Exemplo n.º 1
0
void IndexContentTestCase::testIndexContent_DL()
{
    Index* pIndex;
    IndexReaderPtr pReader;

    const Term* pTerm;
    TermIteratorPtr pTermIter;
    int	docCount = 0;
    int	termCount = 0;
    uint32_t i;
    uint32_t indexTermId;
    string fileName;

    //Check posting list
    Path indexPath = TestHelper::getTestDataPath();
    indexPath.makeDirectory();
    indexPath.pushDirectory(_T("test_dlindex"));    
    pIndex = new Index(indexPath.toString().c_str(), Index::READ, NULL);
    auto_ptr<Index> indexPtr(pIndex);
    pReader = pIndex->acquireReader();
    TermReaderPtr pTermReader = pReader->termReader();

    pTermIter = pTermReader->termIterator("BODY");

    StoredFieldsReaderPtr pDocReader = pReader->createStoredFieldsReader();
    //Iterator all terms
    while(pTermIter->next())
    {
        pTerm = pTermIter->term();
		
        CPPUNIT_ASSERT(pTermReader->seek(pTerm));
				
        indexTermId = (pTerm->cast<int32_t>())->getValue();
        docCount = 0;
        TermPostingIteratorPtr pTermDocFreqs = pTermReader->termPostings();
        while(pTermDocFreqs->nextDoc())
        {
            DocumentPtr pDoc = pDocReader->document(pTermDocFreqs->doc());
            docCount++;
            // 获取文件路径
            fileName.assign(pDoc->getField("PATH")->getValue().c_str());

            TermList* pTermIdList = m_pDocScanner->getTermListOfFile(fileName);
            CPPUNIT_ASSERT(pTermIdList != NULL);

            for(i = 0, termCount = 0; i < pTermIdList->getSize(); i++)
            {
                if(indexTermId == pTermIdList->getValue(i))
                {
                    termCount++;
                }
            }
			
            CPPUNIT_ASSERT_EQUAL((tf_t)termCount, pTermDocFreqs->freq());

        }//end while nextDoc()
        CPPUNIT_ASSERT_EQUAL((df_t)docCount, pTermDocFreqs->getDocFreq());
    }
    CPPUNIT_ASSERT(m_pDocScanner->getTotalTermCount() == pReader->getNumTerms());
}
static bool verifyIndex(DirectoryPtr directory, int32_t startAt)
{
    bool fail = false;
    IndexReaderPtr reader = IndexReader::open(directory, true);

    int32_t max = reader->maxDoc();
    for (int32_t i = 0; i < max; ++i)
    {
        DocumentPtr temp = reader->document(i);
        if (temp->getField(L"count")->stringValue() != StringUtils::toString(i + startAt))
            fail = true;
    }
    reader->close();
    return fail;
}
Exemplo n.º 3
0
void IndexContentTestCase::testIndexContent_WL()
{
    Index* pIndex;
    IndexReaderPtr pReader;

    const Term* pTerm;
    TermIteratorPtr pTermIter;
    int	docCount = 0;
    int	termCount = 0;
    int	pos = -1;
    uint32_t indexTermId;
    string fileName;

    //Check posting list
    Path indexPath = TestHelper::getTestDataPath();
    indexPath.makeDirectory();
    indexPath.pushDirectory(_T("test_wlindex"));    
    pIndex = new Index(indexPath.toString().c_str(), Index::READ, NULL);
    auto_ptr<Index> indexPtr(pIndex);

    CPPUNIT_ASSERT(pIndex != NULL);

    pReader = pIndex->acquireReader();
    TermReaderPtr pTermReader = pReader->termReader();

    pTermIter = pTermReader->termIterator("BODY");

    StoredFieldsReaderPtr pDocReader = pReader->createStoredFieldsReader();
    //Iterator all terms
    while(pTermIter->next())
    {
        pTerm = pTermIter->term();

        CPPUNIT_ASSERT(pTermReader->seek(pTerm));
		
        indexTermId = (pTerm->cast<int32_t>())->getValue();
        TermPositionIteratorPtr pPositions = pTermReader->termPositions();
        docCount = 0;

        while(pPositions->nextDoc())
        {
            DocumentPtr pDoc = pDocReader->document(pPositions->doc());
            docCount++;

            fileName.assign(pDoc->getField("PATH")->getValue().c_str());

            TermList* pTermIdList = m_pDocScanner->getTermListOfFile(fileName);
            CPPUNIT_ASSERT(pTermIdList != NULL);

            pos = pPositions->nextPosition();
            termCount = 0;

            while(pos != -1)
            {
                termCount++;
                CPPUNIT_ASSERT(indexTermId == pTermIdList->getValue(pos));
                pos = pPositions->nextPosition();
            }
            CPPUNIT_ASSERT(termCount == pPositions->freq());
        }//end while nextDoc()
        CPPUNIT_ASSERT(docCount == pPositions->getDocFreq());
    }
    CPPUNIT_ASSERT_EQUAL((int64_t)m_pDocScanner->getTotalTermCount(), 
                         (int64_t)pReader->getNumTerms());
}