permute.cpp

来自「一个google的OCR源码」· C++ 代码 · 共 1,706 行 · 第 1/3 页

  },
  {                              //0xa0=.
    0, 0, 0
  },
  {                              //0xa1=.
    0, 0, 0
  },
  {                              //0xa2=.
    0, 0, 0
  },
  {                              //0xa3=.
    0, 0, 0
  },
  {                              //0xa4=.
    0, 0, 0
  },
  {                              //0xa5=.
    0, 0, 0
  },
  {                              //0xa6=.
    0, 0, 0
  },
  {                              //0xa7=.
    0, 0, 0
  },
  {                              //0xa8=.
    0, 0, 0
  },
  {                              //0xa9=.
    0, 0, 0
  },
  {                              //0xaa=.
    0, 0, 0
  },
  {                              //0xab=.
    0, 0, 0
  },
  {                              //0xac=.
    0, 0, 0
  },
  {                              //0xad=.
    0, 0, 0
  },
  {                              //0xae=.
    0, 0, 0
  },
  {                              //0xaf=.
    0, 0, 0
  },
  {                              //0xb0=.
    0, 0, 0
  },
  {                              //0xb1=.
    0, 0, 0
  },
  {                              //0xb2=.
    0, 0, 0
  },
  {                              //0xb3=.
    0, 0, 0
  },
  {                              //0xb4=.
    0, 0, 0
  },
  {                              //0xb5=.
    0, 0, 0
  },
  {                              //0xb6=.
    0, 0, 0
  },
  {                              //0xb7=.
    0, 0, 0
  },
  {                              //0xb8=.
    0, 0, 0
  },
  {                              //0xb9=.
    0, 0, 0
  },
  {                              //0xba=.
    0, 0, 0
  },
  {                              //0xbb=.
    0, 0, 0
  },
  {                              //0xbc=.
    0, 0, 0
  },
  {                              //0xbd=.
    0, 0, 0
  },
  {                              //0xbe=.
    0, 0, 0
  },
  {                              //0xbf=.
    0, 0, 0
  },
  {                              //0xc0=.
    0, 0, 0
  },
  {                              //0xc1=.
    0, 0, 0
  },
  {                              //0xc2=.
    0, 0, 0
  },
  {                              //0xc3=.
    0, 0, 0
  },
  {                              //0xc4=.
    0, 0, 0
  },
  {                              //0xc5=.
    0, 0, 0
  },
  {                              //0xc6=.
    0, 0, 0
  },
  {                              //0xc7=.
    0, 0, 0
  },
  {                              //0xc8=.
    0, 0, 0
  },
  {                              //0xc9=.
    0, 0, 0
  },
  {                              //0xca=.
    0, 0, 0
  },
  {                              //0xcb=.
    0, 0, 0
  },
  {                              //0xcc=.
    0, 0, 0
  },
  {                              //0xcd=.
    0, 0, 0
  },
  {                              //0xce=.
    0, 0, 0
  },
  {                              //0xcf=.
    0, 0, 0
  },
  {                              //0xd0=.
    0, 0, 0
  },
  {                              //0xd1=.
    0, 0, 0
  },
  {                              //0xd2=.
    0, 0, 0
  },
  {                              //0xd3=.
    0, 0, 0
  },
  {                              //0xd4=.
    0, 0, 0
  },
  {                              //0xd5=.
    0, 0, 0
  },
  {                              //0xd6=.
    0, 0, 0
  },
  {                              //0xd7=.
    0, 0, 0
  },
  {                              //0xd8=.
    0, 0, 0
  },
  {                              //0xd9=.
    0, 0, 0
  },
  {                              //0xda=.
    0, 0, 0
  },
  {                              //0xdb=.
    0, 0, 0
  },
  {                              //0xdc=.
    0, 0, 0
  },
  {                              //0xdd=.
    0, 0, 0
  },
  {                              //0xde=.
    0, 0, 0
  },
  {                              //0xdf=.
    0, 0, 0
  },
  {                              //0xe0=.
    0, 0, 0
  },
  {                              //0xe1=.
    0, 0, 0
  },
  {                              //0xe2=.
    0, 0, 0
  },
  {                              //0xe3=.
    0, 0, 0
  },
  {                              //0xe4=.
    0, 0, 0
  },
  {                              //0xe5=.
    0, 0, 0
  },
  {                              //0xe6=.
    0, 0, 0
  },
  {                              //0xe7=.
    0, 0, 0
  },
  {                              //0xe8=.
    0, 0, 0
  },
  {                              //0xe9=.
    0, 0, 0
  },
  {                              //0xea=.
    0, 0, 0
  },
  {                              //0xeb=.
    0, 0, 0
  },
  {                              //0xec=.
    0, 0, 0
  },
  {                              //0xed=.
    0, 0, 0
  },
  {                              //0xee=.
    0, 0, 0
  },
  {                              //0xef=.
    0, 0, 0
  },
  {                              //0xf0=.
    0, 0, 0
  },
  {                              //0xf1=.
    0, 0, 0
  },
  {                              //0xf2=.
    0, 0, 0
  },
  {                              //0xf3=.
    0, 0, 0
  },
  {                              //0xf4=.
    0, 0, 0
  },
  {                              //0xf5=.
    0, 0, 0
  },
  {                              //0xf6=.
    0, 0, 0
  },
  {                              //0xf7=.
    0, 0, 0
  },
  {                              //0xf8=.
    0, 0, 0
  },
  {                              //0xf9=.
    0, 0, 0
  },
  {                              //0xfa=.
    0, 0, 0
  },
  {                              //0xfb=.
    0, 0, 0
  },
  {                              //0xfc=.
    0, 0, 0
  },
  {                              //0xfd=.
    0, 0, 0
  },
  {                              //0xfe=.
    0, 0, 0
  },
  {                              //0xff=.
    0, 0, 0
  },
};
#endif

//extern "C" double permuter_pending_threshold;

                                 /* Similarity matcher values */
#define SIM_CERTAINTY_SCALE  -10.0
                                 /* Similarity matcher values */
#define SIM_CERTAINTY_OFFSET -10.0
                                 /* Worst E*L product to stop on */
#define SIMILARITY_FLOOR     100.0
/*----------------------------------------------------------------------
              F u n c t i o n s
----------------------------------------------------------------------*/

/**********************************************************************
 * good_choice
 *
 * Return TRUE if a good answer is found for the unknown blob rating.
 **********************************************************************/
int good_choice(A_CHOICE *choice) {
  register float certainty;
  if (choice == NULL)
    return (FALSE);
  if (similarity_enable) {
    if ((class_probability (choice) + 1) * class_certainty (choice) >
      SIMILARITY_FLOOR)
      return (FALSE);
    certainty =
      SIM_CERTAINTY_OFFSET +
      class_probability (choice) * SIM_CERTAINTY_SCALE;
  }

  else {
    certainty = class_certainty (choice);
  }
  if (certainty > certainty_threshold) {
    return (TRUE);
  }

  else {
    return (FALSE);
  }
}


/**********************************************************************
 * add_document_word
 *
 * Add a word found on this document to the document specific
 * dictionary.
 **********************************************************************/
void add_document_word(A_CHOICE *best_choice) {
  char filename[CHARS_PER_LINE];
  FILE *doc_word_file;
  char *string;
  char *lengths;
  int stringlen;                 //length of word

  string = class_string (best_choice);
  lengths = class_lengths (best_choice);
  stringlen = strlen (lengths);

  // Skip if using external dictionary.
  if (letter_is_okay != &def_letter_is_okay) return;

  if (!doc_dict_enable
    || valid_word (string) || CurrentWordAmbig () || stringlen < 2)
    return;

  if (!good_choice (best_choice) || stringlen == 2) {
    if (class_certainty (best_choice) < permuter_pending_threshold)
      return;
    if (!word_in_dawg (pending_words, string)) {
      if (stringlen > 2 ||
          (stringlen >= 2 && unicharset.get_isupper (string, lengths[0]) &&
           unicharset.get_isupper (string + lengths[0], lengths[1])))
        add_word_to_dawg(pending_words,
                         string,
                         MAX_DOC_EDGES,
                         RESERVED_DOC_EDGES);
      return;
    }
  }

  if (save_doc_words) {
    strcpy(filename, imagefile);
    strcat (filename, ".doc");
    doc_word_file = open_file (filename, "a");
    fprintf (doc_word_file, "%s\n", string);
    fclose(doc_word_file);
  }
  add_word_to_dawg(document_words, string, MAX_DOC_EDGES, RESERVED_DOC_EDGES);
}


/**********************************************************************
 * adjust_non_word
 *
 * Assign an adjusted value to a string that is a non-word.  The value
 * that this word choice has is based on case and punctuation rules.
 **********************************************************************/
void
adjust_non_word (A_CHOICE * best_choice, float certainties[]) {
  char *this_word;
  float adjust_factor;

  if (adjust_debug)
    cprintf ("%s %4.2f ",
      class_string (best_choice), class_probability (best_choice));

  this_word = class_string (best_choice);

  class_probability (best_choice) += RATING_PAD;
  if (case_ok (this_word, class_lengths (best_choice))
      && punctuation_ok (this_word, class_lengths (best_choice)) != -1) {
    class_probability (best_choice) *= non_word;
    adjust_factor = non_word;
    if (adjust_debug)
      cprintf (", %4.2f ", non_word);
  }
  else {
    class_probability (best_choice) *= garbage;
    adjust_factor = garbage;
    if (adjust_debug) {
      if (!case_ok (this_word, class_lengths (best_choice)))
        cprintf (", C");
      if (punctuation_ok (this_word, class_lengths (best_choice)) == -1)
        cprintf (", P");
      cprintf (", %4.2f ", garbage);
    }
  }

  class_probability (best_choice) -= RATING_PAD;

  LogNewWordChoice(best_choice, adjust_factor, certainties);

  if (adjust_debug)
    cprintf (" --> %4.2f\n", class_probability (best_choice));
}


/**********************************************************************
 * init_permute
 *
 * Initialize anything that needs to be set up for the permute
 * functions.
 **********************************************************************/
void init_permute_vars() {
  make_adjust_debug();
  make_compound_debug();
  make_non_word();
  make_garbage();
  make_doc_words();
  make_doc_dict();

  init_permdawg_vars();
  init_permnum();
}

void init_permute() {
  if (word_dawg != NULL)
    end_permute();
  init_permdawg();
  STRING name;
  name = language_data_path_prefix;
  name += "word-dawg";
  word_dawg = read_squished_dawg(name.string());

  document_words =
    (EDGE_ARRAY) memalloc (sizeof (EDGE_RECORD) * MAX_DOC_EDGES);
  initialize_dawg(document_words, MAX_DOC_EDGES);

  pending_words =
    (EDGE_ARRAY) memalloc (sizeof (EDGE_RECORD) * MAX_DOC_EDGES);
  initialize_dawg(pending_words, MAX_DOC_EDGES);

  user_words = (EDGE_ARRAY) memalloc (sizeof (EDGE_RECORD) * MAX_USER_EDGES);
  name = language_data_path_prefix;
  name += "user-words";
  read_word_list(name.string(), user_words, MAX_USER_EDGES, USER_RESERVED_EDGES);
}

void end_permute() {
  if (word_dawg == NULL)
    return;  // Not safe to call twice.
  memfree(word_dawg);
  word_dawg = NULL;
  memfree(document_words);
  document_words =  NULL;
  memfree(pending_words);
  pending_words = NULL;
  memfree(user_words);
  user_words = NULL;
  end_permdawg();
}

/**********************************************************************
 * permute_all
 *
 * Permute all the characters together using all of the different types
 * of permuters/selectors available.  Each of the characters must have
 * a non-NIL choice list.
 **********************************************************************/
A_CHOICE *permute_all(CHOICES_LIST char_choices,
                      float rating_limit,
                      A_CHOICE *raw_choice) {
  A_CHOICE *result_1;
  A_CHOICE *result_2 = NULL;
  BOOL8 any_alpha;

  result_1 = permute_top_choice (char_choices, rating_limit, raw_choice,
    &any_alpha);

  if (ngram_permuter_activated)
    return ngram_permute_and_select(char_choices, rating_limit, word_dawg);

  if (result_1 == NULL)
    return (NULL);
  if (permute_only_top)
    return result_1;
  if (any_alpha && array_count (char_choices) <= MAX_WERD_LENGTH) {
    result_2 = permute_words (char_choices, rating_limit);
    if (class_probability (result_1) < class_probability (result_2)
    || class_string (result_2) == NULL) {
      free_choice(result_2);
    }
    else {
      free_choice(result_1);
      result_1 = result_2;
    }
  }

  result_2 = number_permute_and_select (char_choices, rating_limit);

  if (class_probability (result_1) < class_probability (result_2)
  || class_string (result_2) == NULL) {
    free_choice(result_2);
  }
  else {
    free_choice(result_1);
    result_1 = result_2;
  }

  result_2 = permute_compound_words (char_choices, rating_limit);

  if (!result_2 ||
    class_probability (result_1) < class_probability (result_2)
  || class_string (result_2) == NULL) {
    free_choice(result_2);
  }
  else {
    free_choice(result_1);
    result_1 = result_2;
  }

  return (result_1);
}


/**********************************************************************
 * permute_characters
 *
 * Permute these characters together according to each of the different
 * permuters that are enabled.
 **********************************************************************/
void permute_characters(CHOICES_LIST char_choices,
                        float limit,
                        A_CHOICE *best_choice,
                        A_CHOICE *raw_choice) {
  A_CHOICE *this_choice;

  permutation_count++;           /* Global counter */

  this_choice = permute_all (char_choices, limit, raw_choice);

  if (this_choice &&
  class_probability (this_choice) < class_probability (best_choice)) {
    clone_choice(best_choice, this_choice);
  }