/*
* this is for Georg's accent levels:
* unaccented : FALSE
* accented : TRUE
*/
static s_bool syl_is_accented(const SItem *syl, s_erc *error)
{
const char *accent_feat;
S_CLR_ERR(error);
accent_feat = SItemGetString(syl, "accent", error);
if (S_CHK_ERR(error, S_CONTERR,
"syl_is_accented",
"Call to \"SItemGetString\" failed"))
return FALSE;
if (s_strcmp(accent_feat, "unaccented", error) == 0)
{
S_CHK_ERR(error, S_CONTERR,
"syl_is_accented",
"Call to \"s_strcmp\" failed");
return FALSE;
}
if (s_strcmp(accent_feat, "accented", error) == 0)
{
if (S_CHK_ERR(error, S_CONTERR,
"syl_is_accented",
"Call to \"s_strcmp\" failed"))
return FALSE;
return TRUE;
}
return FALSE; /* unknown */
}
/*
* Content word's "cat" feature is 'content'
*/
static s_bool word_is_content(const SItem *word, s_erc *error)
{
const char *cat_feat;
s_bool comp;
S_CLR_ERR(error);
cat_feat = SItemGetString(word, "cat", error);
if (S_CHK_ERR(error, S_CONTERR,
"word_is_content",
"Call to \"SItemGetString\" failed"))
return FALSE;
comp = s_strcmp(cat_feat, "content", error);
if (S_CHK_ERR(error, S_CONTERR,
"syl_is_stressed",
"Call to \"s_strcmp\" failed"))
return FALSE;
if (comp == 0)
return TRUE;
return FALSE; /* not content word */
}
/*
* this is for Georg's stress levels:
* unstressed : FALSE
* primary : TRUE
* secondary : TRUE
*/
static s_bool syl_is_stressed(const SItem *syl, s_erc *error)
{
const char *stress_feat;
S_CLR_ERR(error);
stress_feat = SItemGetString(syl, "stress", error);
if (S_CHK_ERR(error, S_CONTERR,
"syl_is_stressed",
"Call to \"SItemGetString\" failed"))
return FALSE;
if (s_strcmp(stress_feat, "unstressed", error) == 0)
{
S_CHK_ERR(error, S_CONTERR,
"syl_is_stressed",
"Call to \"s_strcmp\" failed");
return FALSE;
}
if (s_strcmp(stress_feat, "primary", error) == 0)
{
if (S_CHK_ERR(error, S_CONTERR,
"syl_is_stressed",
"Call to \"s_strcmp\" failed"))
return FALSE;
return TRUE;
}
if (s_strcmp(stress_feat, "secondary", error) == 0)
{
if (S_CHK_ERR(error, S_CONTERR,
"syl_is_stressed",
"Call to \"s_strcmp\" failed"))
return FALSE;
return TRUE;
}
return FALSE; /* unknown */
}
/* 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 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 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;
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 SVoice *voice;
const SPhoneset *phoneset;
const char *feature;
const char *begin;
const char *token;
s_bool feature_value;
S_CLR_ERR(error);
if (item == NULL)
return NULL;
voice = SItemVoice(item, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemVoice\" failed"))
return NULL;
if (voice == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"Run",
"Item voice is NULL, voice is required to get phoneset");
return NULL;
}
phoneset = S_PHONESET(SVoiceGetData(voice, "phoneset", error));
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SVoiceGetData\" failed"))
return NULL;
if (phoneset == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"Run",
"Phoneset is NULL, phoneset is required to get phone feature");
return NULL;
}
token = SItemGetString(item, "_phoneset_feature", error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SItemGetString\" failed"))
return NULL;
/* get part of token (actual feature), after _ */
begin = s_strchr(token, '_', error);
if (S_CHK_ERR(error, S_CONTERR,
"SItemPath",
"Call to \"s_strchr\" failed"))
return NULL;
if (begin != NULL)
{
feature = begin + 1;
}
else
{
/* no feature */
return NULL;
}
feature_value = S_PHONESET_CALL(phoneset, phone_has_feature)(phoneset,
SItemGetName(item, error),
feature,
error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"phone_has_feature/SItemGetName\" failed"))
return NULL;
if (feature_value == TRUE)
{
extractedFeat = SObjectSetInt(1, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
return NULL;
}
else
{
extractedFeat = SObjectSetInt(0, error);
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SObjectSetInt\" failed"))
return NULL;
}
return extractedFeat;
return NULL;
S_UNUSED(self);
}
static const SPhoneset *_get_phoneset(const SItem *item, s_bool *multilingual, s_erc *error)
{
const SPhoneset *phoneset;
const SVoice *voice;
s_bool is_present;
S_CLR_ERR(error);
/* get the voice */
voice = SItemVoice(item, error);
if (S_CHK_ERR(error, S_CONTERR,
"_get_phoneset",
"Call to \"SItemVoice\" failed"))
return NULL;
if (voice == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"_get_phoneset",
"Item voice is NULL, voice is required to get phoneset");
return NULL;
}
/*
* do we have a 'voices' feature in the voice,
* i.e. is this a multilingual voice
*/
is_present = SVoiceFeatureIsPresent(voice, "voices", error);
if (S_CHK_ERR(error, S_CONTERR,
"_get_phoneset",
"Call to \"SVoiceFeatureIsPresent\" failed"))
return NULL;
if (is_present)
{
/* This is a multilingual voice.
* Get language feature of item, which is language feature
* of item's token.
*/
const SItem *tokenItem;
const char *lang;
const SMap *voicesMap;
const SVoice *thisVoice;
(*multilingual) = TRUE;
tokenItem = SItemPathToItem(item, "R:SylStructure.parent.R:Token.parent",
error);
if (S_CHK_ERR(error, S_CONTERR,
"_get_phoneset",
"Call to \"SItemPathToItem\" failed"))
return NULL;
if (tokenItem == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"_get_phoneset",
"Failed to find item's token, which is required to get language feature");
return NULL;
}
lang = SItemGetString(tokenItem, "lang", error);
if (S_CHK_ERR(error, S_CONTERR,
"_get_phoneset",
"Call to \"SItemGetString\" failed"))
return NULL;
/* now get the phoneset */
voicesMap = (const SMap*)SVoiceGetFeature(voice, "voices", error);
if (S_CHK_ERR(error, S_CONTERR,
"_get_phoneset",
"Call to \"SVoiceGetFeature\" failed"))
return NULL;
thisVoice = (const SVoice*)SMapGetObjectDef(voicesMap, lang, NULL, error);
if (S_CHK_ERR(error, S_CONTERR,
"_get_phoneset",
"Call to \"SMapGetObjectDef\" failed"))
return NULL;
if (thisVoice == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"_get_phoneset",
"Failed to find the voice for language '%s', which is required to get the phoneset", lang);
return NULL;
}
phoneset = S_PHONESET(SVoiceGetData(thisVoice, "phoneset", error));
if (S_CHK_ERR(error, S_CONTERR,
"_get_phoneset",
"Call to \"SVoiceGetData\" failed"))
return NULL;
}
else
{
/* not multilingual voice */
(*multilingual) = FALSE;
phoneset = S_PHONESET(SVoiceGetData(voice, "phoneset", error));
if (S_CHK_ERR(error, S_CONTERR,
"Run",
"Call to \"SVoiceGetData\" failed"))
return NULL;
}
if (phoneset == NULL)
{
S_CTX_ERR(error, S_FAILURE,
"_get_phoneset",
"Item phoneset is NULL, required to extract phone features");
return NULL;
}
return phoneset;
}
