/*!
* Explicit constructor.
*/
AnnotatorPerformanceAnnotation::AnnotatorPerformanceAnnotation(CAS& aCas, const UnicodeString& annotatorName, long elapsedMillis)
: AnnotationWrapper(aCas)
{
FSIndexRepository& indexRep = aCas.getIndexRepository();
annotation = aCas.createAnnotation(tAnnotatorPerformanceAnnotation, 0, 0);
setComponentName(annotatorName);
setElapsedTime(elapsedMillis);
indexRep.addFS(annotation);
}
SentenceAnnotation::SentenceAnnotation(CAS& aCas, size_t begin, size_t end, const TokenAnnotation& firstToken, const TokenAnnotation& lastToken) :
ContextAreaAnnotation(aCas)
{
FSIndexRepository& indexRep = aCas.getIndexRepository();
annotation = aCas.createAnnotation(tSentenceAnnotation, begin, end);
setFirstToken(firstToken);
setLastToken(lastToken);
indexRep.addFS(annotation);
}
// Look for "EnglishDocument" sofa and read it as a stream
TyErrorId process(CAS & rCas, ResultSpecification const & crResultSpecification) {
cout << "SofaDataAnnotator: process() begins" << endl;
/** get the CAS view of the sofa */
CAS * tcas = rCas.getView("EnglishDocument");
/** get the handle to the index repository */
FSIndexRepository & indexRep = tcas->getIndexRepository();
/** get the default text sofa */
SofaFS textSofa = tcas->getSofa();
/** get the handle to the sofa data stream */
SofaDataStream * pStream = textSofa.getSofaDataStream();
/** open the stream */
int rc = pStream->open();
if (rc != 0) {
cout << "open failed " << rc << endl;
return (TyErrorId)UIMA_ERR_USER_ANNOTATOR_COULD_NOT_PROCESS;
}
/** get the total stream size */
size_t streamSize = pStream->getTotalStreamSizeInBytes();
/** read file contents into a buffer */
char * pBuffer = new char[streamSize+1];
memset(pBuffer,'\n' ,streamSize+1);
int elementsize=1;
pStream->read(pBuffer, elementsize, streamSize);
cout << endl;
cout.write(pBuffer, streamSize);
cout << endl;
/** convert to unicode */
UnicodeString ustrInputText(pBuffer, streamSize+1, "utf-8");
/** find tokens and annotate */
UnicodeString delim(" ");
UChar *myLocalSaveState;
UChar * pInputText = (UChar*) ustrInputText.getBuffer();
const UChar * pToken = pInputText;
const UChar * pNextToken = u_strtok_r((UChar*) pInputText, delim.getBuffer(), &myLocalSaveState);
int start = 1;
int tokenlength=0;
int nTokens = 0;
while ( (pNextToken=u_strtok_r(NULL, delim.getBuffer(), &myLocalSaveState)) ) {
tokenlength = pNextToken - pToken;
AnnotationFS annotFS = tcas->createAnnotation(annot, start, start+tokenlength-2);
indexRep.addFS(annotFS);
++nTokens;
start += tokenlength;
pToken = pNextToken;
}
/* last token */
tokenlength = pNextToken - pToken;
AnnotationFS annotFS = tcas->createAnnotation(annot, start, streamSize);
indexRep.addFS(annotFS);
++nTokens;
cout << endl << " Annotated " << nTokens << " tokens." << endl << endl;
/** close the stream */
pStream->close();
delete pStream;
delete[] pBuffer;
cout << "SofaDataAnnotator: process() ends" << endl;
return (TyErrorId)UIMA_ERR_NONE;
}
