static const char* setSentenceType(const SItem *phrase, SMap *puncMap, s_erc *error)
{
S_CLR_ERR(error);
const char* result = NULL;
/* types: "decl, "excl", "interrog" */
/* stop at sentence's last token */
const SItem *wordFromCurrentPhrase = SItemPathToItem(phrase, "daughtern", error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemPathToItem\" failed"))
return NULL;
SItem *wordAsToken = SItemAs(wordFromCurrentPhrase, "Token", error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemAs\" failed"))
return NULL;
SItem *tokenItem = SItemParent(wordAsToken, error);
tokenItem = SItemNext(tokenItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemNext\" failed"))
return NULL;
const char *punctStr = SItemGetName(tokenItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemGetName\" failed"))
return NULL;
s_bool found= SMapObjectPresent(puncMap, punctStr, error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SMapObjectPresent\" failed"))
return NULL;
result = punctStr;
if( found == TRUE)
{
result = SMapGetString ( puncMap, punctStr, error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SMapGetString\" failed"))
return NULL;
}
else
{
result = NULL;
}
return result;
}
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 SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SItem *itrItem;
sint32 count;
S_CLR_ERR(error);
if (item == NULL)
return NULL;
itrItem = item;
count = -1;
while (itrItem != NULL)
{
count++;
itrItem = SItemNext(itrItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
extractedFeat = SObjectSetInt(count, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
}
static uint32 get_num_frames(SRelp *self, const SRelation *unitRel, s_erc *error)
{
const SItem *unit;
SItem *sourceUnit = NULL;
STrackFloat *sourceTrack = NULL;
uint32 num_frames = 0;
S_CLR_ERR(error);
unit = SRelationHead(unitRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"get_num_frames",
"Call to \"SRelationHead\" failed"))
return 0;
while (unit != NULL)
{
sourceUnit = (SItem*)SItemGetObject(unit, "source-unit", error);
if (S_CHK_ERR(error, S_CONTERR,
"get_num_frames",
"Call to \"SItemGetObject\" failed"))
return 0;
sourceTrack = (STrackFloat*)SItemGetObject(sourceUnit, "lpc-coefs", error);
if (S_CHK_ERR(error, S_CONTERR,
"get_num_frames",
"Call to \"SItemGetObject\" failed"))
return 0;
num_frames += sourceTrack->data->row_count;
unit = SItemNext(unit, error);
if (S_CHK_ERR(error, S_CONTERR,
"get_num_frames",
"Call to \"SItemNext\" failed"))
return 0;
}
self->num_channels = sourceTrack->data->col_count;
return num_frames;
}
static sint32 word_num_syls(const SItem *item, s_erc *error)
{
const SItem *itemInSylStructRel;
const SItem *itrItem;
sint32 count;
S_CLR_ERR(error);
if (item == NULL)
return 0;
itemInSylStructRel = SItemAs(item, "SylStructure", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAs\" failed"))
return 0;
if (itemInSylStructRel == NULL)
return 0;
itrItem = SItemDaughter(itemInSylStructRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemDaughter\" failed"))
return 0;
count = 0;
while (itrItem != NULL)
{
count++;
itrItem = SItemNext(itrItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
return 0;
}
/* all OK here */
return count;
}
static void Run(const SUttProcessor *self, SUtterance *utt,
s_erc *error)
{
SHTSEngineMESynthUttProc105 *HTSsynth = (SHTSEngineMESynthUttProc105*)self;
SPlugin *audioPlugin;
const SRelation *segmentRel;
SAudio *audio = NULL;
s_bool is_present;
char **label_data = NULL;
int label_size;
const SItem *item;
const SItem *itemItr;
int counter;
uint i;
int frame;
int state;
const double rate = HTSsynth->engine.global.fperiod * 1e+7 / HTSsynth->engine.global.sampling_rate;
int nstate;
S_CLR_ERR(error);
/* we require the segment relation */
is_present = SUtteranceRelationIsPresent(utt, "Segment", 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 'Segment' relation in utterance");
goto quit_error;
}
segmentRel = SUtteranceGetRelation(utt, "Segment", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceGetRelation\" failed"))
goto quit_error;
item = SRelationHead(segmentRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationHead\" failed"))
goto quit_error;
itemItr = item;
label_size = 0;
while (itemItr != NULL)
{
label_size++;
itemItr = SItemNext(itemItr, error);
}
label_data = S_CALLOC(char*, label_size);
itemItr = item;
counter = 0;
while (itemItr != NULL)
{
SObject *dFeat;
const char *tmp;
dFeat = SItemPathToFeatProc(itemItr, "hts_labels", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemPathToFeatProc\" failed"))
goto quit_error;
if (dFeat == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"Run",
"Failed to generate hts labels for segment item");
goto quit_error;
}
tmp = SObjectGetString(dFeat, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectGetString\" failed"))
goto quit_error;
label_data[counter++] = s_strdup(tmp, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"s_strdup\" failed"))
goto quit_error;
SItemSetObject((SItem*)itemItr, "hts_label", dFeat, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemSetObject\" failed"))
goto quit_error;
itemItr = SItemNext(itemItr, error);
}
/* speech synthesis part */
HTS_Engine_load_label_from_string_list(&(HTSsynth->engine), label_data, label_size);
check_and_change_rate_volume(HTSsynth, utt, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"check_and_change_rate_volume\" failed"))
goto quit_error;
HTS_Engine_create_sstream(&(HTSsynth->engine));
check_and_change_tone(HTSsynth, utt, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"check_and_change_tone\" failed"))
goto quit_error;
HTS_Engine_create_pstream(&(HTSsynth->engine));
if (HTSsynth->me == TRUE) /* mixed excitation */
{
HTS_Engine_create_gstream_me(&(HTSsynth->engine),
HTSsynth->me_num_filters, HTSsynth->me_filter_order,
HTSsynth->me_filter, HTSsynth->xp_sig, HTSsynth->xn_sig,
HTSsynth->hp, HTSsynth->hn,
HTSsynth->pd_filter, HTSsynth->pd_filter_order);
}
else
{
HTS_Engine_create_gstream(&(HTSsynth->engine));
}
nstate = HTS_Speect_ModelSet_get_nstate(&(HTSsynth->engine));
itemItr = item;
counter = 0;
frame = 0;
state = 0;
while (itemItr != NULL)
{
int j;
int duration = 0;
float tmp;
for (j = 0; j < nstate; j++)
duration += HTS_Speect_SStreamSet_get_duration(&(HTSsynth->engine), state++);
tmp = frame * rate;
SItemSetFloat((SItem*)itemItr, "start", tmp/1e+7, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemSetFloat\" failed"))
goto quit_error;
tmp = (frame + duration) * rate;
SItemSetFloat((SItem*)itemItr, "end", tmp/1e+7, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemSetFloat\" failed"))
goto quit_error;
frame += duration;
itemItr = SItemNext(itemItr, error);
counter++;
}
/* We need to give the utterance the audio plug-in. If we don't do
* this and the voice is deleted before the utterance, then the
* utterance can't do *anything* with the audio. Not even delete
* it (segfault). This should be fast because it is already
* loaded.
* Note that this happens before the audio is set. This is because
* utt features are a list implementation.
*/
audioPlugin = s_pm_load_plugin("audio.spi", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceSetFeature\" failed"))
goto quit_error;
SUtteranceSetFeature(utt, "audio_plugin", S_OBJECT(audioPlugin), error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceSetFeature\" failed"))
{
S_DELETE(audioPlugin, "Run", error);
goto quit_error;
}
/* create an audio object */
audio = S_NEW(SAudio, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Failed to create new 'SAudio' object"))
goto quit_error;
/* set audio feature in utterance */
SUtteranceSetFeature(utt, "audio", S_OBJECT(audio), error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceSetFeature\" failed"))
{
S_DELETE(audio, "Run", error);
goto quit_error;
}
audio->sample_rate = HTSsynth->engine.global.sampling_rate;
audio->num_samples = (uint32)HTS_Speect_GStreamSet_get_total_nsample(&(HTSsynth->engine));
audio->samples = S_MALLOC(float, audio->num_samples);
if (audio->samples == NULL)
{
S_FTL_ERR(error, S_MEMERROR,
"Run",
"Failed to allocate memory for 'float' object");
goto quit_error;
}
/* write data */
for (i = 0; i < audio->num_samples; i++)
audio->samples[i] = (float)(HTS_Speect_GStreamSet_get_speech(&(HTSsynth->engine), i) * 1.0);
for (counter = 0; counter < label_size; counter++)
S_FREE(label_data[counter]);
S_FREE(label_data);
HTS_Engine_refresh(&(HTSsynth->engine));
/* all OK here */
return;
/* error clean-up code */
quit_error:
if (label_data != NULL)
{
for (counter = 0; counter < label_size; counter++)
{
if (label_data[counter] != NULL)
S_FREE(label_data[counter]);
}
S_FREE(label_data);
}
return;
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SItem *phraseItem;
const SItem *wordItem;
const SItem *sylStructWordItem;
const SItem *syllableItem;
sint32 num_stressed = 0;
s_bool is_current_syl = FALSE;
S_CLR_ERR(error);
if (item == NULL)
return NULL;
/* get current phrase */
phraseItem = SItemPathToItem(item, "R:SylStructure.parent.R:Phrase.parent",
error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemPathToItem\" failed"))
goto quit_error;
if (phraseItem == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"Run",
"Failed to get phrase of given syllable");
goto quit_error;
}
wordItem = SItemDaughter(phraseItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemDaughter\" failed"))
goto quit_error;
while (wordItem != NULL)
{
sylStructWordItem = SItemAs(wordItem, "SylStructure", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAs\" failed"))
goto quit_error;
/* get syllables */
syllableItem = SItemDaughter(sylStructWordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemDaughter\" failed"))
goto quit_error;
while (syllableItem != NULL)
{
s_bool is_stressed;
is_current_syl = SItemEqual(syllableItem, item, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemEqual\" failed"))
goto quit_error;
if (is_current_syl)
break;
is_stressed = syl_is_stressed(syllableItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"syl_is_stressed\" failed"))
goto quit_error;
if (is_stressed)
num_stressed++;
syllableItem = SItemNext(syllableItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
if (is_current_syl)
break;
wordItem = SItemNext(wordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
extractedFeat = SObjectSetInt(num_stressed, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
s_bool found = FALSE;
SItem *boundary = SItemLastDaughter ( item, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemLastDaughter\" failed"))
goto quit_error;
boundary = SItemAs (boundary, "Boundaries", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAs\" failed"))
goto quit_error;
SItem * boundaryNext = SItemNext (boundary, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
if (boundaryNext != NULL)
boundary = boundaryNext;
while (!found && boundary != NULL)
{
found = SItemFeatureIsPresent( boundary, "tobi_endtone", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemFeatureIsPresent\" failed"))
goto quit_error;
if (!found)
{
boundary = SItemPrev(boundary, error );
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemPrev\" failed"))
goto quit_error;
}
}
if ( found )
{
const char* tone = SItemGetString( boundary, "tobi_endtone", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemGetFeature\" failed"))
goto quit_error;
extractedFeat = SObjectSetString( tone, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetString\" failed"))
goto quit_error;
}
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
S_UNUSED(item);
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SItem *syllableItem;
sint32 num_syls = 1; /* include current syllable */
S_CLR_ERR(error);
if (item == NULL)
return NULL;
/* get the syllable as in syllable relation */
syllableItem = SItemAs(item, "Syllable", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAs\" failed"))
goto quit_error;
if (syllableItem == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"Run",
"Failed to get the syllable as it is in the 'Syllable' relation");
goto quit_error;
}
/* start with the next syllable w.r.t the given one */
syllableItem = SItemNext(syllableItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
while (syllableItem != NULL)
{
s_bool is_stressed;
is_stressed = syl_is_stressed(syllableItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"syl_is_stressed\" failed"))
goto quit_error;
if (is_stressed)
break;
else
num_syls++;
syllableItem = SItemNext(syllableItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
extractedFeat = SObjectSetInt(num_syls, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SItem *itemInSylStructRel;
const SItem *segment;
const SPhoneset *phoneset;
s_bool multilingual = FALSE;
S_CLR_ERR(error);
if (item == NULL)
return NULL;
/* get the phoneset */
phoneset = _get_phoneset(item, &multilingual, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"_get_phoneset\" failed"))
return NULL;
itemInSylStructRel = SItemAs(item, "SylStructure", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAs\" failed"))
goto quit_error;
segment = SItemDaughter(itemInSylStructRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemDaughter\" failed"))
goto quit_error;
while (segment != NULL)
{
s_bool is_vowel;
const char *item_name;
item_name = SItemGetName(segment, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemGetName\" failed"))
goto quit_error;
if (item_name == NULL)
continue;
is_vowel = S_PHONESET_CALL(phoneset, phone_has_feature)(phoneset,
item_name,
"vowel",
error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"phone_has_feature\" failed"))
goto quit_error;
if (is_vowel)
{
if (multilingual)
{
extractedFeat = SItemPathToFeatProc(segment,
"segment_name_multilingual",
error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemPathToFeatProc\" failed"))
goto quit_error;
}
else
{
extractedFeat = SObjectSetString(item_name, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetString\" failed"))
goto quit_error;
}
return extractedFeat;
}
segment = SItemNext(segment, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
/* did not find a vowel */
extractedFeat = SObjectSetString("novowel", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetString\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SItem *wordItem;
sint32 num_words = 1; /* include current word */
S_CLR_ERR(error);
if (item == NULL)
return NULL;
/* get the word as in the word relation */
wordItem = SItemAs(item, "Word", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAs\" failed"))
goto quit_error;
if (wordItem == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"Run",
"Failed to get given word as in 'Word' relation");
goto quit_error;
}
/* start with next word */
wordItem = SItemNext(wordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
while (wordItem != NULL)
{
s_bool is_content;
is_content = word_is_content(wordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"word_is_content\" failed"))
goto quit_error;
if (is_content)
break;
else
num_words++;
wordItem= SItemNext(wordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
extractedFeat = SObjectSetInt(num_words, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SItem *itrItem;
sint32 count;
SWordsToNextPuncFeatProc *castSelf = S_CAST(self, SWordsToNextPuncFeatProc, error);
if (S_CHK_ERR(error, S_CONTERR,
"Initialize",
"Call to S_CAST failed"))
goto quit_error;
S_CLR_ERR(error);
if (item == NULL)
return NULL;
itrItem = SItemPathToItem (item, "R:Token.parent", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemPathToItem\" failed"))
goto quit_error;
SMap* posPunctuation = S_CAST( SMapGetObject ( castSelf->symbols, "pos punctuation", error ), SMap, error );
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SMapGetObject\" failed"))
goto quit_error;
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"S_CAST\" failed"))
goto quit_error;
count = -1;
s_bool found = FALSE;
while (found == FALSE && itrItem != NULL)
{
count++;
s_bool hasPos = SItemFeatureIsPresent ( itrItem, "POS", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemFeatureIsPresent\" failed"))
goto quit_error;
if (hasPos)
{
const char* keyPos = SItemGetString (itrItem, "POS", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemGetString\" failed"))
goto quit_error;
found= SMapObjectPresent(posPunctuation, keyPos, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SMapObjectPresent\" failed"))
goto quit_error;
}
itrItem = SItemNext(itrItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
extractedFeat = SObjectSetInt(count, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
S_UNUSED(self);
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
}
static int write_relation(SDatasource *ds, const SRelation *rel,
int num_tiers, float end_time, s_erc *error)
{
const char *relname;
const SItem *head = NULL;
const char *tier_header = "\titem [%d]:\n\t\tclass = \"IntervalTier\"\n\t\tname = \"%s\"\n";
const char *tier_start_end = "\t\txmin = 0\n\t\txmax = %f\n";
const char *tier_intervals = "\t\tintervals: size = %d\n";
const char *tier_item = " \t\tintervals [%d]:\n\t\t\txmin = %f\n\t\t\txmax = %f\n\t\t\ttext = \"%s\"\n";
char *buff;
size_t str_size;
float prev_end = 0.0;
const SItem *itr;
float item_start = 0.0;
float item_end = 0.0;
const char *item_name = NULL;
SList *info;
size_t l_size;
int counter;
SIterator *litr;
S_CLR_ERR(error);
if (rel == NULL)
return num_tiers;
num_tiers += 1;
relname = SRelationName(rel, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SRelationName\" failed"))
return 0;
head = SRelationHead(rel, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SRelationHead\" failed"))
return 0;
/* tier header */
s_asprintf(&buff, error, tier_header, num_tiers, relname);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"s_asprintf\" failed"))
return 0;
str_size = s_strsize(buff, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"s_strsize\" failed"))
{
S_FREE(buff);
return 0;
}
SDatasourceWrite(ds, buff, sizeof(char), str_size, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SDatasourceWrite\" failed"))
{
S_FREE(buff);
return 0;
}
S_FREE(buff);
/* start time and end time (always 0 and utt end) */
s_asprintf(&buff, error, tier_start_end, end_time);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"s_asprintf\" failed"))
return 0;
str_size = s_strsize(buff, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"s_strsize\" failed"))
{
S_FREE(buff);
return 0;
}
SDatasourceWrite(ds, buff, sizeof(char), str_size, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SDatasourceWrite\" failed"))
{
S_FREE(buff);
return 0;
}
S_FREE(buff);
info = S_LIST(S_NEW(SListList, error));
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Failed to create new list for info cache"))
return 0;
prev_end = 0.0;
itr = head;
while (itr != NULL)
{
item_start = get_start(itr, relname, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"get_start\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
item_end = get_end(itr, relname, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"get_end\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
if (!s_float_equal(item_start, prev_end))
{
/* prepend extra "SIL" */
add_info(info, prev_end, item_start, "SIL", error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"add_info\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
}
item_name = SItemGetName(itr, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SItemGetName\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
add_info(info, item_start, item_end, item_name, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"add_info\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
prev_end = item_end;
itr = SItemNext(itr, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SItemNext\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
}
if (!s_float_equal(item_end, end_time))
{
add_info(info, item_end, end_time, "SIL", error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"add_info\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
}
l_size = SListSize(info, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SListSize\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
/* tier intervals */
s_asprintf(&buff, error, tier_intervals, l_size);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"s_asprintf\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
str_size = s_strsize(buff, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"s_strsize\" failed"))
{
S_FREE(buff);
S_DELETE(info, "write_relation", error);
return 0;
}
SDatasourceWrite(ds, buff, sizeof(char), str_size, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SDatasourceWrite\" failed"))
{
S_FREE(buff);
S_DELETE(info, "write_relation", error);
return 0;
}
S_FREE(buff);
litr = S_ITERATOR_GET(info, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"S_ITERATOR_GET\" failed"))
{
S_DELETE(info, "write_relation", error);
return 0;
}
for (counter = 1; litr != NULL; litr = SIteratorNext(litr), counter++)
{
SList *itemList;
SObject *tmp;
char *name;
itemList = S_LIST(SIteratorObject(litr, error));
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SIteratorObject\" failed"))
{
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
return 0;
}
tmp = SListPop(itemList, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SIteratorObject\" failed"))
{
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
return 0;
}
item_name = SObjectGetString(tmp, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SObjectGetString\" failed"))
{
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
S_DELETE(tmp, "write_relation", error);
return 0;
}
name = s_strdup(item_name, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"s_strdup\" failed"))
{
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
S_DELETE(tmp, "write_relation", error);
return 0;
}
S_DELETE(tmp, "write_relation", error);
tmp = SListPop(itemList, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SIteratorObject\" failed"))
{
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
S_FREE(name);
return 0;
}
item_end = SObjectGetFloat(tmp, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SObjectGetFloat\" failed"))
{
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
S_DELETE(tmp, "write_relation", error);
S_FREE(name);
return 0;
}
S_DELETE(tmp, "write_relation", error);
tmp = SListPop(itemList, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SIteratorObject\" failed"))
{
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
S_FREE(name);
return 0;
}
item_start = SObjectGetFloat(tmp, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SObjectGetFloat\" failed"))
{
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
S_DELETE(tmp, "write_relation", error);
S_FREE(name);
return 0;
}
S_DELETE(tmp, "write_relation", error);
/* tier item */
s_asprintf(&buff, error, tier_item, counter, item_start, item_end, name);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"s_asprintf\" failed"))
{
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
S_FREE(name);
return 0;
}
str_size = s_strsize(buff, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"s_strsize\" failed"))
{
S_FREE(buff);
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
S_FREE(name);
return 0;
}
SDatasourceWrite(ds, buff, sizeof(char), str_size, error);
if (S_CHK_ERR(error, S_CONTERR,
"write_relation",
"Call to \"SDatasourceWrite\" failed"))
{
S_FREE(buff);
S_DELETE(info, "write_relation", error);
S_DELETE(litr, "write_relation", error);
S_FREE(name);
return 0;
}
S_FREE(buff);
S_FREE(name);
}
S_DELETE(info, "write_relation", error);
return num_tiers;
}
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 crfsuite_instance_t* create_phrase_instance ( SItem* phrase,
crfsuite_dictionary_t* attrs,
crfsuite_dictionary_t* labels,
s_erc *error)
{
crfsuite_instance_t * result = malloc ( sizeof(crfsuite_instance_t) );
int i = 0;
int L = labels->num(labels);
const SItem* itrItem = NULL;
const SItem* itrItemNext = NULL;
const SItem* finishItem = NULL;
const char* lbl[] = {"num", "sym", "cap", "p1", "p2", "p3",
"s1", "s2", "s3", "P1", "P2", "P3", "P4",
"S1", "S2", "S3", "S4", "S5", "S6", "w" };
const int words_length = 19;
const char* words[19] = { NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL};
int position = 0;
int lbl_counter = 0;
char buffer[8192];
crfsuite_instance_init ( result );
itrItemNext = SItemPathToItem (phrase, "daughter.R:Token.parent", error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"SItemPathToItem\" failed"))
return NULL;
finishItem = SItemPathToItem (phrase, "n.daughter.R:Token.parent", error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"SItemPathToItem\" failed"))
return NULL;
int counter = 0;
while ( itrItemNext != finishItem && itrItemNext != NULL && counter < 9)
{
words[counter] = SItemGetName(itrItemNext, error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"SItemGetName\" failed"))
return NULL;
counter++;
itrItemNext = SItemNext(itrItemNext, error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"SItemNext\" failed"))
return NULL;
}
itrItem = SItemPathToItem (phrase, "daughter.R:Token.parent", error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"SItemPathToItem\" failed"))
return NULL;
while ( itrItem != finishItem && itrItem != NULL)
{
/* Extraction of the features for each token */
const char *tokenName = SItemGetName (itrItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"SItemGetName\" failed"))
return NULL;
lbl_counter = 0;
/* Extraction of label's ID */
int tokenID = labels->to_id(labels, "UNK");
int attribute_id;
/* If unknown the set the 0 labels (unknown) */
if (tokenID < 0)
tokenID = L;
crfsuite_item_t itemToken;
crfsuite_attribute_t attribute;
crfsuite_item_init(&itemToken);
const char *feat = NULL;
s_bool found = FALSE;
/* if token contains numbers */
found = hasNumber (tokenName, error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"hasNumber\" failed"))
return NULL;
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 0, found? "Y" : "N");
lbl_counter++;
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
/* if token contains symbols */
found = hasSymbol (tokenName, error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"hasSymbol\" failed"))
return NULL;
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 0, found? "Y" : "N");
lbl_counter++;
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
/* if token contains Capitals */
found = hasCapital (tokenName, error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"hasCapital\" failed"))
return NULL;
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 0, found? "Y" : "N");
lbl_counter++;
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
/* Prefixes of 1 to 3 chars of token */
i = 1;
while ( i < 4 )
{
feat = getFirstChars ( tokenName, i, error );
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"getFirstChars\" failed"))
return NULL;
if ( feat == NULL )
feat = "__nil__";
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 0, feat);
lbl_counter++;
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
i += 1;
}
/* Suffixes of 1 to 3 chars of token */
i = 1;
while ( i < 4 )
{
feat = getLastChars ( tokenName, i, error );
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"getLastChars\" failed"))
return NULL;
if ( feat == NULL )
feat = "__nil__";
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 0, feat);
lbl_counter++;
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
i += 1;
}
/* Prefixes of 1 to 4 chars of token without duplicates adjacent */
i = 1;
while ( i < 5 )
{
feat = getFirstChars ( removeDoubles(tokenName, error), i, error );
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"removeDoubles\" failed"))
return NULL;
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"getFirstChars\" failed"))
return NULL;
if ( feat == NULL )
feat = "__nil__";
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 0, feat);
lbl_counter++;
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
i += 1;
}
/* Suffixes of 1 to 6 chars of token without duplicates adjacent */
i = 1;
while ( i < 7 )
{
feat = getLastChars ( removeDoubles(tokenName, error), i, error );
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"removeDoubles\" failed"))
return NULL;
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"getFirstChars\" failed"))
return NULL;
if ( feat == NULL )
feat = "__nil__";
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 0, feat);
lbl_counter++;
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
i += 1;
}
/* Words features */
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 0, words[(position+0+words_length)%words_length]);
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
const char *tmp = words[(position-1+words_length)%words_length];
if(tmp!=NULL)
{
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], -1, tmp);
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
tmp = words[(position+1+words_length)%words_length];
if(tmp!=NULL)
{
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 1, tmp);
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
tmp = words[(position-2+words_length)%words_length];
if(tmp!=NULL)
{
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], -2, tmp);
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
tmp = words[(position+2+words_length)%words_length];
if(tmp!=NULL)
{
sprintf(buffer, "%s[%d]=%s", lbl[lbl_counter], 2, tmp);
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
const char * tmp1 = NULL;
i = 0;
while ( i < 4 )
{
tmp = words[(position-2+i+words_length)%words_length];
tmp1 = words[(position-1+i+words_length)%words_length];
if( tmp!=NULL && tmp1 != NULL )
{
sprintf(buffer, "%s[%d]|%s[%d]=%s|%s", lbl[lbl_counter], -2+i,
lbl[lbl_counter], -1+i,
tmp, tmp1 );
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
i++;
}
const char * tmp2 = NULL;
i = 0;
while ( i < 3 )
{
tmp = words[(position-2+i+words_length)%words_length];
tmp1 = words[(position-1+i+words_length)%words_length];
tmp2 = words[(position+0+i+words_length)%words_length];
if( tmp!=NULL && tmp1 != NULL && tmp2 != NULL )
{
sprintf(buffer, "%s[%d]|%s[%d]|%s[%d]=%s|%s|%s",
lbl[lbl_counter], -2+i,
lbl[lbl_counter], -1+i,
lbl[lbl_counter], +0+i,
tmp, tmp1, tmp2 );
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
i++;
}
const char * tmp3 = NULL;
i = 0;
while ( i < 2 )
{
tmp = words[(position-2+i+words_length)%words_length];
tmp1 = words[(position-1+i+words_length)%words_length];
tmp2 = words[(position+0+i+words_length)%words_length];
tmp3 = words[(position+1+i+words_length)%words_length];
if( tmp!=NULL && tmp1 != NULL && tmp2 != NULL && tmp3 != NULL )
{
sprintf(buffer, "%s[%d]|%s[%d]|%s[%d]|%s[%d]=%s|%s|%s|%s",
lbl[lbl_counter], -2+i,
lbl[lbl_counter], -1+i,
lbl[lbl_counter], +0+i,
lbl[lbl_counter], +1+i,
tmp, tmp1, tmp2, tmp3 );
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
i++;
}
tmp = words[(position-2+words_length)%words_length];
tmp1 = words[(position-1+words_length)%words_length];
tmp2 = words[(position+0+words_length)%words_length];
tmp3 = words[(position+1+words_length)%words_length];
const char * tmp4 = words[(position+2+words_length)%words_length];
if( tmp!=NULL && tmp1 != NULL && tmp2 != NULL && tmp3 != NULL && tmp4 != NULL )
{
sprintf(buffer, "%s[%d]|%s[%d]|%s[%d]|%s[%d]|%s[%d]=%s|%s|%s|%s|%s",
lbl[lbl_counter], -2,
lbl[lbl_counter], -1,
lbl[lbl_counter], +0,
lbl[lbl_counter], +1,
lbl[lbl_counter], +2,
tmp, tmp1, tmp2, tmp3, tmp4 );
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
tmp = words[(position+0+words_length)%words_length];
i = 1;
while ( i < 10 )
{
tmp1 = words[(position-i+words_length)%words_length];
if ( tmp1 != NULL )
{
sprintf(buffer, "%s[%d]|%s[%d]=%s|%s", lbl[lbl_counter], 0,
lbl[lbl_counter], 0-i,
tmp, tmp1 );
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
i++;
}
i = 1;
while ( i < 10 )
{
tmp1 = words[(position+i+words_length)%words_length];
if ( tmp1 != NULL )
{
sprintf(buffer, "%s[%d]|%s[%d]=%s|%s", lbl[lbl_counter], 0,
lbl[lbl_counter], 0+i,
tmp, tmp1 );
attribute_id = attrs->to_id (attrs,buffer);
crfsuite_attribute_set (&attribute, attribute_id, 1.0);
crfsuite_item_append_attribute(&itemToken, &attribute);
}
i++;
}
/* Update words array */
position++;
if ( itrItemNext != NULL )
{
words[position+9] = SItemGetName (itrItemNext, error);
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"SItemNext\" failed"))
return NULL;
itrItemNext = SItemNext ( itrItemNext, error );
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"SItemNext\" failed"))
return NULL;
}
else
words[position+9] = NULL;
itrItem = SItemNext ( itrItem, error );
if (S_CHK_ERR(error, S_CONTERR,
"create_phrase_instance",
"Call to \"SItemNext\" failed"))
return NULL;
crfsuite_instance_append(result, &itemToken, tokenID);
crfsuite_item_finish(&itemToken);
}
return result;
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SItem *phraseItem;
const SItem *wordItem;
sint32 num_content = 0;
S_CLR_ERR(error);
if (item == NULL)
return NULL;
/* get current phrase */
phraseItem = SItemPathToItem(item, "R:Phrase.parent",
error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemPathToItem\" failed"))
goto quit_error;
if (phraseItem == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"Run",
"Failed to get phrase of given word");
goto quit_error;
}
wordItem = SItemDaughter(phraseItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemDaughter\" failed"))
goto quit_error;
while (wordItem != NULL)
{
s_bool is_content;
s_bool is_current_word;
is_current_word = SItemEqual(wordItem, item, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemEqual\" failed"))
goto quit_error;
if (is_current_word)
break;
is_content = word_is_content(wordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"word_is_content\" failed"))
goto quit_error;
if (is_content)
num_content++;
wordItem = SItemNext(wordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
extractedFeat = SObjectSetInt(num_content, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
}
static void Save(const SObject *object, const char *path, s_erc *error)
{
int rc;
SUtterance *utt = S_UTTERANCE(object);
SDatasource *ds;
xmlTextWriterPtr writer;
xmlOutputBufferPtr out;
const char * xsi = "http://www.w3.org/2001/XMLSchema-instance";
S_CLR_ERR(error);
ds = SFilesourceOpenFile(path, "wt", error);
if (S_CHK_ERR(error, S_CONTERR,
"Save",
"Call to \"SFilesourceOpenFile\" failed"))
return;
out = xmlOutputBufferCreateIO(_ds_write,
_ds_close,
ds,
NULL);
if (out == NULL)
{
S_CTX_ERR(error, S_CONTERR,
"Save",
"Call to \"xmlOutputBufferCreateIO\" failed");
return;
}
writer = xmlNewTextWriter(out);
if (writer == NULL)
{
xmlOutputBufferClose(out);
S_CTX_ERR(error, S_CONTERR,
"Save",
"Call to \"xmlNewTextWriter\" failed");
return;
}
/* Start Document */
rc = xmlTextWriterStartDocument(writer, NULL, ENCODING, NULL);
if (rc < 0) {
S_CTX_ERR(error, S_CONTERR,
"Save",
"Call to \"xmlTextWriterStartDocument\" failed");
goto s_write_utt_exit;
}
/* Write the maryxml namespace */
rc = xmlTextWriterStartElement(writer, BAD_CAST "xml");
if (rc < 0) {
S_CTX_ERR(error, S_CONTERR,
"Save",
"Call to \"xmlTextWriterStartElement\" failed");
goto s_write_utt_exit;
}
rc = xmlTextWriterWriteAttribute(writer, BAD_CAST "xmlns:xsi", BAD_CAST xsi);
if (rc < 0) {
S_CTX_ERR(error, S_CONTERR,
"Save",
"Call to \"xmlTextWriterWriteAttribute\" failed");
goto s_write_utt_exit;
}
rc = xmlTextWriterWriteAttribute(writer, BAD_CAST "version", BAD_CAST "0.5");
if (rc < 0) {
S_CTX_ERR(error, S_CONTERR,
"Save",
"Call to \"xmlTextWriterWriteAttribute\" failed");
goto s_write_utt_exit;
}
// print labels
/* get to the first syllable of the current word */
const SItem* itrSegments = SRelationHead(SUtteranceGetRelation(utt, "Segment", error), error);
if (S_CHK_ERR(error, S_CONTERR,
"Save",
"Call to \"SItemPathToItem\" failed"))
goto s_write_utt_exit;
while (itrSegments != NULL)
{
/* get segment content */
const char* label = SItemGetString(itrSegments, "hts_label", error);
if (S_CHK_ERR(error, S_CONTERR,
"Save",
"Call to \"SItemGetName\" failed"))
goto s_write_utt_exit;
/* get next segment */
itrSegments = SItemNext(itrSegments, error);
if (S_CHK_ERR(error, S_CONTERR,
"Save",
"Call to \"SItemNext\" failed"))
goto s_write_utt_exit;
xmlTextWriterWriteElement(writer, BAD_CAST "label", BAD_CAST label);
}
/* Close the tag xml */
rc = xmlTextWriterEndElement(writer);
if (rc < 0)
{
S_CTX_ERR(error, S_CONTERR,
"Save",
"Call to \"xmlTextWriterEndDocument\" failed");
goto s_write_utt_exit;
}
/* Close the document */
rc = xmlTextWriterEndDocument(writer);
if (rc < 0)
{
S_CTX_ERR(error, S_CONTERR,
"Save",
"Call to \"xmlTextWriterEndDocument\" failed");
goto s_write_utt_exit;
}
s_write_utt_exit:
xmlFreeTextWriter(writer);
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SItem *itemInSylStructRel;
const SItem *itrItem = NULL;
const SObject* feature = NULL;
sint32 count;
S_CLR_ERR(error);
if (item == NULL)
return NULL;
itemInSylStructRel = SItemAs(item, "SylStructure", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAs\" failed"))
goto quit_error;
if (itemInSylStructRel == NULL)
{
extractedFeat = SObjectSetInt(0, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
}
else
{
itrItem = SItemDaughter(itemInSylStructRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemDaughter\" failed"))
goto quit_error;
}
count = 0;
while (itrItem != NULL)
{
feature = SItemPathToFeatProc(itrItem, "syllable_num_phones", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemPathToFeatProc\" failed"))
goto quit_error;
if(feature == NULL) {
S_CTX_ERR(error, S_FAILURE,
"Run",
"Call to \"SItemPathToFeatProc\" returned null");
goto quit_error;
}
count += SObjectGetInt(feature, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectGetInt\" failed"))
goto quit_error;
itrItem = SItemNext(itrItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
extractedFeat = SObjectSetInt(count, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
}
static void Run(const SUttProcessor *self, SUtterance *utt,
s_erc *error)
{
SG2P *g2p = NULL;
SLexicon *lexicon = NULL;
SAddendum *addendum = NULL;
SSyllabification *syllab = NULL;
const SRelation *wordRel;
SRelation *syllableRel = NULL;
SRelation *sylStructRel = NULL;
SRelation *segmentRel = NULL;
SItem *wordItem;
char *downcase_word;
SList *phones;
s_bool syllabified;
SList *syllablesPhones;
SItem *sylStructureWordItem;
SItem *syllableItem;
SItem *sylStructSylItem;
SItem *segmentItem;
SIterator *sylItr = NULL;
SIterator *phoneItr = NULL;
const SObject *phone;
s_bool is_present;
S_CLR_ERR(error);
s_get_lexical_objects(self, utt, &g2p, &lexicon, &addendum, &syllab, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"s_get_lexical_objects\" failed"))
goto quit_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;
/* create relations */
syllableRel = SUtteranceNewRelation(utt, "Syllable", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceNewRelation\" failed"))
goto quit_error;
sylStructRel = SUtteranceNewRelation(utt, "SylStructure", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceNewRelation\" failed"))
goto quit_error;
segmentRel = SUtteranceNewRelation(utt, "Segment", 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.
* iterate over the word relation and fill in the
* phones and the associated structure.
*/
wordItem = (SItem*)SRelationHead(wordRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationHead\" failed"))
goto quit_error;
while (wordItem != NULL)
{
/* get word and downcase it */
downcase_word = s_strlwr(s_strdup(SItemGetName(wordItem, error), error), error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Failed to down-case word item"))
goto quit_error;
if (downcase_word == NULL || s_strcmp(downcase_word, "", error) == 0)
goto continue_cycle;
phones = NULL;
syllabified = FALSE;
/* get phone sequence for word */
if (addendum != NULL)
{
phones = S_ADDENDUM_CALL(addendum, get_word)(addendum,
downcase_word,
NULL,
&syllabified,
error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to method \"get_word\" (SAddendum) failed"))
goto quit_error;
}
if ((phones == NULL) && (lexicon != NULL))
{
phones = S_LEXICON_CALL(lexicon, get_word)(lexicon,
downcase_word,
NULL,
&syllabified,
error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to method \"get_word\" (SLexicon) failed"))
goto quit_error;
}
if ((phones == NULL) && (g2p != NULL))
{
phones = S_G2P_CALL(g2p, apply)(g2p, downcase_word, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to method \"apply\" (SG2P) failed"))
goto quit_error;
}
if (phones == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"Run",
"Failed to get phone sequence for word '%s'", downcase_word);
S_FREE(downcase_word);
continue;
}
S_FREE(downcase_word);
/* syllabify phone sequence */
if (syllabified == FALSE)
{
if (syllab != NULL)
{
syllablesPhones = S_SYLLABIFICATION_CALL(syllab, syllabify)(syllab,
wordItem,
phones,
error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to method \"syllabify\" failed"))
goto quit_error;
S_DELETE(phones, "Run", error);
}
else
{
syllablesPhones = S_LIST(S_NEW(SListList, error));
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Failed to create new 'SList' object"))
goto quit_error;
SListAppend(syllablesPhones, S_OBJECT(phones), error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SListAppend\" failed"))
goto quit_error;
}
}
else
syllablesPhones = (SList*)phones;
/* create new syllable structure word item, shares content
* with word item.
*/
sylStructureWordItem = SRelationAppend(sylStructRel, wordItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationAppend\" failed"))
goto quit_error;
/* iterate over syllables */
sylItr = S_ITERATOR_GET(syllablesPhones, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"S_ITERATOR_GET\" failed"))
goto quit_error;
while (sylItr != NULL)
{
/* new item in syllable relation */
syllableItem = SRelationAppend(syllableRel, NULL, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationAppend\" failed"))
goto quit_error;
SItemSetName(syllableItem, "syl", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemSetName\" failed"))
goto quit_error;
/* daughter of above item, but in SylStructure */
sylStructSylItem = SItemAddDaughter(sylStructureWordItem, syllableItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAddDaughter\" failed"))
goto quit_error;
/* iterate over phones and add segments */
phoneItr = S_ITERATOR_GET((SList*)SIteratorObject(sylItr, error), error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"S_ITERATOR_GET/SIteratorObject\" failed"))
goto quit_error;
while (phoneItr != NULL)
{
phone = SIteratorObject(phoneItr, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SIteratorObject\" failed"))
goto quit_error;
segmentItem = SRelationAppend(segmentRel, NULL, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationAppend\" failed"))
goto quit_error;
SItemSetName(segmentItem, SObjectGetString(phone, error), error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemSetName/SObjectGetString\" failed"))
goto quit_error;
SItemAddDaughter(sylStructSylItem, segmentItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAddDaughter\" failed"))
goto quit_error;
phoneItr = SIteratorNext(phoneItr);
}
sylItr = SIteratorNext(sylItr);
}
S_DELETE(syllablesPhones, "Run", error);
continue_cycle:
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 (syllableRel != NULL)
{
SUtteranceDelRelation(utt, "Syllable", error);
S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceDelRelation\" failed");
}
if (sylStructRel != NULL)
{
SUtteranceDelRelation(utt, "SylStructure", error);
S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceDelRelation\" failed");
}
if (segmentRel != NULL)
{
SUtteranceDelRelation(utt, "Segment", error);
S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceDelRelation\" failed");
}
if (sylItr != NULL)
S_DELETE(sylItr, "Run", error);
if (phoneItr != NULL)
S_DELETE(phoneItr, "Run", error);
self = NULL;
}
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;
}
}
/* setSentenceType should be made out of two parts:
* 1) the first section searchs for the last punctuation element of the sentence
* -> if it is a '.' --> set "decl" type (where should I set this feature value?)
* -> if it is a '!' --> set "excl" type (where should I set this feature value?)
* -> if it is a '?' --> set "interrog" type (where should I set this feature value?)
* 2) if the first part decides for "interrog" type, there should be other controls
* to establish the sentence's complete type
* */
static char* setSentenceType(const SItem *phrase, SMap *prosSymbols, s_erc *error)
{
S_CLR_ERR(error);
char* result = "decl";
/* types: "decl, "excl", "interrog" */
/* stop at sentence's last token */
const SItem *wordFromCurrentPhrase = SItemPathToItem(phrase, "daughter", error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemPathToItem\" failed"))
return NULL;
SItem *wordAsToken = SItemAs(wordFromCurrentPhrase, "Token", error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemAs\" failed"))
return NULL;
SItem *tokenItem = SItemParent(wordAsToken, error);
SItem *firstTokenItem = tokenItem;
s_bool isPunct = SItemFeatureIsPresent(tokenItem, "IsPunctuation", error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemFeatureIsPresent\" failed"))
return NULL;
s_bool isFinalPunct = FALSE;
while (isFinalPunct == FALSE)
{
isPunct = SItemFeatureIsPresent(tokenItem, "IsPunctuation", error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemFeatureIsPresent\" failed"))
return NULL;
if (isPunct)
{
const char *punctStr = SItemGetName(tokenItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemGetName\" failed"))
return NULL;
if (s_strcmp(punctStr, ".", error) == 0)
{
isFinalPunct = TRUE;
result = "decl";
}
else if (s_strcmp(punctStr, "!", error) == 0)
{
isFinalPunct = TRUE;
result = "excl";
}
else if (s_strcmp(punctStr, "?", error) == 0)
{
isFinalPunct = TRUE;
const char *posValueStr = NULL;
char *posValueStr_filtered = NULL;
s_bool currPosInCurrList;
s_bool have_symbols = FALSE;
SMap* valueMap = NULL;
have_symbols = SMapObjectPresent(prosSymbols, "firstPosInQuestionW", error);
if (S_CHK_ERR(error, S_CONTERR,
"SetSentenceType",
"Call to \"SMapObjectPresent\" failed"))
goto quit_error;
if (have_symbols)
{
valueMap = S_CAST(SMapGetObject(prosSymbols, "firstPosInQuestionW", error), SMap, error);
if (S_CHK_ERR(error, S_CONTERR,
"SetSentenceType",
"Call to \"SMapGetObject\" failed"))
goto quit_error;
}
else
goto quit_error;
posValueStr = SItemGetString(firstTokenItem, "POS", error);
if (S_CHK_ERR(error, S_CONTERR,
"SetSentenceType",
"Call to \"SItemGetString\" failed"))
goto quit_error;
/* filter the current POS tag, remember to free the memory
* pointed to by 'posValueStr_filtered' pointer
*/
posValueStr_filtered = filterPosTag(posValueStr, error);
if (S_CHK_ERR(error, S_CONTERR,
"SetSentenceType",
"Call to \"filterPosTag\" failed"))
goto quit_error;
currPosInCurrList = searchStringMap(valueMap, posValueStr_filtered, error);
if (currPosInCurrList == TRUE)
{
result = "interrogW";
}
else
{
result = "interrog";
}
quit_error:
if (posValueStr_filtered)
{
S_FREE(posValueStr_filtered);
}
break;
}
}
tokenItem = SItemNext(tokenItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"setSentenceType",
"Call to \"SItemNext\" failed"))
return NULL;
if(tokenItem == NULL) {
isFinalPunct = TRUE;
}
}
return result;
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SUtterance *utt;
const SRelation *sylRel;
const SItem *itr;
sint32 count;
S_CLR_ERR(error);
if (item == NULL)
return NULL;
/* get utterance */
utt = SItemUtterance(item, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemUtterance\" failed"))
goto quit_error;
if (utt == NULL)
return NULL;
/* get Syllable relation */
sylRel = SUtteranceGetRelation(utt, "Syllable", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SUtteranceGetRelation\" failed"))
goto quit_error;
if (sylRel == NULL)
return NULL;
itr = SRelationHead(sylRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SRelationHead\" failed"))
goto quit_error;
count = 0;
while (itr != NULL)
{
count++;
itr = SItemNext(itr, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
extractedFeat = SObjectSetInt(count, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
}
static void read_relation_items(SEbmlRead *ebmlReader, SRelation *rel,
s_hash_table *items_content_table, s_erc *error)
{
uint32 id;
SItem *item = NULL;
s_hash_table *relation_items = NULL;
s_bool container_exhausted;
S_CLR_ERR(error);
/* read S_UTT_EBML_RELATION_ITEMS container */
id = S_EBMLREAD_CALL(ebmlReader, container)(ebmlReader, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"container\" failed"))
goto quit_error;
/* 2^8 = 256 items */
relation_items = s_hash_table_new(&relation_items_table_free_fp, 8, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"relation_items\" failed"))
goto quit_error;
while (1)
{
container_exhausted = S_EBMLREAD_CALL(ebmlReader, container_at_end)(ebmlReader, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"container_at_end\" failed"))
goto quit_error;
if (container_exhausted)
{
SItem *current;
SItem *next;
/* connect relation head and tail items
* find the first item.
*/
current = get_item(relation_items, 1, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"get_item\" failed"))
goto quit_error;
rel->head = current;
/* find the last item*/
while (current != NULL)
{
next = SItemNext(current, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"SItemNext\" failed"))
goto quit_error;
if (next == NULL)
{
rel->tail = current;
break;
}
current = next;
}
s_hash_table_delete(relation_items, error);
S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"s_hash_table_delete\" failed");
break;
}
/* peek id */
id = S_EBMLREAD_CALL(ebmlReader, peek_id)(ebmlReader, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"peek_id\" failed"))
goto quit_error;
switch(id)
{
case S_UTT_EBML_RELATION_ITEM_NODE:
{
uint32 item_node_number;
item_node_number = S_EBMLREAD_CALL(ebmlReader, read_uint)(ebmlReader, &id, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"read_uint\" failed"))
goto quit_error;
/* find the item. */
item = get_item(relation_items, item_node_number, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"get_item\" failed"))
goto quit_error;
item->relation = rel;
break;
}
case S_UTT_EBML_RELATION_ITEM_CONTENT_ID:
{
uint32 item_content_id;
const s_hash_element *he;
s_items_content_container *ic;
item_content_id = S_EBMLREAD_CALL(ebmlReader, read_uint)(ebmlReader, &id, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"read_uint\" failed"))
goto quit_error;
/* find the item content */
he = s_hash_table_find(items_content_table, (void*)&item_content_id,
sizeof(uint32), error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"s_hash_table_find\" failed"))
goto quit_error;
if (he == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"read_relation_items",
"Failed to find item content for read content id");
goto quit_error;
}
ic = (s_items_content_container *)s_hash_element_get_data(he, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"s_hash_element_get_data\" failed"))
goto quit_error;
item->content = ic->content;
ic->used = 1;
/* add the item in item's content relations. This is for
* the SItemAs and SItemIn item functions.
*/
SItmContentAdd(item->content, rel->name, item, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"SItmContentAdd\" failed"))
goto quit_error;
break;
}
case S_UTT_EBML_RELATION_ITEM_UP_NODE:
{
uint32 item_node_number;
SItem *itemParent;
item_node_number = S_EBMLREAD_CALL(ebmlReader, read_uint)(ebmlReader, &id, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"read_uint\" failed"))
goto quit_error;
/* find the item. */
itemParent = get_item(relation_items, item_node_number, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"get_item\" failed"))
goto quit_error;
item->up = itemParent;
break;
}
case S_UTT_EBML_RELATION_ITEM_DOWN_NODE:
{
uint32 item_node_number;
SItem *itemDaughter;
item_node_number = S_EBMLREAD_CALL(ebmlReader, read_uint)(ebmlReader, &id, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"read_uint\" failed"))
goto quit_error;
/* find the item. */
itemDaughter = get_item(relation_items, item_node_number, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"get_item\" failed"))
goto quit_error;
item->down = itemDaughter;
break;
}
case S_UTT_EBML_RELATION_ITEM_NEXT_NODE:
{
uint32 item_node_number;
SItem *itemNext;
item_node_number = S_EBMLREAD_CALL(ebmlReader, read_uint)(ebmlReader, &id, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"read_uint\" failed"))
goto quit_error;
/* find the item. */
itemNext = get_item(relation_items, item_node_number, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"get_item\" failed"))
goto quit_error;
item->next = itemNext;
break;
}
case S_UTT_EBML_RELATION_ITEM_PREV_NODE:
{
uint32 item_node_number;
SItem *itemPrev;
item_node_number = S_EBMLREAD_CALL(ebmlReader, read_uint)(ebmlReader, &id, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"read_uint\" failed"))
goto quit_error;
/* find the item. */
itemPrev = get_item(relation_items, item_node_number, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"Call to \"get_item\" failed"))
goto quit_error;;
item->prev = itemPrev;
break;
}
default:
/* unknown elements, skip */
S_WARNING(S_FAILURE,
"read_relation_items",
"Skipping element with unknown id '0x%x'",
id);
S_EBMLREAD_CALL(ebmlReader, element_skip)(ebmlReader, error);
if (S_CHK_ERR(error, S_CONTERR,
"read_relation_items",
"ebmlRead method \"element_skip\" failed"))
goto quit_error;
}
}
/* here everything went OK */
return;
/* errors here, clean up code */
quit_error:
/* disconnect items from relation */
rel->head = NULL;
rel->tail = NULL;
/* delete them, can't use normal s_hash_table_delete
* because the delete function pointer
* (relation_items_table_free_fp) does not delete the items.
*/
if (relation_items != NULL)
{
s_hash_element *hte;
s_hash_element *next;
s_erc local_err = S_SUCCESS;
hte = (s_hash_element *)s_hash_table_first(relation_items, &local_err);
if (S_CHK_ERR(&local_err, S_CONTERR,
"read_relation_items",
"Call to \"s_hash_table_first\" failed"))
return;
while (hte != NULL)
{
SItem *item;
uint32 *item_node_number;
next = (s_hash_element *)s_hash_element_next(hte, &local_err);
if (S_CHK_ERR(&local_err, S_CONTERR,
"read_relation_items",
"Call to \"s_hash_element_next\" failed"))
next = NULL;
item_node_number = (uint32 *)s_hash_element_key(hte, &local_err);
if (!S_CHK_ERR(&local_err, S_CONTERR,
"read_relation_items",
"Call to \"s_hash_element_key\" failed"))
S_FREE(item_node_number);
item = (SItem *)s_hash_element_get_data(hte, &local_err);
if (!S_CHK_ERR(&local_err, S_CONTERR,
"read_relation_items",
"Call to \"s_hash_element_get_data\" failed"))
S_FORCE_DELETE(item, "read_relation_items", &local_err);
s_hash_element_unlink(hte, &local_err);
S_CHK_ERR(&local_err, S_CONTERR,
"read_relation_items",
"Call to \"s_hash_element_unlink\" failed");
hte = next;
}
s_hash_table_delete(relation_items, &local_err);
S_CHK_ERR(&local_err, S_CONTERR,
"read_relation_items",
"Call to \"s_hash_table_delete\" failed");
}
}
static void copy_lpc_residuals(SRelp *self, const SRelation *unitRel, s_erc *error)
{
const SItem *unit;
SItem *sourceUnit;
STrackFloat *sourceTrack;
SArrayInt *sourceResidual;
uint32 i;
uint32 j;
float prev_time = 0.0;
S_CLR_ERR(error);
unit = SRelationHead(unitRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"copy_lpc_residuals",
"Call to \"SRelationHead\" failed"))
return;
i = 0;
while (unit != NULL)
{
sourceUnit = (SItem*)SItemGetObject(unit, "source-unit", error);
if (S_CHK_ERR(error, S_CONTERR,
"copy_lpc_residuals",
"Call to \"SItemGetObject\" failed"))
return;
sourceTrack = (STrackFloat*)SItemGetObject(sourceUnit, "lpc-coefs", error);
if (S_CHK_ERR(error, S_CONTERR,
"copy_lpc_residuals",
"Call to \"SItemGetObject\" failed"))
return;
sourceResidual = (SArrayInt*)SItemGetObject(sourceUnit, "residuals", error);
if (S_CHK_ERR(error, S_CONTERR,
"copy_lpc_residuals",
"Call to \"SItemGetObject\" failed"))
return;
window_residuals(self, sourceResidual, sourceTrack, i, error);
if (S_CHK_ERR(error, S_CONTERR,
"copy_lpc_residuals",
"Call to \"window_residuals\" failed"))
return;
for (j = 0; j < sourceTrack->data->row_count; j++, i++)
{
self->track->data->f[i] = S_CALLOC(float, self->num_channels);
if (self->track->data->f[i] == NULL)
{
S_FTL_ERR(error, S_MEMERROR,
"copy_lpc_residuals",
"Failed to allocate memory for 'float' object");
return;
}
memcpy(self->track->data->f[i], sourceTrack->data->f[j],
self->num_channels * sizeof(float));
self->track->time[i] = sourceTrack->time[j] + prev_time;
}
prev_time = self->track->time[i - 1];
SItemSetFloat(sourceUnit, "source_end", prev_time, error);
if (S_CHK_ERR(error, S_CONTERR,
"copy_lpc_residuals",
"Call to \"SItemSetFloat\" failed"))
return;
SItemSetInt(sourceUnit, "num_frames", sourceTrack->data->row_count, error);
if (S_CHK_ERR(error, S_CONTERR,
"copy_lpc_residuals",
"Call to \"SItemSetInt\" failed"))
return;
unit = SItemNext(unit, error);
if (S_CHK_ERR(error, S_CONTERR,
"copy_lpc_residuals",
"Call to \"SItemNext\" failed"))
return;
}
}
static SObject *Run(const SFeatProcessor *self, const SItem *item,
s_erc *error)
{
SObject *extractedFeat = NULL;
const SItem *itemInSylStructRel;
const SItem *syllableItem;
const SItem *itr;
sint32 count;
S_CLR_ERR(error);
if (item == NULL)
return NULL;
itemInSylStructRel = SItemAs(item, "SylStructure", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemAs\" failed"))
goto quit_error;
if (itemInSylStructRel == NULL)
{
extractedFeat = SObjectSetInt(0, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
return extractedFeat;
}
syllableItem = SItemParent(itemInSylStructRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemParent\" failed"))
goto quit_error;
if (syllableItem == NULL)
{
extractedFeat = SObjectSetInt(0, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
return extractedFeat;
}
itr = SItemDaughter(syllableItem, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemDaughter\" failed"))
goto quit_error;
count = 0;
while (itr != NULL)
{
s_bool is_equal;
is_equal = SItemEqual(itr, itemInSylStructRel, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemEqual\" failed"))
goto quit_error;
if (is_equal)
break;
count++;
itr = SItemNext(itr, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemNext\" failed"))
goto quit_error;
}
extractedFeat = SObjectSetInt(count, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
goto quit_error;
/* all OK here */
return extractedFeat;
/* error cleanup */
quit_error:
if (extractedFeat != NULL)
S_DELETE(extractedFeat, "Run", error);
return NULL;
S_UNUSED(self);
}
/**
* Prepare the structure for the next call to hunpos.
* @private
*
* @param relation_head starting SItem for this new phrase. It must be a Phrase SItem if @p is_phrase_present is true, a Token otherwise.
* @param data data structure to fill with new SItem pointers.
* @param is_phrase_present tells if we're using phrases or directly tokens
* @param error Error code.
*
*/
static void call_hunpos(const SHunposUttProc *hunposProc, const SItem* relation_head, const SItem** data, s_bool is_phrase_present, s_erc *error)
{
const SItem* phrase_start_item;
const SItem* current_token;
const SItem* start_token; /* start point of the next phrase */
const SItem* last_safe_cut_token; /* last token from where we can cut for the next phrase */
int last_safe_cut_count;
int tokens_count;
S_CLR_ERR(error);
phrase_start_item = relation_head;
/* we already have the starting item if we have only tokens, else grab the token */
start_token = phrase_start_item;
if (is_phrase_present)
{
/* go grab the first token of this phrase */
start_token = SItemPathToItem(relation_head, "R:Phrase.daughter.R:Token.parent", error);
if (S_CHK_ERR(error, S_CONTERR,
"call_hunpos",
"Call to \"SItemPathToItem\" failed"))
return;
}
current_token = start_token;
/* for each phrase */
while (current_token != NULL)
{
tokens_count = 0;
last_safe_cut_count = 0;
start_token = current_token;
last_safe_cut_token = current_token;
/* we stop collecting tokens if the next item is null or if we reached the maximum number of tokens allowed */
while (current_token != NULL && tokens_count < hunposProc->max_tokens_number)
{
data[tokens_count] = current_token;
/* check if it's a safe cut point */
s_bool is_present = SItemFeatureIsPresent(current_token, "IsPunctuation", error);
if (S_CHK_ERR(error, S_CONTERR,
"call_hunpos",
"Call to \"SItemFeatureIsPresent\" failed"))
return;
if (is_present)
{
/* is it a punctiation token? */
sint32 is_punctuation = SItemGetInt(current_token, "IsPunctuation", error);
if (is_punctuation > 0)
{
last_safe_cut_token = current_token;
last_safe_cut_count = tokens_count + 1;
}
}
current_token = SItemNext(current_token, error);
if (S_CHK_ERR(error, S_CONTERR,
"call_hunpos",
"Call to \"SItemNext\" failed"))
return;
/* if we're using phrases, check if the current phrase finished */
if (is_phrase_present)
{
if (current_token != NULL)
{
const SItem* next_token_parent = SItemPathToItem(current_token, "R:Token.daughter.R:Phrase.parent", error);
if (S_CHK_ERR(error, S_CONTERR,
"call_hunpos",
"Call to \"SItemPathToItem\" failed"))
return;
if (next_token_parent != phrase_start_item)
{
/* save the next phrase start and stop the cycle*/
phrase_start_item = next_token_parent;
start_token = current_token;
current_token = NULL;
}
}
else
{
phrase_start_item = NULL;
start_token = NULL;
}
}
tokens_count++;
}
/* do we need to cut it? */
if (current_token != NULL && last_safe_cut_token != start_token)
{
tokens_count = last_safe_cut_count;
current_token = SItemNext(last_safe_cut_token, error);
if (S_CHK_ERR(error, S_CONTERR,
"call_hunpos",
"Call to \"SItemNext\" failed"))
return;
}
/* do the tagging */
int hunpos_error = 0;
hunpos_tagger_tag(hunposProc->hunpos_instance, tokens_count, data, &read_token, data, &set_tag, &hunpos_error);
if (hunpos_error !=0)
{
S_CTX_ERR(error, S_FAILURE,
"call_hunpos",
"Call to \"hunpos_tagger_tag\" failed");
return;
}
/* if we're using phrases, go on with the next one */
if (is_phrase_present && current_token == NULL)
current_token = start_token;
}
return;
}
