void DataSystem::query(const std::string q)
{
try {
// Start an enquire session.
Xapian::Enquire enquire(db);
Xapian::Query query = qp.parse_query(q);
printf("Parsed query is: %s", query.get_description().c_str());
// Find the top 10 results for the query.
enquire.set_query(query);
Xapian::MSet matches = enquire.get_mset(0, 10);
// Display the results.
printf("%d results found.\n", matches.get_matches_estimated());
printf("Matches 1-$d:\n", matches.size());
for (Xapian::MSetIterator i = matches.begin(); i != matches.end(); ++i)
{
printf("%d: %d\% docid= [%s]\n\n",
i.get_rank() + 1,
i.get_percent(),
i.get_document().get_data().c_str());
}
} catch (const Xapian::Error &e)
{
printf("DataSystem, query error: %s", e.get_description());
}
}
int main(int argc, char **argv)
{
// Simplest possible options parsing: we just require two or more
// parameters.
if (argc < 3) {
cout << "usage: " << argv[0] << " <path to database> <search terms>" << endl;
exit(1);
}
// Catch any Xapian::Error exceptions thrown
try {
// Make the database
Xapian::Database db(argv[1]);
// Start an enquire session
Xapian::Enquire enquire(db);
// Set percent and/or weight cutoffs
enquire.set_cutoff(90,0.2);
// Set weighting schema
BM25Weight bm1(1.0,0.0,1.0,0.5,0.3);
enquire.set_weighting_scheme(bm1);
// Build the query object
Xapian::Query query(Xapian::Query::OP_AND, argv + 2, argv + argc);
cout << "Performing query" << query.get_description() << "'" << endl;
// Set Stopper
string stop[8]={"的","了","呵","吧","就","你","我","他"};
SimpleStopper *ss=new SimpleStopper;
for(int i=0;i<8;i++){
ss->add(stop[i]);
}
QueryParser qparser;
qparser.set_stopper(ss);
qparser.set_database(db);
// Give the query object to the enquire session
enquire.set_query(query);
// Get the top 10 results of the query
Xapian::MSet matches = enquire.get_mset(0, 10); //最多返回10个文档
// Display the results
cout << matches.size() << " results found" << endl;
for (Xapian::MSetIterator i = matches.begin();i != matches.end(); ++i) {
Xapian::Document doc = i.get_document();
cout << "Document ID " << *i << "\nPercent " <<i.get_percent() << "%\n" << doc.get_data() << "\n" << endl;
}
db.close();
} catch(const Xapian::Error &error) {
cout << "Exception: " << error.get_msg() << endl;
}
}
bool XapianEngine::queryDatabase(Xapian::Database *pIndex, Xapian::Query &query,
const string &stemLanguage, unsigned int startDoc, const QueryProperties &queryProps)
{
Timer timer;
unsigned int maxResultsCount = queryProps.getMaximumResultsCount();
bool completedQuery = false;
if (pIndex == NULL)
{
return false;
}
// Start an enquire session on the database
Xapian::Enquire enquire(*pIndex);
timer.start();
try
{
AbstractGenerator abstractGen(pIndex, 50);
vector<string> seedTerms;
// Give the query object to the enquire session
enquire.set_query(query);
// How should results be sorted ?
if (queryProps.getSortOrder() == QueryProperties::RELEVANCE)
{
// By relevance, only
enquire.set_sort_by_relevance_then_value(4);
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: sorting by relevance first" << endl;
#endif
}
else if (queryProps.getSortOrder() == QueryProperties::DATE)
{
// By date, and then by relevance
enquire.set_sort_by_value_then_relevance(4);
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: sorting by date and time first" << endl;
#endif
}
// Get the top results of the query
Xapian::MSet matches = enquire.get_mset(startDoc, maxResultsCount, (2 * maxResultsCount) + 1);
m_resultsCountEstimate = matches.get_matches_estimated();
if (matches.empty() == false)
{
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: found " << matches.size() << "/" << maxResultsCount
<< " results found from position " << startDoc << endl;
cout << "XapianEngine::queryDatabase: estimated " << matches.get_matches_lower_bound()
<< "/" << m_resultsCountEstimate << "/" << matches.get_matches_upper_bound() << endl;
#endif
// Get the results
for (Xapian::MSetIterator mIter = matches.begin(); mIter != matches.end(); ++mIter)
{
Xapian::docid docId = *mIter;
Xapian::Document doc(mIter.get_document());
// What terms did this document match ?
seedTerms.clear();
for (Xapian::TermIterator termIter = enquire.get_matching_terms_begin(docId);
termIter != enquire.get_matching_terms_end(docId); ++termIter)
{
char firstChar = (*termIter)[0];
if (isupper(((int)firstChar)) == 0)
{
seedTerms.push_back(*termIter);
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: matched term " << *termIter << endl;
#endif
}
else if (firstChar == 'Z')
{
string stemmed((*termIter).substr(1));
string::size_type stemmedLen = stemmed.length();
// Which of this document's terms stem to this ?
Xapian::TermIterator docTermIter = pIndex->termlist_begin(docId);
if (docTermIter != pIndex->termlist_end(docId))
{
for (docTermIter.skip_to(stemmed);
docTermIter != pIndex->termlist_end(docId); ++docTermIter)
{
// Is this a potential unstem ?
if (strncasecmp((*docTermIter).c_str(), stemmed.c_str(), stemmedLen) != 0)
{
// No, no point looking at the next terms
break;
}
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: matched unstem " << *docTermIter << endl;
#endif
// FIXME: check this term stems to stemmed !
seedTerms.push_back(*docTermIter);
}
}
}
}
DocumentInfo thisResult;
thisResult.setExtract(abstractGen.generateAbstract(docId, seedTerms));
thisResult.setScore((float)mIter.get_percent());
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: found document ID " << docId << endl;
#endif
XapianDatabase::recordToProps(doc.get_data(), &thisResult);
// XapianDatabase stored the language in English
thisResult.setLanguage(Languages::toLocale(thisResult.getLanguage()));
string url(thisResult.getLocation());
if (url.empty() == true)
{
// Hmmm this shouldn't be empty...
// Use this instead, even though the document isn't cached in the index
thisResult.setLocation(XapianDatabase::buildUrl(m_databaseName, docId));
}
// We don't know the index ID, just the document ID
thisResult.setIsIndexed(0, docId);
// Add this result
m_resultsList.push_back(thisResult);
}
}
completedQuery = true;
}
catch (const Xapian::Error &error)
{
cerr << "Couldn't run query: " << error.get_type() << ": " << error.get_msg() << endl;
}
cout << "Ran query \"" << queryProps.getFreeQuery() << "\" in " << timer.stop() << " ms" << endl;
try
{
m_expandTerms.clear();
// Expand the query ?
if (m_expandDocuments.empty() == false)
{
Xapian::RSet expandDocs;
for (set<string>::const_iterator docIter = m_expandDocuments.begin();
docIter != m_expandDocuments.end(); ++docIter)
{
string uniqueTerm(string("U") + XapianDatabase::limitTermLength(Url::escapeUrl(Url::canonicalizeUrl(*docIter)), true));
// Only one document may have this term
Xapian::PostingIterator postingIter = pIndex->postlist_begin(uniqueTerm);
if (postingIter != pIndex->postlist_end(uniqueTerm))
{
expandDocs.add_document(*postingIter);
}
}
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: expand from " << expandDocs.size() << " documents" << endl;
#endif
// Get 10 non-prefixed terms
string allowedPrefixes("RS");
TermDecider expandDecider(pIndex, ((stemLanguage.empty() == true) ? NULL : &m_stemmer),
FileStopper::get_stopper(Languages::toCode(stemLanguage)),
allowedPrefixes, query);
Xapian::ESet expandTerms = enquire.get_eset(10, expandDocs, &expandDecider);
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: " << expandTerms.size() << " expand terms" << endl;
#endif
for (Xapian::ESetIterator termIter = expandTerms.begin();
termIter != expandTerms.end(); ++termIter)
{
string expandTerm(*termIter);
char firstChar = expandTerm[0];
// Is this prefixed ?
if (allowedPrefixes.find(firstChar) != string::npos)
{
expandTerm.erase(0, 1);
}
m_expandTerms.insert(expandTerm);
}
}
}
catch (const Xapian::Error &error)
{
cerr << "Couldn't run query: " << error.get_type() << ": " << error.get_msg() << endl;
}
// Be tolerant of errors as long as we got some results
if ((completedQuery == true) ||
(m_resultsList.empty() == false))
{
return true;
}
return false;
}
bool XapianEngine::queryDatabase(Xapian::Database *pIndex, Xapian::Query &query)
{
bool completedQuery = false;
if (pIndex == NULL)
{
return false;
}
// Start an enquire session on the database
Xapian::Enquire enquire(*pIndex);
try
{
AbstractGenerator abstractGen(pIndex, 50);
vector<string> seedTerms;
// Give the query object to the enquire session
enquire.set_query(query);
// Get the top results of the query
Xapian::MSet matches = enquire.get_mset(0, m_maxResultsCount);
if (matches.empty() == false)
{
// Get the results
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: " << matches.get_matches_estimated()
<< "/" << m_maxResultsCount << " results found" << endl;
#endif
for (Xapian::MSetIterator mIter = matches.begin(); mIter != matches.end(); ++mIter)
{
Xapian::docid docId = *mIter;
Xapian::Document doc(mIter.get_document());
// What terms did this document match ?
seedTerms.clear();
for (Xapian::TermIterator termIter = enquire.get_matching_terms_begin(docId);
termIter != enquire.get_matching_terms_end(docId); ++termIter)
{
seedTerms.push_back(*termIter);
}
DocumentInfo thisResult;
thisResult.setExtract(abstractGen.generateAbstract(docId, seedTerms));
thisResult.setScore((float)mIter.get_percent());
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: found document ID " << docId << endl;
#endif
XapianDatabase::recordToProps(doc.get_data(), &thisResult);
string url(thisResult.getLocation());
if (url.empty() == true)
{
// Hmmm this shouldn't be empty...
// Use this instead, even though the document isn't cached in the index
thisResult.setLocation(XapianDatabase::buildUrl(m_databaseName, docId));
}
// We don't know the index ID, just the document ID
thisResult.setIsIndexed(0, docId);
// Add this result
m_resultsList.push_back(thisResult);
}
}
completedQuery = true;
}
catch (const Xapian::Error &error)
{
cerr << "XapianEngine::queryDatabase: " << error.get_type() << ": " << error.get_msg() << endl;
}
try
{
m_expandTerms.clear();
// Expand the query ?
if (m_relevantDocuments.empty() == false)
{
Xapian::RSet relevantDocs;
unsigned int count = 0;
for (set<unsigned int>::const_iterator docIter = m_relevantDocuments.begin();
docIter != m_relevantDocuments.end(); ++docIter)
{
relevantDocs.add_document(*docIter);
}
// Get 10 non-prefixed terms
Xapian::ESet expandTerms = enquire.get_eset(20, relevantDocs);
for (Xapian::ESetIterator termIter = expandTerms.begin();
(termIter != expandTerms.end()) && (count < 10); ++termIter)
{
if (isupper((int)((*termIter)[0])) == 0)
{
m_expandTerms.insert(*termIter);
++count;
}
}
}
}
catch (const Xapian::Error &error)
{
cerr << "XapianEngine::queryDatabase: " << error.get_type() << ": " << error.get_msg() << endl;
}
// Be tolerant of errors as long as we got some results
if ((completedQuery == true) ||
(m_resultsList.empty() == false))
{
return true;
}
return false;
}
bool XapianEngine::queryDatabase(Xapian::Query &query)
{
bool bStatus = false;
XapianDatabase *pDatabase = XapianDatabaseFactory::getDatabase(m_databaseName, true);
if (pDatabase == NULL)
{
return false;
}
Xapian::Database *pIndex = pDatabase->readLock();
if (pIndex != NULL)
{
try
{
// Start an enquire session on the database
Xapian::Enquire enquire(*pIndex);
// Give the query object to the enquire session
enquire.set_query(query);
// Get the top results of the query
Xapian::MSet matches = enquire.get_mset(0, m_maxResultsCount);
// Get the results
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: " << matches.get_matches_estimated() << "/" << m_maxResultsCount << " results found" << endl;
#endif
for (Xapian::MSetIterator mIter = matches.begin(); mIter != matches.end(); ++mIter)
{
// Get the document data
string record = mIter.get_document().get_data();
// Get the title
string title = StringManip::extractField(record, "caption=", "\n");
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: found omindex title " << title << endl;
#endif
// Get the URL
string url = StringManip::extractField(record, "url=", "\n");
if (url.empty() == true)
{
// Hmmm this shouldn't be empty...
// Use this instead, even though the document isn't cached in the index
url = buildUrl(m_databaseName, *mIter);
}
else
{
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: found omindex URL " << url << endl;
#endif
url = Url::canonicalizeUrl(url);
}
// Get the summary and the type
string summary = StringManip::extractField(record, "sample=", "\n");
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: found omindex summary " << summary << endl;
#endif
string type = StringManip::extractField(record, "type=", "\n");
// ...and finally the language, if available
string language = StringManip::extractField(record, "language=", "\n");
// Add this result
Result thisResult(url, title, summary, language, (float)mIter.get_percent());
m_resultsList.push_back(thisResult);
}
bStatus = true;
}
catch (const Xapian::Error &error)
{
cout << "XapianEngine::queryDatabase: couldn't run query: " << error.get_msg() << endl;
}
}
pDatabase->unlock();
return bStatus;
}
bool XapianEngine::queryDatabase(Xapian::Query &query)
{
bool bStatus = false;
XapianDatabase *pDatabase = XapianDatabaseFactory::getDatabase(m_databaseName, true);
if (pDatabase == NULL)
{
return false;
}
// Get the latest revision...
pDatabase->reopen();
Xapian::Database *pIndex = pDatabase->readLock();
if (pIndex != NULL)
{
try
{
// Start an enquire session on the database
Xapian::Enquire enquire(*pIndex);
// Give the query object to the enquire session
enquire.set_query(query);
// Get the top results of the query
Xapian::MSet matches = enquire.get_mset(0, m_maxResultsCount);
if (matches.empty() == false)
{
multimap<Xapian::weight, string> queryTerms;
vector<string> seedTerms;
Xapian::weight maxWeight = matches.get_max_attained();
// Sort query terms by weight
for (Xapian::TermIterator termIter = query.get_terms_begin();
termIter != query.get_terms_end(); ++termIter)
{
string termName(*termIter);
Xapian::weight termWeight = maxWeight - matches.get_termweight(termName);
queryTerms.insert(pair<Xapian::weight, string>(termWeight, termName));
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: term " << termName
<< " has weight " << matches.get_termweight(termName) << "/" << maxWeight << endl;
#endif
}
for (multimap<Xapian::weight, string>::iterator weightIter = queryTerms.begin();
weightIter != queryTerms.end(); ++weightIter)
{
seedTerms.push_back(weightIter->second);
}
// Get the results
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: " << matches.get_matches_estimated() << "/" << m_maxResultsCount << " results found" << endl;
#endif
for (Xapian::MSetIterator mIter = matches.begin(); mIter != matches.end(); ++mIter)
{
Xapian::docid docId = *mIter;
Xapian::Document doc(mIter.get_document());
string record = doc.get_data();
// Get the title
string title = StringManip::extractField(record, "caption=", "\n");
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: found omindex title " << title << endl;
#endif
// Get the URL
string url = StringManip::extractField(record, "url=", "\n");
if (url.empty() == true)
{
// Hmmm this shouldn't be empty...
// Use this instead, even though the document isn't cached in the index
url = XapianDatabase::buildUrl(m_databaseName, *mIter);
}
else
{
#ifdef DEBUG
cout << "XapianEngine::queryDatabase: found omindex URL " << url << endl;
#endif
url = Url::canonicalizeUrl(url);
}
// Get the type
string type = StringManip::extractField(record, "type=", "\n");
// ...and the language, if available
string language = StringManip::extractField(record, "language=", "\n");
// Finally, get a summary
string summary = StringManip::extractField(record, "sample=", "\n");
if (summary.empty() == true)
{
AbstractGenerator abstractGen(pIndex, 50);
// Generate an abstract based on the query's terms
summary = abstractGen.generateAbstract(seedTerms, docId);
}
// Add this result
Result thisResult(url, title, summary, language, (float)mIter.get_percent());
m_resultsList.push_back(thisResult);
}
}
m_expandTerms.clear();
// Expand the query ?
if (m_relevantDocuments.empty() == false)
{
Xapian::RSet relevantDocs;
unsigned int count = 0;
for (set<unsigned int>::const_iterator docIter = m_relevantDocuments.begin();
docIter != m_relevantDocuments.end(); ++docIter)
{
relevantDocs.add_document(*docIter);
}
// Get 10 non-prefixed terms
Xapian::ESet expandTerms = enquire.get_eset(20, relevantDocs);
for (Xapian::ESetIterator termIter = expandTerms.begin();
(termIter != expandTerms.end()) && (count < 10); ++termIter)
{
if (isupper((int)((*termIter)[0])) == 0)
{
m_expandTerms.insert(*termIter);
++count;
}
}
}
bStatus = true;
}
catch (const Xapian::Error &error)
{
cerr << "XapianEngine::queryDatabase: " << error.get_type() << ": " << error.get_msg() << endl;
}
}
pDatabase->unlock();
return bStatus;
}
int main(int argc, char **argv)
{
if(argc < 2) {
usage(argv);
return 1;
}
try {
char *action = argv[1];
char *db_path = argv[2];
if(!strcmp(action, "index")) {
Xapian::WritableDatabase db(db_path, Xapian::DB_CREATE_OR_OPEN);
Xapian::TermGenerator indexer;
Xapian::Stem stemmer("english");
indexer.set_stemmer(stemmer);
std::string doc_txt;
while(true) {
if(std::cin.eof()) break;
std::string line;
getline(std::cin, line);
doc_txt += line;
}
if(!doc_txt.empty()) {
Xapian::Document doc;
doc.set_data(doc_txt);
indexer.set_document(doc);
indexer.index_text(doc_txt);
db.add_document(doc);
std::cout << "Indexed: " << indexer.get_description() << std::endl;
}
db.commit();
} else if(!strcmp(action, "search")) {
if(argc < 4) {
std::cerr << "You must supply a query string" << std::endl;
return 1;
}
Xapian::Database db(db_path);
Xapian::Enquire enquire(db);
std::string query_str = argv[3];
argv+= 4;
while(*argv) {
query_str += ' ';
query_str += *argv++;
}
Xapian::QueryParser qp;
Xapian::Stem stemmer("english");
qp.set_stemmer(stemmer);
qp.set_database(db);
qp.set_stemming_strategy(Xapian::QueryParser::STEM_SOME);
Xapian::Query query = qp.parse_query(query_str);
std::cout << "Parsed query is: " << query.get_description() <<
std::endl;
enquire.set_query(query);
Xapian::MSet matches = enquire.get_mset(0, 10);
std::cout << matches.get_matches_estimated() << " results found.\n";
std::cout << "Matches 1-" << matches.size() << ":\n" << std::endl;
for (Xapian::MSetIterator i = matches.begin();
i != matches.end(); ++i) {
std::cout << i.get_rank() + 1 << ": " << i.get_percent() <<
"% docid=" << *i << " [" <<
i.get_document().get_data()<< "]" << std::endl <<
std::endl;
}
} else {
std::cerr << "Invalid action " << action << std::endl;
usage(argv);
return 1;
}
} catch (const Xapian::Error &error) {
std::cout << "Exception: " << error.get_msg() << std::endl;
}
}
