/// Validates a query and extracts its terms.
bool XapianEngine::validateQuery(QueryProperties& queryProps, bool includePrefixed,
vector<string> &terms)
{
bool goodQuery = false;
try
{
Xapian::Query fullQuery = parseQuery(NULL, queryProps, "", true);
if (fullQuery.empty() == false)
{
for (Xapian::TermIterator termIter = fullQuery.get_terms_begin();
termIter != fullQuery.get_terms_end(); ++termIter)
{
// Skip prefixed terms unless instructed otherwise
if ((includePrefixed == true) ||
(isupper((int)((*termIter)[0])) == 0))
{
terms.push_back(*termIter);
}
}
goodQuery = true;
}
}
catch (const Xapian::Error &error)
{
cerr << "XapianEngine::validateQuery: " << error.get_type() << ": " << error.get_msg() << endl;
}
return goodQuery;
}
TermDecider(Xapian::Database *pIndex,
Xapian::Stem *pStemmer,
Xapian::Stopper *pStopper,
const string &allowedPrefixes,
Xapian::Query &query) :
Xapian::ExpandDecider(),
m_pIndex(pIndex),
m_pStemmer(pStemmer),
m_pStopper(pStopper),
m_allowedPrefixes(allowedPrefixes),
m_pTermsToAvoid(NULL)
{
m_pTermsToAvoid = new set<string>();
for (Xapian::TermIterator termIter = query.get_terms_begin();
termIter != query.get_terms_end(); ++termIter)
{
string term(*termIter);
if (isupper((int)(term[0])) == 0)
{
m_pTermsToAvoid->insert(term);
if (m_pStemmer != NULL)
{
string stem((*m_pStemmer)(term));
m_pTermsToAvoid->insert(stem);
}
}
else if (term[0] == 'Z')
{
m_pTermsToAvoid->insert(term.substr(1));
}
}
#ifdef DEBUG
cout << "TermDecider: avoiding " << m_pTermsToAvoid->size() << " terms" << endl;
#endif
}
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;
}
/// Runs a query; true if success.
bool XapianEngine::runQuery(QueryProperties& queryProps)
{
// Clear the results list
m_resultsList.clear();
XapianDatabase *pDatabase = XapianDatabaseFactory::getDatabase(m_databaseName, true);
if (pDatabase == NULL)
{
return false;
}
// Get the latest revision...
pDatabase->reopen();
Xapian::Database *pIndex = pDatabase->readLock();
try
{
string stemLanguage;
unsigned int searchStep = 1;
bool followOperators = true;
// Searches are run in this order :
// 1. follow operators and don't stem terms
// 2. if no results, follow operators and stem terms
// 3. if no results, don't follow operators and don't stem terms
// 4. if no results, don't follow operators and stem terms
// Steps 2 and 4 depend on a language being defined for the query
Xapian::Query freeQuery = parseQuery(pIndex, queryProps, "", followOperators);
while (freeQuery.empty() == false)
{
#ifdef DEBUG
cout << "XapianEngine::runQuery: query terms are " << endl;
for (Xapian::TermIterator termIter = freeQuery.get_terms_begin();
termIter != freeQuery.get_terms_end(); ++termIter)
{
cout << " " << *termIter << endl;
}
#endif
// Query the database
if (queryDatabase(pIndex, freeQuery) == false)
{
break;
}
if (m_resultsList.empty() == true)
{
// The search did succeed but didn't return anything
// Try the next step
switch (++searchStep)
{
case 2:
followOperators = true;
stemLanguage = queryProps.getLanguage();
if (stemLanguage.empty() == false)
{
break;
}
++searchStep;
case 3:
followOperators = false;
stemLanguage.clear();
break;
case 4:
followOperators = false;
stemLanguage = queryProps.getLanguage();
if (stemLanguage.empty() == false)
{
break;
}
++searchStep;
default:
pDatabase->unlock();
return true;
}
#ifdef DEBUG
cout << "XapianEngine::runQuery: trying step " << searchStep << endl;
#endif
freeQuery = parseQuery(pIndex, queryProps,
Languages::toEnglish(stemLanguage), followOperators);
continue;
}
pDatabase->unlock();
return true;
}
}
catch (const Xapian::Error &error)
{
cerr << "XapianEngine::runQuery: " << error.get_type() << ": " << error.get_msg() << endl;
}
pDatabase->unlock();
return false;
}
