void searchPhrase(const string &indexPath, istream &in, bool fuzzy) {
IndexSearcher is(indexPath);
string line;
while ( getline(in, line) ) {
line = util::trim(line);
if ( line.length() == 0 ) continue;
cout << line << endl;
searchPhrase(is, line, fuzzy);
}
}
bool LCHMFile::searchQuery( const QString& inquery, QStringList * searchresults, unsigned int limit )
{
QStringList words_must_exist, words_must_not_exist, words_highlight;
QVector<QStringList> phrases_must_exist;
QString query = inquery;
bool query_valid = true;
LCHMSearchProgressResults results;
int pos;
int i;
/*
* Parse the search query with a simple state machine.
* Query should consist of one of more words separated by a space with a possible prefix.
* A prefix may be:
* + indicates that the word is required; any page without this word is excluded from the result.
* - indicates that the word is required to be absent; any page with this word is excluded from
* the result.
* "." indicates a phrase. Anything between quotes indicates a phrase, which is set of space-separated
* words. Will be in result only if the words in phrase are in page in the same sequence, and
* follow each other.
* If there is no prefix, the word considered as required.
*/
QRegExp rxphrase( "\"(.*)\"" );
QRegExp rxword( "([^\\s]+)" );
rxphrase.setMinimal( TRUE );
// First, get the phrase queries
while ( (pos = rxphrase.indexIn (query, 0)) != -1 )
{
// A phrase query found. Locate its boundaries, and parse it.
QStringList plist = rxphrase.cap ( 1 ).split ( QRegExp ("\\s+") );
validateWords ( plist, query_valid );
if ( plist.size() > 0 )
phrases_must_exist.push_back( plist );
query.remove (pos, rxphrase.matchedLength());
}
// Then, parse the rest query
while ( (pos = rxword.indexIn( query, 0 )) != -1 )
{
// A phrase query found. Locate its boundaries, and parse it.
QString word = rxword.cap ( 1 );
QChar type = '+';
if ( word[0] == '-' || word[0] == '+' )
{
type = word[0];
word.remove (0, 1);
}
validateWord ( word, query_valid );
if ( type == '-' )
words_must_not_exist.push_back ( word );
else
words_must_exist.push_back ( word );
query.remove (pos, rxword.matchedLength());
}
#if defined (DUMP_SEARCH_QUERY)
// Dump the search query
QString qdump;
for ( i = 0; i < phrases_must_exist.size(); i++ )
qdump += QString(" \"") + phrases_must_exist[i].join (" ") + QString ("\"");
for ( i = 0; i < words_must_not_exist.size(); i++ )
qdump += QString (" -") + words_must_not_exist[i];
for ( i = 0; i < words_must_exist.size(); i++ )
qdump += QString (" +") + words_must_exist[i];
qDebug ("Search query dump: %s", qdump.ascii());
#endif
// First search for phrases
if ( phrases_must_exist.size() > 0 )
{
LCHMSearchProgressResults tempres;
for ( i = 0; i < phrases_must_exist.size(); i++ )
{
if ( !searchPhrase ( impl(), phrases_must_exist[i], tempres ) )
return false;
mergeResults ( results, tempres, true );
}
}
for ( i = 0; i < words_must_exist.size(); i++ )
{
LCHMSearchProgressResults tempres;
if ( !m_impl->searchWord ( words_must_exist[i], true, false, tempres, false ) )
return false;
mergeResults ( results, tempres, true );
}
for ( i = 0; i < words_must_not_exist.size(); i++ )
{
LCHMSearchProgressResults tempres;
m_impl->searchWord ( words_must_not_exist[i], true, false, tempres, false );
mergeResults ( results, tempres, false );
}
m_impl->getSearchResults( results, searchresults, limit );
return true;
}