int mode_filter(int argc, char** argv)
{
uint argo = 0;
uint negative = FALSE;
if (strcmp(argv[2 + argo], "-n") == 0) {
negative = TRUE;
argo++;
}
uint update = FALSE;
if (strcmp(argv[2 + argo], "-u") == 0) {
update = TRUE;
argo++;
}
char* bloom_file = argv[2 + argo];
bf_t* filter = bf_load(bloom_file, BLOOM_HASH_FUNC);
if (!filter) {
printf("File %s has wrong format or doesn't exist.\n", bloom_file);
return 2;
}
char input_buffer[BF_KEY_BUFFER_SIZE] = { 0 };
uint actual_length = 0;
while (gets_skip(input_buffer, BF_KEY_BUFFER_SIZE, &actual_length)) {
uint has = bf_has(filter, (const char*)input_buffer, actual_length);
if (negative) {
if (!has) {
printf("%s\n", input_buffer);
fflush(stdout);
if (update) {
bf_add(filter, input_buffer, actual_length);
}
}
} else {
if (has) {
printf("%s\n", input_buffer);
fflush(stdout);
} else {
if (update) {
bf_add(filter, input_buffer, actual_length);
}
}
}
}
if (update) {
bf_save(filter, bloom_file);
}
bf_destroy(filter);
return 0;
}
int mode_info(int argc, char** argv)
{
char* bloom_file = argv[2];
bf_t* filter = bf_load(bloom_file, BLOOM_HASH_FUNC);
if (!filter) {
printf("File %s has wrong format or doesn't exist.\n", bloom_file);
return 2;
}
printf("capacity: %ld\n", filter->bits_count);
printf("used: %ld\n", filter->bits_used);
printf("error rate: %.3f%%\n", exp(- filter->bits_count / (float)filter->bits_used * BF_LOG2_2));
printf("hash/item: %d\n", filter->hashes_count);
bf_destroy(filter);
}
int mode_create(int argc, char** argv)
{
uint argo = 0;
uint progress = FALSE;
if (strcmp(argv[2], "-p") == 0) {
argo++;
progress = TRUE;
}
if (argc - argo < 5) {
show_help(argv[0]);
return 1;
}
char* bloom_file = argv[2 + argo];
double error_rate = atof(argv[3 + argo]);
bf_index_t key_count = atol(argv[4 + argo]);
bf_t* filter = bf_create(error_rate, key_count, BLOOM_HASH_FUNC);
char input_buffer[BF_KEY_BUFFER_SIZE] = { 0 };
bf_index_t key = 0;
uint actual_length;
while (gets_skip(input_buffer, BF_KEY_BUFFER_SIZE, &actual_length)) {
bf_add(filter, (const char*)input_buffer, actual_length);
key = key + 1;
if (progress && key % 1000000 == 0) {
printf("%s: %s <- %ld\n", argv[0], bloom_file, key);
}
}
bf_save(filter, bloom_file);
bf_destroy(filter);
return 0;
}
int mode_merge(int argc, char** argv)
{
if (argc < 4) {
show_help(argv[0]);
return 1;
}
char* bloom_file = argv[3];
bf_t* filter = bf_load(bloom_file, BLOOM_HASH_FUNC);
if (!filter) {
printf("File %s has wrong format or doesn't exist.\n", bloom_file);
return 2;
}
bf_t* merge_with;
uint file_index;
for (file_index = 4; file_index < argc; file_index++) {
merge_with = bf_load(argv[file_index], BLOOM_HASH_FUNC);
if (!merge_with) {
printf("File %s has wrong format or doesn't exist.\n", argv[file_index]);
return 2;
}
bf_merge(filter, merge_with);
bf_destroy(merge_with);
}
char* merged_bloom_file = argv[2];
bf_save(filter, merged_bloom_file);
return 0;
}
int
main(
int argc,
char ** argv )
{
prf_model_t * model;
bfile_t * bfile;
int i;
if ( argc <= 3 ) {
printf( "Usage: %s <inputfile> <outputfile> <opcode> [<opcode> ...]\n",
argv[0] );
return -1;
}
bfile = bf_create_r( argv[1] );
if ( ! bfile ) {
printf( "Error: Could not open '%s' for loading.\n", argv[1] );
return -1;
}
prf_init();
model = prf_model_create();
assert( model != NULL );
prf_model_poolmem( model );
if ( ! prf_model_load( model, bfile ) ) {
printf( "Error: Couldn't parse model.\n" );
bf_destroy( bfile );
prf_exit();
return -1;
}
bf_destroy( bfile );
for ( i = 3; i < argc; i++ ) {
prf_nodeinfo_t * info;
info = prf_nodeinfo_get( atoi( argv[i] ) );
if ( ! info ) {
fprintf( stderr, "Warning: couldn't find nodeinfo for '%s'\n",
argv[i] );
} else {
info->save_f = dont_save;
}
}
bfile = bf_create_w( argv[2] );
if ( ! bfile ) {
printf( "Error: Could not open '%s' for writing.\n", argv[2] );
prf_model_destroy( model );
prf_exit();
return -1;
}
prf_model_save( model, bfile );
bf_destroy( bfile );
prf_model_destroy( model );
prf_exit();
return 0;
} /* main() */
