public void testUpdateSameDoc() throws Exception {
final Directory dir = newDirectory();
final LineFileDocs docs = new LineFileDocs(random());
for (int r = 0; r < 3; r++) {
final IndexWriter w =
new IndexWriter(
dir, newIndexWriterConfig(new MockAnalyzer(random())).setMaxBufferedDocs(2));
final int numUpdates = atLeast(20);
int numThreads = TestUtil.nextInt(random(), 2, 6);
IndexingThread[] threads = new IndexingThread[numThreads];
for (int i = 0; i < numThreads; i++) {
threads[i] = new IndexingThread(docs, w, numUpdates);
threads[i].start();
}
for (int i = 0; i < numThreads; i++) {
threads[i].join();
}
w.close();
}
IndexReader open = DirectoryReader.open(dir);
assertEquals(1, open.numDocs());
open.close();
docs.close();
dir.close();
}
private void doTest(Random random, PrintWriter out, boolean useCompoundFiles, int MAX_DOCS)
throws Exception {
Directory directory = newDirectory();
Analyzer analyzer = new MockAnalyzer(random);
IndexWriterConfig conf = newIndexWriterConfig(analyzer);
final MergePolicy mp = conf.getMergePolicy();
mp.setNoCFSRatio(useCompoundFiles ? 1.0 : 0.0);
IndexWriter writer = new IndexWriter(directory, conf);
if (VERBOSE) {
System.out.println("TEST: now build index MAX_DOCS=" + MAX_DOCS);
}
for (int j = 0; j < MAX_DOCS; j++) {
Document d = new Document();
d.add(newTextField(PRIORITY_FIELD, HIGH_PRIORITY, Field.Store.YES));
d.add(newTextField(ID_FIELD, Integer.toString(j), Field.Store.YES));
writer.addDocument(d);
}
writer.close();
// try a search without OR
IndexReader reader = DirectoryReader.open(directory);
IndexSearcher searcher = newSearcher(reader);
Query query = new TermQuery(new Term(PRIORITY_FIELD, HIGH_PRIORITY));
out.println("Query: " + query.toString(PRIORITY_FIELD));
if (VERBOSE) {
System.out.println("TEST: search query=" + query);
}
final Sort sort = new Sort(SortField.FIELD_SCORE, new SortField(ID_FIELD, SortField.Type.INT));
ScoreDoc[] hits = searcher.search(query, null, MAX_DOCS, sort).scoreDocs;
printHits(out, hits, searcher);
checkHits(hits, MAX_DOCS, searcher);
// try a new search with OR
searcher = newSearcher(reader);
hits = null;
BooleanQuery booleanQuery = new BooleanQuery();
booleanQuery.add(
new TermQuery(new Term(PRIORITY_FIELD, HIGH_PRIORITY)), BooleanClause.Occur.SHOULD);
booleanQuery.add(
new TermQuery(new Term(PRIORITY_FIELD, MED_PRIORITY)), BooleanClause.Occur.SHOULD);
out.println("Query: " + booleanQuery.toString(PRIORITY_FIELD));
hits = searcher.search(booleanQuery, null, MAX_DOCS, sort).scoreDocs;
printHits(out, hits, searcher);
checkHits(hits, MAX_DOCS, searcher);
reader.close();
directory.close();
}
public void testTermDocsEnum() throws Exception {
Directory dir = newDirectory();
IndexWriter w =
new IndexWriter(
dir, newIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(random())));
Document d = new Document();
d.add(newStringField("f", "j", Field.Store.NO));
w.addDocument(d);
w.commit();
w.addDocument(d);
IndexReader r = w.getReader();
w.close();
DocsEnum de = MultiFields.getTermDocsEnum(r, null, "f", new BytesRef("j"));
assertEquals(0, de.nextDoc());
assertEquals(1, de.nextDoc());
assertEquals(DocIdSetIterator.NO_MORE_DOCS, de.nextDoc());
r.close();
dir.close();
}
public void testSeparateEnums() throws Exception {
Directory dir = newDirectory();
IndexWriter w =
new IndexWriter(
dir, newIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(random())));
Document d = new Document();
d.add(newStringField("f", "j", Field.Store.NO));
w.addDocument(d);
w.commit();
w.addDocument(d);
IndexReader r = w.getReader();
w.close();
DocsEnum d1 = _TestUtil.docs(random(), r, "f", new BytesRef("j"), null, null, 0);
DocsEnum d2 = _TestUtil.docs(random(), r, "f", new BytesRef("j"), null, null, 0);
assertEquals(0, d1.nextDoc());
assertEquals(0, d2.nextDoc());
r.close();
dir.close();
}
// public SortingReader(IndexReader oldReader, int[] oldToNew) { // TODO MC
public SortingReader(IndexReader oldReader, DocScore[] newToOld) { // TODO MC
super(oldReader);
this.newToOld = newToOld;
this.oldToNew = new int[oldReader.maxDoc()];
int newDoc = 0;
while (newDoc < newToOld.length) {
int oldDoc = newToOld[newDoc].doc;
oldToNew[oldDoc] = newDoc;
newDoc++;
}
}
public void sort(File directory) throws IOException {
LOG.info("IndexSorter: starting.");
Date start = new Date();
int termIndexInterval = getConf().getInt("indexer.termIndexInterval", 128);
IndexReader reader = IndexReader.open(new File(directory, "index"));
Searcher searcher =
new IndexSearcher(new File(directory, "index").getAbsolutePath()); // TODO MC
SortingReader sorter = new SortingReader(reader, newToOld(reader, searcher)); // TODO MC
IndexWriter writer = new IndexWriter(new File(directory, "index-sorted"), null, true);
writer.setTermIndexInterval(termIndexInterval);
writer.setUseCompoundFile(false);
writer.addIndexes(new IndexReader[] {sorter});
writer.close();
Date end = new Date();
LOG.info("IndexSorter: done, " + (end.getTime() - start.getTime()) + " total milliseconds");
}
public void testRandom() throws Exception {
int num = atLeast(2);
for (int iter = 0; iter < num; iter++) {
if (VERBOSE) {
System.out.println("TEST: iter=" + iter);
}
Directory dir = newDirectory();
IndexWriter w =
new IndexWriter(
dir,
newIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(random()))
.setMergePolicy(NoMergePolicy.COMPOUND_FILES));
_TestUtil.keepFullyDeletedSegments(w);
Map<BytesRef, List<Integer>> docs = new HashMap<BytesRef, List<Integer>>();
Set<Integer> deleted = new HashSet<Integer>();
List<BytesRef> terms = new ArrayList<BytesRef>();
int numDocs = _TestUtil.nextInt(random(), 1, 100 * RANDOM_MULTIPLIER);
Document doc = new Document();
Field f = newStringField("field", "", Field.Store.NO);
doc.add(f);
Field id = newStringField("id", "", Field.Store.NO);
doc.add(id);
boolean onlyUniqueTerms = random().nextBoolean();
if (VERBOSE) {
System.out.println("TEST: onlyUniqueTerms=" + onlyUniqueTerms + " numDocs=" + numDocs);
}
Set<BytesRef> uniqueTerms = new HashSet<BytesRef>();
for (int i = 0; i < numDocs; i++) {
if (!onlyUniqueTerms && random().nextBoolean() && terms.size() > 0) {
// re-use existing term
BytesRef term = terms.get(random().nextInt(terms.size()));
docs.get(term).add(i);
f.setStringValue(term.utf8ToString());
} else {
String s = _TestUtil.randomUnicodeString(random(), 10);
BytesRef term = new BytesRef(s);
if (!docs.containsKey(term)) {
docs.put(term, new ArrayList<Integer>());
}
docs.get(term).add(i);
terms.add(term);
uniqueTerms.add(term);
f.setStringValue(s);
}
id.setStringValue("" + i);
w.addDocument(doc);
if (random().nextInt(4) == 1) {
w.commit();
}
if (i > 0 && random().nextInt(20) == 1) {
int delID = random().nextInt(i);
deleted.add(delID);
w.deleteDocuments(new Term("id", "" + delID));
if (VERBOSE) {
System.out.println("TEST: delete " + delID);
}
}
}
if (VERBOSE) {
List<BytesRef> termsList = new ArrayList<BytesRef>(uniqueTerms);
Collections.sort(termsList, BytesRef.getUTF8SortedAsUTF16Comparator());
System.out.println("TEST: terms in UTF16 order:");
for (BytesRef b : termsList) {
System.out.println(" " + UnicodeUtil.toHexString(b.utf8ToString()) + " " + b);
for (int docID : docs.get(b)) {
if (deleted.contains(docID)) {
System.out.println(" " + docID + " (deleted)");
} else {
System.out.println(" " + docID);
}
}
}
}
IndexReader reader = w.getReader();
w.close();
if (VERBOSE) {
System.out.println("TEST: reader=" + reader);
}
Bits liveDocs = MultiFields.getLiveDocs(reader);
for (int delDoc : deleted) {
assertFalse(liveDocs.get(delDoc));
}
for (int i = 0; i < 100; i++) {
BytesRef term = terms.get(random().nextInt(terms.size()));
if (VERBOSE) {
System.out.println(
"TEST: seek term=" + UnicodeUtil.toHexString(term.utf8ToString()) + " " + term);
}
DocsEnum docsEnum = _TestUtil.docs(random(), reader, "field", term, liveDocs, null, 0);
assertNotNull(docsEnum);
for (int docID : docs.get(term)) {
if (!deleted.contains(docID)) {
assertEquals(docID, docsEnum.nextDoc());
}
}
assertEquals(DocIdSetIterator.NO_MORE_DOCS, docsEnum.nextDoc());
}
reader.close();
dir.close();
}
}
@Test
public void testRollingUpdates() throws Exception {
Random random = new Random(random().nextLong());
final BaseDirectoryWrapper dir = newDirectory();
// test checks for no unref'ed files with the IW helper method, which isn't aware of "tried to
// delete files"
if (dir instanceof MockDirectoryWrapper) {
((MockDirectoryWrapper) dir).setEnableVirusScanner(false);
}
final LineFileDocs docs = new LineFileDocs(random, true);
// provider.register(new MemoryCodec());
if (random().nextBoolean()) {
Codec.setDefault(
TestUtil.alwaysPostingsFormat(
new MemoryPostingsFormat(random().nextBoolean(), random.nextFloat())));
}
MockAnalyzer analyzer = new MockAnalyzer(random());
analyzer.setMaxTokenLength(TestUtil.nextInt(random(), 1, IndexWriter.MAX_TERM_LENGTH));
final IndexWriter w = new IndexWriter(dir, newIndexWriterConfig(analyzer));
final int SIZE = atLeast(20);
int id = 0;
IndexReader r = null;
IndexSearcher s = null;
final int numUpdates =
(int)
(SIZE * (2 + (TEST_NIGHTLY ? 200 * random().nextDouble() : 5 * random().nextDouble())));
if (VERBOSE) {
System.out.println("TEST: numUpdates=" + numUpdates);
}
int updateCount = 0;
// TODO: sometimes update ids not in order...
for (int docIter = 0; docIter < numUpdates; docIter++) {
final Document doc = docs.nextDoc();
final String myID = Integer.toString(id);
if (id == SIZE - 1) {
id = 0;
} else {
id++;
}
if (VERBOSE) {
System.out.println(" docIter=" + docIter + " id=" + id);
}
((Field) doc.getField("docid")).setStringValue(myID);
Term idTerm = new Term("docid", myID);
final boolean doUpdate;
if (s != null && updateCount < SIZE) {
TopDocs hits = s.search(new TermQuery(idTerm), 1);
assertEquals(1, hits.totalHits);
doUpdate = !w.tryDeleteDocument(r, hits.scoreDocs[0].doc);
if (VERBOSE) {
if (doUpdate) {
System.out.println(" tryDeleteDocument failed");
} else {
System.out.println(" tryDeleteDocument succeeded");
}
}
} else {
doUpdate = true;
if (VERBOSE) {
System.out.println(" no searcher: doUpdate=true");
}
}
updateCount++;
if (doUpdate) {
if (random().nextBoolean()) {
w.updateDocument(idTerm, doc);
} else {
// It's OK to not be atomic for this test (no separate thread reopening readers):
w.deleteDocuments(new TermQuery(idTerm));
w.addDocument(doc);
}
} else {
w.addDocument(doc);
}
if (docIter >= SIZE && random().nextInt(50) == 17) {
if (r != null) {
r.close();
}
final boolean applyDeletions = random().nextBoolean();
if (VERBOSE) {
System.out.println("TEST: reopen applyDeletions=" + applyDeletions);
}
r = w.getReader(applyDeletions);
if (applyDeletions) {
s = newSearcher(r);
} else {
s = null;
}
assertTrue(
"applyDeletions=" + applyDeletions + " r.numDocs()=" + r.numDocs() + " vs SIZE=" + SIZE,
!applyDeletions || r.numDocs() == SIZE);
updateCount = 0;
}
}
if (r != null) {
r.close();
}
w.commit();
assertEquals(SIZE, w.numDocs());
w.close();
TestIndexWriter.assertNoUnreferencedFiles(dir, "leftover files after rolling updates");
docs.close();
// LUCENE-4455:
SegmentInfos infos = SegmentInfos.readLatestCommit(dir);
long totalBytes = 0;
for (SegmentCommitInfo sipc : infos) {
totalBytes += sipc.sizeInBytes();
}
long totalBytes2 = 0;
for (String fileName : dir.listAll()) {
if (IndexFileNames.CODEC_FILE_PATTERN.matcher(fileName).matches()) {
totalBytes2 += dir.fileLength(fileName);
}
}
assertEquals(totalBytes2, totalBytes);
dir.close();
}
public static void main(String[] args) throws Exception {
// the IndexReader object is the main handle that will give you
// all the documents, terms and inverted index
IndexReader r = IndexReader.open(FSDirectory.open(new File("index")));
// You can figure out the number of documents using the maxDoc() function
System.out.println("The number of documents in this index is: " + r.maxDoc());
int i = 0;
// You can find out all the terms that have been indexed using the terms() function
TermEnum t = r.terms();
while (t.next()) {
// Since there are so many terms, let us try printing only term #100000-#100010
if (i > 100000) System.out.println("[" + i + "] " + t.term().text());
if (++i > 100010) break;
}
// You can create your own query terms by calling the Term constructor, with the field
// 'contents'
// In the following example, the query term is 'brute'
Term te = new Term("contents", "brute");
// You can also quickly find out the number of documents that have term t
System.out.println("Number of documents with the word 'brute' is: " + r.docFreq(te));
// You can use the inverted index to find out all the documents that contain the term 'brute'
// by using the termDocs function
TermDocs td = r.termDocs(te);
while (td.next()) {
System.out.println(
"Document number ["
+ td.doc()
+ "] contains the term 'brute' "
+ td.freq()
+ " time(s).");
}
// You can find the URL of the a specific document number using the document() function
// For example, the URL for document number 14191 is:
Document d = r.document(14191);
String url =
d.getFieldable("path")
.stringValue(); // the 'path' field of the Document object holds the URL
System.out.println(url.replace("%%", "/"));
// -------- Now let us use all of the functions above to make something useful --------
// The following bit of code is a worked out example of how to get a bunch of documents
// in response to a query and show them (without ranking them according to TF/IDF)
Scanner sc = new Scanner(System.in);
String str = "";
System.out.print("query> ");
while (!(str = sc.nextLine()).equals("quit")) {
String[] terms = str.split("\\s+");
for (String word : terms) {
Term term = new Term("contents", word);
TermDocs tdocs = r.termDocs(term);
while (tdocs.next()) {
String d_url =
r.document(tdocs.doc()).getFieldable("path").stringValue().replace("%%", "/");
System.out.println("[" + tdocs.doc() + "] " + d_url);
}
}
System.out.print("query> ");
}
}
// private static int[] oldToNew(IndexReader reader, Searcher searcher) throws IOException {
private static DocScore[] newToOld(IndexReader reader, Searcher searcher) throws IOException {
int readerMax = reader.maxDoc();
DocScore[] newToOld = new DocScore[readerMax];
// use site, an indexed, un-tokenized field to get boost
// byte[] boosts = reader.norms("site"); TODO MC
/* TODO MC */
Document docMeta;
Pattern includes = Pattern.compile("\\|");
String value = NutchConfiguration.create().get(INCLUDE_EXTENSIONS_KEY, "");
String includeExtensions[] = includes.split(value);
Hashtable<String, Boolean> validExtensions = new Hashtable<String, Boolean>();
for (int i = 0; i < includeExtensions.length; i++) {
validExtensions.put(includeExtensions[i], true);
System.out.println("extension boosted " + includeExtensions[i]);
}
/* TODO MC */
for (int oldDoc = 0; oldDoc < readerMax; oldDoc++) {
float score;
if (reader.isDeleted(oldDoc)) {
// score = 0.0f;
score = -1f; // TODO MC
} else {
// score = Similarity.decodeNorm(boosts[oldDoc]); TODO MC
/* TODO MC */
docMeta = searcher.doc(oldDoc);
if (validExtensions.get(docMeta.get("subType"))
== null) { // searched extensions will have higher scores
score = -0.5f;
} else {
score = Integer.parseInt(docMeta.get("inlinks"));
/*
if (score==0) {
score=0.001f; // TODO MC - to not erase
}
*/
}
/* TODO MC */
// System.out.println("Score for old document "+oldDoc+" is "+score+" and type
// "+docMeta.get("subType")); // TODO MC debug remove
}
DocScore docScore = new DocScore();
docScore.doc = oldDoc;
docScore.score = score;
newToOld[oldDoc] = docScore;
}
System.out.println("Sorting " + newToOld.length + " documents.");
Arrays.sort(newToOld);
// HeapSorter.sort(newToOld); // TODO MC - due to the lack of space
/* TODO MC
int[] oldToNew = new int[readerMax];
for (int newDoc = 0; newDoc < readerMax; newDoc++) {
DocScore docScore = newToOld[newDoc];
//oldToNew[docScore.oldDoc] = docScore.score > 0.0f ? newDoc : -1; // TODO MC
oldToNew[docScore.oldDoc] = newDoc; // TODO MC
}
*/
/* TODO MC *
for (int newDoc = 0; newDoc < readerMax; newDoc++) {
DocScore docScore = newToOld[newDoc];
System.out.println("Score for new document "+newDoc+" is "+docScore.score); // TODO MC debug remove
}
* TODO MC */
// return oldToNew; TODO MC
return newToOld; // TODO MC
}
public static void main(String[] args) throws Exception {
if (args.length < 2) {
System.err.println("TermDumper [-c|-v value] field <index...>");
System.exit(1);
}
boolean count = false;
String value = null;
boolean all = false;
int i = 0;
for (; i < args.length; i++) {
String arg = args[i];
if ("-h".equals(arg) || "--help".equals(arg)) {
System.err.println("TermDumper [-c|-v value] field <index...>");
System.exit(1);
} else if ("-c".equals(arg) || "--count".equals(arg)) {
count = true;
} else if ("-v".equals(arg) || "--vaue".equals(arg)) {
value = args[++i];
} else if ("-a".equals(arg) || "--all".equals(arg)) {
all = true;
} else {
break;
}
}
String field = args[i++];
java.util.ArrayList<IndexReader> readers =
new java.util.ArrayList<IndexReader>(args.length - 1);
for (; i < args.length; i++) {
String arg = args[i];
try {
IndexReader reader = IndexReader.open(new MMapDirectory(new File(arg)), true);
readers.add(reader);
} catch (IOException ioe) {
System.err.println("Error reading: " + arg);
}
}
for (IndexReader reader : readers) {
TermDocs termDocs = reader.termDocs();
TermEnum termEnum = reader.terms(new Term(field));
try {
do {
Term term = termEnum.term();
if (term == null || !field.equals(term.field())) break;
if (value == null) {
if (count) {
termDocs.seek(termEnum);
int c = 0;
for (; termDocs.next(); c++) ;
System.out.print(c + " ");
}
System.out.println(term.text());
} else if (value.equals(term.text())) {
termDocs.seek(termEnum);
while (termDocs.next()) {
if (all) {
Document d = reader.document(termDocs.doc());
System.out.println(termDocs.doc());
for (Object o : d.getFields()) {
Field f = (Field) o;
System.out.println(f.name() + " " + d.get(f.name()));
}
} else {
System.out.println(
termDocs.doc() + " " + reader.document(termDocs.doc()).get("url"));
}
}
}
} while (termEnum.next());
} finally {
termDocs.close();
termEnum.close();
}
}
}
