LimaStatusCode EntityTrackerXmlLogger::process(
AnalysisContent& analysis) const
{
SELOGINIT;
LDEBUG << "EntityTrackerXmlLogger::process";
TimeUtils::updateCurrentTime();
/* permet de récupérer les annotations */
//AnnotationData* annotationData = static_cast< AnnotationData* >(analysis.getData("AnnotationData"));
/* recupérer le graph après l'analyse */
//const LinguisticAnalysisStructure::AnalysisGraph& graph = *(static_cast<LinguisticAnalysisStructure::AnalysisGraph*>(analysis.getData(m_graph)));
LinguisticMetaData* metadata=static_cast<LinguisticMetaData*>(analysis.getData("LinguisticMetaData"));
if (metadata == 0) {
SELOGINIT;
LERROR << "no LinguisticMetaData ! abort";
return MISSING_DATA;
}
CoreferenceData* corefData=static_cast<CoreferenceData*>(analysis.getData("CoreferenceData"));
if (corefData == 0) {
SELOGINIT;
LERROR << "no CoreferenceData ! abort";
return MISSING_DATA;
}
ofstream out;
if (!openLogFile(out,metadata->getMetaData("FileName"))) {
SELOGINIT;
LERROR << "Can't open log file '" << metadata->getMetaData("FileName") << "'";
return UNKNOWN_ERROR;
}
out << "<coreference>" << endl;
for (CoreferenceData::const_iterator it=corefData->begin(),
it_end=corefData->end(); it != it_end; it++)
{
out << "<entity mentions=\"" << (*it).size() << "\">" << endl;
for (vector<Token>::const_iterator it2=(*it).begin(), it2_end=(*it).end();
it2 != it2_end; it2++)
{
out << " <entity_mention>"
<< limastring2utf8stdstring((*it2).stringForm())
<<"</entity_mention>";
}
out << "<entity>" <<endl;
}
out.close();
return SUCCESS_ID;
}
void CoreXmlReaderClient::handle(const DocumentsReader::ContentStructuredDocument &contentDocument,
const Lima::LimaString &text, unsigned long int offset, const string tagName)
{
#ifdef DEBUG_LP
XMLREADERCLIENTLOGINIT;
#endif
if(std::string(text.toUtf8().constData()).find_first_not_of(m_emptyTextChars.toUtf8().constData()) == string::npos) {
#ifdef DEBUG_LP
LDEBUG << "CoreXmlReaderClient::empty text, not analyzed";
#endif
return;
}
AbstractStructuredDocumentElement* absElement = contentDocument.back();
DocumentsReader::IndexingDocumentElement* element =
dynamic_cast<DocumentsReader::IndexingDocumentElement*>(absElement);
std::string elementName = element->getElementName().toUtf8().constData();
#ifdef DEBUG_LP
if( logger.loggingLevel() == QsLogging::DebugLevel )
{
LDEBUG << "CoreXmlReaderClient::handle"
<< "[" << text << "], offset =" << offset
<< ", tagName =" << tagName << ", element name =" << elementName ;
}
else if( logger.loggingLevel() == QsLogging::InfoLevel )
#endif
{
// Chercher les analyses diponibles
XMLREADERCLIENTLOGINIT;
LINFO << "CoreXmlReaderClient::handle"
<< "[" << text.left(50) << "], offset =" << offset
<< ", tagName =" << tagName ;
}
ostringstream os;
os << offset;
m_docMetaData["StartOffset"] = os.str();
m_docMetaData["ElementName"] = tagName;
// Set the language to the one associated at init time to the current tag
if (m_mapTagMedia.find(elementName) != m_mapTagMedia.end())
{
#ifdef DEBUG_LP
LDEBUG << "CoreXmlReaderClient::handle using media" << m_mapTagMedia[elementName];
#endif
m_docMetaData["Lang"] = m_mapTagMedia[elementName];
}
else if (!m_defaultMedia.empty())
{
#ifdef DEBUG_LP
LDEBUG << "CoreXmlReaderClient::handle using default media" << m_defaultMedia;
#endif
m_docMetaData["Lang"] = m_defaultMedia;
}
else
{
XMLREADERCLIENTLOGINIT;
LERROR << "CoreXmlReaderClient::handle no media associated to tag"
<< elementName
<< "and no default media is set. metadata Lang will not be set.";
}
// cast element to GenericDocumentProperties
// Common::Misc::GenericDocumentProperties &props = *element;
// get byte offset after end of element
unsigned long offsetIndexingNode = element->getIntValue("offBegPrpty").first;
ostringstream os2;
os2 << offsetIndexingNode;
m_docMetaData["StartOffsetIndexingNode"] = os2.str();
string strText = limastring2utf8stdstring(text);
// size_t posEmptyTextChars = strText.find_first_not_of(m_emptyTextChars.toUtf8().constData());
// if (posEmptyTextChars!=string::npos)
// strText=strText.substr(posEmptyTextChars,strText.length()-posEmptyTextChars);
m_handler->handleProc(
tagName,
strText,
m_docMetaData,
m_docMetaData["pipeline"],
m_mapHandlers,
std::set<std::string>());
}
void SyntacticAnalysisXmlLogger::outputVertex(const LinguisticGraphVertex v,
const LinguisticGraph& graph,
const uint64_t offsetBegin,
const SyntacticData* syntacticData,
std::ostream& xmlStream,
std::map< LinguisticAnalysisStructure::Token*, uint64_t >& tokens,
std::vector< bool >& alreadyDumpedTokens) const
{
if (v == syntacticData->iterator()->firstVertex() ||
v == syntacticData->iterator()->lastVertex())
{
xmlStream << "<vertex id=\"" << v << "\" />" << std::endl;
return;
}
Token* token = get(vertex_token, graph, v);
uint64_t tokenId = (*(tokens.find(token))).second;
// bool alreadyDumped = alreadyDumpedTokens[tokenId];
xmlStream << "<vertex id=\"" << v << "\" form=\"" << limastring2utf8stdstring(token->stringForm()) << "\" pos=\"" << getPosition(token->position(),offsetBegin) << "\" ";
const VertexChainIdProp& chains = get(vertex_chain_id, graph,v);
xmlStream << " >" << std::endl;
if (chains.size() > 0)
{
xmlStream << "<chains>" << std::endl;
VertexChainIdProp::const_iterator itChains, itChains_end;
itChains = chains.begin();
itChains_end = chains.end();
for (; itChains != itChains_end; itChains++)
{
const ChainIdStruct& ids = (*itChains);
xmlStream << "<chain type=\"";
if (ids.chainType() == Common::MediaticData::NO_CHAIN_TYPE)
xmlStream << "0";
else if (ids.chainType() == Common::MediaticData::NOMINAL)
xmlStream << "N";
else
xmlStream << "V";
xmlStream << "\" id=\"" << (ids.chainId()) << "\" />" << std::endl;
}
xmlStream << "</chains>" << std::endl;
}
const DependencyGraph* depGraph = syntacticData->dependencyGraph();
DependencyGraphVertex depV = syntacticData->depVertexForTokenVertex(v);
if (out_degree(depV, *depGraph) > 0)
{
xmlStream << "<dependents>" << std::endl;
DependencyGraphOutEdgeIt depIt, depIt_end;
boost::tie(depIt, depIt_end) = out_edges(depV, *depGraph);
for (; depIt != depIt_end; depIt++)
{
DependencyGraphVertex depTargV = target(*depIt, *depGraph);
LinguisticGraphVertex targV = syntacticData-> tokenVertexForDepVertex(depTargV);
// CEdgeDepChainIdPropertyMap chainsMap = get(edge_depchain_id, *depGraph);
CEdgeDepRelTypePropertyMap relTypeMap = get(edge_deprel_type, *depGraph);
xmlStream << "<dep v=\"" << targV;
// xmlStream << "\" c=\"" << chainsMap[*depIt];
std::string relName=static_cast<const Common::MediaticData::LanguageData&>(Common::MediaticData::MediaticData::single().mediaData(m_language)).getSyntacticRelationName(relTypeMap[*depIt]);
if (relName.empty())
{
relName="UNKNOWN";
}
xmlStream << "\" t=\"" << relName << "\" />" << std::endl;
}
xmlStream << "</dependents>" << std::endl;
}
const FsaStringsPool& sp=Common::MediaticData::MediaticData::single().stringsPool(m_language);
MorphoSyntacticData* word = get(vertex_data, graph, v);
word->outputXml(xmlStream,*m_propertyCodeManager,sp);
xmlStream << "<ref>" << tokenId << "</ref>" << std::endl;
alreadyDumpedTokens[tokenId] = true;
xmlStream << "</vertex>" << std::endl;
}
LimaStatusCode CorefSolvingNormalizedXmlLogger::process(
AnalysisContent& analysis) const
{
// COREFSOLVERLOGINIT;
TimeUtils::updateCurrentTime();
AnnotationData* annotationData = static_cast<AnnotationData*>(analysis.getData("AnnotationData"));
const LinguisticAnalysisStructure::AnalysisGraph& graph = *(static_cast<LinguisticAnalysisStructure::AnalysisGraph*>(analysis.getData(m_graph)));
// LinguisticGraph* lingGraph = const_cast<LinguisticGraph*>(graph.getGraph());
LinguisticMetaData* metadata=static_cast<LinguisticMetaData*>(analysis.getData("LinguisticMetaData"));
if (metadata == 0)
{
COREFSOLVERLOGINIT;
LERROR << "no LinguisticMetaData ! abort" << LENDL;
return MISSING_DATA;
}
ofstream out;
if (!openLogFile(out,metadata->getMetaData("FileName")))
{
COREFSOLVERLOGINIT;
LERROR << "Can't open log file " << LENDL;
return UNKNOWN_ERROR;
}
out << "<coreferences>" << endl;
// LDEBUG << "CorefSolvingNormalizedXmlLogger on graph " << m_graph << LENDL;
AnnotationGraphVertexIt itv, itv_end;
boost::tie(itv, itv_end) = vertices(annotationData->getGraph());
for (; itv != itv_end; itv++)
{
// process
//LDEBUG << "CorefSolvingNormalizedXmlLogger on annotation vertex " << *itv << LENDL;
if (annotationData->hasAnnotation(*itv,utf8stdstring2limastring("Coreferent")))
//if (annotationData->hasAnnotation(*itv,utf8stdstring2limastring("Coreferent")))
{
CoreferentAnnotation* annot ;
try
{
annot = annotationData->annotation(*itv,utf8stdstring2limastring("Coreferent"))
.pointerValue<CoreferentAnnotation>();
}
catch (const boost::bad_any_cast& )
{
COREFSOLVERLOGINIT;
LERROR << "One annotation on vertex " << *itv << " you are trying to cast is not a Coreference; Coreference not logged" << LENDL;
for (int i = 0; i < 19 ; i++)
{
LERROR << "annot "<< i << " : " << limastring2utf8stdstring(annotationData->annotationName(i)) << LENDL ;
}
continue;
}
LinguisticProcessing::LinguisticAnalysisStructure::Token* token = get(vertex_token, *graph.getGraph(), annot->morphVertex());
if (token == 0)
{
COREFSOLVERLOGINIT;
LERROR << "Vertex " << *itv << " has no entry in the analysis graph token map. This should not happen !!" << LENDL;
}
else
{
CoreferentAnnotation* antecedent;
// bool hasAntecedent = false;
AnnotationGraphOutEdgeIt it, it_end;
boost::tie(it, it_end) = boost::out_edges(static_cast<AnnotationGraphVertex>(*itv), annotationData->getGraph());
for (; it != it_end; it++)
{
if (annotationData->hasAnnotation(target(*it,annotationData->getGraph()),utf8stdstring2limastring("Coreferent")))
{
try
{
antecedent = annotationData->annotation(target(*it, annotationData->getGraph()), utf8stdstring2limastring("Coreferent")).pointerValue<CoreferentAnnotation>();
// hasAntecedent = true;
}
catch (const boost::bad_any_cast& )
{
COREFSOLVERLOGINIT;
LERROR << "One annotation on vertex you are trying to cast resulting from an edge out of " << *itv << " is not a Coreference; Coreference not logged" << LENDL;
continue;
}
}
}
out << " <reference>\n"
<< " <pos>" << get(vertex_token,*graph.getGraph(),annot->morphVertex())->position() << "</pos>\n"
<< " <len>" << token->stringForm().length() << "</len>\n"
<< " <string>"<< limastring2utf8stdstring(transcodeToXmlEntities(token->stringForm())) << "</string>\n"
<< " <npId>" << annot->id() << "</npId>\n"
<< " <posVertex>" << annot->morphVertex() << "</posVertex>\n";
//if (hasAntecedent)
if (false)
{
out << " <npRef>" << antecedent->id() << "</npRef>\n";
out << " <refPosVertex>" << antecedent->morphVertex() << "</refPosVertex>\n";
}
out << " <categ>" << annot->categ() << "</categ>\n"
<< " </reference>\n"
<< endl;
}
}
}
out << "</coreferences>" << endl;
out.close();
TimeUtils::logElapsedTime("CorefSolvingNormalizedXmlLogger");
return SUCCESS_ID;
}
