatamanager.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 #include "trie.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 #include "unicharcompress.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
|
23 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 <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
|
25 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 <sys/stat.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 <sys/types.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
|
28 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
29 #if defined(_WIN32) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 <direct.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 #endif |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 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 // Helper makes a filename (<output_dir>/<lang>/<lang><suffix>) and writes 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
|
36 // to the file, using writer if not null, otherwise, a default writer. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
37 // Default writer will overwrite any existing file, but a supplied writer |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
38 // can do its own thing. If lang is empty, returns true but does 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
|
39 // NOTE that suffix should contain any required . for the filename. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
40 bool WriteFile(const std::string &output_dir, const std::string &lang, const std::string &suffix, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
41 const std::vector<char> &data, FileWriter writer) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
42 if (lang.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
|
43 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
|
44 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
45 std::string dirname = output_dir + "/" + lang; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
46 // Attempt to make the directory, but ignore errors, as it may not be 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
|
47 // standard filesystem, and the writer will complain if not successful. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 defined(_WIN32) |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
49 _mkdir(dirname.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
|
50 #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
|
51 mkdir(dirname.c_str(), S_IRWXU | S_IRWXG); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
52 #endif |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
53 std::string filename = dirname + "/" + lang + suffix; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
54 if (writer == 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
|
55 return SaveDataToFile(data, filename.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
|
56 } 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
|
57 return (*writer)(data, filename.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
|
58 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
59 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 // Helper reads a file with optional reader and returns a 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
|
62 // On failure emits a warning message and returns an empty 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
|
63 std::string ReadFile(const std::string &filename, FileReader reader) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
64 if (filename.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
|
65 return std::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
|
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 std::vector<char> 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
|
68 bool read_result; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
69 if (reader == 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
|
70 read_result = LoadDataFromFile(filename.c_str(), &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
|
71 } 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
|
72 read_result = (*reader)(filename.c_str(), &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
|
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 if (read_result) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
75 return std::string(&data[0], data.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
|
76 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
77 tprintf("Failed to read data from: %s\n", filename.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
|
78 return std::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
|
79 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
80 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
81 // Helper writes the unicharset to file and to the traineddata. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
82 bool WriteUnicharset(const UNICHARSET &unicharset, const std::string &output_dir, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
83 const std::string &lang, FileWriter writer, TessdataManager *traineddata) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
84 std::vector<char> unicharset_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
|
85 TFile fp; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
86 fp.OpenWrite(&unicharset_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
|
87 if (!unicharset.save_to_file(&fp)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 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
|
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 traineddata->OverwriteEntry(TESSDATA_LSTM_UNICHARSET, &unicharset_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
|
91 unicharset_data.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
|
92 return WriteFile(output_dir, lang, ".unicharset", unicharset_data, writer); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
93 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 // Helper creates the recoder and writes it to the traineddata, and a human- |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
96 // readable form to 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
|
97 bool WriteRecoder(const UNICHARSET &unicharset, bool pass_through, const std::string &output_dir, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
98 const std::string &lang, FileWriter writer, std::string *radical_table_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
|
99 TessdataManager *traineddata) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
100 UnicharCompress recoder; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
101 // Where the unicharset is carefully setup already to contain a good |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
102 // compact encoding, use a pass-through recoder that does 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
|
103 // For scripts that have a large number of unicodes (Han, Hangul) we want |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
104 // to use the recoder to compress the symbol space by re-encoding 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
|
105 // unicode as multiple codes from a smaller 'alphabet' that are related to 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
|
106 // shapes in the character. Hangul Jamo is a perfect example of this. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
107 // See the Hangul Syllables section, sub-section "Equivalence" 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
|
108 // http://www.unicode.org/versions/Unicode10.0.0/ch18.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
|
109 if (pass_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
|
110 recoder.SetupPassThrough(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
|
111 } 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
|
112 int null_char = unicharset.has_special_codes() ? UNICHAR_BROKEN : unicharset.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
|
113 tprintf("Null char=%d\n", null_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
|
114 if (!recoder.ComputeEncoding(unicharset, null_char, radical_table_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
|
115 tprintf("Creation of encoded unicharset failed!!\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 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
|
117 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
118 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
119 TFile fp; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
120 std::vector<char> recoder_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
|
121 fp.OpenWrite(&recoder_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
|
122 if (!recoder.Serialize(&fp)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
123 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
|
124 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
125 traineddata->OverwriteEntry(TESSDATA_LSTM_RECODER, &recoder_data[0], recoder_data.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
|
126 std::string encoding = recoder.GetEncodingAsString(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
|
127 recoder_data.resize(encoding.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
|
128 memcpy(&recoder_data[0], &encoding[0], encoding.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
|
129 std::string suffix; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
130 suffix += ".charset_size=" + std::to_string(recoder.code_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
|
131 suffix += ".txt"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
132 return WriteFile(output_dir, lang, suffix, recoder_data, writer); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
133 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 // Helper builds a dawg from the given words, using the unicharset as coding, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
136 // and reverse_policy for LTR/RTL, and overwrites file_type in the traineddata. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
137 static bool WriteDawg(const std::vector<std::string> &words, const 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
|
138 Trie::RTLReversePolicy reverse_policy, TessdataType file_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
|
139 TessdataManager *traineddata) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
140 // The first 3 arguments are not used in this case. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
141 Trie trie(DAWG_TYPE_WORD, "", SYSTEM_DAWG_PERM, unicharset.size(), 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
|
142 trie.add_word_list(words, unicharset, reverse_policy); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
143 tprintf("Reducing Trie to SquishedDawg\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
|
144 std::unique_ptr<SquishedDawg> dawg(trie.trie_to_dawg()); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
145 if (dawg == nullptr || dawg->NumEdges() == 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
|
146 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
|
147 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
148 TFile fp; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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<char> dawg_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
|
150 fp.OpenWrite(&dawg_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
|
151 if (!dawg->write_squished_dawg(&fp)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
152 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
|
153 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
154 traineddata->OverwriteEntry(file_type, &dawg_data[0], dawg_data.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
|
155 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
|
156 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
157 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
158 // Builds and writes the dawgs, given a set of words, 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
|
159 // patterns, number patterns, to the traineddata. Encoding uses the given |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
160 // unicharset, and the punc dawgs is reversed if lang_is_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
|
161 static bool WriteDawgs(const std::vector<std::string> &words, const std::vector<std::string> &puncs, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
162 const std::vector<std::string> &numbers, bool lang_is_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
|
163 const UNICHARSET &unicharset, TessdataManager *traineddata) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
164 if (puncs.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
|
165 tprintf("Must have non-empty puncs list to use language 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
|
166 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
|
167 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
168 // For each of the dawg types, make the dawg, and write to traineddata. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
169 // Dawgs are reversed as follows: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
170 // Words: According to the word content. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
171 // Puncs: According to lang_is_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
|
172 // Numbers: Never. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
173 // System dawg (main wordlist). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
174 if (!words.empty() && !WriteDawg(words, unicharset, Trie::RRP_REVERSE_IF_HAS_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
|
175 TESSDATA_LSTM_SYSTEM_DAWG, traineddata)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
176 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
|
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 // punc/punc-dawg. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
179 Trie::RTLReversePolicy reverse_policy = |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
180 lang_is_rtl ? Trie::RRP_FORCE_REVERSE : Trie::RRP_DO_NO_REVERSE; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
181 if (!WriteDawg(puncs, unicharset, reverse_policy, TESSDATA_LSTM_PUNC_DAWG, traineddata)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
182 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
|
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 // numbers/number-dawg. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
185 if (!numbers.empty() && !WriteDawg(numbers, unicharset, Trie::RRP_DO_NO_REVERSE, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
186 TESSDATA_LSTM_NUMBER_DAWG, traineddata)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
187 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
|
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 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
|
190 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
191 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
192 // The main function for combine_lang_model.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
|
193 // Returns EXIT_SUCCESS or EXIT_FAILURE for error. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
194 int CombineLangModel(const UNICHARSET &unicharset, const std::string &script_dir, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
195 const std::string &version_str, const std::string &output_dir, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
196 const std::string &lang, bool pass_through_recoder, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
197 const std::vector<std::string> &words, const std::vector<std::string> &puncs, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
198 const std::vector<std::string> &numbers, bool lang_is_rtl, FileReader reader, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
199 FileWriter writer) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
200 // Build the traineddata 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
|
201 TessdataManager traineddata; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
202 if (!version_str.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
|
203 traineddata.SetVersionString(traineddata.VersionString() + ":" + version_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
|
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 // Unicharset and recoder. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
206 if (!WriteUnicharset(unicharset, output_dir, lang, writer, &traineddata)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
207 tprintf("Error writing 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
|
208 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
|
209 } 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
|
210 tprintf("Config file is optional, continuing...\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
|
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 // If there is a config file, read it and add to traineddata. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
213 std::string config_filename = script_dir + "/" + lang + "/" + lang + ".config"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
214 std::string config_file = ReadFile(config_filename, reader); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
215 if (config_file.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
|
216 traineddata.OverwriteEntry(TESSDATA_LANG_CONFIG, &config_file[0], config_file.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
|
217 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
218 std::string radical_filename = script_dir + "/radical-stroke.txt"; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
219 std::string radical_data = ReadFile(radical_filename, reader); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
220 if (radical_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
|
221 tprintf("Error reading radical code table %s\n", radical_filename.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
|
222 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
|
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 if (!WriteRecoder(unicharset, pass_through_recoder, output_dir, lang, writer, &radical_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
|
225 &traineddata)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
226 tprintf("Error writing recoder!!\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
|
227 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
228 if (!words.empty() || !puncs.empty() || !numbers.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
|
229 if (!WriteDawgs(words, puncs, numbers, lang_is_rtl, unicharset, &traineddata)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
230 tprintf("Error during conversion of wordlists to DAWGs!!\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
|
231 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
|
232 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
233 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
234 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
235 // Traineddata 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
|
236 std::vector<char> traineddata_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
|
237 traineddata.Serialize(&traineddata_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
|
238 if (!WriteFile(output_dir, lang, ".traineddata", traineddata_data, writer)) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
239 tprintf("Error writing output traineddata 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
|
240 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
|
241 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
242 tprintf("Created %s/%s/%s.traineddata", output_dir.c_str(), lang.c_str(), lang.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
|
243 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
|
244 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
245 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
246 } // namespace tesseract |
