int
Dictionary::get_document_index (ustring dictionaryname)
{
DocMap::iterator dm = m_document_map.find(dictionaryname);
int dindex;
if (dm == m_document_map.end()) {
dindex = m_documents.size();
m_document_map[dictionaryname] = dindex;
pugi::xml_document *doc = new pugi::xml_document;
m_documents.push_back (doc);
pugi::xml_parse_result parse_result;
if (boost::ends_with (dictionaryname.string(), ".xml")) {
// xml file -- read it
parse_result = doc->load_file (dictionaryname.c_str());
} else {
// load xml directly from the string
parse_result = doc->load_buffer (dictionaryname.c_str(),
dictionaryname.length());
}
if (! parse_result) {
m_context->error ("XML parsed with errors: %s, at offset %d",
parse_result.description(),
parse_result.offset);
m_document_map[dictionaryname] = -1;
return -1;
}
} else {
dindex = dm->second;
}
DASSERT (dindex < (int)m_documents.size());
return dindex;
}
Document::UP read(DocumentIdT lid, const document::DocumentTypeRepo &) const override {
DocMap::const_iterator itr = _docs.find(lid);
if (itr != _docs.end()) {
Document::UP retval(itr->second->clone());
return retval;
}
return Document::UP();
}
int findTheDot(QueryMap q_obj, DocMap w)
{ int dot=0;
for (QueryMap::iterator l=q_obj.begin(); l != q_obj.end(); ++l)
for (DocMap::iterator p = w.begin(); p != w.end(); ++p)
if (l->first == p->first)
dot+=dotprod(l->second, p->second);
return dot;
}
Doc* DocSet::findOrMakeDoc(const std::string& doc) {
DocMap::iterator it = docMap_.find(doc);
if (it != docMap_.end()) {
return it->second;
}
Doc* newDoc = new Doc(NULL);
docMap_[doc] = newDoc;
return newDoc;
}
void remove(uint64_t syncToken, DocumentIdT lid) override {
_lastSyncToken = syncToken;
_docs.erase(lid);
}
void DocSet::produceClusters(DocSet::ClusterMap& outClusters) {
for (DocMap::iterator it = docMap_.begin(); it != docMap_.end(); ++it) {
Doc* rep = it->second->findRep();
outClusters[rep].push_back(it->first);
}
}
float Doc_mod(DocMap w)
{ float store=0;
for (DocMap::iterator p = w.begin(); p != w.end(); ++p)
store+=(p->second*p->second);
return sqrt(store);
}
