/**
* static boolean hasPendingDeletes(SegmentInfos infos) { for (SegmentInfo info : infos) { if
* (info.deletes.any()) { return true; } } return false; }
*/
void part2(IndexWriter writer, RangeMergePolicy fsmp) throws Exception {
for (int x = 20; x < 25; x++) {
writer.addDocument(DocHelper.createDocument(x, "5", 2));
// System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
writer.flush(false, false);
for (int x = 25; x < 30; x++) {
writer.addDocument(DocHelper.createDocument(x, "5", 2));
// System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
writer.flush(false, false);
// System.out.println("infos3:"+writer.segmentInfos);
Term delterm = new Term("id", "8");
writer.deleteDocuments(delterm);
// System.out.println("segdels3:" + writer.docWriter.deletesToString());
fsmp.doMerge = true;
fsmp.start = 1;
fsmp.length = 2;
writer.maybeMerge();
// deletes for info1, the newly created segment from the
// merge should have no deletes because they were applied in
// the merge
// SegmentInfo info1 = writer.segmentInfos.info(1);
// assertFalse(exists(info1, writer.docWriter.segmentDeletes));
// System.out.println("infos4:"+writer.segmentInfos);
// System.out.println("segdels4:" + writer.docWriter.deletesToString());
}
public void testIndexDivisor() throws IOException {
dir = new MockRAMDirectory();
testDoc = new Document();
DocHelper.setupDoc(testDoc);
DocHelper.writeDoc(dir, testDoc);
testTermDocs(2);
testBadSeek(2);
testSkipTo(2);
}
@Override
public void setUp() throws Exception {
super.setUp();
mergedDir = newDirectory();
merge1Dir = newDirectory();
merge2Dir = newDirectory();
DocHelper.setupDoc(doc1);
SegmentCommitInfo info1 = DocHelper.writeDoc(random(), merge1Dir, doc1);
DocHelper.setupDoc(doc2);
SegmentCommitInfo info2 = DocHelper.writeDoc(random(), merge2Dir, doc2);
reader1 = new SegmentReader(info1, newIOContext(random()));
reader2 = new SegmentReader(info2, newIOContext(random()));
}
public void testIndexDivisorAfterLoad() throws IOException {
dir = new MockRAMDirectory();
testDoc = new Document();
DocHelper.setupDoc(testDoc);
SegmentInfo si = DocHelper.writeDoc(dir, testDoc);
SegmentReader reader = SegmentReader.get(si);
assertEquals(1, reader.docFreq(new Term("keyField", "Keyword")));
try {
reader.setTermInfosIndexDivisor(2);
fail("did not hit IllegalStateException exception");
} catch (IllegalStateException ise) {
// expected
}
}
@BeforeClass
public static void beforeClass() throws Exception {
fieldInfos = new FieldInfos.Builder();
DocHelper.setupDoc(testDoc);
for (IndexableField field : testDoc) {
fieldInfos.addOrUpdate(field.name(), field.fieldType());
}
dir = newDirectory();
IndexWriterConfig conf =
newIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(random()))
.setMergePolicy(newLogMergePolicy());
((LogMergePolicy) conf.getMergePolicy()).setUseCompoundFile(false);
IndexWriter writer = new IndexWriter(dir, conf);
writer.addDocument(testDoc);
writer.close();
FaultyIndexInput.doFail = false;
}
@BeforeClass
public static void beforeClass() throws Exception {
testDoc = new Document();
fieldInfos = new FieldInfos.Builder();
DocHelper.setupDoc(testDoc);
for (IndexableField field : testDoc.getFields()) {
FieldInfo fieldInfo = fieldInfos.getOrAdd(field.name());
IndexableFieldType ift = field.fieldType();
fieldInfo.setIndexOptions(ift.indexOptions());
if (ift.omitNorms()) {
fieldInfo.setOmitsNorms();
}
fieldInfo.setDocValuesType(ift.docValuesType());
}
dir = newDirectory();
IndexWriterConfig conf =
newIndexWriterConfig(new MockAnalyzer(random())).setMergePolicy(newLogMergePolicy());
conf.getMergePolicy().setNoCFSRatio(0.0);
IndexWriter writer = new IndexWriter(dir, conf);
writer.addDocument(testDoc);
writer.close();
}
protected void setUp() throws Exception {
super.setUp();
DocHelper.setupDoc(testDoc);
info = DocHelper.writeDoc(dir, testDoc);
}
public void testDeletes1() throws Exception {
// IndexWriter.debug2 = System.out;
Directory dir = new MockDirectoryWrapper(new Random(random().nextLong()), new RAMDirectory());
IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
iwc.setMergeScheduler(new SerialMergeScheduler());
iwc.setMaxBufferedDocs(5000);
iwc.setRAMBufferSizeMB(100);
RangeMergePolicy fsmp = new RangeMergePolicy(false);
iwc.setMergePolicy(fsmp);
IndexWriter writer = new IndexWriter(dir, iwc);
for (int x = 0; x < 5; x++) {
writer.addDocument(DocHelper.createDocument(x, "1", 2));
// System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
// System.out.println("commit1");
writer.commit();
assertEquals(1, writer.segmentInfos.size());
for (int x = 5; x < 10; x++) {
writer.addDocument(DocHelper.createDocument(x, "2", 2));
// System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
// System.out.println("commit2");
writer.commit();
assertEquals(2, writer.segmentInfos.size());
for (int x = 10; x < 15; x++) {
writer.addDocument(DocHelper.createDocument(x, "3", 2));
// System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs());
}
writer.deleteDocuments(new Term("id", "1"));
writer.deleteDocuments(new Term("id", "11"));
// flushing without applying deletes means
// there will still be deletes in the segment infos
writer.flush(false, false);
assertTrue(writer.bufferedUpdatesStream.any());
// get reader flushes pending deletes
// so there should not be anymore
IndexReader r1 = writer.getReader();
assertFalse(writer.bufferedUpdatesStream.any());
r1.close();
// delete id:2 from the first segment
// merge segments 0 and 1
// which should apply the delete id:2
writer.deleteDocuments(new Term("id", "2"));
writer.flush(false, false);
fsmp = (RangeMergePolicy) writer.getConfig().getMergePolicy();
fsmp.doMerge = true;
fsmp.start = 0;
fsmp.length = 2;
writer.maybeMerge();
assertEquals(2, writer.segmentInfos.size());
// id:2 shouldn't exist anymore because
// it's been applied in the merge and now it's gone
IndexReader r2 = writer.getReader();
int[] id2docs = toDocsArray(new Term("id", "2"), null, r2);
assertTrue(id2docs == null);
r2.close();
/**
* // added docs are in the ram buffer for (int x = 15; x < 20; x++) {
* writer.addDocument(TestIndexWriterReader.createDocument(x, "4", 2));
* System.out.println("numRamDocs(" + x + ")" + writer.numRamDocs()); }
* assertTrue(writer.numRamDocs() > 0); // delete from the ram buffer writer.deleteDocuments(new
* Term("id", Integer.toString(13)));
*
* <p>Term id3 = new Term("id", Integer.toString(3));
*
* <p>// delete from the 1st segment writer.deleteDocuments(id3);
*
* <p>assertTrue(writer.numRamDocs() > 0);
*
* <p>//System.out // .println("segdels1:" + writer.docWriter.deletesToString());
*
* <p>//assertTrue(writer.docWriter.segmentDeletes.size() > 0);
*
* <p>// we cause a merge to happen fsmp.doMerge = true; fsmp.start = 0; fsmp.length = 2;
* System.out.println("maybeMerge "+writer.segmentInfos);
*
* <p>SegmentInfo info0 = writer.segmentInfos.info(0); SegmentInfo info1 =
* writer.segmentInfos.info(1);
*
* <p>writer.maybeMerge(); System.out.println("maybeMerge after "+writer.segmentInfos); // there
* should be docs in RAM assertTrue(writer.numRamDocs() > 0);
*
* <p>// assert we've merged the 1 and 2 segments // and still have a segment leftover == 2
* assertEquals(2, writer.segmentInfos.size()); assertFalse(segThere(info0,
* writer.segmentInfos)); assertFalse(segThere(info1, writer.segmentInfos));
*
* <p>//System.out.println("segdels2:" + writer.docWriter.deletesToString());
*
* <p>//assertTrue(writer.docWriter.segmentDeletes.size() > 0);
*
* <p>IndexReader r = writer.getReader(); IndexReader r1 = r.getSequentialSubReaders()[0];
* printDelDocs(r1.getLiveDocs()); int[] docs = toDocsArray(id3, null, r);
* System.out.println("id3 docs:"+Arrays.toString(docs)); // there shouldn't be any docs for
* id:3 assertTrue(docs == null); r.close();
*
* <p>part2(writer, fsmp);
*/
// System.out.println("segdels2:"+writer.docWriter.segmentDeletes.toString());
// System.out.println("close");
writer.close();
dir.close();
}
public void testMerge() throws IOException {
final Codec codec = Codec.getDefault();
final SegmentInfo si =
new SegmentInfo(
mergedDir,
Version.LATEST,
mergedSegment,
-1,
false,
codec,
Collections.emptyMap(),
StringHelper.randomId(),
new HashMap<>());
SegmentMerger merger =
new SegmentMerger(
Arrays.<CodecReader>asList(reader1, reader2),
si,
InfoStream.getDefault(),
mergedDir,
new FieldInfos.FieldNumbers(),
newIOContext(random(), new IOContext(new MergeInfo(-1, -1, false, -1))));
MergeState mergeState = merger.merge();
int docsMerged = mergeState.segmentInfo.maxDoc();
assertTrue(docsMerged == 2);
// Should be able to open a new SegmentReader against the new directory
SegmentReader mergedReader =
new SegmentReader(
new SegmentCommitInfo(mergeState.segmentInfo, 0, -1L, -1L, -1L),
newIOContext(random()));
assertTrue(mergedReader != null);
assertTrue(mergedReader.numDocs() == 2);
Document newDoc1 = mergedReader.document(0);
assertTrue(newDoc1 != null);
// There are 2 unstored fields on the document
assertTrue(
DocHelper.numFields(newDoc1) == DocHelper.numFields(doc1) - DocHelper.unstored.size());
Document newDoc2 = mergedReader.document(1);
assertTrue(newDoc2 != null);
assertTrue(
DocHelper.numFields(newDoc2) == DocHelper.numFields(doc2) - DocHelper.unstored.size());
PostingsEnum termDocs =
TestUtil.docs(
random(), mergedReader, DocHelper.TEXT_FIELD_2_KEY, new BytesRef("field"), null, 0);
assertTrue(termDocs != null);
assertTrue(termDocs.nextDoc() != DocIdSetIterator.NO_MORE_DOCS);
int tvCount = 0;
for (FieldInfo fieldInfo : mergedReader.getFieldInfos()) {
if (fieldInfo.hasVectors()) {
tvCount++;
}
}
// System.out.println("stored size: " + stored.size());
assertEquals("We do not have 3 fields that were indexed with term vector", 3, tvCount);
Terms vector = mergedReader.getTermVectors(0).terms(DocHelper.TEXT_FIELD_2_KEY);
assertNotNull(vector);
assertEquals(3, vector.size());
TermsEnum termsEnum = vector.iterator();
int i = 0;
while (termsEnum.next() != null) {
String term = termsEnum.term().utf8ToString();
int freq = (int) termsEnum.totalTermFreq();
// System.out.println("Term: " + term + " Freq: " + freq);
assertTrue(DocHelper.FIELD_2_TEXT.indexOf(term) != -1);
assertTrue(DocHelper.FIELD_2_FREQS[i] == freq);
i++;
}
TestSegmentReader.checkNorms(mergedReader);
mergedReader.close();
}