LTR_Text::LTR_Text(const LTR_Text& text):
std::vector<LTR_Token*>()
{
this->reserve(text.size());
for (LTR_Text::const_iterator itTok = text.begin();
itTok != text.end(); ++ itTok) {
this->push_back(new LTR_Token(**itTok));
}
for (SENTENCE_BOUNDS_T::const_iterator itSentBd = text.m_sentenceBounds.begin();
itSentBd != text.m_sentenceBounds.end(); ++ itSentBd) {
m_sentenceBounds.push_back(*itSentBd);
}
}
LimaStatusCode LinearTextRepresentationDumper::process(
AnalysisContent& analysis) const {
DUMPERLOGINIT;
// get metadata
LinguisticMetaData* metadata=dynamic_cast<LinguisticMetaData*>(analysis.getData("LinguisticMetaData"));
if (metadata == 0) {
LERROR << "LinearTextRepresentationDumper::process: no LinguisticMetaData ! abort" << LENDL;
return MISSING_DATA;
}
// get the analysis graph
AnalysisGraph* anaGraph = dynamic_cast<AnalysisGraph*>(analysis.getData("PosGraph"));
if (anaGraph == 0) {
LERROR << "LinearTextRepresentationDumper::process: no AnalysisGraph ! abort" << LENDL;
return MISSING_DATA;
}
// get sentence boundaries
SegmentationData* sb = dynamic_cast<SegmentationData*>(analysis.getData("SentenceBoundaries"));
if (sb == 0) {
LERROR << "LinearTextRepresentationDumper::process: no SentenceBounds ! abort" << LENDL;
return MISSING_DATA;
}
// build LTRText
LTR_Text textRep;
LTRTextBuilder builder(m_language, m_stopList);
builder.buildLTRTextFrom(
*(anaGraph->getGraph()),
sb,
anaGraph->lastVertex(),
&textRep,
metadata->getStartOffset());
// write LTR_Text
LDEBUG << "handler will be: " << m_handler << LENDL;
// MediaId langid = static_cast<const Common::MediaticData::LanguageData&>(Common::MediaticData::MediaticData::single().mediaData(metadata->getMetaData("Lang"))).getMedia();
AnalysisHandlerContainer* h = static_cast<AnalysisHandlerContainer*>(analysis.getData("AnalysisHandlerContainer"));
AbstractTextualAnalysisHandler* handler = static_cast<AbstractTextualAnalysisHandler*>(h->getHandler(m_handler));
if (handler == 0) {
LERROR << "LinearTextRepresentationDumper::process: handler " << m_handler << " has not been given to the core client" << LENDL;
return MISSING_DATA;
}
handler->startAnalysis();
HandlerStreamBuf hsb(handler);
ostream out(&hsb);
LDEBUG << textRep << LENDL;
textRep.binaryWriteOn(out);
out.flush();
handler->endAnalysis();
return SUCCESS_ID;
}
void LTR_Text::appendLTR_Text(const LTR_Text& ltr) {
LTR_Text::size_type thiSize = this->size();
// copy of tokens
this->reserve(thiSize + ltr.size());
for (LTR_Text::const_iterator itTok = ltr.begin();
itTok != ltr.end(); ++ itTok) {
LTR_Token* tok = new LTR_Token(**itTok);
this->push_back(tok);
}
// add sentence bounds
for (LTR_Text::SENTS_CONST_ITER_T itSentBd = ltr.beginSentenceBounds();
itSentBd != ltr.endSentenceBounds(); ++ itSentBd) {
m_sentenceBounds.push_back(*itSentBd + thiSize);
}
}
LimaStatusCode LinearTextRepresentationLogger::process(
AnalysisContent& analysis) const {
DUMPERLOGINIT;
// get metadata
LinguisticMetaData* metadata=dynamic_cast<LinguisticMetaData*>(analysis.getData("LinguisticMetaData"));
if (metadata == 0) {
LERROR << "no LinguisticMetaData ! abort";
return MISSING_DATA;
}
// get the analysis graph
AnalysisGraph* anaGraph = dynamic_cast<AnalysisGraph*>(analysis.getData("PosGraph"));
if (anaGraph == 0) {
LERROR << "no AnalysisGraph ! abort";
return MISSING_DATA;
}
// get sentence boundaries
SegmentationData* sb = dynamic_cast<SegmentationData*>(analysis.getData("SentenceBoundaries"));
if (sb == 0) {
LDEBUG << "LinearTextRepresentationDumper::process: no SentenceBounds available: ignored";
// sentence bounds ignored: null pointer passed to LTRTextBuilder will be handled there
}
// build LTRText
LTR_Text textRep;
LTRTextBuilder builder(m_language, m_stopList);
builder.buildLTRTextFrom(
*(anaGraph->getGraph()),
sb,
anaGraph->firstVertex(),
anaGraph->lastVertex(),
&textRep,
metadata->getStartOffset());
// write LTR_Text
string textFileName = metadata->getMetaData("FileName");
string outputFile = textFileName + m_outputSuffix;
ofstream out(outputFile.c_str(), std::ofstream::binary);
if (!out.good()) {
throw runtime_error("can't open file " + outputFile);
}
textRep.binaryWriteOn(out);
out.flush();
out.close();
return SUCCESS_ID;
}
