// returns a new instance of FieldInfoList which represents field entities in a document index for a specific field, you must delete the instance later. @see FieldInfoList
// Note that not all index types support fields - those that do should override this method.
lemur::api::FieldInfoList *lemur::index::LemurIndriIndex::fieldInfoList(lemur::api::DOCID_T docID, int fieldID) const {
// get the index for this document
indri::collection::Repository::index_state indexes = _repository->indexes();
indri::index::Index* index = _indexWithDocument( indexes, docID );
// ensure we do have an index (i.e. if the docID was invalid...)
if (!index) return NULL;
// and the indri term-list
const indri::index::TermList *tList=index->termList((int)docID);
// ensure we have a term list!
if (!tList) return NULL;
// create a blank field info list object
lemur::api::IndriFieldInfoList *retVal=new lemur::api::IndriFieldInfoList();
// loop through our fields and insert those that match the field ID
const indri::utility::greedy_vector< indri::index::FieldExtent > fieldVec=tList->fields();
int numFields=fieldVec.size();
for (int i=0; i < numFields; i++) {
indri::index::FieldExtent thisField=fieldVec[i];
if (thisField.id==fieldID) {
retVal->add(thisField);
}
}
delete tList;
// and return
return retVal;
}
lemur::api::COUNT_T lemur::index::LemurIndriIndex::docLength( lemur::api::DOCID_T documentID ) const {
indri::collection::Repository::index_state indexes = _repository->indexes();
indri::index::Index* index = _indexWithDocument( indexes, documentID );
if( index ) {
indri::thread::ScopedLock lock( index->statisticsLock() );
return index->documentLength( documentID );
}
return 0;
}
int indri::server::LocalQueryServer::documentLength( lemur::api::DOCID_T documentID ) {
indri::collection::Repository::index_state indexes = _repository.indexes();
indri::index::Index* index = _indexWithDocument( indexes, documentID );
if( index ) {
indri::thread::ScopedLock lock( index->statisticsLock() );
return index->documentLength( documentID );
}
return 0;
}
lemur::api::TermInfoList* lemur::index::LemurIndriIndex::termInfoListSeq(lemur::api::DOCID_T docID) const {
indri::collection::Repository::index_state indexes = _repository->indexes();
indri::index::Index* index = _indexWithDocument( indexes, docID );
lemur::api::TermInfoList *list = NULL;
if( index ) {
indri::thread::ScopedLock lock( index->statisticsLock() );
const indri::index::TermList* termList = index->termList( docID );
list = new indri::index::PositionList(termList);
}
return list;
}
indri::server::QueryServerMetadataResponse* indri::server::LocalQueryServer::pathNames( const std::vector<lemur::api::DOCID_T>& documentIDs, const std::vector<int>& pathBegins, const std::vector<int>& pathEnds ) {
int lastDoc = 0;
indri::index::DocumentStructure docStruct;
std::vector<std::string> result;
std::vector<std::pair<lemur::api::DOCID_T, int> > docSorted;
for( size_t i=0; i<documentIDs.size(); i++ ) {
docSorted.push_back( std::make_pair( documentIDs[i], i ) );
}
std::sort( docSorted.begin(), docSorted.end() );
for( size_t i=0; i<docSorted.size(); i++ ) {
indri::collection::Repository::index_state indexes = _repository.indexes();
bool docStructLoaded = true;
lemur::api::DOCID_T documentID = docSorted[i].first;
if ( documentID != lastDoc ) {
indri::index::Index * index = _indexWithDocument(indexes, documentID);
const indri::index::TermList * termList = index->termList( documentID );
if ( termList != 0 ) {
docStruct.setIndex( *index );
docStruct.loadStructure( termList->fields() );
delete termList;
lastDoc = docStructLoaded;
} else {
docStructLoaded = false;
}
}
std::string path = "";
if ( docStructLoaded ) {
path = docStruct.path( docStruct.findLeaf( pathBegins[docSorted[i].second],
pathEnds[docSorted[i].second] ) );
}
result.push_back( path );
}
std::vector<std::string> actual;
actual.resize( documentIDs.size() );
for( size_t i=0; i<docSorted.size(); i++ ) {
actual[docSorted[i].second] = result[i];
}
return new indri::server::LocalQueryServerMetadataResponse( actual );
}
indri::server::QueryServerVectorsResponse* indri::server::LocalQueryServer::documentVectors( const std::vector<lemur::api::DOCID_T>& documentIDs ) {
indri::server::LocalQueryServerVectorsResponse* response = new indri::server::LocalQueryServerVectorsResponse( (int)documentIDs.size() );
indri::collection::Repository::index_state indexes = _repository.indexes();
std::map<int, std::string> termIDStringMap;
for( size_t i=0; i<documentIDs.size(); i++ ) {
indri::index::Index* index = _indexWithDocument( indexes, documentIDs[i] );
{
indri::thread::ScopedLock lock( index->statisticsLock() );
const indri::index::TermList* termList = index->termList( documentIDs[i] );
indri::api::DocumentVector* result = new indri::api::DocumentVector( index, termList, termIDStringMap );
delete termList;
response->addVector( result );
}
}
return response;
}
// returns a new instance of FieldInfoList which represents all field entities in a document index, you must delete the instance later. @see FieldInfoList
// Note that not all index types support fields - those that do should override this method.
lemur::api::FieldInfoList *lemur::index::LemurIndriIndex::fieldInfoList(lemur::api::DOCID_T docID) const {
// get the index for this document
indri::collection::Repository::index_state indexes = _repository->indexes();
indri::index::Index* index = _indexWithDocument( indexes, docID );
// ensure we do have an index (i.e. if the docID was invalid...)
if (!index) return NULL;
// and the indri term-list
const indri::index::TermList *tList=index->termList((int)docID);
// ensure we have a term list!
if (!tList) return NULL;
// create our field info list object
lemur::api::IndriFieldInfoList *retVal=new lemur::api::IndriFieldInfoList(tList->fields());
delete tList;
// and return
return retVal;
}
lemur::api::TermInfoList* lemur::index::LemurIndriIndex::termInfoListSeq(lemur::api::DOCID_T docID) const {
indri::collection::Repository::index_state indexes = _repository->indexes();
indri::index::Index* index = _indexWithDocument( indexes, docID );
lemur::api::TermInfoList *list = NULL;
if( index ) {
indri::thread::ScopedLock lock( index->statisticsLock() );
const indri::index::TermList* termList = index->termList( docID );
indri::utility::greedy_vector<lemur::api::TERMID_T> termIDs = termList->terms();
std::vector<lemur::index::LocatedTerm> locs;
// indri indexes stopwords as [OOV], so skip those
for( int i = 0; i < termIDs.size(); i++) {
if (termIDs[i] != 0) {
lemur::index::LocatedTerm lt;
lt.term = termIDs[i];
lt.loc = i;
locs.push_back(lt);
}
}
list = new lemur::index::InvFPTermList(docID, locs.size(), locs);
delete(termList);
}
return list;
}
