mirror/tesseract
1
0
Fork 0
mirror of https://github.com/tesseract-ocr/tesseract.git synced 2024-12-04 01:39:16 +08:00
Code Issues Actions 12 Packages Projects Releases Wiki Activity

ccmain: Use C++-11 code instead of TessCallback1

Browse Source 
Signed-off-by: Stefan Weil <sw@weilnetz.de>
This commit is contained in:
Stefan Weil 2019-07-01 21:59:48 +02:00
parent eeec9c66d4
commit ded24d0367
7 changed files with 65 additions and 88 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
src/ccmain/control.cpp
View File

@ -91,11 +91,11 @@ bool Tesseract::recog_interactive(PAGE_RES_IT* pr_it) {
#ifndef DISABLED_LEGACY_ENGINE
if (tessedit_debug_quality_metrics) {
WERD_RES* word_res = pr_it->word();
word_char_quality(word_res, pr_it->row()->row, &char_qual, &good_char_qual);
word_char_quality(word_res, &char_qual, &good_char_qual);
tprintf("\n%d chars; word_blob_quality: %d; outline_errs: %d; "
"char_quality: %d; good_char_quality: %d\n",
word_res->reject_map.length(),
word_blob_quality(word_res, pr_it->row()->row),
word_blob_quality(word_res),
word_outline_errs(word_res), char_qual, good_char_qual);
}
#endif // ndef DISABLED_LEGACY_ENGINE
@ -647,14 +647,13 @@ void Tesseract::rejection_passes(PAGE_RES* page_res,
const int chars_in_word = word->reject_map.length();
const int rejects_in_word = word->reject_map.reject_count();
const int blob_quality = word_blob_quality(word, page_res_it.row()->row);
const int blob_quality = word_blob_quality(word);
stats_.doc_blob_quality += blob_quality;
const int outline_errs = word_outline_errs(word);
stats_.doc_outline_errs += outline_errs;
int16_t all_char_quality;
int16_t accepted_all_char_quality;
word_char_quality(word, page_res_it.row()->row,
&all_char_quality, &accepted_all_char_quality);
word_char_quality(word, &all_char_quality, &accepted_all_char_quality);
stats_.doc_char_quality += all_char_quality;
const uint8_t permuter_type = word->best_choice->permuter();
if ((permuter_type == SYSTEM_DAWG_PERM) ||

123
src/ccmain/docqual.cpp
View File

@ -1,8 +1,7 @@
/******************************************************************
* File: docqual.cpp (Formerly docqual.c)
* Description: Document Quality Metrics
* Author: Phil Cheatle
* Created: Mon May 9 11:27:28 BST 1994
* Author: Phil Cheatle
*
* (C) Copyright 1994, Hewlett-Packard Ltd.
** Licensed under the Apache License, Version 2.0 (the "License");
@ -17,39 +16,31 @@
*
**********************************************************************/
#include <cctype>
#include "docqual.h"
#include "reject.h"
#include "tesscallback.h"
#include "tessvars.h"
#include "tesseractclass.h"
#include <cctype>
#include "docqual.h"
#include "reject.h"
#include "tessvars.h"
#include "tesseractclass.h"
namespace tesseract{
// A little class to provide the callbacks as we have no pre-bound args.
struct DocQualCallbacks {
explicit DocQualCallbacks(WERD_RES* word0)
: word(word0), match_count(0), accepted_match_count(0) {}
static void countMatchingBlobs(int16_t& match_count, int /*index*/) {
++match_count;
}
void CountMatchingBlobs(int index) {
++match_count;
static void countAcceptedBlobs(WERD_RES* word, int16_t& match_count,
int16_t& accepted_match_count, int index) {
if (word->reject_map[index].accepted()) {
++accepted_match_count;
}
++match_count;
}
void CountAcceptedBlobs(int index) {
if (word->reject_map[index].accepted())
++accepted_match_count;
++match_count;
static void acceptIfGoodQuality(WERD_RES* word, int index) {
if (word->reject_map[index].accept_if_good_quality()) {
word->reject_map[index].setrej_quality_accept();
}
void AcceptIfGoodQuality(int index) {
if (word->reject_map[index].accept_if_good_quality())
word->reject_map[index].setrej_quality_accept();
}
WERD_RES* word;
int16_t match_count;
int16_t accepted_match_count;
};
}
/*************************************************************************
* word_blob_quality()
@ -57,16 +48,16 @@ struct DocQualCallbacks {
* ASSUME blobs in both initial word and box_word are in ascending order of
* left hand blob edge.
*************************************************************************/
int16_t Tesseract::word_blob_quality(WERD_RES *word, ROW *row) {
if (word->bln_boxes == nullptr ||
word->rebuild_word == nullptr || word->rebuild_word->blobs.empty())
return 0;
DocQualCallbacks cb(word);
word->bln_boxes->ProcessMatchedBlobs(
*word->rebuild_word,
NewPermanentTessCallback(&cb, &DocQualCallbacks::CountMatchingBlobs));
return cb.match_count;
int16_t Tesseract::word_blob_quality(WERD_RES* word) {
int16_t match_count = 0;
if (word->bln_boxes != nullptr && word->rebuild_word != nullptr &&
!word->rebuild_word->blobs.empty()) {
using namespace std::placeholders; // for _1
word->bln_boxes->ProcessMatchedBlobs(
*word->rebuild_word,
std::bind(countMatchingBlobs, match_count, _1));
}
return match_count;
}
int16_t Tesseract::word_outline_errs(WERD_RES *word) {
@ -89,38 +80,31 @@ int16_t Tesseract::word_outline_errs(WERD_RES *word) {
* Combination of blob quality and outline quality - how many good chars are
* there? - I.e chars which pass the blob AND outline tests.
*************************************************************************/
void Tesseract::word_char_quality(WERD_RES *word,
ROW *row,
int16_t *match_count,
int16_t *accepted_match_count) {
if (word->bln_boxes == nullptr || word->rebuild_word == nullptr ||
word->rebuild_word->blobs.empty()) {
*match_count = 0;
*accepted_match_count = 0;
return;
void Tesseract::word_char_quality(WERD_RES* word, int16_t* match_count,
int16_t* accepted_match_count) {
*match_count = 0;
*accepted_match_count = 0;
if (word->bln_boxes != nullptr && word->rebuild_word != nullptr &&
!word->rebuild_word->blobs.empty()) {
using namespace std::placeholders; // for _1
word->bln_boxes->ProcessMatchedBlobs(
*word->rebuild_word,
std::bind(countAcceptedBlobs,
word, *match_count, *accepted_match_count, _1));
}
DocQualCallbacks cb(word);
word->bln_boxes->ProcessMatchedBlobs(
*word->rebuild_word,
NewPermanentTessCallback(&cb, &DocQualCallbacks::CountAcceptedBlobs));
*match_count = cb.match_count;
*accepted_match_count = cb.accepted_match_count;
}
/*************************************************************************
* unrej_good_chs()
* Unreject POTENTIAL rejects if the blob passes the blob and outline checks
*************************************************************************/
void Tesseract::unrej_good_chs(WERD_RES *word, ROW *row) {
if (word->bln_boxes == nullptr ||
word->rebuild_word == nullptr || word->rebuild_word->blobs.empty())
return;
DocQualCallbacks cb(word);
word->bln_boxes->ProcessMatchedBlobs(
*word->rebuild_word,
NewPermanentTessCallback(&cb, &DocQualCallbacks::AcceptIfGoodQuality));
void Tesseract::unrej_good_chs(WERD_RES* word) {
if (word->bln_boxes != nullptr && word->rebuild_word != nullptr &&
word->rebuild_word->blobs.empty()) {
using namespace std::placeholders; // for _1
word->bln_boxes->ProcessMatchedBlobs(
*word->rebuild_word, std::bind(acceptIfGoodQuality, word, _1));
}
}
int16_t Tesseract::count_outline_errs(char c, int16_t outline_count) {
@ -186,12 +170,12 @@ void Tesseract::unrej_good_quality_words( //unreject potential
word->best_choice->unichar_string().string(),
word->best_choice->unichar_lengths().string())
!= AC_UNACCEPTABLE)) {
unrej_good_chs(word, page_res_it.row ()->row);
unrej_good_chs(word);
}
page_res_it.forward ();
}
else {
/* Skip to end of dodgy row */
// Skip to end of dodgy row.
current_row = page_res_it.row ();
while ((page_res_it.word () != nullptr) &&
(page_res_it.row () == current_row))
@ -285,10 +269,8 @@ void Tesseract::doc_and_block_rejection( //reject big chunks
word->best_choice->unichar_string().string(),
word->best_choice->unichar_lengths().string()) !=
AC_UNACCEPTABLE) {
word_char_quality(word, page_res_it.row()->row,
&char_quality,
&accepted_char_quality);
rej_word = char_quality != word->reject_map.length();
word_char_quality(word, &char_quality, &accepted_char_quality);
rej_word = char_quality != word->reject_map.length();
}
} else {
rej_word = true;
@ -356,8 +338,7 @@ void Tesseract::doc_and_block_rejection( //reject big chunks
word->best_choice->unichar_string().string(),
word->best_choice->unichar_lengths().string()) !=
AC_UNACCEPTABLE) {
word_char_quality(word, page_res_it.row()->row,
&char_quality,
word_char_quality(word, &char_quality,
&accepted_char_quality);
rej_word = char_quality != word->reject_map.length();
}

3
src/ccmain/docqual.h
View File

@ -2,7 +2,6 @@
* File: docqual.h (Formerly docqual.h)
* Description: Document Quality Metrics
* Author: Phil Cheatle
* Created: Mon May 9 11:27:28 BST 1994
*
* (C) Copyright 1994, Hewlett-Packard Ltd.
** Licensed under the Apache License, Version 2.0 (the "License");
@ -34,7 +33,7 @@ enum GARBAGE_LEVEL
G_TERRIBLE
};
int16_t word_blob_quality(WERD_RES *word, ROW *row);
int16_t word_blob_quality(WERD_RES* word);
void reject_whole_page(PAGE_RES_IT &page_res_it);
#endif

2
src/ccmain/reject.cpp
View File

@ -596,7 +596,7 @@ bool Tesseract::repeated_nonalphanum_wd(WERD_RES* word, ROW* row) {
if (word->best_choice->unichar_id(i) != uch_id) return false;
}
word_char_quality(word, row, &char_quality, &accepted_char_quality);
word_char_quality(word, &char_quality, &accepted_char_quality);
if ((word->best_choice->unichar_lengths().length () == char_quality) &&
(char_quality == accepted_char_quality))

6
src/ccmain/tesseractclass.h
View File

@ -648,10 +648,10 @@ class Tesseract : public Wordrec {
void quality_based_rejection(PAGE_RES_IT& page_res_it, bool good_quality_doc);
void convert_bad_unlv_chs(WERD_RES* word_res);
void tilde_delete(PAGE_RES_IT& page_res_it);
int16_t word_blob_quality(WERD_RES* word, ROW* row);
void word_char_quality(WERD_RES* word, ROW* row, int16_t* match_count,
int16_t word_blob_quality(WERD_RES* word);
void word_char_quality(WERD_RES* word, int16_t* match_count,
int16_t* accepted_match_count);
void unrej_good_chs(WERD_RES* word, ROW* row);
void unrej_good_chs(WERD_RES* word);
int16_t count_outline_errs(char c, int16_t outline_count);
int16_t word_outline_errs(WERD_RES* word);
#ifndef DISABLED_LEGACY_ENGINE

5
src/ccstruct/boxword.cpp
View File

@ -188,13 +188,12 @@ void BoxWord::ComputeBoundingBox() {
// for each blob index where the bounding boxes match.
// The callback is deleted on completion.
void BoxWord::ProcessMatchedBlobs(const TWERD& other,
TessCallback1<int>* cb) const {
std::function<void(int)> cb) const {
for (int i = 0; i < length_ && i < other.NumBlobs(); ++i) {
TBOX blob_box = other.blobs[i]->bounding_box();
if (blob_box == boxes_[i])
cb->Run(i);
cb(i);
}
delete cb;
}
} // namespace tesseract.

5
src/ccstruct/boxword.h
View File

@ -20,6 +20,7 @@
#ifndef TESSERACT_CSTRUCT_BOXWORD_H_
#define TESSERACT_CSTRUCT_BOXWORD_H_
#include <functional> // for std::function
#include "genericvector.h" // for GenericVector
#include "rect.h" // for TBOX
@ -28,8 +29,6 @@ class WERD;
struct TWERD;
template <class A1> class TessCallback1;
namespace tesseract {
// Class to hold an array of bounding boxes for an output word and
@ -75,7 +74,7 @@ class BoxWord {
// This and other putatively are the same, so call the (permanent) callback
// for each blob index where the bounding boxes match.
// The callback is deleted on completion.
void ProcessMatchedBlobs(const TWERD& other, TessCallback1<int>* cb) const;
void ProcessMatchedBlobs(const TWERD& other, std::function<void(int)> cb) const;
const TBOX& bounding_box() const {
return bbox_;