C++ BooleanQuery类代码示例

//static
Query* Query::mergeBooleanQueries(Query** queries) {
    CL_NS(util)::CLVector<BooleanClause*> allClauses;
    int32_t i = 0;
    while ( queries[i] != NULL ){
		BooleanQuery* bq = (BooleanQuery*)queries[i];
		
		int32_t size = bq->getClauseCount();
		BooleanClause** clauses = _CL_NEWARRAY(BooleanClause*, size);
		bq->getClauses(clauses);
		
		for (int32_t j = 0;j<size;++j ){
			allClauses.push_back(clauses[j]);
			j++;
		}
		_CLDELETE_ARRAY(clauses);
		i++;
    }

    BooleanQuery* result = _CLNEW BooleanQuery();
    CL_NS(util)::CLVector<BooleanClause*>::iterator itr = allClauses.begin();
    while (itr != allClauses.end() ) {
		result->add(*itr);
    }
    return result;
}

	void testSrchWildcard(CuTest *tc ){
#ifdef NO_WILDCARD_QUERY
		CuNotImpl(tc,_T("Wildcard"));
#else
		CuAssert(tc,_T("Searcher was not open"),s!=NULL);
		//testWildcard
		_TestSearchesRun(tc, &a,s, _T("term*") );
		_TestSearchesRun(tc, &a,s, _T("term*^2") );
		_TestSearchesRun(tc, &a,s, _T("term~") );
		_TestSearchesRun(tc, &a,s, _T("term^2~") );
		_TestSearchesRun(tc, &a,s, _T("term~^2") );
		_TestSearchesRun(tc, &a,s, _T("term*germ") );
		_TestSearchesRun(tc, &a,s, _T("term*germ^3") );

		//test problem reported by Gary Mangum
		BooleanQuery* bq = _CLNEW BooleanQuery();
		Term* upper = _CLNEW Term(_T("contents"),_T("0105"));
		Term* lower = _CLNEW Term(_T("contents"),_T("0105"));
		RangeQuery* rq=_CLNEW RangeQuery(lower,upper,true);
		bq->add(rq,true,true,false);
		_CLDECDELETE(upper);
		_CLDECDELETE(lower);

		Term* prefix = _CLNEW Term(_T("contents"),_T("reuters21578"));
		PrefixQuery* pq = _CLNEW PrefixQuery(prefix);
		_CLDECDELETE(prefix);
		bq->add(pq,true,true,false);

		Hits* h = NULL;
		try{
			h = s->search( bq );
		}_CLFINALLY(
		_CLDELETE(h);
		_CLDELETE(bq);
		);

//static 
Query* MultiFieldQueryParser::parse(const TCHAR* query, const TCHAR** fields, const uint8_t* flags, Analyzer* analyzer)
{
    BooleanQuery* bQuery = _CLNEW BooleanQuery();
    int32_t i = 0;
    while ( fields[i] != NULL )
    {
		Query* q = QueryParser::parse(query, fields[i], analyzer);
        uint8_t flag = flags[i];
        switch (flag)
        {
			case MultiFieldQueryParser::REQUIRED_FIELD:
                bQuery->add(q, true, true, false);
                break;
            case MultiFieldQueryParser::PROHIBITED_FIELD:
                bQuery->add(q, true, false, true);
                break;
            default:
                bQuery->add(q, true, false, false);
                break;
        }

        i++;
    }
    return bQuery;
}

    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 );
    }

Query*
CLuceneIndexReader::Private::createMultiFieldQuery(const Strigi::Query& query) {
    BooleanQuery* bq = _CLNEW BooleanQuery();
    for (vector<string>::const_iterator i = query.fields().begin();
            i != query.fields().end(); ++i) {
        Query* q = createSingleFieldQuery(*i, query);
        bq->add(q, true, false, false);
    }
    return bq;
}

void TestBasics::testBoolean2()
{
    Term * term1 = _CLNEW Term( _T( "field" ), _T( "sevento" ));
    Term * term2 = _CLNEW Term( _T( "field" ), _T( "sevenly" ));

    BooleanQuery * query = _CLNEW BooleanQuery();
    query->add( _CLNEW TermQuery( term1 ), true, BooleanClause::MUST );
    query->add( _CLNEW TermQuery( term2 ), true, BooleanClause::MUST );
    
    _CLLDECDELETE( term1 );
    _CLLDECDELETE( term2 );

    checkHits( query, NULL, 0 );
    _CLLDELETE( query );
}

	/**
     * FIXME: Describe <code>rewrite</code> method here.
     *
     * @param reader an <code>IndexReader</code> value
     * @return a <code>Query</code> value
     * @exception IOException if an error occurs
     */
    Query* RangeQuery::rewrite(IndexReader* reader){

        BooleanQuery* query = _CLNEW BooleanQuery;
        TermEnum* enumerator = reader->terms(lowerTerm);
		Term* lastTerm = NULL;
        try {
            bool checkLower = false;
            if (!inclusive) // make adjustments to set to exclusive
                checkLower = true;

            const TCHAR* testField = getField();
            do {
                lastTerm = enumerator->term();
                if (lastTerm != NULL && lastTerm->field() == testField ) {
                    if (!checkLower || _tcscmp(lastTerm->text(),lowerTerm->text()) > 0) {
                        checkLower = false;
                        if (upperTerm != NULL) {
                            int compare = _tcscmp(upperTerm->text(),lastTerm->text());
                            /* if beyond the upper term, or is exclusive and
                             * this is equal to the upper term, break out */
                            if ((compare < 0) || (!inclusive && compare == 0))
                                break;
                        }
                        TermQuery* tq = _CLNEW TermQuery(lastTerm); // found a match
                        tq->setBoost(getBoost()); // set the boost
                        query->add(tq, true, false, false); // add to query
                    }
                }else {
                    break;
                }
				_CLDECDELETE(lastTerm);
            }
            while (enumerator->next());
		}catch(...){
			_CLDECDELETE(lastTerm); //always need to delete this
			_CLDELETE(query); //in case of error, delete the query
            enumerator->close();
			_CLDELETE(enumerator);
			throw; //rethrow
		}
		_CLDECDELETE(lastTerm); //always need to delete this
		enumerator->close();
		_CLDELETE(enumerator);

        return query;
    }

void TestSpansAdvanced::doTestBooleanQueryWithSpanQueries( const float_t expectedScore )
{
    Term * t1 = _CLNEW Term( field_text, _T( "work" ));
    Query * spanQuery = _CLNEW SpanTermQuery( t1 );
    BooleanQuery * query = _CLNEW BooleanQuery();

    query->add( spanQuery, false, BooleanClause::MUST );
    query->add( spanQuery, false, BooleanClause::MUST );

    TCHAR * expectedIds[] = { _T( "1" ), _T( "2" ), _T( "3" ), _T( "4" ) };
    float_t expectedScores[] = { expectedScore, expectedScore, expectedScore, expectedScore };

    assertHits( query, _T( "two span queries" ), expectedIds, expectedScores, 4 );

    _CLLDECDELETE( t1 );
    _CLLDELETE( spanQuery );
    _CLLDELETE( query );
}

/// TestBooleanScorer.java, ported 5/9/2009
void testBooleanScorer(CuTest *tc) {
    const TCHAR* FIELD = _T("category");
    RAMDirectory directory;

    TCHAR* values[] = { _T("1"), _T("2"), _T("3"), _T("4"), NULL};

    try {
        WhitespaceAnalyzer a;
        IndexWriter* writer = _CLNEW IndexWriter(&directory, &a, true);
        for (size_t i = 0; values[i]!=NULL; i++) {
            Document* doc = _CLNEW Document();
            doc->add(*_CLNEW Field(FIELD, values[i], Field::STORE_YES | Field::INDEX_TOKENIZED));
            writer->addDocument(doc);
            _CLLDELETE(doc);
        }
        writer->close();
        _CLLDELETE(writer);

        BooleanQuery* booleanQuery1 = _CLNEW BooleanQuery();
        Term *t = _CLNEW Term(FIELD, _T("1"));
        booleanQuery1->add(_CLNEW TermQuery(t), true, BooleanClause::SHOULD);
        _CLDECDELETE(t);
        t = _CLNEW Term(FIELD, _T("2"));
        booleanQuery1->add(_CLNEW TermQuery(t), true, BooleanClause::SHOULD);
        _CLDECDELETE(t);

        BooleanQuery* query = _CLNEW BooleanQuery();
        query->add(booleanQuery1, true, BooleanClause::MUST);
        t = _CLNEW Term(FIELD, _T("9"));
        query->add(_CLNEW TermQuery(t), true, BooleanClause::MUST_NOT);
        _CLDECDELETE(t);

        IndexSearcher *indexSearcher = _CLNEW IndexSearcher(&directory);
        Hits *hits = indexSearcher->search(query);
        CLUCENE_ASSERT(2 == hits->length()); // Number of matched documents
        _CLLDELETE(hits);
        _CLLDELETE(indexSearcher);

        _CLLDELETE(query);
    }
    catch (CLuceneError& e) {
        CuFail(tc, e.twhat());
    }
}

   Query* PrefixQuery::rewrite(IndexReader* reader){
    BooleanQuery* query = _CLNEW BooleanQuery();
    TermEnum* enumerator = reader->terms(prefix);
	Term* lastTerm = NULL;
    try {
      const TCHAR* prefixText = prefix->text();
      const TCHAR* prefixField = prefix->field();
      const TCHAR* tmp;
      size_t i;
	  int32_t prefixLen = prefix->textLength();
      do {
        lastTerm = enumerator->term();
		if (lastTerm != NULL && lastTerm->field() == prefixField ){
		  
		  //now see if term->text() starts with prefixText
		  int32_t termLen = lastTerm->textLength();
		  if ( prefixLen>termLen )
			  break; //the prefix is longer than the term, can't be matched

            tmp = lastTerm->text();
            
            //check for prefix match in reverse, since most change will be at the end
            for ( i=prefixLen-1;i!=-1;--i ){
                if ( tmp[i] != prefixText[i] ){
                    tmp=NULL;//signals inequality
                    break;
                }
            }
            if ( tmp == NULL )
                break;

          TermQuery* tq = _CLNEW TermQuery(lastTerm);	  // found a match
          tq->setBoost(getBoost());                // set the boost
          query->add(tq,true,false, false);		  // add to query
        } else
          break;
		_CLDECDELETE(lastTerm);
      } while (enumerator->next());
    }_CLFINALLY(
      enumerator->close();
	  _CLDELETE(enumerator);
	  _CLDECDELETE(lastTerm);
	);

// static
Query* MultiFieldQueryParser::parse(const TCHAR* query, const TCHAR** _fields, const uint8_t* flags, Analyzer* analyzer) {
	BooleanQuery* bQuery = _CLNEW BooleanQuery();
	for (size_t i = 0; _fields[i]!=NULL; i++) {
	  //TODO: this is really confusing... why not refactor _fields and flags to use a array object.
	  //flags can be NULL since NULL == 0...
		/*if (flags[i] == NULL) {
			_CLLDELETE(bQuery);
			_CLTHROWA(CL_ERR_IllegalArgument, "_fields.length != flags.length");
		}*/
		QueryParser* qp = _CLNEW QueryParser(_fields[i], analyzer);
		Query* q = qp->parse(query);
		if (q!=NULL && // q never null, just being defensive
			(!(q->instanceOf(BooleanQuery::getClassName())) || ((BooleanQuery*)q)->getClauseCount()>0)) {
				bQuery->add(q, true, (BooleanClause::Occur)flags[i]);
		} else
			_CLLDELETE(q);
		_CLLDELETE(qp);
	}
	return bQuery;
}

/**
 * Tests a single span query that matches multiple documents.
 */
void TestSpansAdvanced2::testMultipleDifferentSpanQueries()
{
    Term * t1 = _CLNEW Term( field_text, _T( "should" ));
    Term * t2 = _CLNEW Term( field_text, _T( "we" ));
    Query * spanQuery1 = _CLNEW SpanTermQuery( t1 );
    Query * spanQuery2 = _CLNEW SpanTermQuery( t2 );
    BooleanQuery * query = _CLNEW BooleanQuery();
    query->add( spanQuery1, true, BooleanClause::MUST );
    query->add( spanQuery2, true, BooleanClause::MUST );
    
    TCHAR * expectedIds[] = { _T( "D" ), _T( "A" ) };
    // these values were pre LUCENE-413
    // final float[] expectedScores = new float[] { 0.93163157f, 0.20698164f };
    float_t expectedScores[] = { 1.0191123f, 0.93163157f };
    assertHits( query, _T( "multiple different span queries" ), expectedIds, expectedScores, 2 );

    _CLDELETE( query );
    _CLLDECDELETE( t1 );
    _CLLDECDELETE( t2 );
}

    void testBooleanOrderUnAffected()
    {
        // NOTE: uses index build in *this* setUp

        IndexReader * pReader = IndexReader::open( m_pSmall );
	    IndexSearcher * pSearch = _CLNEW IndexSearcher( pReader );

        // first do a regular RangeQuery which uses term expansion so
        // docs with more terms in range get higher scores
        Term * pLower = _CLNEW Term( _T( "data" ), _T( "1" ));
        Term * pUpper = _CLNEW Term( _T( "data" ), _T( "4" ));
        Query * rq = _CLNEW RangeQuery( pLower, pUpper, true );
        _CLLDECDELETE( pUpper );
        _CLLDECDELETE( pLower );

        Hits * pExpected = pSearch->search( rq );
        size_t numHits = pExpected->length();
 
        // now do a boolean where which also contains a
        // ConstantScoreRangeQuery and make sure the order is the same
        
        BooleanQuery * q = _CLNEW BooleanQuery();
        q->add( rq, true, BooleanClause::MUST );
        q->add( csrq( _T( "data" ), _T( "1" ), _T( "6" ), true, true ), true, BooleanClause::MUST );
 
        Hits * pActual = pSearch->search( q );
        assertEqualsMsg( _T( "wrong number of hits" ), numHits, pActual->length() );
        for( size_t i = 0; i < numHits; i++ )
        {
            assertEqualsMsg( _T( "mismatch in docid for a hit" ), pExpected->id( i ), pActual->id( i ));
        }
        _CLDELETE( pActual );
        _CLDELETE( pExpected );
        _CLDELETE( q );

        pSearch->close();
        _CLDELETE( pSearch );

        pReader->close();
        _CLDELETE( pReader );
    }

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;
}

    /////////////////////////////////////////////////////////////////////////////
    // CLucene specific 
    // Visual Studio 2005 shows memory leaks for this test, but some other 
    // tools do not detect any memory leaks. So what is right?
    // IN VC80 shows memory leaks ONLY if both sub-queries are added as 
    // MUST BooleanClauses. 
    void testBooleanMemLeaks()
    {
        IndexReader * pReader = IndexReader::open( m_pSmall );
	    IndexSearcher * pSearch = _CLNEW IndexSearcher( pReader );
 
        Query * q1 = csrq( _T( "data" ), _T( "A" ), _T( "A" ), true, true );  // matches document #0
        Query * q2 = csrq( _T( "data" ), _T( "Z" ), _T( "Z" ), true, true );  // matches document #1
        BooleanQuery * bq = _CLNEW BooleanQuery( true );
        bq->add( q1, true, BooleanClause::MUST );
        bq->add( q2, true, BooleanClause::MUST );

        Hits * pResult = pSearch->search( bq );

        _CLDELETE( pResult );
        _CLDELETE( bq );

        pSearch->close();
        _CLDELETE( pSearch );

        pReader->close();
        _CLDELETE( pReader );
    }