bool MultipleTermPositions::next()
{
if (termPositionsQueue->empty())
return false;
posList->clear();
_doc = termPositionsQueue->top()->doc();
TermPositionsPtr tp;
do
{
tp = termPositionsQueue->top();
for (int32_t i = 0; i < tp->freq(); ++i)
posList->add(tp->nextPosition());
if (tp->next())
termPositionsQueue->updateTop();
else
{
termPositionsQueue->pop();
tp->close();
}
}
while (!termPositionsQueue->empty() && termPositionsQueue->top()->doc() == _doc);
posList->sort();
_freq = posList->size();
return true;
}
TEST_F(LazyProxSkippingTest, testSeek) {
DirectoryPtr directory = newLucene<RAMDirectory>();
IndexWriterPtr writer = newLucene<IndexWriter>(directory, newLucene<WhitespaceAnalyzer>(), true, IndexWriter::MaxFieldLengthLIMITED);
for (int32_t i = 0; i < 10; ++i) {
DocumentPtr doc = newLucene<Document>();
doc->add(newLucene<Field>(field, L"a b", Field::STORE_YES, Field::INDEX_ANALYZED));
writer->addDocument(doc);
}
writer->close();
IndexReaderPtr reader = IndexReader::open(directory, true);
TermPositionsPtr tp = reader->termPositions();
tp->seek(newLucene<Term>(field, L"b"));
for (int32_t i = 0; i < 10; ++i) {
tp->next();
EXPECT_EQ(tp->doc(), i);
EXPECT_EQ(tp->nextPosition(), 1);
}
tp->seek(newLucene<Term>(field, L"a"));
for (int32_t i = 0; i < 10; ++i) {
tp->next();
EXPECT_EQ(tp->doc(), i);
EXPECT_EQ(tp->nextPosition(), 0);
}
}
bool TermPositionsQueue::lessThan(const TermPositionsPtr& first, const TermPositionsPtr& second)
{
return (first->doc() < second->doc());
}
/// Builds an index with payloads in the given Directory and performs different
/// tests to verify the payload encoding
static void encodingTest(const DirectoryPtr& dir) {
PayloadAnalyzerPtr analyzer = newLucene<PayloadAnalyzer>();
IndexWriterPtr writer = newLucene<IndexWriter>(dir, analyzer, true, IndexWriter::MaxFieldLengthLIMITED);
// should be in sync with value in TermInfosWriter
int32_t skipInterval = 16;
int32_t numTerms = 5;
String fieldName = L"f1";
int32_t numDocs = skipInterval + 1;
// create content for the test documents with just a few terms
Collection<TermPtr> terms = generateTerms(fieldName, numTerms);
StringStream sb;
for (Collection<TermPtr>::iterator term = terms.begin(); term != terms.end(); ++term) {
sb << (*term)->text() << L" ";
}
String content = sb.str();
int32_t payloadDataLength = numTerms * numDocs * 2 + numTerms * numDocs * (numDocs - 1) / 2;
ByteArray payloadData = generateRandomData(payloadDataLength);
DocumentPtr d = newLucene<Document>();
d->add(newLucene<Field>(fieldName, content, Field::STORE_NO, Field::INDEX_ANALYZED));
// add the same document multiple times to have the same payload lengths for all
// occurrences within two consecutive skip intervals
int32_t offset = 0;
for (int32_t i = 0; i < 2 * numDocs; ++i) {
analyzer->setPayloadData(fieldName, payloadData, offset, 1);
offset += numTerms;
writer->addDocument(d);
}
// make sure we create more than one segment to test merging
writer->commit();
for (int32_t i = 0; i < numDocs; ++i) {
analyzer->setPayloadData(fieldName, payloadData, offset, i);
offset += i * numTerms;
writer->addDocument(d);
}
writer->optimize();
// flush
writer->close();
// Verify the index
IndexReaderPtr reader = IndexReader::open(dir, true);
ByteArray verifyPayloadData(ByteArray::newInstance(payloadDataLength));
offset = 0;
Collection<TermPositionsPtr> tps = Collection<TermPositionsPtr>::newInstance(numTerms);
for (int32_t i = 0; i < numTerms; ++i) {
tps[i] = reader->termPositions(terms[i]);
}
while (tps[0]->next()) {
for (int32_t i = 1; i < numTerms; ++i) {
tps[i]->next();
}
int32_t freq = tps[0]->freq();
for (int32_t i = 0; i < freq; ++i) {
for (int32_t j = 0; j < numTerms; ++j) {
tps[j]->nextPosition();
tps[j]->getPayload(verifyPayloadData, offset);
offset += tps[j]->getPayloadLength();
}
}
}
for (int32_t i = 0; i < numTerms; ++i) {
tps[i]->close();
}
EXPECT_TRUE(payloadData.equals(verifyPayloadData));
// test lazy skipping
TermPositionsPtr tp = reader->termPositions(terms[0]);
tp->next();
tp->nextPosition();
// now we don't read this payload
tp->nextPosition();
EXPECT_EQ(1, tp->getPayloadLength());
ByteArray payload = tp->getPayload(ByteArray(), 0);
EXPECT_EQ(payload[0], payloadData[numTerms]);
tp->nextPosition();
// we don't read this payload and skip to a different document
tp->skipTo(5);
tp->nextPosition();
EXPECT_EQ(1, tp->getPayloadLength());
payload = tp->getPayload(ByteArray(), 0);
EXPECT_EQ(payload[0], payloadData[5 * numTerms]);
// Test different lengths at skip points
tp->seek(terms[1]);
tp->next();
tp->nextPosition();
EXPECT_EQ(1, tp->getPayloadLength());
tp->skipTo(skipInterval - 1);
tp->nextPosition();
EXPECT_EQ(1, tp->getPayloadLength());
tp->skipTo(2 * skipInterval - 1);
tp->nextPosition();
EXPECT_EQ(1, tp->getPayloadLength());
tp->skipTo(3 * skipInterval - 1);
tp->nextPosition();
EXPECT_EQ(3 * skipInterval - 2 * numDocs - 1, tp->getPayloadLength());
// Test multiple call of getPayload()
tp->getPayload(ByteArray(), 0);
// it is forbidden to call getPayload() more than once without calling nextPosition()
try {
tp->getPayload(ByteArray(), 0);
} catch (IOException& e) {
EXPECT_TRUE(check_exception(LuceneException::IO)(e));
}
reader->close();
// test long payload
analyzer = newLucene<PayloadAnalyzer>();
writer = newLucene<IndexWriter>(dir, analyzer, true, IndexWriter::MaxFieldLengthLIMITED);
String singleTerm = L"lucene";
d = newLucene<Document>();
d->add(newLucene<Field>(fieldName, singleTerm, Field::STORE_NO, Field::INDEX_ANALYZED));
// add a payload whose length is greater than the buffer size of BufferedIndexOutput
payloadData = generateRandomData(2000);
analyzer->setPayloadData(fieldName, payloadData, 100, 1500);
writer->addDocument(d);
writer->optimize();
// flush
writer->close();
reader = IndexReader::open(dir, true);
tp = reader->termPositions(newLucene<Term>(fieldName, singleTerm));
tp->next();
tp->nextPosition();
verifyPayloadData.resize(tp->getPayloadLength());
tp->getPayload(verifyPayloadData, 0);
ByteArray portion(ByteArray::newInstance(1500));
MiscUtils::arrayCopy(payloadData.get(), 100, portion.get(), 0, 1500);
EXPECT_TRUE(portion.equals(verifyPayloadData));
reader->close();
}
