SEXP generateSnippets(bool html){
vector<string> snippetString;
vector< indri::api::ParsedDocument* > pdocuments = environment.documents(documentIDs);
indri::api::SnippetBuilder sp(html);
for( size_t row=0; row < documentIDs.size(); row++ )
snippetString.push_back(sp.build(documentIDs[row], pdocuments[row], qa));
CharacterVector c = wrap(snippetString);
c.attr("names") = extDocIDs;
return c;
}
void _printResultRegion( std::stringstream& output, std::string queryIndex, int start, int end ) {
std::vector<std::string> documentNames;
std::vector<indri::api::ParsedDocument*> documents;
std::vector<indri::api::ScoredExtentResult> resultSubset;
resultSubset.assign( _results.begin() + start, _results.begin() + end );
// Fetch document data for printing
if( _printDocuments || _printPassages || _printSnippets ) {
// Need document text, so we'll fetch the whole document
documents = _environment.documents( resultSubset );
documentNames.clear();
for( size_t i=0; i<resultSubset.size(); i++ ) {
indri::api::ParsedDocument* doc = documents[i];
std::string documentName;
indri::utility::greedy_vector<indri::parse::MetadataPair>::iterator iter = std::find_if( documents[i]->metadata.begin(),
documents[i]->metadata.end(),
indri::parse::MetadataPair::key_equal( "docno" ) );
if( iter != documents[i]->metadata.end() )
documentName = (char*) iter->value;
// store the document name in a separate vector so later code can find it
documentNames.push_back( documentName );
}
} else {
// We only want document names, so the documentMetadata call may be faster
documentNames = _environment.documentMetadata( resultSubset, "docno" );
}
std::vector<std::string> pathNames;
if ( _inexFormat ) {
// retrieve path names
pathNames = _environment.pathNames( resultSubset );
}
// Print results
for( size_t i=0; i < resultSubset.size(); i++ ) {
int rank = start+i+1;
std::string queryNumber = queryIndex;
if( _trecFormat ) {
// TREC formatted output: queryNumber, Q0, documentName, rank, score, runID
output << queryNumber << " "
<< "Q0 "
<< documentNames[i] << " "
<< rank << " "
<< resultSubset[ i ].score << " "
<< _runID << std::endl;
} else if( _inexFormat ) {
output << " <result>" << std::endl
<< " <file>" << documentNames[i] << "</file>" << std::endl
<< " <path>" << pathNames[i] << "</path>" << std::endl
<< " <rsv>" << resultSubset[i].score << "</rsv>" << std::endl
<< " </result>" << std::endl;
}
else {
// score, documentName, firstWord, lastWord
output << resultSubset[i].score << "\t"
<< documentNames[i] << "\t"
<< resultSubset[i].begin << "\t"
<< resultSubset[i].end << std::endl;
}
if( _printDocuments ) {
output << documents[i]->text << std::endl;
}
if( _printPassages ) {
int byteBegin = documents[i]->positions[ resultSubset[i].begin ].begin;
int byteEnd = documents[i]->positions[ resultSubset[i].end-1 ].end;
output.write( documents[i]->text + byteBegin, byteEnd - byteBegin );
output << std::endl;
}
if( _printSnippets ) {
indri::api::SnippetBuilder builder(false);
output << builder.build( resultSubset[i].document, documents[i], _annotation ) << std::endl;
}
if( documents.size() )
delete documents[i];
}
}