void calculate_ams_all_serial( word_count *casted_t, gpointer key_t ) {
double count_t_c;
gpointer key_c, value_t_c;
word_count *casted_c;
GHashTableIter iter_c;
g_hash_table_iter_init( &iter_c, casted_t->links );
casted_t->entropy = calculate_entropy(casted_t->count, casted_t->links);
while( g_hash_table_iter_next( &iter_c, &key_c, &value_t_c ) ){
casted_c = g_hash_table_lookup( c_dict, key_c );
count_t_c = *((double *)value_t_c);
calculate_and_print_am( (int *)key_t, casted_t->id, (int *)key_c,
casted_c->id, count_t_c, casted_t->count, casted_c->count,
casted_t->entropy, casted_c->entropy );
}
}
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;
}
double calculate_entropy_file(PE_FILE *pe)
{
unsigned int byte_count[256] = {0};
unsigned int n, size;
unsigned char buffer[1024];
n = size = 0;
rewind(pe->handle);
while((n = fread(buffer, 1, 1024, pe->handle)) != 0)
{
for (unsigned int i = 0; i < n; i++)
byte_count[(int) buffer[i]]++, size++;
}
return calculate_entropy(byte_count, size);
}
