1 package org.apache.lucene.queryParser.standard;
4 * Licensed to the Apache Software Foundation (ASF) under one or more
5 * contributor license agreements. See the NOTICE file distributed with
6 * this work for additional information regarding copyright ownership.
7 * The ASF licenses this file to You under the Apache License, Version 2.0
8 * (the "License"); you may not use this file except in compliance with
9 * the License. You may obtain a copy of the License at
11 * http://www.apache.org/licenses/LICENSE-2.0
13 * Unless required by applicable law or agreed to in writing, software
14 * distributed under the License is distributed on an "AS IS" BASIS,
15 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
16 * See the License for the specific language governing permissions and
17 * limitations under the License.
20 import java.io.IOException;
21 import java.io.Reader;
22 import java.text.Collator;
23 import java.text.DateFormat;
24 import java.util.Calendar;
25 import java.util.Date;
26 import java.util.GregorianCalendar;
27 import java.util.HashMap;
28 import java.util.List;
29 import java.util.Locale;
31 import java.util.Collections;
33 import org.apache.lucene.analysis.Analyzer;
34 import org.apache.lucene.analysis.KeywordAnalyzer;
35 import org.apache.lucene.analysis.LowerCaseTokenizer;
36 import org.apache.lucene.analysis.MockAnalyzer;
37 import org.apache.lucene.analysis.MockTokenizer;
38 import org.apache.lucene.analysis.SimpleAnalyzer;
39 import org.apache.lucene.analysis.StopAnalyzer;
40 import org.apache.lucene.analysis.StopFilter;
41 import org.apache.lucene.analysis.TokenFilter;
42 import org.apache.lucene.analysis.TokenStream;
43 import org.apache.lucene.analysis.Tokenizer;
44 import org.apache.lucene.analysis.WhitespaceAnalyzer;
45 import org.apache.lucene.analysis.standard.StandardAnalyzer;
46 import org.apache.lucene.analysis.tokenattributes.CharTermAttribute;
47 import org.apache.lucene.analysis.tokenattributes.OffsetAttribute;
48 import org.apache.lucene.analysis.tokenattributes.PositionIncrementAttribute;
49 import org.apache.lucene.document.DateField;
50 import org.apache.lucene.document.DateTools;
51 import org.apache.lucene.document.Document;
52 import org.apache.lucene.document.Field;
53 import org.apache.lucene.index.IndexWriter;
54 import org.apache.lucene.index.IndexReader;
55 import org.apache.lucene.index.Term;
56 import org.apache.lucene.messages.MessageImpl;
57 import org.apache.lucene.queryParser.core.QueryNodeException;
58 import org.apache.lucene.queryParser.core.messages.QueryParserMessages;
59 import org.apache.lucene.queryParser.core.nodes.FuzzyQueryNode;
60 import org.apache.lucene.queryParser.core.nodes.QueryNode;
61 import org.apache.lucene.queryParser.core.processors.QueryNodeProcessorImpl;
62 import org.apache.lucene.queryParser.core.processors.QueryNodeProcessorPipeline;
63 import org.apache.lucene.queryParser.standard.config.StandardQueryConfigHandler;
64 import org.apache.lucene.queryParser.standard.config.StandardQueryConfigHandler.Operator;
65 import org.apache.lucene.queryParser.standard.nodes.WildcardQueryNode;
66 import org.apache.lucene.search.BooleanClause;
67 import org.apache.lucene.search.BooleanQuery;
68 import org.apache.lucene.search.FuzzyQuery;
69 import org.apache.lucene.search.IndexSearcher;
70 import org.apache.lucene.search.MatchAllDocsQuery;
71 import org.apache.lucene.search.MultiPhraseQuery;
72 import org.apache.lucene.search.MultiTermQuery;
73 import org.apache.lucene.search.PhraseQuery;
74 import org.apache.lucene.search.PrefixQuery;
75 import org.apache.lucene.search.Query;
76 import org.apache.lucene.search.ScoreDoc;
77 import org.apache.lucene.search.TermQuery;
78 import org.apache.lucene.search.TermRangeQuery;
79 import org.apache.lucene.search.WildcardQuery;
80 import org.apache.lucene.store.Directory;
81 import org.apache.lucene.util.LuceneTestCase;
84 * This test case is a copy of the core Lucene query parser test, it was adapted
85 * to use new QueryParserHelper instead of the old query parser.
89 public class TestQPHelper extends LuceneTestCase {
91 public static Analyzer qpAnalyzer = new QPTestAnalyzer();
93 public static final class QPTestFilter extends TokenFilter {
94 private final CharTermAttribute termAtt = addAttribute(CharTermAttribute.class);
95 private final OffsetAttribute offsetAtt = addAttribute(OffsetAttribute.class);
98 * Filter which discards the token 'stop' and which expands the token
99 * 'phrase' into 'phrase1 phrase2'
101 public QPTestFilter(TokenStream in) {
105 private boolean inPhrase = false;
106 private int savedStart = 0;
107 private int savedEnd = 0;
110 public boolean incrementToken() throws IOException {
114 termAtt.setEmpty().append("phrase2");
115 offsetAtt.setOffset(savedStart, savedEnd);
118 while (input.incrementToken()) {
119 if (termAtt.toString().equals("phrase")) {
121 savedStart = offsetAtt.startOffset();
122 savedEnd = offsetAtt.endOffset();
123 termAtt.setEmpty().append("phrase1");
124 offsetAtt.setOffset(savedStart, savedEnd);
126 } else if (!termAtt.toString().equals("stop"))
133 public void reset() throws IOException {
135 this.inPhrase = false;
141 public static final class QPTestAnalyzer extends Analyzer {
143 /** Filters LowerCaseTokenizer with StopFilter. */
145 public final TokenStream tokenStream(String fieldName, Reader reader) {
146 return new QPTestFilter(new LowerCaseTokenizer(TEST_VERSION_CURRENT, reader));
150 public static class QPTestParser extends StandardQueryParser {
151 public QPTestParser(Analyzer a) {
152 ((QueryNodeProcessorPipeline)getQueryNodeProcessor())
153 .add(new QPTestParserQueryNodeProcessor());
158 private static class QPTestParserQueryNodeProcessor extends
159 QueryNodeProcessorImpl {
162 protected QueryNode postProcessNode(QueryNode node)
163 throws QueryNodeException {
170 protected QueryNode preProcessNode(QueryNode node)
171 throws QueryNodeException {
173 if (node instanceof WildcardQueryNode || node instanceof FuzzyQueryNode) {
175 throw new QueryNodeException(new MessageImpl(
176 QueryParserMessages.EMPTY_MESSAGE));
185 protected List<QueryNode> setChildrenOrder(List<QueryNode> children)
186 throws QueryNodeException {
196 private int originalMaxClauses;
199 public void setUp() throws Exception {
201 originalMaxClauses = BooleanQuery.getMaxClauseCount();
204 public StandardQueryParser getParser(Analyzer a) throws Exception {
206 a = new SimpleAnalyzer(TEST_VERSION_CURRENT);
207 StandardQueryParser qp = new StandardQueryParser();
210 qp.setDefaultOperator(StandardQueryConfigHandler.Operator.OR);
216 public Query getQuery(String query, Analyzer a) throws Exception {
217 return getParser(a).parse(query, "field");
220 public Query getQueryAllowLeadingWildcard(String query, Analyzer a) throws Exception {
221 StandardQueryParser parser = getParser(a);
222 parser.setAllowLeadingWildcard(true);
223 return parser.parse(query, "field");
226 public void assertQueryEquals(String query, Analyzer a, String result)
228 Query q = getQuery(query, a);
229 String s = q.toString("field");
230 if (!s.equals(result)) {
231 fail("Query /" + query + "/ yielded /" + s + "/, expecting /" + result
236 public void assertQueryEqualsAllowLeadingWildcard(String query, Analyzer a, String result)
238 Query q = getQueryAllowLeadingWildcard(query, a);
239 String s = q.toString("field");
240 if (!s.equals(result)) {
241 fail("Query /" + query + "/ yielded /" + s + "/, expecting /" + result
246 public void assertQueryEquals(StandardQueryParser qp, String field,
247 String query, String result) throws Exception {
248 Query q = qp.parse(query, field);
249 String s = q.toString(field);
250 if (!s.equals(result)) {
251 fail("Query /" + query + "/ yielded /" + s + "/, expecting /" + result
256 public void assertEscapedQueryEquals(String query, Analyzer a, String result)
258 String escapedQuery = QueryParserUtil.escape(query);
259 if (!escapedQuery.equals(result)) {
260 fail("Query /" + query + "/ yielded /" + escapedQuery + "/, expecting /"
265 public void assertWildcardQueryEquals(String query, boolean lowercase,
266 String result, boolean allowLeadingWildcard) throws Exception {
267 StandardQueryParser qp = getParser(null);
268 qp.setLowercaseExpandedTerms(lowercase);
269 qp.setAllowLeadingWildcard(allowLeadingWildcard);
270 Query q = qp.parse(query, "field");
271 String s = q.toString("field");
272 if (!s.equals(result)) {
273 fail("WildcardQuery /" + query + "/ yielded /" + s + "/, expecting /"
278 public void assertWildcardQueryEquals(String query, boolean lowercase,
279 String result) throws Exception {
280 assertWildcardQueryEquals(query, lowercase, result, false);
283 public void assertWildcardQueryEquals(String query, String result)
285 StandardQueryParser qp = getParser(null);
286 Query q = qp.parse(query, "field");
287 String s = q.toString("field");
288 if (!s.equals(result)) {
289 fail("WildcardQuery /" + query + "/ yielded /" + s + "/, expecting /"
294 public Query getQueryDOA(String query, Analyzer a) throws Exception {
296 a = new SimpleAnalyzer(TEST_VERSION_CURRENT);
297 StandardQueryParser qp = new StandardQueryParser();
299 qp.setDefaultOperator(Operator.AND);
301 return qp.parse(query, "field");
305 public void assertQueryEqualsDOA(String query, Analyzer a, String result)
307 Query q = getQueryDOA(query, a);
308 String s = q.toString("field");
309 if (!s.equals(result)) {
310 fail("Query /" + query + "/ yielded /" + s + "/, expecting /" + result
315 public void testConstantScoreAutoRewrite() throws Exception {
316 StandardQueryParser qp = new StandardQueryParser(new MockAnalyzer(random, MockTokenizer.WHITESPACE, false));
317 Query q = qp.parse("foo*bar", "field");
318 assertTrue(q instanceof WildcardQuery);
319 assertEquals(MultiTermQuery.CONSTANT_SCORE_AUTO_REWRITE_DEFAULT, ((MultiTermQuery) q).getRewriteMethod());
321 q = qp.parse("foo*", "field");
322 assertTrue(q instanceof PrefixQuery);
323 assertEquals(MultiTermQuery.CONSTANT_SCORE_AUTO_REWRITE_DEFAULT, ((MultiTermQuery) q).getRewriteMethod());
325 q = qp.parse("[a TO z]", "field");
326 assertTrue(q instanceof TermRangeQuery);
327 assertEquals(MultiTermQuery.CONSTANT_SCORE_AUTO_REWRITE_DEFAULT, ((MultiTermQuery) q).getRewriteMethod());
330 public void testCJK() throws Exception {
331 // Test Ideographic Space - As wide as a CJK character cell (fullwidth)
332 // used google to translate the word "term" to japanese -> ??
333 assertQueryEquals("term\u3000term\u3000term", null,
334 "term\u0020term\u0020term");
335 assertQueryEqualsAllowLeadingWildcard("??\u3000??\u3000??", null, "??\u0020??\u0020??");
338 public void testCJKTerm() throws Exception {
339 // individual CJK chars as terms
340 StandardAnalyzer analyzer = new StandardAnalyzer(TEST_VERSION_CURRENT);
342 BooleanQuery expected = new BooleanQuery();
343 expected.add(new TermQuery(new Term("field", "ä¸")), BooleanClause.Occur.SHOULD);
344 expected.add(new TermQuery(new Term("field", "国")), BooleanClause.Occur.SHOULD);
346 assertEquals(expected, getQuery("ä¸å›½", analyzer));
349 public void testCJKBoostedTerm() throws Exception {
350 // individual CJK chars as terms
351 StandardAnalyzer analyzer = new StandardAnalyzer(TEST_VERSION_CURRENT);
353 BooleanQuery expected = new BooleanQuery();
354 expected.setBoost(0.5f);
355 expected.add(new TermQuery(new Term("field", "ä¸")), BooleanClause.Occur.SHOULD);
356 expected.add(new TermQuery(new Term("field", "国")), BooleanClause.Occur.SHOULD);
358 assertEquals(expected, getQuery("ä¸å›½^0.5", analyzer));
361 public void testCJKPhrase() throws Exception {
362 // individual CJK chars as terms
363 StandardAnalyzer analyzer = new StandardAnalyzer(TEST_VERSION_CURRENT);
365 PhraseQuery expected = new PhraseQuery();
366 expected.add(new Term("field", "ä¸"));
367 expected.add(new Term("field", "国"));
369 assertEquals(expected, getQuery("\"ä¸å›½\"", analyzer));
372 public void testCJKBoostedPhrase() throws Exception {
373 // individual CJK chars as terms
374 StandardAnalyzer analyzer = new StandardAnalyzer(TEST_VERSION_CURRENT);
376 PhraseQuery expected = new PhraseQuery();
377 expected.setBoost(0.5f);
378 expected.add(new Term("field", "ä¸"));
379 expected.add(new Term("field", "国"));
381 assertEquals(expected, getQuery("\"ä¸å›½\"^0.5", analyzer));
384 public void testCJKSloppyPhrase() throws Exception {
385 // individual CJK chars as terms
386 StandardAnalyzer analyzer = new StandardAnalyzer(TEST_VERSION_CURRENT);
388 PhraseQuery expected = new PhraseQuery();
390 expected.add(new Term("field", "ä¸"));
391 expected.add(new Term("field", "国"));
393 assertEquals(expected, getQuery("\"ä¸å›½\"~3", analyzer));
396 public void testSimple() throws Exception {
397 assertQueryEquals("\"term germ\"~2", null, "\"term germ\"~2");
398 assertQueryEquals("term term term", null, "term term term");
399 assertQueryEquals("t�rm term term", new MockAnalyzer(random, MockTokenizer.WHITESPACE, false),
401 assertQueryEquals("�mlaut", new MockAnalyzer(random, MockTokenizer.WHITESPACE, false), "�mlaut");
403 assertQueryEquals("\"\"", new KeywordAnalyzer(), "");
404 assertQueryEquals("foo:\"\"", new KeywordAnalyzer(), "foo:");
406 assertQueryEquals("a AND b", null, "+a +b");
407 assertQueryEquals("(a AND b)", null, "+a +b");
408 assertQueryEquals("c OR (a AND b)", null, "c (+a +b)");
410 assertQueryEquals("a AND NOT b", null, "+a -b");
412 assertQueryEquals("a AND -b", null, "+a -b");
414 assertQueryEquals("a AND !b", null, "+a -b");
416 assertQueryEquals("a && b", null, "+a +b");
418 assertQueryEquals("a && ! b", null, "+a -b");
420 assertQueryEquals("a OR b", null, "a b");
421 assertQueryEquals("a || b", null, "a b");
423 assertQueryEquals("a OR !b", null, "a -b");
425 assertQueryEquals("a OR ! b", null, "a -b");
427 assertQueryEquals("a OR -b", null, "a -b");
429 assertQueryEquals("+term -term term", null, "+term -term term");
430 assertQueryEquals("foo:term AND field:anotherTerm", null,
431 "+foo:term +anotherterm");
432 assertQueryEquals("term AND \"phrase phrase\"", null,
433 "+term +\"phrase phrase\"");
434 assertQueryEquals("\"hello there\"", null, "\"hello there\"");
435 assertTrue(getQuery("a AND b", null) instanceof BooleanQuery);
436 assertTrue(getQuery("hello", null) instanceof TermQuery);
437 assertTrue(getQuery("\"hello there\"", null) instanceof PhraseQuery);
439 assertQueryEquals("germ term^2.0", null, "germ term^2.0");
440 assertQueryEquals("(term)^2.0", null, "term^2.0");
441 assertQueryEquals("(germ term)^2.0", null, "(germ term)^2.0");
442 assertQueryEquals("term^2.0", null, "term^2.0");
443 assertQueryEquals("term^2", null, "term^2.0");
444 assertQueryEquals("\"germ term\"^2.0", null, "\"germ term\"^2.0");
445 assertQueryEquals("\"term germ\"^2", null, "\"term germ\"^2.0");
447 assertQueryEquals("(foo OR bar) AND (baz OR boo)", null,
448 "+(foo bar) +(baz boo)");
449 assertQueryEquals("((a OR b) AND NOT c) OR d", null, "(+(a b) -c) d");
450 assertQueryEquals("+(apple \"steve jobs\") -(foo bar baz)", null,
451 "+(apple \"steve jobs\") -(foo bar baz)");
452 assertQueryEquals("+title:(dog OR cat) -author:\"bob dole\"", null,
453 "+(title:dog title:cat) -author:\"bob dole\"");
457 public void testPunct() throws Exception {
458 Analyzer a = new MockAnalyzer(random, MockTokenizer.WHITESPACE, false);
459 assertQueryEquals("a&b", a, "a&b");
460 assertQueryEquals("a&&b", a, "a&&b");
461 assertQueryEquals(".NET", a, ".NET");
464 public void testSlop() throws Exception {
466 assertQueryEquals("\"term germ\"~2", null, "\"term germ\"~2");
467 assertQueryEquals("\"term germ\"~2 flork", null, "\"term germ\"~2 flork");
468 assertQueryEquals("\"term\"~2", null, "term");
469 assertQueryEquals("\" \"~2 germ", null, "germ");
470 assertQueryEquals("\"term germ\"~2^2", null, "\"term germ\"~2^2.0");
473 public void testNumber() throws Exception {
474 // The numbers go away because SimpleAnalzyer ignores them
475 assertQueryEquals("3", null, "");
476 assertQueryEquals("term 1.0 1 2", null, "term");
477 assertQueryEquals("term term1 term2", null, "term term term");
479 Analyzer a = new MockAnalyzer(random, MockTokenizer.WHITESPACE, false);
480 assertQueryEquals("3", a, "3");
481 assertQueryEquals("term 1.0 1 2", a, "term 1.0 1 2");
482 assertQueryEquals("term term1 term2", a, "term term1 term2");
485 public void testWildcard() throws Exception {
486 assertQueryEquals("term*", null, "term*");
487 assertQueryEquals("term*^2", null, "term*^2.0");
488 assertQueryEquals("term~", null, "term~0.5");
489 assertQueryEquals("term~0.7", null, "term~0.7");
491 assertQueryEquals("term~^2", null, "term~0.5^2.0");
493 assertQueryEquals("term^2~", null, "term~0.5^2.0");
494 assertQueryEquals("term*germ", null, "term*germ");
495 assertQueryEquals("term*germ^3", null, "term*germ^3.0");
497 assertTrue(getQuery("term*", null) instanceof PrefixQuery);
498 assertTrue(getQuery("term*^2", null) instanceof PrefixQuery);
499 assertTrue(getQuery("term~", null) instanceof FuzzyQuery);
500 assertTrue(getQuery("term~0.7", null) instanceof FuzzyQuery);
501 FuzzyQuery fq = (FuzzyQuery) getQuery("term~0.7", null);
502 assertEquals(0.7f, fq.getMinSimilarity(), 0.1f);
503 assertEquals(FuzzyQuery.defaultPrefixLength, fq.getPrefixLength());
504 fq = (FuzzyQuery) getQuery("term~", null);
505 assertEquals(0.5f, fq.getMinSimilarity(), 0.1f);
506 assertEquals(FuzzyQuery.defaultPrefixLength, fq.getPrefixLength());
508 assertQueryNodeException("term~1.1"); // value > 1, throws exception
510 assertTrue(getQuery("term*germ", null) instanceof WildcardQuery);
513 * Tests to see that wild card terms are (or are not) properly lower-cased
514 * with propery parser configuration
516 // First prefix queries:
517 // by default, convert to lowercase:
518 assertWildcardQueryEquals("Term*", true, "term*");
519 // explicitly set lowercase:
520 assertWildcardQueryEquals("term*", true, "term*");
521 assertWildcardQueryEquals("Term*", true, "term*");
522 assertWildcardQueryEquals("TERM*", true, "term*");
523 // explicitly disable lowercase conversion:
524 assertWildcardQueryEquals("term*", false, "term*");
525 assertWildcardQueryEquals("Term*", false, "Term*");
526 assertWildcardQueryEquals("TERM*", false, "TERM*");
527 // Then 'full' wildcard queries:
528 // by default, convert to lowercase:
529 assertWildcardQueryEquals("Te?m", "te?m");
530 // explicitly set lowercase:
531 assertWildcardQueryEquals("te?m", true, "te?m");
532 assertWildcardQueryEquals("Te?m", true, "te?m");
533 assertWildcardQueryEquals("TE?M", true, "te?m");
534 assertWildcardQueryEquals("Te?m*gerM", true, "te?m*germ");
535 // explicitly disable lowercase conversion:
536 assertWildcardQueryEquals("te?m", false, "te?m");
537 assertWildcardQueryEquals("Te?m", false, "Te?m");
538 assertWildcardQueryEquals("TE?M", false, "TE?M");
539 assertWildcardQueryEquals("Te?m*gerM", false, "Te?m*gerM");
541 assertWildcardQueryEquals("Term~", "term~0.5");
542 assertWildcardQueryEquals("Term~", true, "term~0.5");
543 assertWildcardQueryEquals("Term~", false, "Term~0.5");
546 // TODO: implement this on QueryParser
547 // Q0002E_INVALID_SYNTAX_CANNOT_PARSE: Syntax Error, cannot parse '[A TO
548 // C]': Lexical error at line 1, column 1. Encountered: "[" (91), after
550 assertWildcardQueryEquals("[A TO C]", "[a TO c]");
551 assertWildcardQueryEquals("[A TO C]", true, "[a TO c]");
552 assertWildcardQueryEquals("[A TO C]", false, "[A TO C]");
553 // Test suffix queries: first disallow
555 assertWildcardQueryEquals("*Term", true, "*term");
557 } catch (QueryNodeException pe) {
558 // expected exception
561 assertWildcardQueryEquals("?Term", true, "?term");
563 } catch (QueryNodeException pe) {
564 // expected exception
566 // Test suffix queries: then allow
567 assertWildcardQueryEquals("*Term", true, "*term", true);
568 assertWildcardQueryEquals("?Term", true, "?term", true);
571 public void testLeadingWildcardType() throws Exception {
572 StandardQueryParser qp = getParser(null);
573 qp.setAllowLeadingWildcard(true);
574 assertEquals(WildcardQuery.class, qp.parse("t*erm*", "field").getClass());
575 assertEquals(WildcardQuery.class, qp.parse("?term*", "field").getClass());
576 assertEquals(WildcardQuery.class, qp.parse("*term*", "field").getClass());
579 public void testQPA() throws Exception {
580 assertQueryEquals("term term^3.0 term", qpAnalyzer, "term term^3.0 term");
581 assertQueryEquals("term stop^3.0 term", qpAnalyzer, "term term");
583 assertQueryEquals("term term term", qpAnalyzer, "term term term");
584 assertQueryEquals("term +stop term", qpAnalyzer, "term term");
585 assertQueryEquals("term -stop term", qpAnalyzer, "term term");
587 assertQueryEquals("drop AND (stop) AND roll", qpAnalyzer, "+drop +roll");
588 assertQueryEquals("term +(stop) term", qpAnalyzer, "term term");
589 assertQueryEquals("term -(stop) term", qpAnalyzer, "term term");
591 assertQueryEquals("drop AND stop AND roll", qpAnalyzer, "+drop +roll");
592 assertQueryEquals("term phrase term", qpAnalyzer,
593 "term phrase1 phrase2 term");
595 assertQueryEquals("term AND NOT phrase term", qpAnalyzer,
596 "+term -(phrase1 phrase2) term");
598 assertQueryEquals("stop^3", qpAnalyzer, "");
599 assertQueryEquals("stop", qpAnalyzer, "");
600 assertQueryEquals("(stop)^3", qpAnalyzer, "");
601 assertQueryEquals("((stop))^3", qpAnalyzer, "");
602 assertQueryEquals("(stop^3)", qpAnalyzer, "");
603 assertQueryEquals("((stop)^3)", qpAnalyzer, "");
604 assertQueryEquals("(stop)", qpAnalyzer, "");
605 assertQueryEquals("((stop))", qpAnalyzer, "");
606 assertTrue(getQuery("term term term", qpAnalyzer) instanceof BooleanQuery);
607 assertTrue(getQuery("term +stop", qpAnalyzer) instanceof TermQuery);
610 public void testRange() throws Exception {
611 assertQueryEquals("[ a TO z]", null, "[a TO z]");
612 assertEquals(MultiTermQuery.CONSTANT_SCORE_AUTO_REWRITE_DEFAULT, ((TermRangeQuery)getQuery("[ a TO z]", null)).getRewriteMethod());
614 StandardQueryParser qp = new StandardQueryParser();
616 qp.setMultiTermRewriteMethod(MultiTermQuery.SCORING_BOOLEAN_QUERY_REWRITE);
617 assertEquals(MultiTermQuery.SCORING_BOOLEAN_QUERY_REWRITE,((TermRangeQuery)qp.parse("[ a TO z]", "field")).getRewriteMethod());
619 assertQueryEquals("[ a TO z ]", null, "[a TO z]");
620 assertQueryEquals("{ a TO z}", null, "{a TO z}");
621 assertQueryEquals("{ a TO z }", null, "{a TO z}");
622 assertQueryEquals("{ a TO z }^2.0", null, "{a TO z}^2.0");
623 assertQueryEquals("[ a TO z] OR bar", null, "[a TO z] bar");
624 assertQueryEquals("[ a TO z] AND bar", null, "+[a TO z] +bar");
625 assertQueryEquals("( bar blar { a TO z}) ", null, "bar blar {a TO z}");
626 assertQueryEquals("gack ( bar blar { a TO z}) ", null,
627 "gack (bar blar {a TO z})");
630 public void testFarsiRangeCollating() throws Exception {
631 Directory ramDir = newDirectory();
632 IndexWriter iw = new IndexWriter(ramDir, newIndexWriterConfig(TEST_VERSION_CURRENT, new WhitespaceAnalyzer(TEST_VERSION_CURRENT)));
633 Document doc = new Document();
634 doc.add(newField("content", "\u0633\u0627\u0628", Field.Store.YES,
635 Field.Index.NOT_ANALYZED));
638 IndexSearcher is = new IndexSearcher(ramDir, true);
640 StandardQueryParser qp = new StandardQueryParser();
641 qp.setAnalyzer(new WhitespaceAnalyzer(TEST_VERSION_CURRENT));
643 // Neither Java 1.4.2 nor 1.5.0 has Farsi Locale collation available in
644 // RuleBasedCollator. However, the Arabic Locale seems to order the
646 // characters properly.
647 Collator c = Collator.getInstance(new Locale("ar"));
648 qp.setRangeCollator(c);
650 // Unicode order would include U+0633 in [ U+062F - U+0698 ], but Farsi
651 // orders the U+0698 character before the U+0633 character, so the
653 // index Term below should NOT be returned by a ConstantScoreRangeQuery
654 // with a Farsi Collator (or an Arabic one for the case when Farsi is
658 // Test ConstantScoreRangeQuery
659 qp.setMultiTermRewriteMethod(MultiTermQuery.CONSTANT_SCORE_FILTER_REWRITE);
660 ScoreDoc[] result = is.search(qp.parse("[ \u062F TO \u0698 ]", "content"),
661 null, 1000).scoreDocs;
662 assertEquals("The index Term should not be included.", 0, result.length);
664 result = is.search(qp.parse("[ \u0633 TO \u0638 ]", "content"), null, 1000).scoreDocs;
665 assertEquals("The index Term should be included.", 1, result.length);
668 qp.setMultiTermRewriteMethod(MultiTermQuery.SCORING_BOOLEAN_QUERY_REWRITE);
669 result = is.search(qp.parse("[ \u062F TO \u0698 ]", "content"), null, 1000).scoreDocs;
670 assertEquals("The index Term should not be included.", 0, result.length);
672 result = is.search(qp.parse("[ \u0633 TO \u0638 ]", "content"), null, 1000).scoreDocs;
673 assertEquals("The index Term should be included.", 1, result.length);
679 /** for testing legacy DateField support */
680 private String getLegacyDate(String s) throws Exception {
681 DateFormat df = DateFormat.getDateInstance(DateFormat.SHORT);
682 return DateField.dateToString(df.parse(s));
685 /** for testing DateTools support */
686 private String getDate(String s, DateTools.Resolution resolution)
688 DateFormat df = DateFormat.getDateInstance(DateFormat.SHORT);
689 return getDate(df.parse(s), resolution);
692 /** for testing DateTools support */
693 private String getDate(Date d, DateTools.Resolution resolution)
695 if (resolution == null) {
696 return DateField.dateToString(d);
698 return DateTools.dateToString(d, resolution);
702 private String escapeDateString(String s) {
703 if (s.contains(" ")) {
704 return "\"" + s + "\"";
710 private String getLocalizedDate(int year, int month, int day) {
711 DateFormat df = DateFormat.getDateInstance(DateFormat.SHORT);
712 Calendar calendar = new GregorianCalendar();
714 calendar.set(year, month, day);
715 calendar.set(Calendar.HOUR_OF_DAY, 23);
716 calendar.set(Calendar.MINUTE, 59);
717 calendar.set(Calendar.SECOND, 59);
718 calendar.set(Calendar.MILLISECOND, 999);
719 return df.format(calendar.getTime());
722 /** for testing legacy DateField support */
723 public void testLegacyDateRange() throws Exception {
724 String startDate = getLocalizedDate(2002, 1, 1);
725 String endDate = getLocalizedDate(2002, 1, 4);
726 Calendar endDateExpected = new GregorianCalendar();
727 endDateExpected.clear();
728 endDateExpected.set(2002, 1, 4, 23, 59, 59);
729 endDateExpected.set(Calendar.MILLISECOND, 999);
730 assertQueryEquals("[ " + escapeDateString(startDate) + " TO " + escapeDateString(endDate) + "]", null, "["
731 + getLegacyDate(startDate) + " TO "
732 + DateField.dateToString(endDateExpected.getTime()) + "]");
733 assertQueryEquals("{ " + escapeDateString(startDate) + " " + escapeDateString(endDate) + " }", null, "{"
734 + getLegacyDate(startDate) + " TO " + getLegacyDate(endDate) + "}");
737 public void testDateRange() throws Exception {
738 String startDate = getLocalizedDate(2002, 1, 1);
739 String endDate = getLocalizedDate(2002, 1, 4);
740 Calendar endDateExpected = new GregorianCalendar();
741 endDateExpected.clear();
742 endDateExpected.set(2002, 1, 4, 23, 59, 59);
743 endDateExpected.set(Calendar.MILLISECOND, 999);
744 final String defaultField = "default";
745 final String monthField = "month";
746 final String hourField = "hour";
747 StandardQueryParser qp = new StandardQueryParser();
749 // Don't set any date resolution and verify if DateField is used
750 assertDateRangeQueryEquals(qp, defaultField, startDate, endDate,
751 endDateExpected.getTime(), null);
753 Map<CharSequence, DateTools.Resolution> dateRes = new HashMap<CharSequence, DateTools.Resolution>();
755 // set a field specific date resolution
756 dateRes.put(monthField, DateTools.Resolution.MONTH);
757 qp.setDateResolution(dateRes);
759 // DateField should still be used for defaultField
760 assertDateRangeQueryEquals(qp, defaultField, startDate, endDate,
761 endDateExpected.getTime(), null);
763 // set default date resolution to MILLISECOND
764 qp.setDateResolution(DateTools.Resolution.MILLISECOND);
766 // set second field specific date resolution
767 dateRes.put(hourField, DateTools.Resolution.HOUR);
768 qp.setDateResolution(dateRes);
770 // for this field no field specific date resolution has been set,
771 // so verify if the default resolution is used
772 assertDateRangeQueryEquals(qp, defaultField, startDate, endDate,
773 endDateExpected.getTime(), DateTools.Resolution.MILLISECOND);
775 // verify if field specific date resolutions are used for these two
777 assertDateRangeQueryEquals(qp, monthField, startDate, endDate,
778 endDateExpected.getTime(), DateTools.Resolution.MONTH);
780 assertDateRangeQueryEquals(qp, hourField, startDate, endDate,
781 endDateExpected.getTime(), DateTools.Resolution.HOUR);
784 public void assertDateRangeQueryEquals(StandardQueryParser qp,
785 String field, String startDate, String endDate, Date endDateInclusive,
786 DateTools.Resolution resolution) throws Exception {
787 assertQueryEquals(qp, field, field + ":[" + escapeDateString(startDate) + " TO " + escapeDateString(endDate)
788 + "]", "[" + getDate(startDate, resolution) + " TO "
789 + getDate(endDateInclusive, resolution) + "]");
790 assertQueryEquals(qp, field, field + ":{" + escapeDateString(startDate) + " TO " + escapeDateString(endDate)
791 + "}", "{" + getDate(startDate, resolution) + " TO "
792 + getDate(endDate, resolution) + "}");
795 public void testEscaped() throws Exception {
796 Analyzer a = new MockAnalyzer(random, MockTokenizer.WHITESPACE, false);
799 * assertQueryEquals("\\[brackets", a, "\\[brackets");
800 * assertQueryEquals("\\[brackets", null, "brackets");
801 * assertQueryEquals("\\\\", a, "\\\\"); assertQueryEquals("\\+blah", a,
802 * "\\+blah"); assertQueryEquals("\\(blah", a, "\\(blah");
804 * assertQueryEquals("\\-blah", a, "\\-blah"); assertQueryEquals("\\!blah",
805 * a, "\\!blah"); assertQueryEquals("\\{blah", a, "\\{blah");
806 * assertQueryEquals("\\}blah", a, "\\}blah"); assertQueryEquals("\\:blah",
807 * a, "\\:blah"); assertQueryEquals("\\^blah", a, "\\^blah");
808 * assertQueryEquals("\\[blah", a, "\\[blah"); assertQueryEquals("\\]blah",
809 * a, "\\]blah"); assertQueryEquals("\\\"blah", a, "\\\"blah");
810 * assertQueryEquals("\\(blah", a, "\\(blah"); assertQueryEquals("\\)blah",
811 * a, "\\)blah"); assertQueryEquals("\\~blah", a, "\\~blah");
812 * assertQueryEquals("\\*blah", a, "\\*blah"); assertQueryEquals("\\?blah",
813 * a, "\\?blah"); //assertQueryEquals("foo \\&\\& bar", a,
814 * "foo \\&\\& bar"); //assertQueryEquals("foo \\|| bar", a,
815 * "foo \\|| bar"); //assertQueryEquals("foo \\AND bar", a,
819 assertQueryEquals("\\*", a, "*");
821 assertQueryEquals("\\a", a, "a");
823 assertQueryEquals("a\\-b:c", a, "a-b:c");
824 assertQueryEquals("a\\+b:c", a, "a+b:c");
825 assertQueryEquals("a\\:b:c", a, "a:b:c");
826 assertQueryEquals("a\\\\b:c", a, "a\\b:c");
828 assertQueryEquals("a:b\\-c", a, "a:b-c");
829 assertQueryEquals("a:b\\+c", a, "a:b+c");
830 assertQueryEquals("a:b\\:c", a, "a:b:c");
831 assertQueryEquals("a:b\\\\c", a, "a:b\\c");
833 assertQueryEquals("a:b\\-c*", a, "a:b-c*");
834 assertQueryEquals("a:b\\+c*", a, "a:b+c*");
835 assertQueryEquals("a:b\\:c*", a, "a:b:c*");
837 assertQueryEquals("a:b\\\\c*", a, "a:b\\c*");
839 assertQueryEquals("a:b\\-?c", a, "a:b-?c");
840 assertQueryEquals("a:b\\+?c", a, "a:b+?c");
841 assertQueryEquals("a:b\\:?c", a, "a:b:?c");
843 assertQueryEquals("a:b\\\\?c", a, "a:b\\?c");
845 assertQueryEquals("a:b\\-c~", a, "a:b-c~0.5");
846 assertQueryEquals("a:b\\+c~", a, "a:b+c~0.5");
847 assertQueryEquals("a:b\\:c~", a, "a:b:c~0.5");
848 assertQueryEquals("a:b\\\\c~", a, "a:b\\c~0.5");
850 // TODO: implement Range queries on QueryParser
851 assertQueryEquals("[ a\\- TO a\\+ ]", null, "[a- TO a+]");
852 assertQueryEquals("[ a\\: TO a\\~ ]", null, "[a: TO a~]");
853 assertQueryEquals("[ a\\\\ TO a\\* ]", null, "[a\\ TO a*]");
856 "[\"c\\:\\\\temp\\\\\\~foo0.txt\" TO \"c\\:\\\\temp\\\\\\~foo9.txt\"]",
857 a, "[c:\\temp\\~foo0.txt TO c:\\temp\\~foo9.txt]");
859 assertQueryEquals("a\\\\\\+b", a, "a\\+b");
861 assertQueryEquals("a \\\"b c\\\" d", a, "a \"b c\" d");
862 assertQueryEquals("\"a \\\"b c\\\" d\"", a, "\"a \"b c\" d\"");
863 assertQueryEquals("\"a \\+b c d\"", a, "\"a +b c d\"");
865 assertQueryEquals("c\\:\\\\temp\\\\\\~foo.txt", a, "c:\\temp\\~foo.txt");
867 assertQueryNodeException("XY\\"); // there must be a character after the
870 // test unicode escaping
871 assertQueryEquals("a\\u0062c", a, "abc");
872 assertQueryEquals("XY\\u005a", a, "XYZ");
873 assertQueryEquals("XY\\u005A", a, "XYZ");
874 assertQueryEquals("\"a \\\\\\u0028\\u0062\\\" c\"", a, "\"a \\(b\" c\"");
876 assertQueryNodeException("XY\\u005G"); // test non-hex character in escaped
878 assertQueryNodeException("XY\\u005"); // test incomplete escaped unicode
881 // Tests bug LUCENE-800
882 assertQueryEquals("(item:\\\\ item:ABCD\\\\)", a, "item:\\ item:ABCD\\");
883 assertQueryNodeException("(item:\\\\ item:ABCD\\\\))"); // unmatched closing
885 assertQueryEquals("\\*", a, "*");
886 assertQueryEquals("\\\\", a, "\\"); // escaped backslash
888 assertQueryNodeException("\\"); // a backslash must always be escaped
891 assertQueryEquals("(\"a\\\\\") or (\"b\")", a, "a\\ or b");
894 public void testQueryStringEscaping() throws Exception {
895 Analyzer a = new MockAnalyzer(random, MockTokenizer.WHITESPACE, false);
897 assertEscapedQueryEquals("a-b:c", a, "a\\-b\\:c");
898 assertEscapedQueryEquals("a+b:c", a, "a\\+b\\:c");
899 assertEscapedQueryEquals("a:b:c", a, "a\\:b\\:c");
900 assertEscapedQueryEquals("a\\b:c", a, "a\\\\b\\:c");
902 assertEscapedQueryEquals("a:b-c", a, "a\\:b\\-c");
903 assertEscapedQueryEquals("a:b+c", a, "a\\:b\\+c");
904 assertEscapedQueryEquals("a:b:c", a, "a\\:b\\:c");
905 assertEscapedQueryEquals("a:b\\c", a, "a\\:b\\\\c");
907 assertEscapedQueryEquals("a:b-c*", a, "a\\:b\\-c\\*");
908 assertEscapedQueryEquals("a:b+c*", a, "a\\:b\\+c\\*");
909 assertEscapedQueryEquals("a:b:c*", a, "a\\:b\\:c\\*");
911 assertEscapedQueryEquals("a:b\\\\c*", a, "a\\:b\\\\\\\\c\\*");
913 assertEscapedQueryEquals("a:b-?c", a, "a\\:b\\-\\?c");
914 assertEscapedQueryEquals("a:b+?c", a, "a\\:b\\+\\?c");
915 assertEscapedQueryEquals("a:b:?c", a, "a\\:b\\:\\?c");
917 assertEscapedQueryEquals("a:b?c", a, "a\\:b\\?c");
919 assertEscapedQueryEquals("a:b-c~", a, "a\\:b\\-c\\~");
920 assertEscapedQueryEquals("a:b+c~", a, "a\\:b\\+c\\~");
921 assertEscapedQueryEquals("a:b:c~", a, "a\\:b\\:c\\~");
922 assertEscapedQueryEquals("a:b\\c~", a, "a\\:b\\\\c\\~");
924 assertEscapedQueryEquals("[ a - TO a+ ]", null, "\\[ a \\- TO a\\+ \\]");
925 assertEscapedQueryEquals("[ a : TO a~ ]", null, "\\[ a \\: TO a\\~ \\]");
926 assertEscapedQueryEquals("[ a\\ TO a* ]", null, "\\[ a\\\\ TO a\\* \\]");
929 assertEscapedQueryEquals("|| abc ||", a, "\\|\\| abc \\|\\|");
930 assertEscapedQueryEquals("&& abc &&", a, "\\&\\& abc \\&\\&");
933 public void testTabNewlineCarriageReturn() throws Exception {
934 assertQueryEqualsDOA("+weltbank +worlbank", null, "+weltbank +worlbank");
936 assertQueryEqualsDOA("+weltbank\n+worlbank", null, "+weltbank +worlbank");
937 assertQueryEqualsDOA("weltbank \n+worlbank", null, "+weltbank +worlbank");
938 assertQueryEqualsDOA("weltbank \n +worlbank", null, "+weltbank +worlbank");
940 assertQueryEqualsDOA("+weltbank\r+worlbank", null, "+weltbank +worlbank");
941 assertQueryEqualsDOA("weltbank \r+worlbank", null, "+weltbank +worlbank");
942 assertQueryEqualsDOA("weltbank \r +worlbank", null, "+weltbank +worlbank");
944 assertQueryEqualsDOA("+weltbank\r\n+worlbank", null, "+weltbank +worlbank");
945 assertQueryEqualsDOA("weltbank \r\n+worlbank", null, "+weltbank +worlbank");
946 assertQueryEqualsDOA("weltbank \r\n +worlbank", null, "+weltbank +worlbank");
947 assertQueryEqualsDOA("weltbank \r \n +worlbank", null,
948 "+weltbank +worlbank");
950 assertQueryEqualsDOA("+weltbank\t+worlbank", null, "+weltbank +worlbank");
951 assertQueryEqualsDOA("weltbank \t+worlbank", null, "+weltbank +worlbank");
952 assertQueryEqualsDOA("weltbank \t +worlbank", null, "+weltbank +worlbank");
955 public void testSimpleDAO() throws Exception {
956 assertQueryEqualsDOA("term term term", null, "+term +term +term");
957 assertQueryEqualsDOA("term +term term", null, "+term +term +term");
958 assertQueryEqualsDOA("term term +term", null, "+term +term +term");
959 assertQueryEqualsDOA("term +term +term", null, "+term +term +term");
960 assertQueryEqualsDOA("-term term term", null, "-term +term +term");
963 public void testBoost() throws Exception {
964 StandardAnalyzer oneStopAnalyzer = new StandardAnalyzer(TEST_VERSION_CURRENT, Collections.singleton("on"));
965 StandardQueryParser qp = new StandardQueryParser();
966 qp.setAnalyzer(oneStopAnalyzer);
968 Query q = qp.parse("on^1.0", "field");
970 q = qp.parse("\"hello\"^2.0", "field");
972 assertEquals(q.getBoost(), (float) 2.0, (float) 0.5);
973 q = qp.parse("hello^2.0", "field");
975 assertEquals(q.getBoost(), (float) 2.0, (float) 0.5);
976 q = qp.parse("\"on\"^1.0", "field");
979 StandardQueryParser qp2 = new StandardQueryParser();
980 qp2.setAnalyzer(new StandardAnalyzer(TEST_VERSION_CURRENT));
982 q = qp2.parse("the^3", "field");
983 // "the" is a stop word so the result is an empty query:
985 assertEquals("", q.toString());
986 assertEquals(1.0f, q.getBoost(), 0.01f);
989 public void assertQueryNodeException(String queryString) throws Exception {
991 getQuery(queryString, null);
992 } catch (QueryNodeException expected) {
995 fail("ParseException expected, not thrown");
998 public void testException() throws Exception {
999 assertQueryNodeException("*leadingWildcard"); // disallowed by default
1000 assertQueryNodeException("\"some phrase");
1001 assertQueryNodeException("(foo bar");
1002 assertQueryNodeException("foo bar))");
1003 assertQueryNodeException("field:term:with:colon some more terms");
1004 assertQueryNodeException("(sub query)^5.0^2.0 plus more");
1005 assertQueryNodeException("secret AND illegal) AND access:confidential");
1008 public void testCustomQueryParserWildcard() {
1010 new QPTestParser(new MockAnalyzer(random, MockTokenizer.WHITESPACE, false)).parse("a?t", "contents");
1011 fail("Wildcard queries should not be allowed");
1012 } catch (QueryNodeException expected) {
1013 // expected exception
1017 public void testCustomQueryParserFuzzy() throws Exception {
1019 new QPTestParser(new MockAnalyzer(random, MockTokenizer.WHITESPACE, false)).parse("xunit~", "contents");
1020 fail("Fuzzy queries should not be allowed");
1021 } catch (QueryNodeException expected) {
1022 // expected exception
1026 public void testBooleanQuery() throws Exception {
1027 BooleanQuery.setMaxClauseCount(2);
1029 StandardQueryParser qp = new StandardQueryParser();
1030 qp.setAnalyzer(new MockAnalyzer(random, MockTokenizer.WHITESPACE, false));
1031 qp.parse("one two three", "field");
1032 fail("ParseException expected due to too many boolean clauses");
1033 } catch (QueryNodeException expected) {
1034 // too many boolean clauses, so ParseException is expected
1039 * This test differs from TestPrecedenceQueryParser
1041 public void testPrecedence() throws Exception {
1042 StandardQueryParser qp = new StandardQueryParser();
1043 qp.setAnalyzer(new MockAnalyzer(random, MockTokenizer.WHITESPACE, false));
1045 Query query1 = qp.parse("A AND B OR C AND D", "field");
1046 Query query2 = qp.parse("+A +B +C +D", "field");
1048 assertEquals(query1, query2);
1051 public void testLocalDateFormat() throws IOException, QueryNodeException {
1052 Directory ramDir = newDirectory();
1053 IndexWriter iw = new IndexWriter(ramDir, newIndexWriterConfig(TEST_VERSION_CURRENT, new WhitespaceAnalyzer(TEST_VERSION_CURRENT)));
1054 addDateDoc("a", 2005, 12, 2, 10, 15, 33, iw);
1055 addDateDoc("b", 2005, 12, 4, 22, 15, 00, iw);
1057 IndexSearcher is = new IndexSearcher(ramDir, true);
1058 assertHits(1, "[12/1/2005 TO 12/3/2005]", is);
1059 assertHits(2, "[12/1/2005 TO 12/4/2005]", is);
1060 assertHits(1, "[12/3/2005 TO 12/4/2005]", is);
1061 assertHits(1, "{12/1/2005 TO 12/3/2005}", is);
1062 assertHits(1, "{12/1/2005 TO 12/4/2005}", is);
1063 assertHits(0, "{12/3/2005 TO 12/4/2005}", is);
1068 public void testStarParsing() throws Exception {
1069 // final int[] type = new int[1];
1070 // StandardQueryParser qp = new StandardQueryParser("field", new
1071 // WhitespaceAnalyzer()) {
1072 // protected Query getWildcardQuery(String field, String termStr) throws
1074 // // override error checking of superclass
1076 // return new TermQuery(new Term(field,termStr));
1078 // protected Query getPrefixQuery(String field, String termStr) throws
1080 // // override error checking of superclass
1082 // return new TermQuery(new Term(field,termStr));
1085 // protected Query getFieldQuery(String field, String queryText) throws
1088 // return super.getFieldQuery(field, queryText);
1094 // tq = (TermQuery)qp.parse("foo:zoo*");
1095 // assertEquals("zoo",tq.getTerm().text());
1096 // assertEquals(2,type[0]);
1098 // tq = (TermQuery)qp.parse("foo:zoo*^2");
1099 // assertEquals("zoo",tq.getTerm().text());
1100 // assertEquals(2,type[0]);
1101 // assertEquals(tq.getBoost(),2,0);
1103 // tq = (TermQuery)qp.parse("foo:*");
1104 // assertEquals("*",tq.getTerm().text());
1105 // assertEquals(1,type[0]); // could be a valid prefix query in the
1108 // tq = (TermQuery)qp.parse("foo:*^2");
1109 // assertEquals("*",tq.getTerm().text());
1110 // assertEquals(1,type[0]);
1111 // assertEquals(tq.getBoost(),2,0);
1113 // tq = (TermQuery)qp.parse("*:foo");
1114 // assertEquals("*",tq.getTerm().field());
1115 // assertEquals("foo",tq.getTerm().text());
1116 // assertEquals(3,type[0]);
1118 // tq = (TermQuery)qp.parse("*:*");
1119 // assertEquals("*",tq.getTerm().field());
1120 // assertEquals("*",tq.getTerm().text());
1121 // assertEquals(1,type[0]); // could be handled as a prefix query in the
1124 // tq = (TermQuery)qp.parse("(*:*)");
1125 // assertEquals("*",tq.getTerm().field());
1126 // assertEquals("*",tq.getTerm().text());
1127 // assertEquals(1,type[0]);
1131 public void testStopwords() throws Exception {
1132 StandardQueryParser qp = new StandardQueryParser();
1134 new StopAnalyzer(TEST_VERSION_CURRENT, StopFilter.makeStopSet(TEST_VERSION_CURRENT, "the", "foo" )));
1136 Query result = qp.parse("a:the OR a:foo", "a");
1137 assertNotNull("result is null and it shouldn't be", result);
1138 assertTrue("result is not a BooleanQuery", result instanceof BooleanQuery);
1139 assertTrue(((BooleanQuery) result).clauses().size() + " does not equal: "
1140 + 0, ((BooleanQuery) result).clauses().size() == 0);
1141 result = qp.parse("a:woo OR a:the", "a");
1142 assertNotNull("result is null and it shouldn't be", result);
1143 assertTrue("result is not a TermQuery", result instanceof TermQuery);
1145 "(fieldX:xxxxx OR fieldy:xxxxxxxx)^2 AND (fieldx:the OR fieldy:foo)",
1147 assertNotNull("result is null and it shouldn't be", result);
1148 assertTrue("result is not a BooleanQuery", result instanceof BooleanQuery);
1150 System.out.println("Result: " + result);
1151 assertTrue(((BooleanQuery) result).clauses().size() + " does not equal: "
1152 + 2, ((BooleanQuery) result).clauses().size() == 2);
1155 public void testPositionIncrement() throws Exception {
1156 StandardQueryParser qp = new StandardQueryParser();
1158 new StopAnalyzer(TEST_VERSION_CURRENT, StopFilter.makeStopSet(TEST_VERSION_CURRENT, "the", "in", "are", "this" )));
1160 qp.setEnablePositionIncrements(true);
1162 String qtxt = "\"the words in poisitions pos02578 are stopped in this phrasequery\"";
1164 int expectedPositions[] = { 1, 3, 4, 6, 9 };
1165 PhraseQuery pq = (PhraseQuery) qp.parse(qtxt, "a");
1166 // System.out.println("Query text: "+qtxt);
1167 // System.out.println("Result: "+pq);
1168 Term t[] = pq.getTerms();
1169 int pos[] = pq.getPositions();
1170 for (int i = 0; i < t.length; i++) {
1171 // System.out.println(i+". "+t[i]+" pos: "+pos[i]);
1172 assertEquals("term " + i + " = " + t[i] + " has wrong term-position!",
1173 expectedPositions[i], pos[i]);
1177 public void testMatchAllDocs() throws Exception {
1178 StandardQueryParser qp = new StandardQueryParser();
1179 qp.setAnalyzer(new MockAnalyzer(random, MockTokenizer.WHITESPACE, false));
1181 assertEquals(new MatchAllDocsQuery(), qp.parse("*:*", "field"));
1182 assertEquals(new MatchAllDocsQuery(), qp.parse("(*:*)", "field"));
1183 BooleanQuery bq = (BooleanQuery) qp.parse("+*:* -*:*", "field");
1184 assertTrue(bq.getClauses()[0].getQuery() instanceof MatchAllDocsQuery);
1185 assertTrue(bq.getClauses()[1].getQuery() instanceof MatchAllDocsQuery);
1188 private void assertHits(int expected, String query, IndexSearcher is)
1189 throws IOException, QueryNodeException {
1190 StandardQueryParser qp = new StandardQueryParser();
1191 qp.setAnalyzer(new MockAnalyzer(random, MockTokenizer.WHITESPACE, false));
1192 qp.setLocale(Locale.ENGLISH);
1194 Query q = qp.parse(query, "date");
1195 ScoreDoc[] hits = is.search(q, null, 1000).scoreDocs;
1196 assertEquals(expected, hits.length);
1199 private void addDateDoc(String content, int year, int month, int day,
1200 int hour, int minute, int second, IndexWriter iw) throws IOException {
1201 Document d = new Document();
1202 d.add(newField("f", content, Field.Store.YES, Field.Index.ANALYZED));
1203 Calendar cal = Calendar.getInstance(Locale.ENGLISH);
1204 cal.set(year, month - 1, day, hour, minute, second);
1205 d.add(newField("date", DateField.dateToString(cal.getTime()),
1206 Field.Store.YES, Field.Index.NOT_ANALYZED));
1211 public void tearDown() throws Exception {
1212 BooleanQuery.setMaxClauseCount(originalMaxClauses);
1216 private class CannedTokenStream extends Tokenizer {
1217 private int upto = 0;
1218 private final PositionIncrementAttribute posIncr = addAttribute(PositionIncrementAttribute.class);
1219 private final CharTermAttribute term = addAttribute(CharTermAttribute.class);
1222 public boolean incrementToken() {
1228 posIncr.setPositionIncrement(1);
1229 term.setEmpty().append("a");
1230 } else if (upto == 1) {
1231 posIncr.setPositionIncrement(1);
1232 term.setEmpty().append("b");
1233 } else if (upto == 2) {
1234 posIncr.setPositionIncrement(0);
1235 term.setEmpty().append("c");
1237 posIncr.setPositionIncrement(0);
1238 term.setEmpty().append("d");
1245 public void reset() throws IOException {
1251 private class CannedAnalyzer extends Analyzer {
1253 public TokenStream tokenStream(String ignored, Reader alsoIgnored) {
1254 return new CannedTokenStream();
1258 public void testMultiPhraseQuery() throws Exception {
1259 Directory dir = newDirectory();
1260 IndexWriter w = new IndexWriter(dir, newIndexWriterConfig(TEST_VERSION_CURRENT, new CannedAnalyzer()));
1261 Document doc = new Document();
1262 doc.add(newField("field", "", Field.Store.NO, Field.Index.ANALYZED));
1264 IndexReader r = IndexReader.open(w, true);
1265 IndexSearcher s = newSearcher(r);
1267 Query q = new StandardQueryParser(new CannedAnalyzer()).parse("\"a\"", "field");
1268 assertTrue(q instanceof MultiPhraseQuery);
1269 assertEquals(1, s.search(q, 10).totalHits);