Python2/PyMuPDF: mupdf-source/thirdparty/tesseract/src/ccmain/docqual.cpp annotate

annotate mupdf-source/thirdparty/tesseract/src/ccmain/docqual.cpp @ 46:7ee69f120f19 default tip

>>>>> tag v1.26.5+1 for changeset b74429b0f5c4

author	Franz Glasner <fzglas.hg@dom66.de>
date	Sat, 11 Oct 2025 17:17:30 +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: docqual.cpp (Formerly docqual.c)
b50eed0cc0ef ADD: MuPDF v1.26.7: 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: Document Quality Metrics
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	4 * Author: Phil Cheatle
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	5 *
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	6 * (C) Copyright 1994, Hewlett-Packard Ltd.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	7 ** 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	8 ** 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	9 ** 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	10 ** 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	11 ** 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	12 ** 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	13 ** 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	14 ** 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	15 ** 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	16 *
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 #include "docqual.h"
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	20 #include <cctype>
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	21 #include "reject.h"
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	22 #include "tesseractclass.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 #include "tessvars.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	24
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	25 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	26
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	27 static void countMatchingBlobs(int16_t &match_count, int /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	28 ++match_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	29 }
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 static void countAcceptedBlobs(WERD_RES *word, int16_t &match_count, int16_t &accepted_match_count,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	32 int 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	33 if (word->reject_map[index].accepted()) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	34 ++accepted_match_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	35 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	36 ++match_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	37 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	38
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	39 static void acceptIfGoodQuality(WERD_RES *word, int 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	40 if (word->reject_map[index].accept_if_good_quality()) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	41 word->reject_map[index].setrej_quality_accept();
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	44
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	45 /*************************************************************************
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	46 * word_blob_quality()
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 many blobs in the box_word are identical to those of the inword?
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	48 * ASSUME blobs in both initial word and box_word are in ascending order of
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	49 * left hand blob edge.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	50 *************************************************************************/
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	51 int16_t Tesseract::word_blob_quality(WERD_RES *word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	52 int16_t match_count = 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	53 if (word->bln_boxes != nullptr && word->rebuild_word != 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	54 !word->rebuild_word->blobs.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	55 using namespace std::placeholders; // for _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	56 word->bln_boxes->ProcessMatchedBlobs(*word->rebuild_word,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	57 std::bind(countMatchingBlobs, match_count, _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	58 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	59 return match_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: 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
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	62 int16_t Tesseract::word_outline_errs(WERD_RES *word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	63 int16_t err_count = 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	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 if (word->rebuild_word != 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	66 int16_t i = 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	67 for (unsigned b = 0; b < word->rebuild_word->NumBlobs(); ++b) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	68 TBLOB *blob = word->rebuild_word->blobs[b];
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	69 err_count += count_outline_errs(word->best_choice->unichar_string()[i], blob->NumOutlines());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	70 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	71 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	72 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	73 return err_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	74 }
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 /*************************************************************************
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	77 * word_char_quality()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	78 * Combination of blob quality and outline quality - how many good chars 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	79 * there? - I.e chars which pass the blob AND outline tests.
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 void Tesseract::word_char_quality(WERD_RES word, int16_t match_count,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	82 int16_t *accepted_match_count) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	83 *match_count = 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	84 *accepted_match_count = 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	85 if (word->bln_boxes != nullptr && word->rebuild_word != 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	86 !word->rebuild_word->blobs.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	87 using namespace std::placeholders; // for _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	88 word->bln_boxes->ProcessMatchedBlobs(
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	89 *word->rebuild_word,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	90 std::bind(countAcceptedBlobs, word, match_count, accepted_match_count, _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	91 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	92 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	93
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	94 /*************************************************************************
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	95 * unrej_good_chs()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	96 * Unreject POTENTIAL rejects if the blob passes the blob and outline checks
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	97 *************************************************************************/
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	98 void Tesseract::unrej_good_chs(WERD_RES *word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	99 if (word->bln_boxes != nullptr && word->rebuild_word != 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	100 word->rebuild_word->blobs.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	101 using namespace std::placeholders; // for _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	102 word->bln_boxes->ProcessMatchedBlobs(*word->rebuild_word,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	103 std::bind(acceptIfGoodQuality, word, _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	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 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	106
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	107 int16_t Tesseract::count_outline_errs(char c, int16_t outline_count) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	108 int expected_outline_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 if (outlines_odd.contains(c)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	111 return 0; // Don't use this 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	112 } else if (outlines_2.contains(c)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	113 expected_outline_count = 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	114 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	115 expected_outline_count = 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	116 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	117 return abs(outline_count - expected_outline_count);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	118 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	119
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	120 void Tesseract::quality_based_rejection(PAGE_RES_IT &page_res_it, bool good_quality_doc) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	121 if ((tessedit_good_quality_unrej && good_quality_doc)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	122 unrej_good_quality_words(page_res_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	123 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	124 doc_and_block_rejection(page_res_it, good_quality_doc);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	125 if (unlv_tilde_crunching) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	126 tilde_crunch(page_res_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	127 tilde_delete(page_res_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	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	129 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	130
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	131 /*************************************************************************
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	132 * unrej_good_quality_words()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	133 * Accept potential rejects in words which pass the following checks:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	134 * - Contains a potential reject
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	135 * - Word looks like a sensible alpha word.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	136 * - Word segmentation is the same as the original image
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	137 * - All characters have the expected number of outlines
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	138 * NOTE - the rejection counts are recalculated after unrejection
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	139 * - CAN'T do it in a single pass without a bit of fiddling
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	140 * - keep it simple but inefficient
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	141 *************************************************************************/
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	142 void Tesseract::unrej_good_quality_words( // unreject potential
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	143 PAGE_RES_IT &page_res_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	144 WERD_RES *word;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	145 ROW_RES *current_row;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	146 BLOCK_RES *current_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	147 int 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	148
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	149 page_res_it.restart_page();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	150 while (page_res_it.word() != 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	151 check_debug_pt(page_res_it.word(), 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	152 if (bland_unrej) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	153 word = page_res_it.word();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	154 for (i = 0; i < word->reject_map.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	155 if (word->reject_map[i].accept_if_good_quality()) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	156 word->reject_map[i].setrej_quality_accept();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	157 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	158 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	159 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	160 } else if ((page_res_it.row()->char_count > 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	161 ((page_res_it.row()->rej_count /
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	162 static_cast<float>(page_res_it.row()->char_count)) <= quality_rowrej_pc)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	163 word = page_res_it.word();
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 (word->reject_map.quality_recoverable_rejects() &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	165 (tessedit_unrej_any_wd \|\|
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	166 acceptable_word_string(*word->uch_set, word->best_choice->unichar_string().c_str(),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	167 word->best_choice->unichar_lengths().c_str()) !=
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	168 AC_UNACCEPTABLE)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	169 unrej_good_chs(word);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	170 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	171 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	172 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	173 // Skip to end of dodgy row.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	174 current_row = page_res_it.row();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	175 while ((page_res_it.word() != nullptr) && (page_res_it.row() == current_row)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	176 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	177 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	178 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	179 check_debug_pt(page_res_it.word(), 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	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 page_res_it.restart_page();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	182 page_res_it.page_res->char_count = 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	183 page_res_it.page_res->rej_count = 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	184 current_block = 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	185 current_row = 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	186 while (page_res_it.word() != 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	187 if (current_block != page_res_it.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	188 current_block = page_res_it.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	189 current_block->char_count = 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	190 current_block->rej_count = 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	191 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	192 if (current_row != page_res_it.row()) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	193 current_row = page_res_it.row();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	194 current_row->char_count = 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	195 current_row->rej_count = 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	196 current_row->whole_word_rej_count = 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	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 page_res_it.rej_stat_word();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	199 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	202
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 * doc_and_block_rejection()
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 * If the page has too many rejects - reject all of 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	207 * If any block has too many rejects - reject all words in the 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	208 *************************************************************************/
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	209
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	210 void Tesseract::doc_and_block_rejection( // reject big chunks
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	211 PAGE_RES_IT &page_res_it, bool good_quality_doc) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	212 BLOCK_RES *current_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	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 int16_t char_quality = 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	215 int16_t accepted_char_quality;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	216
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	217 if (page_res_it.page_res->rej_count * 100.0 / page_res_it.page_res->char_count >
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	218 tessedit_reject_doc_percent) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	219 reject_whole_page(page_res_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	220 if (tessedit_debug_doc_rejection) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	221 tprintf("REJECT ALL #chars: %d #Rejects: %d; \n", page_res_it.page_res->char_count,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	222 page_res_it.page_res->rej_count);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	223 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	224 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	225 if (tessedit_debug_doc_rejection) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	226 tprintf("NO PAGE REJECTION #chars: %d # Rejects: %d; \n", page_res_it.page_res->char_count,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	227 page_res_it.page_res->rej_count);
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 /* Walk blocks testing for block rejection */
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	231
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	232 page_res_it.restart_page();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	233 WERD_RES *word;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	234 while ((word = page_res_it.word()) != 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	235 current_block = page_res_it.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	236 int16_t block_no = current_block->block->pdblk.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	237 if (current_block->char_count > 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	238 (current_block->rej_count * 100.0 / current_block->char_count) >
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	239 tessedit_reject_block_percent) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	240 if (tessedit_debug_block_rejection) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	241 tprintf("REJECTING BLOCK %d #chars: %d; #Rejects: %d\n", block_no,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	242 current_block->char_count, current_block->rej_count);
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 bool prev_word_rejected = 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	245 while ((word = page_res_it.word()) != nullptr && (page_res_it.block() == current_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	246 bool rej_word;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	247 if (tessedit_preserve_blk_rej_perfect_wds) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	248 rej_word = word->reject_map.reject_count() > 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	249 word->reject_map.length() < tessedit_preserve_min_wd_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	250 if (rej_word && tessedit_dont_blkrej_good_wds &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	251 word->reject_map.length() >= tessedit_preserve_min_wd_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 acceptable_word_string(*word->uch_set, word->best_choice->unichar_string().c_str(),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	253 word->best_choice->unichar_lengths().c_str()) !=
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	254 AC_UNACCEPTABLE) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	255 word_char_quality(word, &char_quality, &accepted_char_quality);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	256 rej_word = char_quality != word->reject_map.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	257 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	258 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	259 rej_word = 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	260 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	261 if (rej_word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 Reject spacing if both current and prev words are rejected.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	264 NOTE - this is NOT restricted to FUZZY spaces. - When tried this
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	265 generated more space errors.
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 if (tessedit_use_reject_spaces && prev_word_rejected &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	268 page_res_it.prev_row() == page_res_it.row() && word->word->space() == 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	269 word->reject_spaces = 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	270 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	271 word->reject_map.rej_word_block_rej();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	272 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	273 prev_word_rejected = rej_word;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	274 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	275 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	276 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	277 if (tessedit_debug_block_rejection) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	278 tprintf("NOT REJECTING BLOCK %d #chars: %d # Rejects: %d; \n", block_no,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	279 page_res_it.block()->char_count, page_res_it.block()->rej_count);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	280 }
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 /* Walk rows in block testing for row rejection */
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	283 int16_t row_no = 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	284 while (page_res_it.word() != nullptr && page_res_it.block() == current_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	285 ROW_RES *current_row = page_res_it.row();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	286 row_no++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	287 /* Reject whole row 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	288 fraction of chars on row which are rejected exceed a limit 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	289 fraction rejects which occur in WHOLE WERD rejects is LESS THAN a
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	290 limit
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	291 */
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 (current_row->char_count > 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	293 (current_row->rej_count * 100.0 / current_row->char_count) >
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	294 tessedit_reject_row_percent &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	295 (current_row->whole_word_rej_count * 100.0 / current_row->rej_count) <
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	296 tessedit_whole_wd_rej_row_percent) {
b50eed0cc0ef ADD: MuPDF v1.26.7: 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 (tessedit_debug_block_rejection) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	298 tprintf("REJECTING ROW %d #chars: %d; #Rejects: %d\n", row_no,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	299 current_row->char_count, current_row->rej_count);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	300 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	301 bool prev_word_rejected = 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	302 while ((word = page_res_it.word()) != nullptr && page_res_it.row() == current_row) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	303 /* Preserve words on good docs unless they are mostly rejected*/
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	304 bool rej_word;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	305 if (!tessedit_row_rej_good_docs && good_quality_doc) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	306 rej_word = word->reject_map.reject_count() /
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	307 static_cast<float>(word->reject_map.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	308 tessedit_good_doc_still_rowrej_wd;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	309 } else if (tessedit_preserve_row_rej_perfect_wds) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	310 /* Preserve perfect words anyway */
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	311 rej_word = word->reject_map.reject_count() > 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	312 word->reject_map.length() < tessedit_preserve_min_wd_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	313 if (rej_word && tessedit_dont_rowrej_good_wds &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	314 word->reject_map.length() >= tessedit_preserve_min_wd_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	315 acceptable_word_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	316 *word->uch_set, word->best_choice->unichar_string().c_str(),
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	317 word->best_choice->unichar_lengths().c_str()) != AC_UNACCEPTABLE) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	318 word_char_quality(word, &char_quality, &accepted_char_quality);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	319 rej_word = char_quality != word->reject_map.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	320 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	321 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	322 rej_word = 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	323 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	324 if (rej_word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	325 /*
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	326 Reject spacing if both current and prev words are rejected.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	327 NOTE - this is NOT restricted to FUZZY spaces. - When tried
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	328 this generated more space errors.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	329 */
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	330 if (tessedit_use_reject_spaces && prev_word_rejected &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	331 page_res_it.prev_row() == page_res_it.row() && word->word->space() == 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	332 word->reject_spaces = 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	333 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	334 word->reject_map.rej_word_row_rej();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	335 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	336 prev_word_rejected = rej_word;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	337 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	338 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	339 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	340 if (tessedit_debug_block_rejection) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	341 tprintf("NOT REJECTING ROW %d #chars: %d # Rejects: %d; \n", row_no,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	342 current_row->char_count, current_row->rej_count);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	343 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	344 while (page_res_it.word() != nullptr && page_res_it.row() == current_row) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	345 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	346 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	347 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	348 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	349 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	350 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	351 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	352 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	353
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	354 /*************************************************************************
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	355 * reject_whole_page()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	356 * Don't believe any of it - set the reject map to 00..00 in all words
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	357 *
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	358 *************************************************************************/
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	359
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	360 void reject_whole_page(PAGE_RES_IT &page_res_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	361 page_res_it.restart_page();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	362 while (page_res_it.word() != 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	363 page_res_it.word()->reject_map.rej_word_doc_rej();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	364 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	365 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	366 // whole page is rejected
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	367 page_res_it.page_res->rejected = 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	368 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	369
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	370 void Tesseract::tilde_crunch(PAGE_RES_IT &page_res_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	371 WERD_RES *word;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	372 GARBAGE_LEVEL garbage_level;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	373 PAGE_RES_IT copy_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	374 bool prev_potential_marked = 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	375 bool found_terrible_word = 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	376 bool ok_dict_word;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	377
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	378 page_res_it.restart_page();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	379 while (page_res_it.word() != 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	380 POLY_BLOCK *pb = page_res_it.block()->block->pdblk.poly_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	381 if (pb != nullptr && !pb->IsText()) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	382 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	383 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	384 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	385 word = page_res_it.word();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	386
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	387 if (crunch_early_convert_bad_unlv_chs) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	388 convert_bad_unlv_chs(word);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	389 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	390
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	391 if (crunch_early_merge_tess_fails) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	392 word->merge_tess_fails();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	393 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	394
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	395 if (word->reject_map.accept_count() != 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	396 found_terrible_word = 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	397 // Forget earlier potential crunches
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	398 prev_potential_marked = 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	399 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	400 ok_dict_word = safe_dict_word(word);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	401 garbage_level = garbage_word(word, ok_dict_word);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	402
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	403 if ((garbage_level != G_NEVER_CRUNCH) && (terrible_word_crunch(word, garbage_level))) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	404 if (crunch_debug > 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	405 tprintf("T CRUNCHING: \"%s\"\n", word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	406 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	407 word->unlv_crunch_mode = CR_KEEP_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	408 if (prev_potential_marked) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	409 while (copy_it.word() != word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	410 if (crunch_debug > 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	411 tprintf("P1 CRUNCHING: \"%s\"\n",
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	412 copy_it.word()->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	413 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	414 copy_it.word()->unlv_crunch_mode = CR_KEEP_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	415 copy_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	416 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	417 prev_potential_marked = 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	418 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	419 found_terrible_word = 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	420 } else if ((garbage_level != G_NEVER_CRUNCH) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	421 (potential_word_crunch(word, garbage_level, ok_dict_word))) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	422 if (found_terrible_word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	423 if (crunch_debug > 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	424 tprintf("P2 CRUNCHING: \"%s\"\n", word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	425 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	426 word->unlv_crunch_mode = CR_KEEP_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	427 } else if (!prev_potential_marked) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	428 copy_it = page_res_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	429 prev_potential_marked = 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	430 if (crunch_debug > 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	431 tprintf("P3 CRUNCHING: \"%s\"\n", word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	432 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	433 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	434 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	435 found_terrible_word = 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	436 // Forget earlier potential crunches
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	437 prev_potential_marked = 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	438 if (crunch_debug > 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	439 tprintf("NO CRUNCH: \"%s\"\n", word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	440 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	441 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	442 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	443 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	444 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	445 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	446
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	447 bool Tesseract::terrible_word_crunch(WERD_RES *word, GARBAGE_LEVEL garbage_level) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	448 int crunch_mode = 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	449
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	450 if (word->best_choice->unichar_string().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	451 (strspn(word->best_choice->unichar_string().c_str(), " ") ==
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	452 word->best_choice->unichar_string().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	453 crunch_mode = 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	454 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	455 int adjusted_len = word->reject_map.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	456 if (adjusted_len > crunch_rating_max) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	457 adjusted_len = crunch_rating_max;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	458 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	459 float rating_per_ch = word->best_choice->rating() / adjusted_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	460
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	461 if (rating_per_ch > crunch_terrible_rating) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	462 crunch_mode = 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	463 } else if (crunch_terrible_garbage && (garbage_level == G_TERRIBLE)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	464 crunch_mode = 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	465 } else if ((word->best_choice->certainty() < crunch_poor_garbage_cert) &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	466 (garbage_level != G_OK)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	467 crunch_mode = 4;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	468 } else if ((rating_per_ch > crunch_poor_garbage_rate) && (garbage_level != G_OK)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	469 crunch_mode = 5;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	470 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	471 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	472 if (crunch_mode > 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	473 if (crunch_debug > 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	474 tprintf("Terrible_word_crunch (%d) on \"%s\"\n", crunch_mode,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	475 word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	476 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	477 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	478 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	479 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	480 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	481 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	482
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	483 bool Tesseract::potential_word_crunch(WERD_RES *word, GARBAGE_LEVEL garbage_level,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	484 bool ok_dict_word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	485 float rating_per_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	486 int adjusted_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	487 const char *str = word->best_choice->unichar_string().c_str();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	488 const char *lengths = word->best_choice->unichar_lengths().c_str();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	489 bool word_crunchable;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	490 int poor_indicator_count = 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	491
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	492 word_crunchable =
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	493 !crunch_leave_accept_strings \|\| word->reject_map.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	494 (acceptable_word_string(*word->uch_set, str, lengths) == AC_UNACCEPTABLE && !ok_dict_word);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	495
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	496 adjusted_len = word->reject_map.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	497 if (adjusted_len > 10) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	498 adjusted_len = 10;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	499 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	500 rating_per_ch = word->best_choice->rating() / adjusted_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	501
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	502 if (rating_per_ch > crunch_pot_poor_rate) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	503 if (crunch_debug > 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	504 tprintf("Potential poor rating on \"%s\"\n", word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	505 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	506 poor_indicator_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	507 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	508
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	509 if (word_crunchable && word->best_choice->certainty() < crunch_pot_poor_cert) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	510 if (crunch_debug > 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	511 tprintf("Potential poor cert on \"%s\"\n", word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	512 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	513 poor_indicator_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	514 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	515
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	516 if (garbage_level != G_OK) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	517 if (crunch_debug > 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	518 tprintf("Potential garbage on \"%s\"\n", word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	519 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	520 poor_indicator_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	521 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	522 return poor_indicator_count >= crunch_pot_indicators;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	523 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	524
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	525 void Tesseract::tilde_delete(PAGE_RES_IT &page_res_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	526 PAGE_RES_IT copy_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	527 bool deleting_from_bol = 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	528 bool marked_delete_point = 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	529 int16_t debug_delete_mode;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	530 CRUNCH_MODE delete_mode;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	531 int16_t x_debug_delete_mode;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	532 CRUNCH_MODE x_delete_mode;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	533
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	534 page_res_it.restart_page();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	535 while (page_res_it.word() != 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	536 WERD_RES *word = page_res_it.word();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	537
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	538 delete_mode = word_deletable(word, debug_delete_mode);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	539 if (delete_mode != CR_NONE) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	540 if (word->word->flag(W_BOL) \|\| deleting_from_bol) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	541 if (crunch_debug > 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	542 tprintf("BOL CRUNCH DELETING(%d): \"%s\"\n", debug_delete_mode,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	543 word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	544 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	545 word->unlv_crunch_mode = delete_mode;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	546 deleting_from_bol = 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	547 } else if (word->word->flag(W_EOL)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	548 if (marked_delete_point) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	549 while (copy_it.word() != word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	550 x_delete_mode = word_deletable(copy_it.word(), x_debug_delete_mode);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	551 if (crunch_debug > 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	552 tprintf("EOL CRUNCH DELETING(%d): \"%s\"\n", x_debug_delete_mode,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	553 copy_it.word()->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	554 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	555 copy_it.word()->unlv_crunch_mode = x_delete_mode;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	556 copy_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	557 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	558 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	559 if (crunch_debug > 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	560 tprintf("EOL CRUNCH DELETING(%d): \"%s\"\n", debug_delete_mode,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	561 word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	562 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	563 word->unlv_crunch_mode = delete_mode;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	564 deleting_from_bol = 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	565 marked_delete_point = 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	566 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	567 if (!marked_delete_point) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	568 copy_it = page_res_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	569 marked_delete_point = 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	570 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	571 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	572 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	573 deleting_from_bol = 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	574 // Forget earlier potential crunches
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	575 marked_delete_point = 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	576 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	577 /*
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	578 The following step has been left till now as the tess fails are used to
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	579 determine if the word is deletable.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	580 */
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	581 if (!crunch_early_merge_tess_fails) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	582 word->merge_tess_fails();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	583 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	584 page_res_it.forward();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	585 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	586 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	587
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	588 void Tesseract::convert_bad_unlv_chs(WERD_RES *word_res) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	589 int 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	590 UNICHAR_ID unichar_dash = word_res->uch_set->unichar_to_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	591 UNICHAR_ID unichar_space = word_res->uch_set->unichar_to_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	592 UNICHAR_ID unichar_tilde = word_res->uch_set->unichar_to_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	593 UNICHAR_ID unichar_pow = word_res->uch_set->unichar_to_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	594 for (i = 0; i < word_res->reject_map.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	595 if (word_res->best_choice->unichar_id(i) == unichar_tilde) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	596 word_res->best_choice->set_unichar_id(unichar_dash, 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	597 if (word_res->reject_map[i].accepted()) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	598 word_res->reject_map[i].setrej_unlv_rej();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	599 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	600 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	601 if (word_res->best_choice->unichar_id(i) == unichar_pow) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	602 word_res->best_choice->set_unichar_id(unichar_space, 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	603 if (word_res->reject_map[i].accepted()) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	604 word_res->reject_map[i].setrej_unlv_rej();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	605 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	606 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	607 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	608 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	609
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	610 GARBAGE_LEVEL Tesseract::garbage_word(WERD_RES *word, bool ok_dict_word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	611 enum STATES {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	612 JUNK,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	613 FIRST_UPPER,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	614 FIRST_LOWER,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	615 FIRST_NUM,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	616 SUBSEQUENT_UPPER,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	617 SUBSEQUENT_LOWER,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	618 SUBSEQUENT_NUM
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	619 };
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	620 const char *str = word->best_choice->unichar_string().c_str();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	621 const char *lengths = word->best_choice->unichar_lengths().c_str();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	622 STATES state = JUNK;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	623 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	624 int isolated_digits = 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	625 int isolated_alphas = 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	626 int bad_char_count = 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	627 int tess_rejs = 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	628 int dodgy_chars = 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	629 int ok_chars;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	630 UNICHAR_ID last_char = -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	631 int alpha_repetition_count = 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	632 int longest_alpha_repetition_count = 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	633 int longest_lower_run_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	634 int lower_string_count = 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	635 int longest_upper_run_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	636 int upper_string_count = 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	637 int total_alpha_count = 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	638 int total_digit_count = 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	639
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	640 for (; str != '\0'; str += (lengths++)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	641 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	642 if (word->uch_set->get_isupper(str, *lengths)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	643 total_alpha_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	644 switch (state) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	645 case SUBSEQUENT_UPPER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	646 case FIRST_UPPER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	647 state = SUBSEQUENT_UPPER;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	648 upper_string_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	649 if (longest_upper_run_len < upper_string_count) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	650 longest_upper_run_len = upper_string_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	651 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	652 if (last_char == word->uch_set->unichar_to_id(str, *lengths)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	653 alpha_repetition_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	654 if (longest_alpha_repetition_count < alpha_repetition_count) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	655 longest_alpha_repetition_count = alpha_repetition_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	656 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	657 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	658 last_char = word->uch_set->unichar_to_id(str, *lengths);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	659 alpha_repetition_count = 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	660 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	661 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	662 case FIRST_NUM:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	663 isolated_digits++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	664 // Fall through.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	665 default:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	666 state = FIRST_UPPER;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	667 last_char = word->uch_set->unichar_to_id(str, *lengths);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	668 alpha_repetition_count = 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	669 upper_string_count = 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	670 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	671 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	672 } else if (word->uch_set->get_islower(str, *lengths)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	673 total_alpha_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	674 switch (state) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	675 case SUBSEQUENT_LOWER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	676 case FIRST_LOWER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	677 state = SUBSEQUENT_LOWER;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	678 lower_string_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	679 if (longest_lower_run_len < lower_string_count) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	680 longest_lower_run_len = lower_string_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	681 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	682 if (last_char == word->uch_set->unichar_to_id(str, *lengths)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	683 alpha_repetition_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	684 if (longest_alpha_repetition_count < alpha_repetition_count) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	685 longest_alpha_repetition_count = alpha_repetition_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	686 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	687 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	688 last_char = word->uch_set->unichar_to_id(str, *lengths);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	689 alpha_repetition_count = 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	690 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	691 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	692 case FIRST_NUM:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	693 isolated_digits++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	694 // Fall through.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	695 default:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	696 state = FIRST_LOWER;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	697 last_char = word->uch_set->unichar_to_id(str, *lengths);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	698 alpha_repetition_count = 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	699 lower_string_count = 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	700 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	701 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	702 } else if (word->uch_set->get_isdigit(str, *lengths)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	703 total_digit_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	704 switch (state) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	705 case FIRST_NUM:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	706 state = SUBSEQUENT_NUM;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	707 case SUBSEQUENT_NUM:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	708 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	709 case FIRST_UPPER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	710 case FIRST_LOWER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	711 isolated_alphas++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	712 // Fall through.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	713 default:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	714 state = FIRST_NUM;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	715 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	716 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	717 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	718 if (lengths == 1 && 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	719 tess_rejs++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	720 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	721 bad_char_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	722 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	723 switch (state) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	724 case FIRST_NUM:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	725 isolated_digits++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	726 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	727 case FIRST_UPPER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	728 case FIRST_LOWER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	729 isolated_alphas++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	730 default:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	731 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	732 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	733 state = JUNK;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	734 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	735 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	736
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	737 switch (state) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	738 case FIRST_NUM:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	739 isolated_digits++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	740 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	741 case FIRST_UPPER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	742 case FIRST_LOWER:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	743 isolated_alphas++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	744 default:
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	745 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	746 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	747
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	748 if (crunch_include_numerals) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	749 total_alpha_count += total_digit_count - isolated_digits;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	750 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	751
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	752 if (crunch_leave_ok_strings && len >= 4 && 2 * (total_alpha_count - isolated_alphas) > 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	753 longest_alpha_repetition_count < crunch_long_repetitions) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	754 if ((crunch_accept_ok &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	755 acceptable_word_string(*word->uch_set, str, lengths) != AC_UNACCEPTABLE) \|\|
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	756 longest_lower_run_len > crunch_leave_lc_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	757 longest_upper_run_len > crunch_leave_uc_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	758 return G_NEVER_CRUNCH;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	759 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	760 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	761 if (word->reject_map.length() > 1 && strpbrk(str, " ") == 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	762 (word->best_choice->permuter() == SYSTEM_DAWG_PERM \|\|
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	763 word->best_choice->permuter() == FREQ_DAWG_PERM \|\|
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	764 word->best_choice->permuter() == USER_DAWG_PERM \|\|
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	765 word->best_choice->permuter() == NUMBER_PERM \|\|
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	766 acceptable_word_string(*word->uch_set, str, lengths) != AC_UNACCEPTABLE \|\| ok_dict_word)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	767 return G_OK;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	768 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	769
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	770 ok_chars = len - bad_char_count - isolated_digits - isolated_alphas - tess_rejs;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	771
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	772 if (crunch_debug > 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	773 tprintf("garbage_word: \"%s\"\n", word->best_choice->unichar_string().c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	774 tprintf("LEN: %d bad: %d iso_N: %d iso_A: %d rej: %d\n", len, bad_char_count,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	775 isolated_digits, isolated_alphas, tess_rejs);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	776 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	777 if (bad_char_count == 0 && tess_rejs == 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	778 (len > isolated_digits + isolated_alphas \|\| len <= 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	779 return G_OK;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	780 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	781
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	782 if (tess_rejs > ok_chars \|\| (tess_rejs > 0 && (bad_char_count + tess_rejs) * 2 > 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	783 return G_TERRIBLE;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	784 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	785
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	786 if (len > 4) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	787 dodgy_chars = 2 * tess_rejs + bad_char_count + isolated_digits + isolated_alphas;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	788 if (dodgy_chars > 5 \|\| (dodgy_chars / static_cast<float>(len)) > 0.5) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	789 return G_DODGY;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	790 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	791 return G_OK;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	792 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	793 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	794 dodgy_chars = 2 * tess_rejs + bad_char_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	795 if ((len == 4 && dodgy_chars > 2) \|\| (len == 3 && dodgy_chars > 2) \|\| dodgy_chars >= 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	796 return G_DODGY;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	797 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	798 return G_OK;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	799 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	800 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	801 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	802
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	803 /*************************************************************************
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	804 * word_deletable()
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	805 * DELETE WERDS AT ENDS OF ROWS 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	806 * Word is crunched &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	807 * ( string length = 0 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	808 * > 50% of chars are "\|" (before merging) 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	809 * certainty < -10 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	810 * rating /char > 60 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	811 * TOP of word is more than 0.5 xht BELOW baseline 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	812 * BOTTOM of word is more than 0.5 xht ABOVE xht 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	813 * length of word < 3xht 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	814 * height of word < 0.7 xht 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	815 * height of word > 3.0 xht 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	816 * >75% of the outline BBs have longest dimension < 0.5xht
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	817 *************************************************************************/
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	818
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	819 CRUNCH_MODE Tesseract::word_deletable(WERD_RES *word, int16_t &delete_mode) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	820 int word_len = word->reject_map.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	821 float rating_per_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	822 TBOX box; // BB of word
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	823
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	824 if (word->unlv_crunch_mode == CR_NONE) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	825 delete_mode = 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	826 return CR_NONE;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	827 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	828
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	829 if (word_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	830 delete_mode = 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	831 return CR_DELETE;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	832 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	833
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	834 if (word->rebuild_word != 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	835 // Cube leaves rebuild_word 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	836 box = word->rebuild_word->bounding_box();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	837 if (box.height() < crunch_del_min_ht * kBlnXHeight) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	838 delete_mode = 4;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	839 return CR_DELETE;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	840 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	841
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	842 if (noise_outlines(word->rebuild_word)) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	843 delete_mode = 5;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	844 return CR_DELETE;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	845 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	846 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	847
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	848 if ((failure_count(word) * 1.5) > word_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	849 delete_mode = 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	850 return CR_LOOSE_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	851 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	852
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	853 if (word->best_choice->certainty() < crunch_del_cert) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	854 delete_mode = 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	855 return CR_LOOSE_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	856 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	857
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	858 rating_per_ch = word->best_choice->rating() / word_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	859
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	860 if (rating_per_ch > crunch_del_rating) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	861 delete_mode = 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	862 return CR_LOOSE_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	863 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	864
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	865 if (box.top() < kBlnBaselineOffset - crunch_del_low_word * kBlnXHeight) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	866 delete_mode = 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	867 return CR_LOOSE_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	868 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	869
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	870 if (box.bottom() > kBlnBaselineOffset + crunch_del_high_word * kBlnXHeight) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	871 delete_mode = 10;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	872 return CR_LOOSE_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	873 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	874
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	875 if (box.height() > crunch_del_max_ht * kBlnXHeight) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	876 delete_mode = 11;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	877 return CR_LOOSE_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	878 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	879
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	880 if (box.width() < crunch_del_min_width * kBlnXHeight) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	881 delete_mode = 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	882 return CR_LOOSE_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	883 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	884
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	885 delete_mode = 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	886 return CR_NONE;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	887 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	888
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	889 int16_t Tesseract::failure_count(WERD_RES *word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	890 const char *str = word->best_choice->unichar_string().c_str();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	891 int tess_rejs = 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	892
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	893 for (; *str != '\0'; 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	894 if (*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	895 tess_rejs++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	896 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	897 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	898 return tess_rejs;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	899 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	900
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	901 bool Tesseract::noise_outlines(TWERD *word) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	902 TBOX box; // BB of outline
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	903 int16_t outline_count = 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	904 int16_t small_outline_count = 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	905 int16_t max_dimension;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	906 float small_limit = kBlnXHeight * crunch_small_outlines_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	907
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	908 for (unsigned b = 0; b < word->NumBlobs(); ++b) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	909 TBLOB *blob = word->blobs[b];
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	910 for (TESSLINE *ol = blob->outlines; ol != nullptr; ol = ol->next) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	911 outline_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	912 box = ol->bounding_box();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	913 if (box.height() > box.width()) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	914 max_dimension = box.height();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	915 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	916 max_dimension = box.width();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	917 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	918 if (max_dimension < small_limit) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	919 small_outline_count++;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	920 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	921 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	922 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	923 return small_outline_count >= outline_count;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	924 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	925
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. Franz Glasner <fzglas.hg@dom66.de> parents: diff changeset	926 } // namespace tesseract

Mercurial > hgrepos > Python2 > PyMuPDF

annotate mupdf-source/thirdparty/tesseract/src/ccmain/docqual.cpp @ 46:7ee69f120f19 default tip