Mercurial > hgrepos > Python2 > PyMuPDF
annotate mupdf-source/thirdparty/tesseract/src/ccutil/unicharcompress.h @ 17:dd9cdb856310
Remove PKG-INFO from the because it is regenerated automatically for the sdist
| author | Franz Glasner <fzglas.hg@dom66.de> |
|---|---|
| date | Thu, 18 Sep 2025 17:40:40 +0200 |
| parents | b50eed0cc0ef |
| children |
| rev | line source |
|---|---|
|
2
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
1 /////////////////////////////////////////////////////////////////////// |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
2 // File: 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
|
3 // Description: Unicode re-encoding using a sequence of smaller numbers 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
|
4 // place of a single large code for CJK, similarly for Indic, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
5 // and dissection of ligatures for other scripts. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
6 // Author: Ray Smith |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
7 // |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
8 // (C) Copyright 2015, 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
|
9 // 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
|
10 // 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
|
11 // 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
|
12 // 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
|
13 // 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
|
14 // 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
|
15 // 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
|
16 // 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
|
17 // 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
|
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 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
21 #ifndef TESSERACT_CCUTIL_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
|
22 #define TESSERACT_CCUTIL_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 <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
|
25 #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
|
26 #include "serialis.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 "unicharset.h" |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
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 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
|
30 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
31 // Trivial class to hold the code for a recoded unichar-id. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
32 class RecodedCharID { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
33 public: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
34 // The maximum length of a 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
|
35 static const int kMaxCodeLen = 9; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 RecodedCharID() : self_normalized_(1), 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
|
38 memset(code_, 0, sizeof(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
|
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 void Truncate(int 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
|
41 length_ = 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
|
42 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
43 // Sets the code value at the given index in the code. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
44 void Set(int index, int value) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
45 code_[index] = value; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
46 if (length_ <= index) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
47 length_ = index + 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
|
48 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
49 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
50 // Shorthand for setting codes of length 3, as all Hangul and Han codes are |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
51 // length 3. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
52 void Set3(int code0, int code1, int code2) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
53 length_ = 3; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
54 code_[0] = code0; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
55 code_[1] = code1; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
56 code_[2] = code2; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
57 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
58 bool empty() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
59 return 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
|
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 // Accessors |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
62 int length() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
63 return 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
|
64 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
65 int operator()(int index) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
66 return code_[index]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
67 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
68 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
69 // Writes to the given file. Returns false in case of 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
|
70 bool Serialize(TFile *fp) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
71 return fp->Serialize(&self_normalized_) && fp->Serialize(&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
|
72 fp->Serialize(&code_[0], 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
|
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 // Reads from the given file. Returns false in case of 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
|
75 bool DeSerialize(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
|
76 return fp->DeSerialize(&self_normalized_) && fp->DeSerialize(&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
|
77 fp->DeSerialize(&code_[0], 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
|
78 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
79 bool operator==(const RecodedCharID &other) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
80 if (length_ != other.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
|
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 for (int i = 0; i < length_; ++i) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
84 if (code_[i] != other.code_[i]) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
85 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
|
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 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
88 return true; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
89 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
90 // Hash functor for RecodedCharID. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
91 struct RecodedCharIDHash { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
92 uint64_t operator()(const RecodedCharID &code) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
93 uint64_t result = 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
|
94 for (int i = 0; i < code.length_; ++i) { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
95 result ^= static_cast<uint64_t>(code(i)) << (7 * i); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
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 return 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
|
98 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
99 }; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
100 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
101 private: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
102 // True if this code is self-normalizing, ie is the master entry for indices |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
103 // that map to the same code. Has boolean value, but int8_t for serialization. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
104 int8_t self_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
|
105 // The number of elements in use in 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
|
106 int32_t 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
|
107 // The re-encoded form of the unichar-id to which this RecodedCharID relates. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
108 int32_t code_[kMaxCodeLen]; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
109 }; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
110 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
111 // Class holds a "compression" of a unicharset to simplify the learning problem |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
112 // for a neural-network-based classifier. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
113 // Objectives: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
114 // 1 (CJK): Ids of a unicharset with a large number of classes are expressed as |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
115 // a sequence of 3 codes with much fewer values. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
116 // This is achieved using the Jamo coding for Hangul and the Unicode |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
117 // Radical-Stroke-index for Han. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
118 // 2 (Indic): Instead of thousands of codes with one for each grapheme, re-code |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
119 // as the unicode sequence (but coded in a more compact space). |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
120 // 3 (the rest): Eliminate multi-path problems with ligatures and fold confusing |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
121 // and not significantly distinct shapes (quotes) together, ie |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
122 // represent the fi ligature as the f-i pair, and fold u+2019 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
|
123 // friends all onto ascii single ' |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
124 // 4 The null character and mapping to target activations: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
125 // To save horizontal coding space, the compressed codes are generally mapped |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
126 // to target network activations without intervening null characters, BUT |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
127 // in the case of ligatures, such as ff, null characters have to be included |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
128 // so existence of repeated codes is detected at codebook-building time, 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
|
129 // null characters are embedded directly into the codes, so the rest 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
|
130 // system doesn't need to worry about the problem (much). There is still an |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
131 // effect on the range of ways in which the target activations can 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
|
132 // generated. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 // The computed code values are compact (no unused values), and, for CJK, |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
135 // unique (each code position uses a disjoint set of values from each other 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
|
136 // position). For non-CJK, the same code value CAN be used in multiple |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
137 // positions, eg the ff ligature is converted to <f> <nullchar> <f>, where <f> |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
138 // is the same code as is used for the single f. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
139 class TESS_API UnicharCompress { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
140 public: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
141 UnicharCompress(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
142 UnicharCompress(const UnicharCompress &src); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
143 ~UnicharCompress(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
144 UnicharCompress &operator=(const UnicharCompress &src); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
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 // The 1st Hangul unicode. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
147 static const int kFirstHangul = 0xac00; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
148 // The number of Hangul unicodes. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
149 static const int kNumHangul = 11172; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
150 // The number of Jamos for each of the 3 parts of a Hangul character, being |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
151 // the Leading consonant, Vowel and Trailing consonant. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
152 static const int kLCount = 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
|
153 static const int kVCount = 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
|
154 static const int kTCount = 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
|
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 // Computes the encoding for the given unicharset. It is a requirement that |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
157 // the file training/langdata/radical-stroke.txt have been read into 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
|
158 // input string radical_stroke_table. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
159 // Returns false if the encoding cannot be constructed. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
160 bool ComputeEncoding(const UNICHARSET &unicharset, int null_id, std::string *radical_stroke_table); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
161 // Sets up an encoder that doesn't change the unichars at all, so it just |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
162 // passes them through unchanged. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
163 void SetupPassThrough(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
|
164 // Sets up an encoder directly using the given encoding vector, which maps |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
165 // unichar_ids to the given codes. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
166 void SetupDirect(const std::vector<RecodedCharID> &codes); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
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 // Returns the number of different values that can be used in a code, ie |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
169 // 1 + the maximum value that will ever be used by an RecodedCharID code 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
|
170 // any position in its array. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
171 int code_range() const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
172 return 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
|
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 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
175 // Encodes a single unichar_id. Returns the length of the code, (or zero if |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
176 // invalid input), and the encoding itself in 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
|
177 int EncodeUnichar(unsigned unichar_id, RecodedCharID *code) const; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
178 // Decodes code, returning the original unichar-id, or |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
179 // INVALID_UNICHAR_ID if the input is invalid. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
180 int DecodeUnichar(const RecodedCharID &code) const; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
181 // Returns true if the given code is a valid start or single 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
|
182 bool IsValidFirstCode(int code) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
183 return is_valid_start_[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
|
184 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
185 // Returns a list of valid non-final next codes for a given prefix 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
|
186 // which may be 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
|
187 const std::vector<int> *GetNextCodes(const RecodedCharID &code) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
188 auto it = next_codes_.find(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
|
189 return it == next_codes_.end() ? nullptr : it->second; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
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 // Returns a list of valid final codes for a given prefix code, which may |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
192 // be 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
|
193 const std::vector<int> *GetFinalCodes(const RecodedCharID &code) const { |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
194 auto it = final_codes_.find(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
|
195 return it == final_codes_.end() ? nullptr : it->second; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
196 } |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
197 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
198 // Writes to the given file. Returns false in case of 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
|
199 bool Serialize(TFile *fp) const; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
200 // Reads from the given file. Returns false in case of 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
|
201 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
202 bool DeSerialize(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
|
203 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
204 // Returns a string containing a text file that describes the encoding thus: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
205 // <index>[,<index>]*<tab><UTF8-str><newline> |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
206 // In words, a comma-separated list of one or more indices, followed by a tab |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
207 // and the UTF-8 string that the code represents per line. Most simple scripts |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
208 // will encode a single index to a UTF8-string, but Chinese, Japanese, Korean |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
209 // and the Indic scripts will contain a many-to-many mapping. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
210 // See the class comment above for details. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
211 std::string GetEncodingAsString(const UNICHARSET &unicharset) const; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
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 // Helper decomposes a Hangul unicode to 3 parts, leading, vowel, trailing. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
214 // Note that the returned values are 0-based indices, NOT unicode Jamo. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
215 // Returns false if the input is not in the Hangul unicode range. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
216 static bool DecomposeHangul(int unicode, int *leading, int *vowel, int *trailing); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 private: |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
219 // Renumbers codes to eliminate unused values. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
220 void DefragmentCodeValues(int encoded_null); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
221 // Computes the value of code_range_ from the encoder_. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
222 void ComputeCodeRange(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
223 // Initializes the decoding hash_map from the encoder_ array. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
224 void SetupDecoder(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
225 // Frees allocated memory. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
226 void Cleanup(); |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: 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 // The encoder that maps a unichar-id to a sequence of small codes. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
229 // encoder_ is the only part that is serialized. The rest is computed on load. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
230 std::vector<RecodedCharID> encoder_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
231 // Decoder converts the output of encoder back to a unichar-id. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
232 std::unordered_map<RecodedCharID, int, RecodedCharID::RecodedCharIDHash> decoder_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
233 // True if the index is a valid single or start 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
|
234 std::vector<bool> is_valid_start_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
235 // Maps a prefix code to a list of valid next codes. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
236 // The map owns the vectors. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
237 std::unordered_map<RecodedCharID, std::vector<int> *, RecodedCharID::RecodedCharIDHash> |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
238 next_codes_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
239 // Maps a prefix code to a list of valid final codes. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
240 // The map owns the vectors. |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
241 std::unordered_map<RecodedCharID, std::vector<int> *, RecodedCharID::RecodedCharIDHash> |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
242 final_codes_; |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
243 // Max of any value in encoder_ + 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
|
244 int 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
|
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 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
247 } // 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
|
248 |
|
b50eed0cc0ef
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff
changeset
|
249 #endif // TESSERACT_CCUTIL_UNICHARCOMPRESS_H_ |
