fixspace.cpp File Reference

#include "fixspace.h"
#include <cstdint>
#include "blobs.h"
#include "boxword.h"
#include "errcode.h"
#include "host.h"
#include "normalis.h"
#include "ocrclass.h"
#include "pageres.h"
#include "params.h"
#include "ratngs.h"
#include "rect.h"
#include "stepblob.h"
#include "strngs.h"
#include "tesseractclass.h"
#include "tessvars.h"
#include "tprintf.h"
#include "unichar.h"
#include "unicharset.h"
#include "werd.h"

Go to the source code of this file.

Namespaces
	tesseract

Macros
#define	PERFECT_WERDS   999
#define	MAXSPACING   128 /*max expected spacing in pix */

Functions
void	initialise_search (WERD_RES_LIST &src_list, WERD_RES_LIST &new_list)
transform_to_next_perm()
Examines the current word list to find the smallest word gap size. Then walks the word list closing any gaps of this size by either inserted new combination words, or extending existing ones. The routine COULD be limited to stop it building words longer than N blobs. If there are no more gaps then it DELETES the entire list and returns the empty list to cause termination.
void	transform_to_next_perm (WERD_RES_LIST &words)
void	fixspace_dbg (WERD_RES *word)

Macro Definition Documentation

MAXSPACING

#define MAXSPACING   128 /*max expected spacing in pix */

Definition at line 47 of file fixspace.cpp.

PERFECT_WERDS

#define PERFECT_WERDS   999

Definition at line 46 of file fixspace.cpp.

Function Documentation

fixspace_dbg()

void fixspace_dbg

(

WERD_RES *

word

)

Definition at line 825 of file fixspace.cpp.

                                  {
  TBOX box = word->word->bounding_box();
  const bool show_map_detail = false;
  int16_t i;

  box.print();
  tprintf(" \"%s\" ", word->best_choice->unichar_string().string());
  tprintf("Blob count: %d (word); %d/%d (rebuild word)\n",
          word->word->cblob_list()->length(),
          word->rebuild_word->NumBlobs(),
          word->box_word->length());
  word->reject_map.print(debug_fp);
  tprintf("\n");
  if (show_map_detail) {
    tprintf("\"%s\"\n", word->best_choice->unichar_string().string());
    for (i = 0; word->best_choice->unichar_string()[i] != '\0'; i++) {
      tprintf("**** \"%c\" ****\n", word->best_choice->unichar_string()[i]);
      word->reject_map[i].full_print(debug_fp);
    }
  }

  tprintf("Tess Accepted: %s\n", word->tess_accepted ? "TRUE" : "FALSE");
  tprintf("Done flag: %s\n\n", word->done ? "TRUE" : "FALSE");
}

initialise_search()

void initialise_search	(	WERD_RES_LIST &	src_list,
		WERD_RES_LIST &	new_list
	)

Definition at line 207 of file fixspace.cpp.

                                                                         {
  WERD_RES_IT src_it(&src_list);
  WERD_RES_IT new_it(&new_list);
  WERD_RES *src_wd;
  WERD_RES *new_wd;

  for (src_it.mark_cycle_pt(); !src_it.cycled_list(); src_it.forward()) {
    src_wd = src_it.data();
    if (!src_wd->combination) {
      new_wd = WERD_RES::deep_copy(src_wd);
      new_wd->combination = false;
      new_wd->part_of_combo = false;
      new_it.add_after_then_move(new_wd);
    }
  }
}

transform_to_next_perm()

void transform_to_next_perm

(

WERD_RES_LIST &

words

)

Definition at line 402 of file fixspace.cpp.

                                                  {
  WERD_RES_IT word_it(&words);
  WERD_RES_IT prev_word_it(&words);
  WERD_RES *word;
  WERD_RES *prev_word;
  WERD_RES *combo;
  WERD *copy_word;
  int16_t prev_right = -INT16_MAX;
  TBOX box;
  int16_t gap;
  int16_t min_gap = INT16_MAX;

  for (word_it.mark_cycle_pt(); !word_it.cycled_list(); word_it.forward()) {
    word = word_it.data();
    if (!word->part_of_combo) {
      box = word->word->bounding_box();
      if (prev_right > -INT16_MAX) {
        gap = box.left() - prev_right;
        if (gap < min_gap)
          min_gap = gap;
      }
      prev_right = box.right();
    }
  }
  if (min_gap < INT16_MAX) {
    prev_right = -INT16_MAX;        // back to start
    word_it.set_to_list(&words);
    // Note: we can't use cycle_pt due to inserted combos at start of list.
    for (; (prev_right == -INT16_MAX) || !word_it.at_first();
         word_it.forward()) {
      word = word_it.data();
      if (!word->part_of_combo) {
        box = word->word->bounding_box();
        if (prev_right > -INT16_MAX) {
          gap = box.left() - prev_right;
          if (gap <= min_gap) {
            prev_word = prev_word_it.data();
            if (prev_word->combination) {
              combo = prev_word;
            } else {
              /* Make a new combination and insert before
               * the first word being joined. */
              copy_word = new WERD;
              *copy_word = *(prev_word->word);
              // deep copy
              combo = new WERD_RES(copy_word);
              combo->combination = TRUE;
              combo->x_height = prev_word->x_height;
              prev_word->part_of_combo = true;
              prev_word_it.add_before_then_move(combo);
            }
            combo->word->set_flag(W_EOL, word->word->flag(W_EOL));
            if (word->combination) {
              combo->word->join_on(word->word);
              // Move blobs to combo
              // old combo no longer needed
              delete word_it.extract();
            } else {
              // Copy current wd to combo
              combo->copy_on(word);
              word->part_of_combo = true;
            }
            combo->done = FALSE;
            combo->ClearResults();
          } else {
            prev_word_it = word_it;  // catch up
          }
        }
        prev_right = box.right();
      }
    }
  } else {
    words.clear();  // signal termination
  }
}