static void set_tag(void* data, int n, const char * tag, int* error)
{
s_erc tmperror;
SItem** nodeList = data;
SItemSetString(nodeList[n], "POS", tag, &tmperror);
if (S_CHK_ERR(&tmperror, S_CONTERR,
"set_tag",
"Call to \"SItemSetString\" failed"))
*error = 1;
}
static void s_compute_stresses ( const SFeatProcessor* proc, SItem* word, s_erc *error )
{
S_CLR_ERR(error);
SItem *wordAsSylStructure = SItemAs(word, "SylStructure", error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_stresses",
"Call to \"SItemAs\" failed"))
return;
SItem *syllable = SItemDaughter(wordAsSylStructure, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_stresses",
"Call to \"SItemDaughter\" failed"))
return;
SObject* result = NULL;
while (syllable != NULL)
{
result = SFeatProcessorRun ( proc, syllable, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_stresses",
"Call to \"SItemPathToFeatProc\" failed"))
goto s_compute_stresses_cleanup;
const char* resultString = SObjectGetString ( result, error );
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_stresses",
"Call to \"SObjectGetInt\" failed"))
goto s_compute_stresses_cleanup;
SItemSetString ( syllable, "stress", resultString, error );
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_stresses",
"Call to \"SItemSetInt\" failed"))
goto s_compute_stresses_cleanup;
syllable = SItemNext ( syllable, error );
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_stresses",
"Call to \"SItemNext\" failed"))
goto s_compute_stresses_cleanup;
if(result != NULL) {
S_DELETE(result, "s_compute_stresses", error);
}
}
s_compute_stresses_cleanup:
if(result != NULL) {
S_DELETE(result, "s_compute_stresses", error);
}
}
static void Run(const SUttProcessor *self, SUtterance *utt,
s_erc *error)
{
const SRelation *wordRel;
SItem *wordItem;
const SItem *tokenItem;
s_bool is_present;
const char *end_punc;
SRelation *phraseRelation = NULL;
SItem *phraseItem = NULL;
const char *post_punc;
SRelation *sentenceRelation = NULL;
SItem *sentenceItem = NULL;
S_CLR_ERR(error);
/* we require the word relation */
is_present = SUtteranceRelationIsPresent(utt, "Word", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceRelationIsPresent\" failed"))
goto quit_error;
if (!is_present)
{
S_CTX_ERR(error, S_FAILURE,
"Run",
"Failed to find 'Word' relation in utterance");
goto quit_error;
}
wordRel = SUtteranceGetRelation(utt, "Word", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceGetRelation\" failed"))
goto quit_error;
/* get phrasing symbols */
s_get_phrasing_symbols(self, &end_punc, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"s_get_phrasing_symbols\" failed"))
goto quit_error;
/* create Phrase relation */
phraseRelation = SUtteranceNewRelation(utt, "Phrase", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceNewRelation\" failed"))
goto quit_error;
/* create Sentence relation */
sentenceRelation = SUtteranceNewRelation(utt, "Sentence", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceNewRelation\" failed"))
goto quit_error;
/* start at the first item in the word relation, cast away
* const, we want to add daughter items
*/
wordItem = (SItem*)SRelationHead(wordRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationHead\" failed"))
goto quit_error;
while (wordItem != NULL)
{
SItem *lastWordInToken;
SItem *wordAsToken;
if (phraseItem == NULL)
{
/* if phrase item is NULL, create a new phrase item (NULL
* shared content) that is appended to phrase
* relation. Will happen in first pass.
*/
sentenceItem = SRelationAppend(sentenceRelation, NULL, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationAppend\" failed"))
goto quit_error;
/* Added on top a sentence item, for now is one on one with the phrase item
* */
phraseItem = SRelationAppend(phraseRelation, NULL, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationAppend\" failed"))
goto quit_error;
/* add an item name, NB, no break */
SItemSetString(phraseItem, "name", "NB", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemSetString\" failed"))
goto quit_error;
SItemAddDaughter(sentenceItem, phraseItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAddDaughter\" failed"))
goto quit_error;
}
/* Create a daughter item for the phrase item. Shared content
* is the word item.
*/
SItemAddDaughter(phraseItem, wordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAddDaughter\" failed"))
goto quit_error;
/* get word as in Token relation */
wordAsToken = SItemAs(wordItem, "Token", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Failed to get word item's as in Token relation"))
goto quit_error;
/*
* get word's token which is the parent of wordAsToken.
*/
tokenItem = SItemParent(wordAsToken, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Failed to get word item's token item"))
goto quit_error;
/* get last word in token */
lastWordInToken = SItemLastDaughter(tokenItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Failed to get last daughter of token item"))
goto quit_error;
/* check if the next token is punctuation */
is_present = FALSE;
tokenItem = SItemNext(tokenItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
if (tokenItem != NULL)
{
is_present = SItemFeatureIsPresent(tokenItem, "IsPunctuation", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemFeatureIsPresent\" failed"))
goto quit_error;
if (is_present)
{
sint32 value = SItemGetInt(tokenItem, "IsPunctuation", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemGetInt\" failed"))
goto quit_error;
is_present = (value > 0);
}
}
if ((is_present) && (wordAsToken == lastWordInToken))
{
char *ptr;
post_punc = SItemGetName(tokenItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemGetName\" failed"))
goto quit_error;
/* check if it is in the end_punc list */
ptr = s_strpbrk(post_punc, end_punc, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"s_strpbrk\" failed"))
goto quit_error;
if (ptr != NULL)
{
/* add a phrase break */
SItemSetString(phraseItem, "name", "BB", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemSetString\" failed"))
goto quit_error;
/* set to NULL so that a new phrase item is created */
phraseItem = NULL;
}
}
wordItem = SItemNext(wordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
/* here all is OK */
return;
/* error clean-up code */
quit_error:
if (phraseRelation != NULL)
{
SUtteranceDelRelation(utt, "Phrase", error);
S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceDelRelation\" failed");
}
}
static void Run(const SUttProcessor *self, SUtterance *utt,
s_erc *error)
{
SCrfSuiteUttProc *crfsuiteProc = (SCrfSuiteUttProc*)self;
crfsuite_model_t * ptr_model = malloc (sizeof(crfsuite_tagger_t));
crfsuite_tagger_t * ptr_tagger = malloc (sizeof(crfsuite_tagger_t));
crfsuite_dictionary_t * ptr_attrs = malloc (sizeof(crfsuite_dictionary_t));
crfsuite_dictionary_t * ptr_labels = malloc (sizeof(crfsuite_dictionary_t));
crfsuite_instance_t * instance = malloc (sizeof(crfsuite_instance_t));
/* Initialize model object */
if ( crfsuite_create_instance_from_file( crfsuiteProc->model_file,
(void**)&ptr_model
) != 0 ) {
goto exit_cleanup;
}
const SRelation* phrase = SUtteranceGetRelation(utt, "Phrase", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceGetRelation\" failed"))
return;
SItem* itrPhrase = SRelationHead( phrase, error );
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationHead\" failed"))
return;
while ( itrPhrase != NULL )
{
/* Obtain the dictionary interface representing the labels in the model. */
if ( ptr_model->get_labels(ptr_model, &ptr_labels) != 0) {
goto exit_cleanup;
}
/* Obtain the dictionary interface representing the attributes in the model. */
if ( ptr_model->get_attrs(ptr_model, &ptr_attrs) != 0) {
goto exit_cleanup;
}
/* Obtain the tagger interface. */
if ( ptr_model->get_tagger(ptr_model, &ptr_tagger) != 0) {
goto exit_cleanup;
}
instance = create_phrase_instance ( itrPhrase, ptr_attrs, ptr_labels, error );
int *output = calloc(sizeof(int), instance->num_items);
floatval_t score = 0;
/* Set the instance to the tagger. */
if ( ptr_tagger->set(ptr_tagger, instance) != 0) {
goto exit_cleanup;
}
/* Obtain the viterbi label sequence. */
if (ptr_tagger->viterbi(ptr_tagger, output, &score) != 0) {
goto exit_cleanup;
}
/* Extract the output and insert in the POS attribute */
const SItem* tokenTMP = SItemPathToItem ( itrPhrase, "daughter.R:Token", error );
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemPathToItem\" failed"))
return;
SItem* token = SItemParent (tokenTMP, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemParent\" failed"))
return;
const SItem* lastToken = SItemPathToItem ( itrPhrase, "n.daughter.R:Token.parent", error );
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemPathToItem\" failed"))
return;
int i = 0;
while ( token != NULL && token != lastToken )
{
const char * str = malloc (sizeof (char)*16);
ptr_labels->to_string (ptr_labels, output[i], &str);
i += 1;
SItemSetString (token, "POS", str, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemSetString\" failed"))
return;
token = SItemNext(token, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
return;
}
free(output);
crfsuite_instance_finish(instance);
itrPhrase = SItemNext(itrPhrase, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
return;
}
/* here all is OK */
S_UNUSED(utt);
exit_cleanup:
if ( ptr_model != NULL )
free ( ptr_model );
if ( ptr_tagger != NULL )
free ( ptr_tagger );
if ( ptr_attrs != NULL )
free ( ptr_attrs );
if ( ptr_labels != NULL )
free ( ptr_labels );
if ( instance != NULL )
free ( instance );
}
static char *s_add_token_to_relation(SRelation *tokenRel, const SToken *token,
const char *token_string, s_erc *error)
{
SItem *tokenItem;
const char *tmp;
char *post_punc;
S_CLR_ERR(error);
/* create item, NULL shared content */
tokenItem = SRelationAppend(tokenRel, NULL, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_add_token_to_relation",
"Call to \"SRelationAppend\" failed"))
return NULL;
/* item's name is the token string */
SItemSetName(tokenItem, token_string, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_add_token_to_relation",
"Call to \"SItemSetName\" failed"))
return NULL;
/* get white-space */
tmp = STokenGetWhitespace(token, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_add_token_to_relation",
"Call to \"STokenGetWhitespace\" failed"))
return NULL;
if (tmp != NULL)
{
SItemSetString(tokenItem, "whitespace", tmp, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_add_token_to_relation",
"Call to \"SItemSetString\" failed"))
return NULL;
}
/* get pre-punctuation */
tmp = STokenGetPrePunc(token, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_add_token_to_relation",
"Call to \"STokenGetPrePunc\" failed"))
return NULL;
if (tmp != NULL)
{
SItemSetString(tokenItem, "prepunc", tmp, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_add_token_to_relation",
"Call to \"SItemSetString\" failed"))
return NULL;
}
/* get post-punctuation */
tmp = STokenGetPostPunc(token, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_add_token_to_relation",
"Call to \"STokenGetPostPunc\" failed"))
return NULL;
if (tmp != NULL)
{
SItemSetString(tokenItem, "postpunc", tmp, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_add_token_to_relation",
"Call to \"SItemSetString\" failed"))
return NULL;
}
post_punc = s_strdup(tmp, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_add_token_to_relation",
"Call to \"s_strdup\" failed"))
return NULL;
return post_punc; /* return the post punctuation */
}
static void s_compute_phonetic_features (SItem* word, s_erc *error )
{
SItem *syllable;
SItem * phone;
char* position_in_syllable_string = NULL;
/* Extract Phoneset from Voice*/
const SVoice* voice = SItemVoice (word, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"SItemGetVoice\" failed"))
return;
const SPhoneset* phoneset = (SPhoneset*)SVoiceGetData(voice, "phoneset", error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"SVoiceGetData\" failed"))
return;
SItem *wordAsSylStructure = SItemAs(word, "SylStructure", error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_stresses",
"Call to \"SItemAs\" failed"))
return;
syllable = SItemDaughter(wordAsSylStructure, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_stresses",
"Call to \"SItemDaughter\" failed"))
return;
while (syllable != NULL)
{
phone = SItemDaughter(syllable, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"SItemDaughter\" failed"))
return;
s_bool nucleusFound = FALSE;
while (phone != NULL)
{
const char* phone_value = SItemGetName(phone, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"SItemGetName\" failed"))
return;
s_bool isVowel = S_PHONESET_CALL(phoneset, phone_has_feature)
(phoneset,
phone_value,
"vowel",
error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"phone_has_feature\" failed"))
return;
if( isVowel )
{
nucleusFound = TRUE;
position_in_syllable_string = "nucleus";
}
else
{
if( nucleusFound == TRUE )
position_in_syllable_string = "coda";
else
position_in_syllable_string = "onset";
}
SItemSetString ( phone, "syllablepart", position_in_syllable_string, error );
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"SItemSetString\" failed"))
return;
s_bool hasLong = S_PHONESET_CALL(phoneset, phone_has_feature)
(phoneset,
phone_value,
"duration_long",
error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"phone_has_feature\" failed"))
return;
s_bool hasShort = S_PHONESET_CALL(phoneset, phone_has_feature)
(phoneset,
phone_value,
"duration_short",
error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"phone_has_feature\" failed"))
return;
const char * feat = NULL;
if( hasLong )
{
feat = "+";
}
else if( hasShort )
{
feat = "-";
}
if(feat != NULL)
{
SItemSetString ( phone, "duration", feat, error );
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"SItemSetString\" failed"))
return;
}
phone = SItemNext ( phone, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"SItemNext\" failed"))
return;
}
syllable = SItemNext (syllable, error);
if (S_CHK_ERR(error, S_CONTERR,
"s_compute_phonetic_features",
"Call to \"SItemNext\" failed"))
return;
}
}