// Helper to set the properties for an input unicharset file, writes to the
// output file. If an appropriate script unicharset can be found in the
// script_dir directory, then the tops and bottoms are expanded using the
// script unicharset.
// If non-empty, xheight data for the fonts are written to the xheights_file.
void SetPropertiesForInputFile(const std::string& script_dir,
const std::string& input_unicharset_file,
const std::string& output_unicharset_file,
const std::string& output_xheights_file) {
UNICHARSET unicharset;
// Load the input unicharset
unicharset.load_from_file(input_unicharset_file.c_str());
tprintf("Loaded unicharset of size %d from file %s\n", unicharset.size(),
input_unicharset_file.c_str());
// Set unichar properties
tprintf("Setting unichar properties\n");
SetupBasicProperties(true, false, &unicharset);
tprintf("Setting script properties\n");
SetScriptProperties(script_dir, &unicharset);
if (!output_xheights_file.empty()) {
std::string xheights_str = GetXheightString(script_dir, unicharset);
File::WriteStringToFileOrDie(xheights_str, output_xheights_file);
}
// Write the output unicharset
tprintf("Writing unicharset to file %s\n", output_unicharset_file.c_str());
unicharset.save_to_file(output_unicharset_file.c_str());
}
int main(int argc, char** argv) {
int option;
const char* output_directory = ".";
STRING unicharset_file_name;
// Special characters are now included by default.
UNICHARSET unicharset;
setlocale(LC_ALL, "");
// Print usage
if (argc <= 1) {
printf("Usage: %s [-D DIRECTORY] FILE...\n", argv[0]);
exit(1);
}
// Parse arguments
while ((option = tessopt(argc, argv, "D" )) != EOF) {
switch (option) {
case 'D':
output_directory = tessoptarg;
++tessoptind;
break;
}
}
// Save file name
unicharset_file_name = output_directory;
unicharset_file_name += "/";
unicharset_file_name += kUnicharsetFileName;
// Load box files
for (; tessoptind < argc; ++tessoptind) {
printf("Extracting unicharset from %s\n", argv[tessoptind]);
FILE* box_file = fopen(argv[tessoptind], "rb");
if (box_file == NULL) {
printf("Cannot open box file %s\n", argv[tessoptind]);
return -1;
}
TBOX box;
STRING unichar_string;
int line_number = 0;
while (ReadNextBox(&line_number, box_file, &unichar_string, &box)) {
unicharset.unichar_insert(unichar_string.string());
set_properties(&unicharset, unichar_string.string());
}
}
// Write unicharset file
if (unicharset.save_to_file(unicharset_file_name.string())) {
printf("Wrote unicharset file %s.\n", unicharset_file_name.string());
}
else {
printf("Cannot save unicharset file %s.\n", unicharset_file_name.string());
return -1;
}
return 0;
}
// Helper to set the properties for an input unicharset file, writes to the
// output file. If an appropriate script unicharset can be found in the
// script_dir directory, then the tops and bottoms are expanded using the
// script unicharset.
// If non-empty, xheight data for the fonts are written to the xheights_file.
void SetPropertiesForInputFile(const string& script_dir,
const string& input_unicharset_file,
const string& output_unicharset_file,
const string& output_xheights_file) {
UNICHARSET unicharset;
// Load the input unicharset
unicharset.load_from_file(input_unicharset_file.c_str());
tprintf("Loaded unicharset of size %d from file %s\n", unicharset.size(),
input_unicharset_file.c_str());
// Set unichar properties
tprintf("Setting unichar properties\n");
SetupBasicProperties(true, false, &unicharset);
string xheights_str;
for (int s = 0; s < unicharset.get_script_table_size(); ++s) {
// Load the unicharset for the script if available.
string filename = script_dir + "/" +
unicharset.get_script_from_script_id(s) + ".unicharset";
UNICHARSET script_set;
if (script_set.load_from_file(filename.c_str())) {
unicharset.SetPropertiesFromOther(script_set);
}
// Load the xheights for the script if available.
filename = script_dir + "/" + unicharset.get_script_from_script_id(s) +
".xheights";
string script_heights;
if (File::ReadFileToString(filename, &script_heights))
xheights_str += script_heights;
}
if (!output_xheights_file.empty())
File::WriteStringToFileOrDie(xheights_str, output_xheights_file);
for (int c = SPECIAL_UNICHAR_CODES_COUNT; c < unicharset.size(); ++c) {
if (unicharset.PropertiesIncomplete(c)) {
tprintf("Warning: properties incomplete for index %d = %s\n",
c, unicharset.id_to_unichar(c));
}
}
// Write the output unicharset
tprintf("Writing unicharset to file %s\n", output_unicharset_file.c_str());
unicharset.save_to_file(output_unicharset_file.c_str());
}