void read_input_file( char *filename ) {
FILE *input = open_file_read( filename );
int r, t_len, c_len;
double count;
t_dict = g_hash_table_new_full( &g_int_hash, &g_int_equal,
NULL, destroy_word_count );
c_dict = g_hash_table_new_full( &g_int_hash, &g_int_equal,
NULL, destroy_word_count );
symbols_dict = g_hash_table_new_full( &g_str_hash, &g_str_equal,
free, free );
inv_symbols_dict = g_hash_table_new_full( &g_int_hash, &g_int_equal,
NULL, NULL );
while( !feof( input ) ) {
char *target_buff = NULL;
char *context_buff = NULL;
// Read the strings and count from the file
r = fscanf( input, "%ms %ms %lf", &target_buff, &context_buff, &count ); // Reads potentially large strings from file
// Shrink the strings to minimal memory, reject too large words
if( r == 3 ) {
t_len = strlen( target_buff );
c_len = strlen( context_buff );
if( t_len <= MAX_S && c_len <= MAX_S ) {
//perra( "Target:\"%s\" Context:\"%s\"\n",target,context );
insert_into_index( t_dict, symbols_dict, inv_symbols_dict, target_buff, context_buff, t_len, c_len, count, &idc_t );
insert_into_index( c_dict, symbols_dict, inv_symbols_dict, context_buff, target_buff, c_len, t_len, count, &idc_c );
n_pairs += count;
}
else {
perra( "Warning: Ignore (\"%s\",\"%s\")\n", target_buff,
context_buff );
perra( "Word length (%d,%d) => max is %d\n", t_len, c_len, MAX_S );
}
}
free( target_buff );
free( context_buff );
target_buff = NULL;
context_buff = NULL;
}
fclose( input );
}
void removeFromBufferAndIndex(int thread_number){
char * file_name;
FILE * file;
char buffer[512];
int line_number = 1;
// Waiting to remove from buffer
//printf("STOP REMOVE!\n");
pthread_mutex_lock(&mutex_one);
//printf("IN REMOVE LOCK\n");
while(buffer_amount == 0){
if(scanning_done){
pthread_mutex_unlock(&mutex_one);
return;
}
pthread_cond_wait(&fill_cond, &mutex_one);
//printf("STUCK IN REMOVE\n");
}
file_name = scan_buffer[buffer_amount - 1];
--buffer_amount;
//printf("--BUFFER_AMOUNT -> in removeFromBufferAndIndex\n");
pthread_cond_signal(&empty_cond);
pthread_mutex_unlock(&mutex_one);
// Has removed from buffer and signal to others who are waiting
//printf("Thread: %d File to be indexed: %s\n", thread_number, file_name);
if ((file = fopen(file_name, "r")) == NULL) {
//printf("Wasn't able to open: %s\n", file_name);
return;
}
while (!feof(file)) {
char * word;
char * saveptr;
fgets(buffer, 512, file);
word = strtok_r(buffer, " \n\t-_!@#$%^&*()_+=,./<>?", &saveptr);
while (word != NULL) {
//printf("Word inserted: %s\n",word);
//TESTING PURPOSES ONLY (using old index.c)
pthread_mutex_lock(&mutex_test);
insert_into_index(word, file_name, line_number);
//TESTING PURPOSES ONLY (using old index.c)
pthread_mutex_unlock(&mutex_test);
word = strtok_r(NULL, " \n\t-_!@#$%^&*()_+=,./<>?",&saveptr);
}
line_number = line_number+1;
}
fclose(file);
}
int main(int argc, char * argv[])
{
index_search_results_t * results;
init_index();
insert_into_index("hello", "test.c", 10);
insert_into_index("hello", "test.c", 20);
insert_into_index("hello", "foo.c", 30);
insert_into_index("goodbye", "test.c", 10);
results = find_in_index("hello");
if (results) {
int i;
for (i = 0; i < results->num_results; i++) {
printf("%s: %d\n",
results->results[i].file_name,
results->results[i].line_number);
}
free(results);
}
find_in_index("goodbye");
return(0);
}
