Mercurial > hgrepos > Python2 > PyMuPDF

annotate mupdf-source/thirdparty/tesseract/src/training/unicharset/normstrngs.cpp @ 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: normstrngs.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: Utilities to normalize and manipulate UTF-32 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
4 * UTF-8 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
5 * Author: Ranjith Unnikrishnan
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 2013, 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 #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
21
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 <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
23 #include <unordered_map>
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 <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
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 <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
27 #include "errcode.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 #include "icuerrorcode.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 "unicode/normalizer2.h" // From libicu
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 "unicode/translit.h" // From libicu
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 "unicode/uchar.h" // From libicu
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
32 #include "unicode/unorm2.h" // From libicu
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
33 #include "unicode/uscript.h" // From libicu
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 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
36
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
37 static bool is_hyphen_punc(const 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
38 static const char32 kHyphenPuncUnicodes[] = {
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 0x2010, // hyphen
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
41 0x2011, // non-breaking hyphen
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
42 0x2012, // figure dash
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
43 0x2013, // en dash
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
44 0x2014, // em dash
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
45 0x2015, // horizontal bar
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
46 // how about 0x2043 hyphen bullet?
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
47 // how about 0x2500 box drawings light horizontal?
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
48 0x207b, // superscript minus
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
49 0x208b, // subscript minus
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
50 0x2212, // minus sign
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
51 0xfe58, // small em dash
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
52 0xfe63, // small hyphen-minus
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
53 0xff0d, // fullwidth hyphen-minus
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
54 0x2e17 // double oblique hyphen (Fraktur)
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 for (int kHyphenPuncUnicode : kHyphenPuncUnicodes) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
57 if (kHyphenPuncUnicode == 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
58 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
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 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
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 bool is_single_quote(const 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
65 static const char32 kSingleQuoteUnicodes[] = {
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 0x2018, // left single quotation mark (English, others)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
68 0x2019, // right single quotation mark (Danish, Finnish, Swedish, Norw.)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
69 // We may have to introduce a comma set with 0x201a
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
70 0x201A, // single low-9 quotation mark (German)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
71 0x201B, // single high-reversed-9 quotation mark (PropList.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
72 0x2032, // prime
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
73 0x300C, // left corner bracket (East Asian languages)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
74 0xFF07 // fullwidth apostrophe
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
75 };
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
76 for (int kSingleQuoteUnicode : kSingleQuoteUnicodes) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
77 if (kSingleQuoteUnicode == 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
78 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
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 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
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
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
84 static bool is_double_quote(const 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
85 static const char32 kDoubleQuoteUnicodes[] = {
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 0x201C, // left double quotation mark (English, others)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
88 0x201D, // right double quotation mark (Danish, Finnish, Swedish, Norw.)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
89 0x201F, // double high-reversed-9 quotation mark (PropList.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
90 0x2033, // double prime
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
91 0x201E, // double low-9 quotation mark (German)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
92 0x301D, // reversed double prime quotation mark (East Asian langs,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
93 // horiz.)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
94 0x301E, // close double prime (East Asian languages written horizontally)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
95 0xFF02 // fullwidth quotation mark
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
96 };
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
97 for (int kDoubleQuoteUnicode : kDoubleQuoteUnicodes) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
98 if (kDoubleQuoteUnicode == 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
99 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
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 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
102 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
103 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
104
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
105 // Helper runs a standard unicode normalization, optional OCR normalization,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
106 // and leaves the result as char32 for subsequent processing.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
107 static void NormalizeUTF8ToUTF32(UnicodeNormMode u_mode, OCRNorm ocr_normalize, const char *str8,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
108 std::vector<char32> *normed32) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
109 // Convert to ICU string for unicode normalization.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
110 icu::UnicodeString uch_str(str8, "UTF-8");
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
111 IcuErrorCode error_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
112 // Convert the enum to the new weird icu representation.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
113 const char *norm_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
114 u_mode == UnicodeNormMode::kNFKD || u_mode == UnicodeNormMode::kNFKC ? "nfkc" : "nfc";
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
115 UNormalization2Mode compose = u_mode == UnicodeNormMode::kNFC || u_mode == UnicodeNormMode::kNFKC
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
116 ? UNORM2_COMPOSE
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
117 : UNORM2_DECOMPOSE;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
118 // Pointer to singleton does not require deletion.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
119 const icu::Normalizer2 *normalizer =
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
120 icu::Normalizer2::getInstance(nullptr, norm_type, compose, error_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
121 error_code.assertSuccess();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
122 error_code.reset();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
123 icu::UnicodeString norm_str = normalizer->normalize(uch_str, error_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
124 error_code.assertSuccess();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
125 // Convert to char32 for output. OCR normalization if required.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
126 normed32->reserve(norm_str.length()); // An approximation.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
127 for (int offset = 0; offset < norm_str.length(); offset = norm_str.moveIndex32(offset, 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
128 char32 ch = norm_str.char32At(offset);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
129 // Skip all ZWS, RTL and LTR 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
130 if (Validator::IsZeroWidthMark(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
131 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
132 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
133 if (ocr_normalize == OCRNorm::kNormalize) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
134 ch = OCRNormalize(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
135 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
136 normed32->push_back(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
137 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
138 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
139
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
140 // Helper removes joiners from strings that contain no letters.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
141 static void StripJoiners(std::vector<char32> *str32) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
142 for (char32 ch : *str32) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
143 if (u_isalpha(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
144 return;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
145 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
146 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
147 int len = 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
148 for (char32 ch : *str32) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
149 if (ch != Validator::kZeroWidthJoiner && ch != Validator::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
150 (*str32)[len++] = 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
151 }
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 str32->resize(len);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
154 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
155
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
156 // Normalizes a UTF8 string according to the given modes. Returns true 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
157 // success. If false is returned, some failure or invalidity was present, 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
158 // the result string is produced on a "best effort" 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
159 bool NormalizeUTF8String(UnicodeNormMode u_mode, OCRNorm ocr_normalize,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
160 GraphemeNorm grapheme_normalize, const char *str8,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
161 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
162 std::vector<char32> normed32;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
163 NormalizeUTF8ToUTF32(u_mode, ocr_normalize, str8, &normed32);
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 (grapheme_normalize == GraphemeNorm::kNormalize) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
165 StripJoiners(&normed32);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
166 std::vector<std::vector<char32>> 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
167 bool success = Validator::ValidateCleanAndSegment(GraphemeNormMode::kSingleString, 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
168 normed32, &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
169 if (graphemes.empty() || graphemes[0].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
170 success = 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
171 } else if (normalized != 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
172 *normalized = UNICHAR::UTF32ToUTF8(graphemes[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
173 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
174 return 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
175 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
176 if (normalized != 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
177 *normalized = UNICHAR::UTF32ToUTF8(normed32);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
178 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
179 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
180 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
181
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
182 // Normalizes a UTF8 string according to the given modes and splits into
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
183 // graphemes according to g_mode. Returns true on success. If false is returned,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
184 // some failure or invalidity was present, and the result string is produced 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
185 // a "best effort" 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
186 bool NormalizeCleanAndSegmentUTF8(UnicodeNormMode u_mode, OCRNorm ocr_normalize,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
187 GraphemeNormMode g_mode, bool report_errors, const char *str8,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
188 std::vector<std::string> *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
189 std::vector<char32> normed32;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
190 NormalizeUTF8ToUTF32(u_mode, ocr_normalize, str8, &normed32);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
191 StripJoiners(&normed32);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
192 std::vector<std::vector<char32>> graphemes32;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
193 bool success = Validator::ValidateCleanAndSegment(g_mode, report_errors, normed32, &graphemes32);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
194 if (g_mode != GraphemeNormMode::kSingleString && 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
195 // If we modified the string to clean it up, the segmentation may not be
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
196 // correct, so check for changes and do it again.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
197 std::vector<char32> cleaned32;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
198 for (const auto &g : graphemes32) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
199 cleaned32.insert(cleaned32.end(), g.begin(), g.end());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
200 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
201 if (cleaned32 != normed32) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
202 graphemes32.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
203 success = Validator::ValidateCleanAndSegment(g_mode, report_errors, cleaned32, &graphemes32);
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
206 graphemes->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
207 graphemes->reserve(graphemes32.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
208 for (const auto &grapheme : graphemes32) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
209 graphemes->push_back(UNICHAR::UTF32ToUTF8(grapheme));
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
210 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
211 return 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
212 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
213
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
214 // Apply just the OCR-specific normalizations and return the normalized 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
215 char32 OCRNormalize(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
216 if (is_hyphen_punc(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
217 return '-';
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
218 } else if (is_single_quote(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
219 return '\'';
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
220 } else if (is_double_quote(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
221 return '"';
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
222 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
223 return 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
224 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
225
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
226 bool IsOCREquivalent(char32 ch1, char32 ch2) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
227 return OCRNormalize(ch1) == OCRNormalize(ch2);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
228 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
229
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
230 bool IsValidCodepoint(const 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
231 // In the range [0, 0xD800) or [0xE000, 0x10FFFF]
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
232 return (static_cast<uint32_t>(ch) < 0xD800) || (ch >= 0xE000 && ch <= 0x10FFFF);
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 bool IsWhitespace(const 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
236 ASSERT_HOST_MSG(IsValidCodepoint(ch), "Invalid Unicode codepoint: 0x%x\n", 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
237 return u_isUWhiteSpace(static_cast<UChar32>(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
238 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
239
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
240 bool IsUTF8Whitespace(const char *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
241 return SpanUTF8Whitespace(text) == strlen(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
242 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
243
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
244 unsigned int SpanUTF8Whitespace(const char *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
245 int n_white = 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
246 for (UNICHAR::const_iterator it = UNICHAR::begin(text, strlen(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
247 it != UNICHAR::end(text, strlen(text)); ++it) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
248 if (!IsWhitespace(*it)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
249 break;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
250 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
251 n_white += it.utf8_len();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
252 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
253 return n_white;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
254 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
255
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
256 unsigned int SpanUTF8NotWhitespace(const char *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
257 int n_notwhite = 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
258 for (UNICHAR::const_iterator it = UNICHAR::begin(text, strlen(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
259 it != UNICHAR::end(text, strlen(text)); ++it) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
260 if (IsWhitespace(*it)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
261 break;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
262 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
263 n_notwhite += it.utf8_len();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
264 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
265 return n_notwhite;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
266 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
267
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
268 bool IsInterchangeValid(const 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
269 return IsValidCodepoint(ch) && !(ch >= 0xFDD0 && ch <= 0xFDEF) && // Noncharacters.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
270 !(ch >= 0xFFFE && ch <= 0xFFFF) && !(ch >= 0x1FFFE && ch <= 0x1FFFF) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
271 !(ch >= 0x2FFFE && ch <= 0x2FFFF) && !(ch >= 0x3FFFE && ch <= 0x3FFFF) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
272 !(ch >= 0x4FFFE && ch <= 0x4FFFF) && !(ch >= 0x5FFFE && ch <= 0x5FFFF) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
273 !(ch >= 0x6FFFE && ch <= 0x6FFFF) && !(ch >= 0x7FFFE && ch <= 0x7FFFF) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
274 !(ch >= 0x8FFFE && ch <= 0x8FFFF) && !(ch >= 0x9FFFE && ch <= 0x9FFFF) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
275 !(ch >= 0xAFFFE && ch <= 0xAFFFF) && !(ch >= 0xBFFFE && ch <= 0xBFFFF) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
276 !(ch >= 0xCFFFE && ch <= 0xCFFFF) && !(ch >= 0xDFFFE && ch <= 0xDFFFF) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
277 !(ch >= 0xEFFFE && ch <= 0xEFFFF) && !(ch >= 0xFFFFE && ch <= 0xFFFFF) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
278 !(ch >= 0x10FFFE && ch <= 0x10FFFF) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
279 (!u_isISOControl(static_cast<UChar32>(ch)) || ch == '\n' || ch == '\f' || ch == '\t' ||
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
280 ch == '\r');
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
281 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
282
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
283 bool IsInterchangeValid7BitAscii(const 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
284 return IsValidCodepoint(ch) && ch <= 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
285 (!u_isISOControl(static_cast<UChar32>(ch)) || ch == '\n' || ch == '\f' || ch == '\t' ||
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
286 ch == '\r');
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
287 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
288
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
289 char32 FullwidthToHalfwidth(const 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
290 // Return unchanged if not in the fullwidth-halfwidth Unicode block.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
291 if (ch < 0xFF00 || ch > 0xFFEF || !IsValidCodepoint(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
292 if (ch != 0x3000) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
293 return 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
294 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
295 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
296 // Special case for fullwidth left and right "white parentheses".
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
297 if (ch == 0xFF5F) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
298 return 0x2985;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
299 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
300 if (ch == 0xFF60) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
301 return 0x2986;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
302 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
303 // Construct a full-to-half width transliterator.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
304 IcuErrorCode error_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
305 icu::UnicodeString uch_str(static_cast<UChar32>(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
306 const icu::Transliterator *fulltohalf =
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
307 icu::Transliterator::createInstance("Fullwidth-Halfwidth", UTRANS_FORWARD, error_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
308 error_code.assertSuccess();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
309 error_code.reset();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
310
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
311 fulltohalf->transliterate(uch_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
312 delete fulltohalf;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
313 ASSERT_HOST(uch_str.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
314 return uch_str[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
315 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
316
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
317 } // namespace tesseract