int
main(int argc, char *argv[])
{
logmath_t *lmath;
ngram_model_t *model;
/* Initialize a logmath object to pass to ngram_read */
lmath = logmath_init(1.0001, 0, 0);
/* Read a language model */
model = ngram_model_read(NULL, LMDIR "/100.arpa.DMP", NGRAM_BIN, lmath);
TEST_ASSERT(model);
ngram_model_casefold(model, NGRAM_UPPER);
TEST_EQUAL(0, strcmp("</s>", ngram_word(model, 5)));
TEST_EQUAL(0, strcmp("BE", ngram_word(model, 42)));
TEST_EQUAL(0, strcmp("FLOORED", ngram_word(model, 130)));
TEST_EQUAL(0, strcmp("ZERO", ngram_word(model, 398)));
TEST_EQUAL(0, strcmp("~", ngram_word(model, 399)));
ngram_model_casefold(model, NGRAM_LOWER);
TEST_EQUAL(0, strcmp("</s>", ngram_word(model, 5)));
TEST_EQUAL(0, strcmp("be", ngram_word(model, 42)));
TEST_EQUAL(0, strcmp("floored", ngram_word(model, 130)));
TEST_EQUAL(0, strcmp("zero", ngram_word(model, 398)));
TEST_EQUAL(0, strcmp("~", ngram_word(model, 399)));
ngram_model_free(model);
logmath_free(lmath);
return 0;
}
int
main(int argc, char *argv[])
{
cmd_ln_t *config;
ngram_model_t *lm = NULL;
logmath_t *lmath;
int itype, otype;
char const *kase;
if ((config = cmd_ln_parse_r(NULL, defn, argc, argv, TRUE)) == NULL)
return 1;
if (cmd_ln_boolean_r(config, "-help")) {
usagemsg(argv[0]);
}
err_set_debug_level(cmd_ln_int32_r(config, "-debug"));
/* Create log math object. */
if ((lmath = logmath_init
(cmd_ln_float64_r(config, "-logbase"), 0, 0)) == NULL) {
E_FATAL("Failed to initialize log math\n");
}
if (cmd_ln_str_r(config, "-i") == NULL || cmd_ln_str_r(config, "-i") == NULL) {
E_ERROR("Please specify both input and output models\n");
goto error_out;
}
/* Load the input language model. */
if (cmd_ln_str_r(config, "-ifmt")) {
if ((itype = ngram_str_to_type(cmd_ln_str_r(config, "-ifmt")))
== NGRAM_INVALID) {
E_ERROR("Invalid input type %s\n", cmd_ln_str_r(config, "-ifmt"));
goto error_out;
}
lm = ngram_model_read(config, cmd_ln_str_r(config, "-i"),
itype, lmath);
}
else {
lm = ngram_model_read(config, cmd_ln_str_r(config, "-i"),
NGRAM_AUTO, lmath);
}
/* Guess or set the output language model type. */
if (cmd_ln_str_r(config, "-ofmt")) {
if ((otype = ngram_str_to_type(cmd_ln_str_r(config, "-ofmt")))
== NGRAM_INVALID) {
E_ERROR("Invalid output type %s\n", cmd_ln_str_r(config, "-ofmt"));
goto error_out;
}
}
else {
otype = ngram_file_name_to_type(cmd_ln_str_r(config, "-o"));
}
/* Recode the language model if desired. */
if (cmd_ln_str_r(config, "-ienc")) {
if (ngram_model_recode(lm, cmd_ln_str_r(config, "-ienc"),
cmd_ln_str_r(config, "-oenc")) != 0) {
E_ERROR("Failed to recode language model from %s to %s\n",
cmd_ln_str_r(config, "-ienc"),
cmd_ln_str_r(config, "-oenc"));
goto error_out;
}
}
/* Case fold if requested. */
if ((kase = cmd_ln_str_r(config, "-case"))) {
if (0 == strcmp(kase, "lower")) {
ngram_model_casefold(lm, NGRAM_LOWER);
}
else if (0 == strcmp(kase, "upper")) {
ngram_model_casefold(lm, NGRAM_UPPER);
}
else {
E_ERROR("Unknown value for -case: %s\n", kase);
goto error_out;
}
}
/* Write the output language model. */
if (ngram_model_write(lm, cmd_ln_str_r(config, "-o"), otype) != 0) {
E_ERROR("Failed to write language model in format %s to %s\n",
ngram_type_to_str(otype), cmd_ln_str_r(config, "-o"));
goto error_out;
}
/* That's all folks! */
ngram_model_free(lm);
return 0;
error_out:
ngram_model_free(lm);
return 1;
}
