Tail *
tail_read (FILE *file)
{
long save_pos;
Tail *t;
TrieIndex i;
uint32 sig;
/* check signature */
save_pos = ftell (file);
if (!file_read_int32 (file, (int32 *) &sig) || TAIL_SIGNATURE != sig)
goto exit_file_read;
if (NULL == (t = (Tail *) malloc (sizeof (Tail))))
goto exit_file_read;
if (!file_read_int32 (file, &t->first_free) ||
!file_read_int32 (file, &t->num_tails))
{
goto exit_tail_created;
}
if (t->num_tails > SIZE_MAX / sizeof (TailBlock))
goto exit_tail_created;
t->tails = (TailBlock *) malloc (t->num_tails * sizeof (TailBlock));
if (!t->tails)
goto exit_tail_created;
for (i = 0; i < t->num_tails; i++) {
int16 length;
if (!file_read_int32 (file, &t->tails[i].next_free) ||
!file_read_int32 (file, &t->tails[i].data) ||
!file_read_int16 (file, &length))
{
goto exit_in_loop;
}
t->tails[i].suffix = (TrieChar *) malloc (length + 1);
if (length > 0) {
if (!file_read_chars (file, (char *)t->tails[i].suffix, length)) {
free (t->tails[i].suffix);
goto exit_in_loop;
}
}
t->tails[i].suffix[length] = '\0';
}
return t;
exit_in_loop:
while (i > 0) {
free (t->tails[--i].suffix);
}
free (t->tails);
exit_tail_created:
free (t);
exit_file_read:
fseek (file, save_pos, SEEK_SET);
return NULL;
}
Tail *
tail_open (const char *path, const char *name, TrieIOMode mode)
{
Tail *t;
TrieIndex i;
uint16 sig;
long file_size;
t = (Tail *) malloc (sizeof (Tail));
t->file = file_open (path, name, ".tl", mode);
if (!t->file)
goto exit1;
file_size = file_length (t->file);
if (file_size != 0 && file_read_int16 (t->file, (int16 *) &sig)
&& sig != TAIL_SIGNATURE)
{
goto exit2;
}
/* init tails data */
if (file_size == 0) {
t->first_free = 0;
t->num_tails = 0;
t->tails = NULL;
t->is_dirty = TRUE;
} else {
file_read_int16 (t->file, &t->first_free);
file_read_int16 (t->file, &t->num_tails);
t->tails = (TailBlock *) malloc (t->num_tails * sizeof (TailBlock));
if (!t->tails)
goto exit2;
for (i = 0; i < t->num_tails; i++) {
int8 length;
file_read_int16 (t->file, &t->tails[i].next_free);
file_read_int16 (t->file, &t->tails[i].data);
file_read_int8 (t->file, &length);
t->tails[i].suffix = (TrieChar *) malloc (length + 1);
if (length > 0)
file_read_chars (t->file, (char *)t->tails[i].suffix, length);
t->tails[i].suffix[length] = '\0';
}
t->is_dirty = FALSE;
}
return t;
exit2:
fclose (t->file);
exit1:
free (t);
return NULL;
}
static bool address_expansion_read(FILE *f, address_expansion_t *expansion) {
if (f == NULL) return false;
if (!file_read_uint32(f, (uint32_t *)&expansion->canonical_index)) {
return false;
}
uint32_t language_len;
if (!file_read_uint32(f, &language_len)) {
return false;
}
if (!file_read_chars(f, expansion->language, language_len)) {
return false;
}
if (!file_read_uint32(f, (uint32_t *)&expansion->num_dictionaries)) {
return false;
}
for (size_t i = 0; i < expansion->num_dictionaries; i++) {
if (!file_read_uint16(f, (uint16_t *)expansion->dictionary_ids + i)) {
return false;
}
}
if (!file_read_uint32(f, &expansion->address_components)) {
return false;
}
if (!file_read_uint8(f, (uint8_t *)&expansion->separable)) {
return false;
}
return true;
}
geodb_t *geodb_init(char *dir) {
if (dir == NULL) return NULL;
geodb_t *gdb = malloc(sizeof(geodb_t));
if (gdb == NULL) return NULL;
char_array *path = char_array_new_size(strlen(dir));
char_array_cat_joined(path, PATH_SEPARATOR, true, 2, dir, GEODB_NAMES_TRIE_FILENAME);
char *names_path = char_array_get_string(path);
gdb->names = trie_load(names_path);
if (gdb->names == NULL) {
goto exit_geodb_created;
}
char_array_clear(path);
char_array_cat_joined(path, PATH_SEPARATOR, true, 2, dir, GEODB_FEATURES_TRIE_FILENAME);
char *features_path = char_array_get_string(path);
gdb->features = trie_load(features_path);
if(gdb->features == NULL) {
goto exit_geodb_created;
}
char_array_clear(path);
char_array_cat_joined(path, PATH_SEPARATOR, true, 2, dir, GEODB_POSTAL_CODES_FILENAME);
char *postal_codes_path = char_array_get_string(path);
FILE *f = fopen(postal_codes_path, "rb");
uint64_t num_postal_strings = 0;
if (!file_read_uint64(f, (uint64_t *)&num_postal_strings)) {
goto exit_geodb_created;
}
size_t postal_codes_str_len;
if (!file_read_uint64(f, (uint64_t *)&postal_codes_str_len)) {
goto exit_geodb_created;
}
char_array *array = char_array_new_size(postal_codes_str_len);
if (!file_read_chars(f, array->a, postal_codes_str_len)) {
goto exit_geodb_created;
}
array->n = postal_codes_str_len;
gdb->postal_codes = cstring_array_from_char_array(array);
if (cstring_array_num_strings(gdb->postal_codes) != num_postal_strings) {
goto exit_geodb_created;
}
fclose(f);
char_array_clear(path);
char_array_cat_joined(path, PATH_SEPARATOR, true, 2, dir, GEODB_HASH_FILENAME);
char *hash_file_path = strdup(char_array_get_string(path));
char_array_clear(path);
char_array_cat_joined(path, PATH_SEPARATOR, true, 2, dir, GEODB_LOG_FILENAME);
char *log_path = char_array_get_string(path);
gdb->hash_reader = NULL;
if ((sparkey_hash_open(&gdb->hash_reader, hash_file_path, log_path)) != SPARKEY_SUCCESS) {
free(hash_file_path);
char_array_destroy(path);
goto exit_geodb_created;
}
free(hash_file_path);
char_array_destroy(path);
gdb->log_iter = NULL;
if ((sparkey_logiter_create(&gdb->log_iter, sparkey_hash_getreader(gdb->hash_reader))) != SPARKEY_SUCCESS) {
goto exit_geodb_created;
}
gdb->value_buf = char_array_new_size(sparkey_logreader_maxvaluelen(sparkey_hash_getreader(gdb->hash_reader)));
if (gdb->value_buf == NULL) {
goto exit_geodb_created;
}
gdb->geoname = geoname_new();
if (gdb->geoname == NULL) {
goto exit_geodb_created;
}
gdb->postal_code = gn_postal_code_new();
if (gdb->postal_code == NULL) {
goto exit_geodb_created;
}
return gdb;
exit_geodb_created:
geodb_destroy(gdb);
return NULL;
}
bool address_dictionary_read(FILE *f) {
if (address_dict != NULL) return false;
uint32_t signature;
if (!file_read_uint32(f, &signature) || signature != ADDRESS_DICTIONARY_SIGNATURE) {
return false;
}
address_dict = malloc(sizeof(address_dictionary_t));
if (address_dict == NULL) return false;
uint32_t canonical_str_len;
if (!file_read_uint32(f, &canonical_str_len)) {
goto exit_address_dict_created;
}
char_array *array = char_array_new_size(canonical_str_len);
if (array == NULL) {
goto exit_address_dict_created;
}
if (!file_read_chars(f, array->a, canonical_str_len)) {
char_array_destroy(array);
goto exit_address_dict_created;
}
array->n = canonical_str_len;
address_dict->canonical = cstring_array_from_char_array(array);
uint32_t num_values;
if (!file_read_uint32(f, &num_values)) {
goto exit_address_dict_created;
}
address_dict->values = address_expansion_value_array_new_size(num_values);
for (uint32_t i = 0; i < num_values; i++) {
address_expansion_value_t *value = address_expansion_value_read(f);
if (value == NULL) {
goto exit_address_dict_created;
}
address_expansion_value_array_push(address_dict->values, value);
}
address_dict->trie = trie_read(f);
if (address_dict->trie == NULL) {
goto exit_address_dict_created;
}
return true;
exit_address_dict_created:
address_dictionary_destroy(address_dict);
return false;
}