/// Build a randomish document for both RAMDirectory and MemoryIndex, and run all the queries against it.
void checkAgainstRAMDirectory() {
StringStream fooField;
StringStream termField;
// add up to 250 terms to field "foo"
int32_t fieldCount = random->nextInt(250) + 1;
for (int32_t i = 0; i < fieldCount; ++i) {
fooField << L" " << randomTerm();
}
// add up to 250 terms to field "foo"
int32_t termCount = random->nextInt(250) + 1;
for (int32_t i = 0; i < termCount; ++i) {
termField << L" " << randomTerm();
}
RAMDirectoryPtr ramdir = newLucene<RAMDirectory>();
AnalyzerPtr analyzer = randomAnalyzer();
IndexWriterPtr writer = newLucene<IndexWriter>(ramdir, analyzer, IndexWriter::MaxFieldLengthUNLIMITED);
DocumentPtr doc = newLucene<Document>();
FieldPtr field1 = newLucene<Field>(L"foo", fooField.str(), Field::STORE_NO, Field::INDEX_ANALYZED);
FieldPtr field2 = newLucene<Field>(L"term", termField.str(), Field::STORE_NO, Field::INDEX_ANALYZED);
doc->add(field1);
doc->add(field2);
writer->addDocument(doc);
writer->close();
MemoryIndexPtr memory = newLucene<MemoryIndex>();
memory->addField(L"foo", fooField.str(), analyzer);
memory->addField(L"term", termField.str(), analyzer);
checkAllQueries(memory, ramdir, analyzer);
}
/// Return a random unicode term, like StressIndexingTest.
String randomString() {
int32_t end = random->nextInt(20);
if (buffer.size() < 1 + end) {
buffer.resize((int32_t)((double)(1 + end) * 1.25));
}
for (int32_t i = 0; i < end; ++i) {
int32_t t = random->nextInt(5);
if (t == 0 && i < end - 1) {
#ifdef LPP_UNICODE_CHAR_SIZE_2
// Make a surrogate pair
// High surrogate
buffer[i++] = (wchar_t)nextInt(0xd800, 0xdc00);
// Low surrogate
buffer[i] = (wchar_t)nextInt(0xdc00, 0xe000);
#else
buffer[i] = (wchar_t)nextInt(0xdc00, 0xe000);
#endif
} else if (t <= 1) {
buffer[i] = (wchar_t)nextInt(0x01, 0x80);
} else if (t == 2) {
buffer[i] = (wchar_t)nextInt(0x80, 0x800);
} else if (t == 3) {
buffer[i] = (wchar_t)nextInt(0x800, 0xd800);
} else if (t == 4) {
buffer[i] = (wchar_t)nextInt(0xe000, 0xfff0);
}
}
return String(buffer.get(), end);
}
void testSorting(int32_t precisionStep) {
RandomPtr rnd = newLucene<Random>();
String field = L"field" + StringUtils::toString(precisionStep);
// 10 random tests, the index order is ascending, so using a reverse sort field should return descending documents
for (int32_t i = 0; i < 10; ++i) {
int32_t lower = (int32_t)(rnd->nextDouble() * noDocs * distance) + startOffset;
int32_t upper = (int32_t)(rnd->nextDouble() * noDocs * distance) + startOffset;
if (lower > upper) {
std::swap(lower, upper);
}
QueryPtr tq = NumericRangeQuery::newIntRange(field, precisionStep, lower, upper, true, true);
TopDocsPtr topDocs = searcher->search(tq, FilterPtr(), noDocs, newLucene<Sort>(newLucene<SortField>(field, SortField::INT, true)));
if (topDocs->totalHits == 0) {
continue;
}
Collection<ScoreDocPtr> sd = topDocs->scoreDocs;
EXPECT_TRUE(sd);
int32_t last = StringUtils::toInt(searcher->doc(sd[0]->doc)->get(field));
for (int32_t j = 1; j < sd.size(); ++j) {
int32_t act = StringUtils::toInt(searcher->doc(sd[j]->doc)->get(field));
EXPECT_TRUE(last > act);
last = act;
}
}
}
void testRangeSplit(int32_t precisionStep) {
RandomPtr rnd = newLucene<Random>();
String field = L"ascfield" + StringUtils::toString(precisionStep);
// 50 random tests
for (int32_t i = 0; i < 50; ++i) {
int32_t lower = (int32_t)(rnd->nextDouble() * noDocs - noDocs / 2.0);
int32_t upper = (int32_t)(rnd->nextDouble() * noDocs - noDocs / 2.0);
if (lower > upper) {
std::swap(lower, upper);
}
// test inclusive range
QueryPtr tq = NumericRangeQuery::newIntRange(field, precisionStep, lower, upper, true, true);
TopDocsPtr tTopDocs = searcher->search(tq, 1);
EXPECT_EQ(upper - lower + 1, tTopDocs->totalHits);
// test exclusive range
tq = NumericRangeQuery::newIntRange(field, precisionStep, lower, upper, false, false);
tTopDocs = searcher->search(tq, 1);
EXPECT_EQ(std::max(upper - lower - 1, (int32_t)0), tTopDocs->totalHits);
// test left exclusive range
tq = NumericRangeQuery::newIntRange(field, precisionStep, lower, upper, false, true);
tTopDocs = searcher->search(tq, 1);
EXPECT_EQ(upper - lower, tTopDocs->totalHits);
// test right exclusive range
tq = NumericRangeQuery::newIntRange(field, precisionStep, lower, upper, true, false);
tTopDocs = searcher->search(tq, 1);
EXPECT_EQ(upper - lower, tTopDocs->totalHits);
}
}
/// half of the time, returns a random term from TEST_TERMS.
/// the other half of the time, returns a random unicode string.
String randomTerm() {
if (random->nextInt() % 2 == 1) {
// return a random TEST_TERM
return TEST_TERMS[random->nextInt(TEST_TERMS.size())];
} else {
// return a random unicode term
return randomString();
}
}
String randomToken()
{
static const wchar_t* alphabet = L"abcdefghijklmnopqrstuvwxyz";
int32_t tl = 1 + rndToken->nextInt(7);
StringStream sb;
for (int32_t cx = 0; cx < tl; ++cx)
sb << alphabet[rndToken->nextInt(25)];
return sb.str();
}
void testRandomTrieAndClassicRangeQuery(int32_t precisionStep) {
RandomPtr rnd = newLucene<Random>();
String field = L"field" + StringUtils::toString(precisionStep);
int32_t termCountT = 0;
int32_t termCountC = 0;
for (int32_t i = 0; i < 50; ++i) {
int32_t lower = (int32_t)(rnd->nextDouble() * noDocs * distance) + startOffset;
int32_t upper = (int32_t)(rnd->nextDouble() * noDocs * distance) + startOffset;
if (lower > upper) {
std::swap(lower, upper);
}
// test inclusive range
NumericRangeQueryPtr tq = NumericRangeQuery::newIntRange(field, precisionStep, lower, upper, true, true);
TermRangeQueryPtr cq = newLucene<TermRangeQuery>(field, NumericUtils::intToPrefixCoded(lower), NumericUtils::intToPrefixCoded(upper), true, true);
TopDocsPtr tTopDocs = searcher->search(tq, 1);
TopDocsPtr cTopDocs = searcher->search(cq, 1);
EXPECT_EQ(cTopDocs->totalHits, tTopDocs->totalHits);
termCountT += tq->getTotalNumberOfTerms();
termCountC += cq->getTotalNumberOfTerms();
// test exclusive range
tq = NumericRangeQuery::newIntRange(field, precisionStep, lower, upper, false, false);
cq = newLucene<TermRangeQuery>(field, NumericUtils::intToPrefixCoded(lower), NumericUtils::intToPrefixCoded(upper), false, false);
tTopDocs = searcher->search(tq, 1);
cTopDocs = searcher->search(cq, 1);
EXPECT_EQ(cTopDocs->totalHits, tTopDocs->totalHits);
termCountT += tq->getTotalNumberOfTerms();
termCountC += cq->getTotalNumberOfTerms();
// test left exclusive range
tq = NumericRangeQuery::newIntRange(field, precisionStep, lower, upper, false, true);
cq = newLucene<TermRangeQuery>(field, NumericUtils::intToPrefixCoded(lower), NumericUtils::intToPrefixCoded(upper), false, true);
tTopDocs = searcher->search(tq, 1);
cTopDocs = searcher->search(cq, 1);
EXPECT_EQ(cTopDocs->totalHits, tTopDocs->totalHits);
termCountT += tq->getTotalNumberOfTerms();
termCountC += cq->getTotalNumberOfTerms();
// test right exclusive range
tq = NumericRangeQuery::newIntRange(field, precisionStep, lower, upper, true, false);
cq = newLucene<TermRangeQuery>(field, NumericUtils::intToPrefixCoded(lower), NumericUtils::intToPrefixCoded(upper), true, false);
tTopDocs = searcher->search(tq, 1);
cTopDocs = searcher->search(cq, 1);
EXPECT_EQ(cTopDocs->totalHits, tTopDocs->totalHits);
termCountT += tq->getTotalNumberOfTerms();
termCountC += cq->getTotalNumberOfTerms();
}
if (precisionStep == INT_MAX) {
EXPECT_EQ(termCountT, termCountC);
}
}
String randomField()
{
int32_t fl = 1 + rndToken->nextInt(3);
StringStream fb;
for (int32_t fx = 0; fx < fl; ++fx)
fb << randomToken() << L" ";
return fb.str();
}
void runReadBytes(IndexInputPtr input, int32_t bufferSize)
{
int32_t pos = 0;
RandomPtr random = newLucene<Random>();
// gradually increasing size
for (int32_t size = 1; size < bufferSize * 10; size = size + size / 200 + 1)
{
checkReadBytes(input, size, pos);
pos += size;
if (pos >= TEST_FILE_LENGTH)
{
// wrap
pos = 0;
input->seek(0);
}
}
// wildly fluctuating size
for (int64_t i = 0; i < 1000; ++i)
{
int32_t size = random->nextInt(10000);
checkReadBytes(input, 1 + size, pos);
pos += 1 + size;
if (pos >= TEST_FILE_LENGTH)
{
// wrap
pos = 0;
input->seek(0);
}
}
// constant small size (7 bytes)
for (int32_t i = 0; i < bufferSize; ++i)
{
checkReadBytes(input, 7, pos);
pos += 7;
if (pos >= TEST_FILE_LENGTH)
{
// wrap
pos = 0;
input->seek(0);
}
}
}
AnalyzerPtr randomAnalyzer() {
switch (random->nextInt(3)) {
case 0:
return newLucene<SimpleAnalyzer>();
case 1:
return newLucene<StopAnalyzer>(LuceneVersion::LUCENE_CURRENT);
default:
return newLucene<StandardAnalyzer>(LuceneVersion::LUCENE_CURRENT);
}
}
virtual bool testPoint(const String& name)
{
if (random->nextInt(4) == 2)
LuceneThread::threadYield();
return true;
}
/// start is inclusive and end is exclusive
int32_t nextInt(int32_t start, int32_t end) {
return start + random->nextInt(end - start);
}
String getTempDir(const String& desc) {
return FileUtils::joinPath(getTempDir(), desc + L"." + StringUtils::toString(randomTest->nextInt()));
}
namespace Lucene {
static RandomPtr randomTest = newLucene<Random>();
static String testDir;
void setTestDir(const String& dir) {
testDir = dir;
}
String getTestDir() {
if (testDir.empty()) {
boost::throw_exception(RuntimeException(L"test directory not set"));
}
return testDir;
}
String getTempDir() {
static String tempDir;
if (tempDir.empty()) {
tempDir = FileUtils::joinPath(getTestDir(), L"temp");
FileUtils::createDirectory(tempDir);
}
return tempDir;
}
String getTempDir(const String& desc) {
return FileUtils::joinPath(getTempDir(), desc + L"." + StringUtils::toString(randomTest->nextInt()));
}
void syncConcurrentMerges(const IndexWriterPtr& writer) {
syncConcurrentMerges(writer->getMergeScheduler());
}
void syncConcurrentMerges(const MergeSchedulerPtr& ms) {
if (MiscUtils::typeOf<ConcurrentMergeScheduler>(ms)) {
boost::dynamic_pointer_cast<ConcurrentMergeScheduler>(ms)->sync();
}
}
String intToEnglish(int32_t i) {
String english(_intToEnglish(i));
boost::trim(english);
return english;
}
String _intToEnglish(int32_t i) {
String english;
if (i == 0) {
return L"zero";
}
if (i < 0) {
english += L"minus ";
i = -i;
}
if (i >= 1000000000) { // billions
english += _intToEnglish(i / 1000000000);
english += L"billion, ";
i = i % 1000000000;
}
if (i >= 1000000) { // millions
english += _intToEnglish(i / 1000000);
english += L"million, ";
i = i % 1000000;
}
if (i >= 1000) { // thousands
english += _intToEnglish(i / 1000);
english += L"thousand, ";
i = i % 1000;
}
if (i >= 100) { // hundreds
english += _intToEnglish(i / 100);
english += L"hundred ";
i = i % 100;
}
if (i >= 20) {
switch (i/10) {
case 9:
english += L"ninety";
break;
case 8:
english += L"eighty";
break;
case 7:
english += L"seventy";
break;
case 6:
english += L"sixty";
break;
case 5:
english += L"fifty";
break;
case 4:
english += L"forty";
break;
case 3:
english += L"thirty";
break;
case 2:
english += L"twenty";
break;
}
i = i % 10;
english += i == 0 ? L" " : L"-";
}
switch (i) {
case 19:
english += L"nineteen ";
break;
case 18:
english += L"eighteen ";
break;
case 17:
english += L"seventeen ";
break;
case 16:
english += L"sixteen ";
break;
case 15:
english += L"fifteen ";
break;
case 14:
english += L"fourteen ";
break;
case 13:
english += L"thirteen ";
break;
case 12:
english += L"twelve ";
break;
case 11:
english += L"eleven ";
break;
case 10:
english += L"ten ";
break;
case 9:
english += L"nine ";
break;
case 8:
english += L"eight ";
break;
case 7:
english += L"seven ";
break;
case 6:
english += L"six ";
break;
case 5:
english += L"five ";
break;
case 4:
english += L"four ";
break;
case 3:
english += L"three ";
break;
case 2:
english += L"two ";
break;
case 1:
english += L"one ";
break;
}
return english;
}
bool checkIndex(const DirectoryPtr& dir) {
CheckIndexPtr checker = newLucene<CheckIndex>(dir);
IndexStatusPtr indexStatus = checker->checkIndex();
if (!indexStatus || !indexStatus->clean) {
boost::throw_exception(RuntimeException(L"CheckIndex failed"));
return false;
}
return true;
}
}
