wordlist2dawg.cpp File Reference

#include "classify.h"
#include "commontraining.h"
#include "dawg.h"
#include "dict.h"
#include "helpers.h"
#include "serialis.h"
#include "trie.h"
#include "unicharset.h"

Go to the source code of this file.

Functions
int	main (int argc, char **argv)

Function Documentation

main()

int main	(	int	argc,
		char **	argv
	)

Definition at line 33 of file wordlist2dawg.cpp.

                                {
  tesseract::CheckSharedLibraryVersion();
 
  if (argc > 1 && (!strcmp(argv[1], "-v") || !strcmp(argv[1], "--version"))) {
    printf("%s\n", tesseract::TessBaseAPI::Version());
    return EXIT_SUCCESS;
  } else if (!(argc == 4 || (argc == 5 && strcmp(argv[1], "-t") == 0) ||
               (argc == 6 && strcmp(argv[1], "-r") == 0))) {
    printf(
        "Usage: %s -v | --version |\n"
        "       %s [-t | -r [reverse policy] ] word_list_file"
        " dawg_file unicharset_file\n",
        argv[0], argv[0]);
    return EXIT_FAILURE;
  }
  tesseract::Classify classify;
  int argv_index = 0;
  if (argc == 5) {
    ++argv_index;
  }
  tesseract::Trie::RTLReversePolicy reverse_policy = tesseract::Trie::RRP_DO_NO_REVERSE;
  if (argc == 6) {
    ++argv_index;
    int tmp_int;
    sscanf(argv[++argv_index], "%d", &tmp_int);
    reverse_policy = static_cast<tesseract::Trie::RTLReversePolicy>(tmp_int);
    tprintf("Set reverse_policy to %s\n", tesseract::Trie::get_reverse_policy_name(reverse_policy));
  }
  const char *wordlist_filename = argv[++argv_index];
  const char *dawg_filename = argv[++argv_index];
  const char *unicharset_file = argv[++argv_index];
  tprintf("Loading unicharset from '%s'\n", unicharset_file);
  if (!classify.getDict().getUnicharset().load_from_file(unicharset_file)) {
    tprintf("Failed to load unicharset from '%s'\n", unicharset_file);
    return EXIT_FAILURE;
  }
  const UNICHARSET &unicharset = classify.getDict().getUnicharset();
  if (argc == 4 || argc == 6) {
    tesseract::Trie trie(
        // the first 3 arguments are not used in this case
        tesseract::DAWG_TYPE_WORD, "", SYSTEM_DAWG_PERM, unicharset.size(),
        classify.getDict().dawg_debug_level);
    tprintf("Reading word list from '%s'\n", wordlist_filename);
    if (!trie.read_and_add_word_list(wordlist_filename, unicharset, reverse_policy)) {
      tprintf("Failed to add word list from '%s'\n", wordlist_filename);
      return EXIT_FAILURE;
    }
    tprintf("Reducing Trie to SquishedDawg\n");
    std::unique_ptr<tesseract::SquishedDawg> dawg(trie.trie_to_dawg());
    if (dawg && dawg->NumEdges() > 0) {
      tprintf("Writing squished DAWG to '%s'\n", dawg_filename);
      dawg->write_squished_dawg(dawg_filename);
    } else {
      tprintf("Dawg is empty, skip producing the output file\n");
    }
  } else if (argc == 5) {
    tprintf("Loading dawg DAWG from '%s'\n", dawg_filename);
    tesseract::SquishedDawg words(dawg_filename,
                                  // these 3 arguments are not used in this case
                                  tesseract::DAWG_TYPE_WORD, "", SYSTEM_DAWG_PERM,
                                  classify.getDict().dawg_debug_level);
    tprintf("Checking word list from '%s'\n", wordlist_filename);
    words.check_for_words(wordlist_filename, unicharset, true);
  } else { // should never get here
    tprintf("Invalid command-line options\n");
    return EXIT_FAILURE;
  }
  return EXIT_SUCCESS;
}