/**
* Init the Sally tool
* @param argc number of arguments
* @param argv arguments
*/
static void sally_init()
{
int ehash;
const char *cfg_str;
if (verbose > 1)
config_print(&cfg);
/* Set delimiters */
config_lookup_string(&cfg, "features.ngram_delim", &cfg_str);
if (strlen(cfg_str) > 0)
fvec_delim_set(cfg_str);
/* Check for TFIDF weighting */
config_lookup_string(&cfg, "features.vect_embed", &cfg_str);
if (!strcasecmp(cfg_str, "tfidf"))
idf_create(input);
/* Load stop words */
config_lookup_string(&cfg, "input.stopword_file", &cfg_str);
if (strlen(cfg_str) > 0)
stopwords_load(cfg_str);
/* Check for feature hash table */
config_lookup_int(&cfg, "features.explicit_hash", &ehash);
config_lookup_string(&cfg, "features.hash_file", &cfg_str);
if (ehash || strlen(cfg_str) > 0) {
info_msg(1, "Enabling feature hash table.");
fhash_init();
}
/* Open input */
config_lookup_string(&cfg, "input.input_format", &cfg_str);
input_config(cfg_str);
info_msg(1, "Opening '%0.40s' with input module '%s'.", input, cfg_str);
entries = input_open(input);
if (entries < 0)
fatal("Could not open input source");
/* Open output */
config_lookup_string(&cfg, "output.output_format", &cfg_str);
output_config(cfg_str);
info_msg(1, "Opening '%0.40s' with output module '%s'.", output, cfg_str);
if (!output_open(output))
fatal("Could not open output destination");
}
/**
* Main function
*/
int main(int argc, char **argv)
{
int err = FALSE;
/* Create config */
config_init(&cfg);
config_check(&cfg);
fhash_init();
err |= test_sorted_ngrams();
err |= test_blended_ngrams();
err |= test_pos_ngrams();
fhash_destroy();
config_destroy(&cfg);
return err;
}
