scaling_bloom_t *new_scaling_bloom(unsigned int capacity, double error_rate, const char *filename)
{
scaling_bloom_t *bloom;
counting_bloom_t *cur_bloom;
int fd;
if ((fd = open(filename, O_RDWR | O_CREAT | O_TRUNC, (mode_t)0600)) < 0) {
perror("Error, Opening File Failed");
fprintf(stderr, " %s \n", filename);
return NULL;
}
bloom = scaling_bloom_init(capacity, error_rate, filename, fd);
if (!(cur_bloom = new_counting_bloom_from_scale(bloom))) {
fprintf(stderr, "Error, Could not create counting bloom\n");
free_scaling_bloom(bloom);
return NULL;
}
cur_bloom->header->count = 0;
cur_bloom->header->id = 0;
bloom->header->mem_seqnum = 1;
return bloom;
}
int test_scaling_accuracy(const char *bloom_file, const char *words_file)
{
FILE *fp;
char word[256];
scaling_bloom_t *bloom;
int i;
struct stats results = { 0 };
printf("\n* test scaling accuracy\n");
if ((fp = fopen(bloom_file, "r"))) {
fclose(fp);
remove(bloom_file);
}
if (!(bloom = new_scaling_bloom(CAPACITY, ERROR_RATE, bloom_file))) {
fprintf(stderr, "ERROR: Could not create bloom filter\n");
return TEST_FAIL;
}
if (!(fp = fopen(words_file, "r"))) {
fprintf(stderr, "ERROR: Could not open words file\n");
return TEST_FAIL;
}
for (i = 0; fgets(word, sizeof(word), fp); i++) {
if (i % 2 == 0) {
chomp_line(word);
scaling_bloom_add(bloom, word, strlen(word), i);
}
}
fseek(fp, 0, SEEK_SET);
for (i = 0; fgets(word, sizeof(word), fp); i++) {
if (i % 2 == 1) {
chomp_line(word);
bloom_score(scaling_bloom_check(bloom, word, strlen(word)), 0, &results, word);
}
}
fclose(fp);
printf("Elements added: %6d" "\n"
"Elements checked: %6d" "\n"
"Total size: %d KiB" "\n\n",
(i + 1) / 2, i / 2,
(int) bloom->num_bytes / 1024);
free_scaling_bloom(bloom);
return print_results(&results);
}
scaling_bloom_t *new_scaling_bloom_from_file(unsigned int capacity, double error_rate, const char *filename)
{
int fd;
off_t size;
uint64_t id;
unsigned int count, offset;
scaling_bloom_t *bloom;
counting_bloom_t *cur_bloom;
if ((fd = open(filename, O_RDWR, (mode_t)0600)) < 0) {
fprintf(stderr, "Error, Could not open file %s with open: \n", filename);
perror("");
return NULL;
}
if ((size = lseek(fd, 0, SEEK_END)) < 0) {
perror("Error, calling lseek() to tell file size");
close(fd);
return NULL;
}
if (size == 0) {
fprintf(stderr, "Error, File size zero\n");
}
bloom = scaling_bloom_init(capacity, error_rate, filename, fd);
offset = sizeof(scaling_bloom_header_t);
size -= offset;
while (size) {
id = ((counting_bloom_header_t *)(bloom->bitmap->array + offset))->id;
count = ((counting_bloom_header_t *)(bloom->bitmap->array + offset))->count;
cur_bloom = new_counting_bloom_from_scale(bloom, id, count);
size -= cur_bloom->num_bytes;
offset += cur_bloom->num_bytes;
if (size < 0) {
free_scaling_bloom(bloom);
fprintf(stderr, "Error, Actual filesize and expected filesize are not equal\n");
return NULL;
}
}
return bloom;
}
scaling_bloom_t *scaling_bloom_init(unsigned int capacity, double error_rate, const char *filename, int fd)
{
scaling_bloom_t *bloom;
if ((bloom = (scaling_bloom_t *)malloc(sizeof(scaling_bloom_t))) == NULL) {
return NULL;
}
if ((bloom->bitmap = new_bitmap(fd, sizeof(scaling_bloom_header_t))) == NULL) {
fprintf(stderr, "Error, Could not create bitmap with file\n");
free_scaling_bloom(bloom);
return NULL;
}
bloom->header = (scaling_bloom_header_t *) bloom->bitmap->array;
bloom->capacity = capacity;
bloom->error_rate = error_rate;
bloom->num_blooms = 0;
bloom->num_bytes = sizeof(scaling_bloom_header_t);
bloom->fd = fd;
bloom->blooms = NULL;
return bloom;
}
scaling_bloom_t *new_scaling_bloom_from_file(unsigned int capacity, double error_rate, const char *filename)
{
int fd;
off_t size;
scaling_bloom_t *bloom;
counting_bloom_t *cur_bloom;
if ((fd = open(filename, O_RDWR, (mode_t)0600)) < 0) {
fprintf(stderr, "Error, Could not open file %s: %s\n", filename, strerror(errno));
return NULL;
}
if ((size = lseek(fd, 0, SEEK_END)) < 0) {
perror("Error, calling lseek() to tell file size");
close(fd);
return NULL;
}
if (size == 0) {
fprintf(stderr, "Error, File size zero\n");
}
bloom = scaling_bloom_init(capacity, error_rate, filename, fd);
size -= sizeof(scaling_bloom_header_t);
while (size) {
cur_bloom = new_counting_bloom_from_scale(bloom);
// leave count and id as they were set in the file
size -= cur_bloom->num_bytes;
if (size < 0) {
free_scaling_bloom(bloom);
fprintf(stderr, "Error, Actual filesize and expected filesize are not equal\n");
return NULL;
}
}
return bloom;
}
static void Dablooms_dealloc(Dablooms *self)
{
free_scaling_bloom(self->filter);
self->ob_type->tp_free((PyObject *)self);
}
void Dablooms_free(DabloomStruct* self)
{
free_scaling_bloom(self->filter);
free(self);
}
