tesseract  5.0.0
ocrclass.h
Go to the documentation of this file.
1 /**********************************************************************
2  * File: ocrclass.h
3  * Description: Class definitions and constants for the OCR API.
4  * Author: Hewlett-Packard Co
5  *
6  * (C) Copyright 1996, Hewlett-Packard Co.
7  ** Licensed under the Apache License, Version 2.0 (the "License");
8  ** you may not use this file except in compliance with the License.
9  ** You may obtain a copy of the License at
10  ** http://www.apache.org/licenses/LICENSE-2.0
11  ** Unless required by applicable law or agreed to in writing, software
12  ** distributed under the License is distributed on an "AS IS" BASIS,
13  ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14  ** See the License for the specific language governing permissions and
15  ** limitations under the License.
16  *
17  **********************************************************************/
18 
19 /**********************************************************************
20  * This file contains typedefs for all the structures used by
21  * the HP OCR interface.
22  * The structures are designed to allow them to be used with any
23  * structure alignment up to 8.
24  **********************************************************************/
25 
26 #ifndef CCUTIL_OCRCLASS_H_
27 #define CCUTIL_OCRCLASS_H_
28 
29 #include <chrono>
30 #include <ctime>
31 
32 namespace tesseract {
33 
34 /**********************************************************************
35  * EANYCODE_CHAR
36  * Description of a single character. The character code is defined by
37  * the character set of the current font.
38  * Output text is sent as an array of these structures.
39  * Spaces and line endings in the output are represented in the
40  * structures of the surrounding characters. They are not directly
41  * represented as characters.
42  * The first character in a word has a positive value of blanks.
43  * Missing information should be set to the defaults in the comments.
44  * If word bounds are known, but not character bounds, then the top and
45  * bottom of each character should be those of the word. The left of the
46  * first and right of the last char in each word should be set. All other
47  * lefts and rights should be set to -1.
48  * If set, the values of right and bottom are left+width and top+height.
49  * Most of the members come directly from the parameters to ocr_append_char.
50  * The formatting member uses the enhancement parameter and combines the
51  * line direction stuff into the top 3 bits.
52  * The coding is 0=RL char, 1=LR char, 2=DR NL, 3=UL NL, 4=DR Para,
53  * 5=UL Para, 6=TB char, 7=BT char. API users do not need to know what
54  * the coding is, only that it is backwards compatible with the previous
55  * version.
56  **********************************************************************/
57 
58 struct EANYCODE_CHAR { /*single character */
59  // It should be noted that the format for char_code for version 2.0 and beyond
60  // is UTF8 which means that ASCII characters will come out as one structure
61  // but other characters will be returned in two or more instances of this
62  // structure with a single byte of the UTF8 code in each, but each will have
63  // the same bounding box. Programs which want to handle languagues with
64  // different characters sets will need to handle extended characters
65  // appropriately, but *all* code needs to be prepared to receive UTF8 coded
66  // characters for characters such as bullet and fancy quotes.
67  uint16_t char_code; /*character itself */
68  int16_t left; /*of char (-1) */
69  int16_t right; /*of char (-1) */
70  int16_t top; /*of char (-1) */
71  int16_t bottom; /*of char (-1) */
72  int16_t font_index; /*what font (0) */
73  uint8_t confidence; /*0=perfect, 100=reject (0/100) */
74  uint8_t point_size; /*of char, 72=i inch, (10) */
75  int8_t blanks; /*no of spaces before this char (1) */
76  uint8_t formatting; /*char formatting (0) */
77 };
78 
79 /**********************************************************************
80  * ETEXT_DESC
81  * Description of the output of the OCR engine.
82  * This structure is used as both a progress monitor and the final
83  * output header, since it needs to be a valid progress monitor while
84  * the OCR engine is storing its output to shared memory.
85  * During progress, all the buffer info is -1.
86  * Progress starts at 0 and increases to 100 during OCR. No other constraint.
87  * Additionally the progress callback contains the bounding box of the word that
88  * is currently being processed.
89  * Every progress callback, the OCR engine must set ocr_alive to 1.
90  * The HP side will set ocr_alive to 0. Repeated failure to reset
91  * to 1 indicates that the OCR engine is dead.
92  * If the cancel function is not null then it is called with the number of
93  * user words found. If it returns true then operation is cancelled.
94  **********************************************************************/
95 class ETEXT_DESC;
96 
97 using CANCEL_FUNC = bool (*)(void *, int);
98 using PROGRESS_FUNC = bool (*)(int, int, int, int, int);
99 using PROGRESS_FUNC2 = bool (*)(ETEXT_DESC *, int, int, int, int);
100 
101 class ETEXT_DESC { // output header
102 public:
103  int16_t count{0};
104  int16_t progress{0};
108  int8_t more_to_come{0};
109  volatile int8_t ocr_alive{0};
110  int8_t err_code{0};
111  CANCEL_FUNC cancel{nullptr};
113  nullptr};
115  void *cancel_this{nullptr};
116  std::chrono::steady_clock::time_point end_time;
120 
121  ETEXT_DESC() : progress_callback2(&default_progress_func) {
122  end_time = std::chrono::time_point<std::chrono::steady_clock,
123  std::chrono::milliseconds>();
124  }
125 
126  // Sets the end time to be deadline_msecs milliseconds from now.
127  void set_deadline_msecs(int32_t deadline_msecs) {
128  if (deadline_msecs > 0) {
129  end_time = std::chrono::steady_clock::now() +
130  std::chrono::milliseconds(deadline_msecs);
131  }
132  }
133 
134  // Returns false if we've not passed the end_time, or have not set a deadline.
135  bool deadline_exceeded() const {
136  if (end_time.time_since_epoch() ==
137  std::chrono::steady_clock::duration::zero()) {
138  return false;
139  }
140  auto now = std::chrono::steady_clock::now();
141  return (now > end_time);
142  }
143 
144 private:
145  static bool default_progress_func(ETEXT_DESC *ths, int left, int right,
146  int top, int bottom) {
147  if (ths->progress_callback != nullptr) {
148  return (*(ths->progress_callback))(ths->progress, left, right, top,
149  bottom);
150  }
151  return true;
152  }
153 };
154 
155 } // namespace tesseract
156 
157 #endif // CCUTIL_OCRCLASS_H_
bool(*)(void *, int) CANCEL_FUNC
Definition: ocrclass.h:97
bool(*)(int, int, int, int, int) PROGRESS_FUNC
Definition: ocrclass.h:98
bool(*)(ETEXT_DESC *, int, int, int, int) PROGRESS_FUNC2
Definition: ocrclass.h:99
volatile int8_t ocr_alive
true if not last
Definition: ocrclass.h:109
bool deadline_exceeded() const
Definition: ocrclass.h:135
void * cancel_this
monitor-aware progress callback
Definition: ocrclass.h:115
PROGRESS_FUNC progress_callback
returns true to cancel
Definition: ocrclass.h:112
EANYCODE_CHAR text[1]
Definition: ocrclass.h:119
PROGRESS_FUNC2 progress_callback2
called whenever progress increases
Definition: ocrclass.h:114
int16_t progress
chars in this buffer(0)
Definition: ocrclass.h:104
void set_deadline_msecs(int32_t deadline_msecs)
Definition: ocrclass.h:127
std::chrono::steady_clock::time_point end_time
this or other data for cancel
Definition: ocrclass.h:116
int8_t more_to_come
percent complete increasing (0-100)
Definition: ocrclass.h:108
ETEXT_DESC()
character data
Definition: ocrclass.h:121
CANCEL_FUNC cancel
for errcode use
Definition: ocrclass.h:111
int8_t err_code
ocr sets to 1, HP 0
Definition: ocrclass.h:110