28static std::unique_ptr<tesseract::Dawg> LoadSquishedDawg(
const UNICHARSET &unicharset,
const char *filename) {
29 const int kDictDebugLevel = 1;
31 if (!dawg_file.
Open(filename,
nullptr)) {
32 tprintf(
"Could not open %s for reading.\n", filename);
35 tprintf(
"Loading word list from %s\n", filename);
38 if (!retval->Load(&dawg_file)) {
39 tprintf(
"Could not read %s\n", filename);
50 fprintf(file_,
"%s\n", word);
59 const char *outfile_name) {
60 FILE *out = fopen(outfile_name,
"wb");
62 tprintf(
"Could not open %s for writing.\n", outfile_name);
66 using namespace std::placeholders;
71int main(
int argc,
char *argv[]) {
72 tesseract::CheckSharedLibraryVersion();
74 if (argc > 1 && (!strcmp(argv[1],
"-v") || !strcmp(argv[1],
"--version"))) {
77 }
else if (argc != 4) {
78 tprintf(
"Print all the words in a given dawg.\n");
80 "Usage: %s -v | --version | %s <unicharset> <dawgfile> "
85 const char *unicharset_file = argv[1];
86 const char *dawg_file = argv[2];
87 const char *wordlist_file = argv[3];
90 tprintf(
"Error loading unicharset from %s.\n", unicharset_file);
93 auto dict = LoadSquishedDawg(unicharset, dawg_file);
94 if (dict ==
nullptr) {
95 tprintf(
"Error loading dictionary from %s.\n", dawg_file);
98 int retval = WriteDawgAsWordlist(unicharset, dict.get(), wordlist_file);
int main(int argc, char *argv[])
void tprintf(const char *format,...)
static const char * Version()
bool Open(const char *filename, FileReader reader)
bool load_from_file(const char *const filename, bool skip_fragments)
void iterate_words(const UNICHARSET &unicharset, std::function< void(const WERD_CHOICE *)> cb) const
void output_word(const char *word)
WordOutputter(FILE *file)