END_TEST
START_TEST (initialize_using_lang_code)
{
int rc;
char *errMsg = NULL;
varnam *handle;
strbuf *tmp;
rc = varnam_init_from_id ("ml", &handle, &errMsg);
if (errMsg != NULL) {
printf ("init_from_lang failed: %s\n", errMsg);
}
assert_success (rc);
tmp = strbuf_init (20);
strbuf_addf (tmp, "%s/ml.vst", strbuf_to_s (varnam_get_symbols_dir()));
ck_assert_str_eq (strbuf_to_s(tmp), varnam_get_scheme_file (handle));
strbuf_destroy (tmp);
tmp = strbuf_init (10);
strbuf_addf (tmp, "%s/.local/share/varnam/suggestions/ml.vst.learnings", getenv ("HOME"));
ck_assert_str_eq (strbuf_to_s (tmp), varnam_get_suggestions_file (handle));
strbuf_destroy (tmp);
varnam_destroy (handle);
}
int
execute_query_int (sqlite3* db, const char* sql)
{
int rc, result;
sqlite3_stmt* stmt;
strbuf* error;
rc = sqlite3_prepare_v2 (db, sql, -1, &stmt, NULL);
if (rc != SQLITE_OK) {
error = strbuf_init (50);
strbuf_addf (error, "Failed to prepare query: %s. Return code was: %d\n", sql, rc);
ck_abort_msg (strbuf_to_s (error));
}
rc = sqlite3_step (stmt);
if (rc != SQLITE_ROW) {
error = strbuf_init (50);
strbuf_addf (error, "Failed to execute query: %s. Return code was: %d\n", sql, rc);
ck_abort_msg (strbuf_to_s (error));
}
result = sqlite3_column_int (stmt, 0);
sqlite3_finalize (stmt);
return result;
}
static int
varnam_learn_internal(varnam *handle, const char *word, int confidence)
{
int rc;
varray *tokens;
strbuf *sanitized_word;
if (handle == NULL || word == NULL)
return VARNAM_ARGS_ERROR;
if (!is_words_store_available(handle)) {
return VARNAM_ERROR;
}
if (!is_utf8 (word)) {
set_last_error (handle, "Incorrect encoding. Expected UTF-8 string");
return VARNAM_ERROR;
}
tokens = get_pooled_array (handle);
/* This removes all starting and trailing special characters from the word */
sanitized_word = sanitize_word (handle, word);
rc = vst_tokenize (handle, strbuf_to_s (sanitized_word), VARNAM_TOKENIZER_VALUE, VARNAM_MATCH_ALL, tokens);
if (rc) return rc;
#ifdef _VARNAM_VERBOSE
printf ("%s\n", "Tokens before reducing noice");
print_tokens_array (tokens);
#endif
/* Tokens may contain more data that we can handle. Reducing noice so that we learn most relevant combinations */
reduce_noise_in_tokens (tokens);
#ifdef _VARNAM_VERBOSE
printf ("%s\n", "Tokens after reducing noice");
print_tokens_array (tokens);
#endif
if (!can_learn_from_tokens (handle, tokens, strbuf_to_s (sanitized_word)))
return VARNAM_ERROR;
return vwt_persist_possibilities (handle,
tokens,
strbuf_to_s (sanitized_word),
confidence);
}
static strbuf*
sanitize_word (varnam *handle, const char *word)
{
size_t i;
bool is_special = false;
strbuf *string, *to_remove;
string = get_pooled_string (handle);
to_remove = get_pooled_string (handle);
strbuf_add (string, word);
for (i = 0; i < string->length; i++)
{
is_special = is_special_character (string->buffer[i]);
if (is_special)
strbuf_addc (to_remove, string->buffer[i]);
else
break;
}
strbuf_remove_from_first (string, strbuf_to_s (to_remove));
strbuf_clear (to_remove);
for (i = string->length - 1; i >= 0; i--)
{
is_special = is_special_character (string->buffer[i]);
if (is_special)
strbuf_addc (to_remove, string->buffer[i]);
else
break;
}
strbuf_remove_from_last (string, strbuf_to_s (to_remove));
language_specific_sanitization (string);
/* Remove trailing ZWNJ and leading ZWJ */
strbuf_remove_from_first (string, ZWNJ());
strbuf_remove_from_last (string, ZWNJ());
strbuf_remove_from_first (string, ZWJ());
return string;
}
END_TEST
START_TEST (numbers_will_be_ignored_while_learning)
{
int rc;
strbuf *string;
rc = varnam_learn (varnam_instance, "01");
assert_error (rc);
string = strbuf_init (50);
strbuf_add (string, "Can't process '0'. One or more characters in '01' are not known");
ck_assert_str_eq (varnam_get_last_error (varnam_instance), strbuf_to_s (string));
rc = varnam_learn (varnam_instance, "१०१");
assert_error (rc);
strbuf_clear (string);
strbuf_add (string, "Nothing to learn from '१०१'");
ck_assert_str_eq (varnam_get_last_error (varnam_instance), strbuf_to_s (string));
strbuf_destroy (string);
}
void
assert_error (int value)
{
strbuf *string = NULL;
if (value != VARNAM_ERROR) {
string = strbuf_init (50);
strbuf_addf (string, "Expected VARNAM_ERROR, but got %d. %s", value,
varnam_get_last_error (varnam_instance));
ck_abort_msg (strbuf_to_s (string));
}
}
END_TEST
START_TEST (words_with_repeating_characters_will_not_be_learned)
{
int rc;
strbuf *string;
const char *word_to_learn = "കകകകകകക";
rc = varnam_learn (varnam_instance, word_to_learn);
assert_error (rc);
string = strbuf_init (50);
strbuf_addf (string, "'%s' looks incorrect. Not learning anything", word_to_learn);
ck_assert_str_eq (varnam_get_last_error (varnam_instance), strbuf_to_s (string));
strbuf_destroy (string);
}
static const char*
get_scheme_details(varnam *handle, const char* key, struct strbuf *buffer)
{
struct strbuf *value = buffer;
if (handle == NULL)
return NULL;
if (strbuf_is_blank (value))
{
vst_get_metadata (handle, key, value);
}
return strbuf_to_s (value);
}
void
ensure_word_list_contains(varray *words, const char *word)
{
int i = 0, found = 0;
vword *w;
strbuf *error;
for (i = 0; i < varray_length (words); i++) {
w = varray_get (words, i);
if (strcmp (w->text, word) == 0) {
found = 1;
break;
}
}
if (!found) {
error = strbuf_init (50);
strbuf_addf (error, "Expected word list to contain '%s'", word);
ck_abort_msg (strbuf_to_s (error));
}
}
void
reinitialize_varnam_instance(const char *filename)
{
int rc;
char *msg;
varnam *handle;
strbuf *error;
if (varnam_instance != NULL) {
varnam_destroy (varnam_instance);
varnam_instance = NULL;
}
rc = varnam_init (filename, &handle, &msg);
if(rc != VARNAM_SUCCESS) {
error = strbuf_init (50);
strbuf_addf (error, "Varnam initialization failed. %s. %s", filename, msg);
ck_abort_msg (strbuf_to_s (error));
}
varnam_instance = handle;
}
END_TEST
START_TEST (varnam_export_full)
{
int rc, pcnt, wcnt, i;
float filecnt;
strbuf* f; strbuf* error;
f = strbuf_init (20);
pcnt = execute_query_int (varnam_instance->internal->known_words, "select count(*) from patterns_content;");
wcnt = execute_query_int (varnam_instance->internal->known_words, "select count(*) from words;");
rc = varnam_export_words (varnam_instance, 2, "output/", VARNAM_EXPORT_FULL, NULL);
assert_success (rc);
filecnt = pcnt / 2;
for (i = 0; i < (int) ceil (filecnt); i++) {
strbuf_clear (f);
strbuf_addf (f, "output/%d.patterns.txt", i);
if (!file_exist (strbuf_to_s (f))) {
error = strbuf_init (10);
strbuf_addf (error, "Failed to find file: %s\n", strbuf_to_s (f));
ck_abort_msg (strbuf_to_s (error));
}
}
filecnt = wcnt / 2;
for (i = 0; i < (int) ceil (filecnt); i++) {
strbuf_clear (f);
strbuf_addf (f, "output/%d.words.txt", i);
if (!file_exist (strbuf_to_s (f))) {
error = strbuf_init (10);
strbuf_addf (error, "Failed to find file: %s\n", strbuf_to_s (f));
ck_abort_msg (strbuf_to_s (error));
}
}
strbuf_destroy (f);
}
int
varnam_learn_from_file(varnam *handle,
const char *filepath,
vlearn_status *status,
void (*callback)(varnam *handle, const char *word, int status_code, void *object),
void *object)
{
int rc;
FILE *infile;
char line_buffer[10000];
strbuf *word;
varray *word_parts;
int confidence;
int parts;
infile = fopen(filepath, "r");
if (!infile) {
set_last_error (handle, "Couldn't open file '%s' for reading.\n", filepath);
return VARNAM_ERROR;
}
if (status != NULL)
{
status->total_words = 0;
status->failed = 0;
}
rc = vwt_optimize_for_huge_transaction(handle);
if (rc) {
fclose (infile);
return rc;
}
/* Learning from file will be mostly new words. Optimizing for that */
v_->_config_mostly_learning_new_words = 1;
varnam_log (handle, "Starting to learn from %s", filepath);
rc = vwt_start_changes (handle);
if (rc) {
vwt_turn_off_optimization_for_huge_transaction(handle);
fclose (infile);
return rc;
}
while (fgets(line_buffer, sizeof(line_buffer), infile))
{
reset_pool (handle);
word = get_pooled_string (handle);
strbuf_add (word, trimwhitespace (line_buffer));
word_parts = strbuf_split (word, handle, ' ');
parts = varray_length (word_parts);
if (parts > 0 && parts <= 2)
{
confidence = 1;
if (parts == 2) {
word = varray_get (word_parts, 1);
confidence = atoi (strbuf_to_s (word));
}
word = varray_get (word_parts, 0);
rc = varnam_learn_internal (handle, strbuf_to_s (word), confidence);
if (rc) {
if (status != NULL) status->failed++;
}
}
else {
rc = VARNAM_ERROR;
if (status != NULL) status->failed++;
}
if (status != NULL) status->total_words++;
if (callback != NULL) callback (handle, strbuf_to_s (word), rc, object);
}
varnam_log (handle, "Writing changes to disk");
rc = vwt_end_changes (handle);
if (rc) {
varnam_log (handle, "Writing changes to disk failed");
}
varnam_log (handle, "Ensuring file integrity");
rc = vwt_turn_off_optimization_for_huge_transaction(handle);
if (rc) {
varnam_log (handle, "Failed to check file integrity");
}
varnam_log (handle, "Compacting file");
rc = vwt_compact_file (handle);
if (rc) return rc;
fclose (infile);
return rc;
}
