void testExtractFromWildcardQuery( CuTest * tc )
{
Directory * pIndex = setUpIndex();
IndexReader * pReader = IndexReader::open( pIndex );
TermSet termSet;
WildcardQuery * wildcard;
Term * t1;
Query * rewrite;
t1 = _CLNEW Term( _T("data"), _T("aaaa?") );
wildcard = _CLNEW WildcardQuery( t1 );
rewrite = wildcard->rewrite( pReader );
rewrite->extractTerms( &termSet );
_CLLDECDELETE( t1 );
assertEqualsMsg( _T( "wrong number of terms" ), 3, termSet.size() );
for( TermSet::iterator itTerms = termSet.begin(); itTerms != termSet.end(); itTerms++ )
{
Term * pTerm = *itTerms;
if( 0 != _tcscmp( _T( "aaaaa" ), pTerm->text())
&& 0 != _tcscmp( _T( "aaaab" ), pTerm->text())
&& 0 != _tcscmp( _T( "aaaac" ), pTerm->text()))
{
assertTrueMsg( _T( "wrong term" ), false );
}
}
clearTermSet( termSet );
if( rewrite != wildcard )
_CLDELETE( rewrite );
_CLDELETE( wildcard );
t1 = _CLNEW Term( _T("data"), _T("aaa*") );
wildcard = _CLNEW WildcardQuery( t1 );
rewrite = wildcard->rewrite( pReader );
rewrite->extractTerms( &termSet );
_CLLDECDELETE( t1 );
assertEqualsMsg( _T( "wrong number of terms" ), 5, termSet.size() );
for( TermSet::iterator itTerms = termSet.begin(); itTerms != termSet.end(); itTerms++ )
{
Term * pTerm = *itTerms;
assertTrueMsg( _T( "wrong term" ), ( 0 == _tcsncmp( _T( "aaa" ), pTerm->text(), 3 )));
}
clearTermSet( termSet );
if( rewrite != wildcard )
_CLDELETE( rewrite );
_CLDELETE( wildcard );
pReader->close();
_CLDELETE( pReader );
closeIndex( pIndex );
pIndex = NULL;
}
void testEqualScores()
{
// NOTE: uses index build in *this* setUp
IndexReader * pReader = IndexReader::open( m_pSmall );
IndexSearcher * pSearch = _CLNEW IndexSearcher( pReader );
Hits * pResult;
// some hits match more terms then others, score should be the same
Query * q = csrq( _T( "data" ), _T( "1" ), _T( "6" ), true, true );
pResult = pSearch->search( q );
size_t numHits = pResult->length();
assertEqualsMsg( _T( "wrong number of results" ), 6, numHits );
float_t score = pResult->score( 0 );
for( size_t i = 1; i < numHits; i++ )
{
assertTrueMsg( _T( "score was not the same" ), score == pResult->score( i ));
}
_CLDELETE( pResult );
_CLDELETE( q );
pSearch->close();
_CLDELETE( pSearch );
pReader->close();
_CLDELETE( pReader );
}
void testBoost()
{
// NOTE: uses index build in *this* setUp
IndexReader * pReader = IndexReader::open( m_pSmall );
IndexSearcher * pSearch = _CLNEW IndexSearcher( pReader );
Hits * pResult;
// test for correct application of query normalization
// must use a non score normalizing method for this.
Query * q = csrq( _T( "data" ), _T( "1" ), _T( "6" ), true, true );
q->setBoost( 100 );
pResult = pSearch->search( q );
for( size_t i = 1; i < pResult->length(); i++ )
{
assertTrueMsg( _T( "score was not was not correct" ), 1.0f == pResult->score( i ));
}
_CLDELETE( pResult );
_CLDELETE( q );
//
// Ensure that boosting works to score one clause of a query higher
// than another.
//
Query * q1 = csrq( _T( "data" ), _T( "A" ), _T( "A" ), true, true ); // matches document #0
q1->setBoost( .1f );
Query * q2 = csrq( _T( "data" ), _T( "Z" ), _T( "Z" ), true, true ); // matches document #1
BooleanQuery * bq = _CLNEW BooleanQuery( true );
bq->add( q1, true, BooleanClause::SHOULD );
bq->add( q2, true, BooleanClause::SHOULD );
pResult = pSearch->search( bq );
assertEquals( 1, pResult->id( 0 ));
assertEquals( 0, pResult->id( 1 ));
assertTrue( pResult->score( 0 ) > pResult->score( 1 ));
_CLDELETE( pResult );
_CLDELETE( bq );
q1 = csrq( _T( "data" ), _T( "A" ), _T( "A" ), true, true ); // matches document #0
q1->setBoost( 10.0f );
q2 = csrq( _T( "data" ), _T( "Z" ), _T( "Z" ), true, true ); // matches document #1
bq = _CLNEW BooleanQuery( true );
bq->add( q1, true, BooleanClause::SHOULD );
bq->add( q2, true, BooleanClause::SHOULD );
pResult = pSearch->search( bq );
assertEquals( 0, pResult->id( 0 ));
assertEquals( 1, pResult->id( 1 ));
assertTrue( pResult->score( 0 ) > pResult->score( 1 ));
_CLDELETE( pResult );
_CLDELETE( bq );
pSearch->close();
_CLDELETE( pSearch );
pReader->close();
_CLDELETE( pReader );
}
void TestSpansAdvanced::assertHits( Query * query, const TCHAR * description, const TCHAR ** expectedIds, float_t * expectedScores, size_t expectedCount )
{
float_t tolerance = 1e-5f;
QueryUtils::check( tc, query, searcher );
// Hits hits = searcher.search(query);
// hits normalizes and throws things off if one score is greater than 1.0
TopDocs * topdocs = searcher->_search( query, NULL, 10000 );
/*****
// display the hits
System.out.println(hits.length() + " hits for search: \"" + description + '\"');
for (int i = 0; i < hits.length(); i++) {
System.out.println(" " + FIELD_ID + ':' + hits.doc(i).get(FIELD_ID) + " (score:" + hits.score(i) + ')');
}
*****/
// did we get the hits we expected
assertEquals( expectedCount, topdocs->totalHits );
for( size_t i = 0; i < expectedCount; i++ )
{
//System.out.println(i + " exp: " + expectedIds[i]);
//System.out.println(i + " field: " + hits.doc(i).get(FIELD_ID));
int32_t id = topdocs->scoreDocs[ i ].doc;
float_t score = topdocs->scoreDocs[ i ].score;
Document doc;
searcher->doc( id, doc );
assertTrueMsg( _T( "actual document does not match the expected one" ), 0 == _tcscmp( expectedIds[ i ], doc.get( field_id )));
assertTrueMsg( _T( "score does not match" ), ( expectedScores[ i ] > score ? expectedScores[ i ] - score : score - expectedScores[ i ] ) < tolerance );
Explanation exp;
searcher->explain( query, id, &exp );
float_t sd = exp.getDetail( 0 )->getValue() - score;
if ( sd < 0 ) sd *= -1;
assertTrueMsg( _T( "explained score does not match" ), sd < tolerance );
}
_CLLDELETE( topdocs );
}
void testExtractFromFuzzyQuery( CuTest * tc )
{
Directory * pIndex = setUpIndex();
IndexReader * pReader = IndexReader::open( pIndex );
TermSet termSet;
FuzzyQuery * fuzzy;
Term * t1;
Query * rewrite;
t1 = _CLNEW Term( _T("data"), _T("aaaab") );
fuzzy = _CLNEW FuzzyQuery( t1, 0.7f );
rewrite = fuzzy->rewrite( pReader );
rewrite->extractTerms( &termSet );
_CLLDECDELETE( t1 );
assertEqualsMsg( _T( "wrong number of terms" ), 4, termSet.size() );
for( TermSet::iterator itTerms = termSet.begin(); itTerms != termSet.end(); itTerms++ )
{
Term * pTerm = *itTerms;
if( 0 != _tcscmp( _T( "aaaaa" ), pTerm->text())
&& 0 != _tcscmp( _T( "aaaab" ), pTerm->text())
&& 0 != _tcscmp( _T( "aaabb" ), pTerm->text())
&& 0 != _tcscmp( _T( "aaaac" ), pTerm->text()))
{
assertTrueMsg( _T( "wrong term" ), false );
}
}
clearTermSet( termSet );
if( rewrite != fuzzy )
_CLDELETE( rewrite );
_CLDELETE( fuzzy );
pReader->close();
_CLDELETE( pReader );
closeIndex( pIndex );
pIndex = NULL;
}
void testExtractFromBooleanQuery( CuTest * tc )
{
Directory * pIndex = setUpIndex();
IndexReader * pReader = IndexReader::open( pIndex );
TermSet termSet;
Term * t1 = _CLNEW Term( _T("data"), _T("aaaab") );
Term * t2 = _CLNEW Term( _T("data"), _T("aaabb") );
Term * t3 = _CLNEW Term( _T("data"), _T("aaabb") );
BooleanQuery * bq = _CLNEW BooleanQuery();
bq->add( _CLNEW TermQuery( t1 ), true, BooleanClause::SHOULD );
bq->add( _CLNEW TermQuery( t2 ), true, BooleanClause::SHOULD );
bq->add( _CLNEW TermQuery( t3 ), true, BooleanClause::SHOULD );
Query * rewrite = bq->rewrite( pReader );
rewrite->extractTerms( &termSet );
assertEqualsMsg( _T( "wrong number of terms" ), 2, termSet.size() );
for( TermSet::iterator itTerms = termSet.begin(); itTerms != termSet.end(); itTerms++ )
{
Term * pTerm = *itTerms;
assertTrueMsg( _T( "wrong term" ), ( 0 == t1->compareTo( pTerm ) || 0 == t2->compareTo( pTerm )));
}
clearTermSet( termSet );
_CLLDECDELETE( t1 );
_CLLDECDELETE( t2 );
_CLLDECDELETE( t3 );
if( rewrite != bq )
_CLDELETE( rewrite );
_CLDELETE( bq );
pReader->close();
_CLDELETE( pReader );
closeIndex( pIndex );
pIndex = NULL;
}
