std::string SemanticRelationsXmlLogger::
vertexStringForSemanticAnnotation(const std::string& vertexRole,
const AnnotationGraphVertex& vertex,
const VertexTokenPropertyMap& tokenMap,
AnnotationData* annotationData,
uint64_t offset) const
{
ostringstream oss;
// get id of the corresponding vertex in analysis graph
LinguisticGraphVertex v;
if (!annotationData->hasIntAnnotation(vertex,Common::Misc::utf8stdstring2limastring(m_graph)))
{
// SEMLOGINIT;
// LDEBUG << *itv << " has no " << m_graph << " annotation. Skeeping it." << LENDL;
return "";
}
v = annotationData->intAnnotation(vertex,Common::Misc::utf8stdstring2limastring(m_graph));
LinguisticAnalysisStructure::Token* vToken = tokenMap[v];
// LDEBUG << "SemanticRelationsXmlLogger tokenMap[" << v << "] = " << vToken << LENDL;
if (vToken == 0)
{
SEMLOGINIT;
LERROR << "Vertex " << v << " has no entry in the analysis graph token map. This should not happen !!" << LENDL;
return "";
}
// get annotation : element in relation can be an entity => get entity type
// otherwise, its type is "token"
std::string type("token");
std::set< uint32_t > matches = annotationData->matches(m_graph,v,"annot");
for (std::set< uint32_t >::const_iterator it = matches.begin();
it != matches.end(); it++)
{
if (annotationData->hasAnnotation(*it,Common::Misc::utf8stdstring2limastring("SpecificEntity"))) {
const SpecificEntityAnnotation* annot = 0;
try {
annot = annotationData->annotation(*it,Common::Misc::utf8stdstring2limastring("SpecificEntity"))
.pointerValue<SpecificEntityAnnotation>();
}
catch (const boost::bad_any_cast& e) {
SEMLOGINIT;
LERROR << "This annotation is not a SemanticAnnotation" << LENDL;
continue;
}
type=Common::Misc::limastring2utf8stdstring(Common::LinguisticData::LinguisticData::single().getEntityName(annot->getType()));
break;
}
}
oss << " <" << vertexRole
<< " type=\"" << type << "\""
<< " pos=\"" << offset+vToken->position() << "\""
<< " len=\"" << vToken->length() << "\""
<< " string=\"" << vToken->stringForm() << "\""
<< "/>" << endl;
return oss.str();
}
void SpecificEntitiesXmlLogger::
outputEntity(std::ostream& out,
LinguisticGraphVertex v,
const SpecificEntityAnnotation* annot,
const VertexTokenPropertyMap& tokenMap,
uint64_t offset) const
{
LinguisticAnalysisStructure::Token* vToken = tokenMap[v];
// LDEBUG << "SpecificEntitiesXmlLogger tokenMap[" << v << "] = " << vToken;
if (vToken == 0)
{
SELOGINIT;
LERROR << "Vertex " << v << " has no entry in the analysis graph token map. This should not happen !!";
}
else
{
if (m_compactFormat) {
// same format as RecognizerResultLogger
out
<< "<entity>"
//<< "<pos>" << offset+annot->getPosition() << "</pos>"
<< "<pos>" << offset+annot->getPosition() << "</pos>"
<< "<len>" << annot->getLength() << "</len>"
//<< "<typeNum>" << (*m).getType() << "</typeNum>"
<< "<type>"
<< Common::MediaticData::MediaticData::single().getEntityName(annot->getType()).toUtf8().data()
<< "</type>"
<< "<string>"<< Common::Misc::transcodeToXmlEntities(vToken->stringForm()).toUtf8().data() << "</string>"
<< "<norm>";
const Automaton::EntityFeatures& features=annot->getFeatures();
for (Automaton::EntityFeatures::const_iterator
featureItr=features.begin(),features_end=features.end();
featureItr!=features_end; featureItr++)
{
out << "<" << featureItr->getName() << ">"
<< Common::Misc::limastring2utf8stdstring(Common::Misc::transcodeToXmlEntities(Common::Misc::utf8stdstring2limastring(featureItr->getValueString())))
<< "</" << featureItr->getName() << ">";
}
out << "</norm>"
<< "</entity>"
<< endl;
}
else {
// recuperer le vertex morph en question dans le graphe morph
// recuperer la chaine, la position et la longueur pour ce vertex morph
out << "<specific_entity>" << endl;
out << " <string>" << Common::Misc::transcodeToXmlEntities(vToken->stringForm()).toUtf8().data() << "</string>" << endl;
out << " <position>" << vToken->position()<< "</position>" << endl;
out << " <length>" << vToken->length() << "</length>" << endl;
out << " <type>"
<< Common::MediaticData::MediaticData::single().getEntityName(annot->getType())
<< "</type>" << endl;
out << "</specific_entity>" << endl;
}
}
}