int main( int argc, char *argv[] ) {
word_count *casted_c;
GHashTableIter iter_c;
//int *id_t, *id_c;
gpointer key_c, value_t_c, value_t, key_t;
if( argc != 2 ) { usage(); }
perr( "Reading input file into hashmap...\n" );
read_input_file( argv[1] );
// File header
perr( "Calculating association scores...\n" );
printf( "target\tid_target\tcontext\tid_context\tf_tc\tf_t\tf_c\t" );
printf( "cond_prob\tpmi\tnpmi\tlmi\ttscore\tzscore\tdice\tchisquare\t" );
printf( "loglike\taffinity\tentropy_target\tentropy_context\n" );
// First calculate all entropies for contexts
g_hash_table_iter_init( &iter_c, c_dict );
while( g_hash_table_iter_next( &iter_c, &key_c, &value_t_c ) ){
casted_c = g_hash_table_lookup( c_dict, key_c );
casted_c->entropy = calculate_entropy(casted_c->count, casted_c->links);
}
g_hash_table_iter_init( &iter_t, t_dict );
nb_targets = g_hash_table_size( t_dict );
if( nb_threads > 1 ) {
run_multi_threaded( &calculate_ams_all, nb_threads );
}
else {
while( g_hash_table_iter_next( &iter_t, &key_t, &value_t ) ){
calculate_ams_all_serial( (word_count *)value_t, key_t );
update_count();
}
}
// Clean and free to avoid memory leaks
perr( "Finished, cleaning up...\n" );
g_hash_table_destroy( t_dict );
g_hash_table_destroy( c_dict );
// MUST be last to be destroyed, otherwise will destroy keys in previous dicts
// and memory will leak from unreachable values
g_hash_table_destroy( symbols_dict );
g_hash_table_destroy( inv_symbols_dict ); // no effect
perra( "Number of targets: %d\n", idc_t );
perra( "Number of contexts: %d\n", idc_c );
perr( "You can now calculate similarities with command below\n");
perr( " ./calculate_similarity [OPTIONS] <out-file>\n\n" );
return 0;
}
int main( int argc, char *argv[] ) {
int argindex = treat_options( argc, argv );
if( argindex != argc - 2 ){
perr( "You must provide two filenames as arguments\n" );
usage();
}
perr( "Reading input profiles into hashmap...\n" );
read_profiles_file( argv[ argindex + 1 ] );
perr( "Generalizing...\n" );
sim_file = open_file_read( argv[ argindex ] );
sim_index = get_index_column_name( sim_file, score_name );
if( sim_index < 4 ) {
fprintf( stderr, "Column named \"%s\" not found!\nYou must ", score_name );
perr( "specify a valid -s option. Chose among SIM-FILE column headers.\n");
perr( "Remember that the 1st 4 fields cannot be used as scores.\n" );
usage();
}
if( nb_threads > 1 ) {
run_multi_threaded( &read_sim_and_generalize, nb_threads );
}
else {
perr( "Not using threads\n" );
read_sim_and_generalize_serial();
}
// Clean and free to avoid memory leaks
perr( "Finished, cleaning up...\n" );
fclose( sim_file );
g_hash_table_destroy( c_dict );
g_hash_table_destroy( t_dict );
if( t_filter ) {
g_hash_table_destroy( t_filter );
}
if( n_filter ) {
g_hash_table_destroy( n_filter );
}
if( c_filter ) {
g_hash_table_destroy( c_filter );
}
return 0;
}
void run() {
run_single_threaded();
run_multi_threaded();
}