FileDocCategorySizeDatePackage
TestQueryParser.javaAPI DocApache Lucene 1.4.319035Fri Oct 01 11:24:12 BST 2004org.apache.lucene.queryParser

TestQueryParser

public class TestQueryParser extends TestCase
Tests QueryParser.

Fields Summary
public static Analyzer
qpAnalyzer
private int
originalMaxClauses
Constructors Summary
Methods Summary
public voidassertQueryEquals(java.lang.String query, org.apache.lucene.analysis.Analyzer a, java.lang.String result)

    Query q = getQuery(query, a);
    String s = q.toString("field");
    if (!s.equals(result)) {
      fail("Query /" + query + "/ yielded /" + s
           + "/, expecting /" + result + "/");
    }
  
public voidassertQueryEqualsDOA(java.lang.String query, org.apache.lucene.analysis.Analyzer a, java.lang.String result)

    Query q = getQueryDOA(query, a);
    String s = q.toString("field");
    if (!s.equals(result)) {
      fail("Query /" + query + "/ yielded /" + s
           + "/, expecting /" + result + "/");
    }
  
public voidassertWildcardQueryEquals(java.lang.String query, boolean lowercase, java.lang.String result)

    QueryParser qp = getParser(null);
    qp.setLowercaseWildcardTerms(lowercase);
    Query q = qp.parse(query);
    String s = q.toString("field");
    if (!s.equals(result)) {
      fail("WildcardQuery /" + query + "/ yielded /" + s
           + "/, expecting /" + result + "/");
    }
  
public java.lang.StringgetDate(java.lang.String s)

    DateFormat df = DateFormat.getDateInstance(DateFormat.SHORT);
    return DateField.dateToString(df.parse(s));
  
public java.lang.StringgetLocalizedDate(int year, int month, int day)

    DateFormat df = DateFormat.getDateInstance(DateFormat.SHORT);
    Calendar calendar = Calendar.getInstance();
    calendar.set(year, month, day);
    return df.format(calendar.getTime());
  
public org.apache.lucene.queryParser.QueryParsergetParser(org.apache.lucene.analysis.Analyzer a)

    if (a == null)
      a = new SimpleAnalyzer();
    QueryParser qp = new QueryParser("field", a);
    qp.setOperator(QueryParser.DEFAULT_OPERATOR_OR);
    return qp;
  
public org.apache.lucene.search.QuerygetQuery(java.lang.String query, org.apache.lucene.analysis.Analyzer a)

    return getParser(a).parse(query);
  
public org.apache.lucene.search.QuerygetQueryDOA(java.lang.String query, org.apache.lucene.analysis.Analyzer a)

    if (a == null)
      a = new SimpleAnalyzer();
    QueryParser qp = new QueryParser("field", a);
    qp.setOperator(QueryParser.DEFAULT_OPERATOR_AND);
    return qp.parse(query);
  
public voidsetUp()

    originalMaxClauses = BooleanQuery.getMaxClauseCount();
  
public voidtearDown()

    BooleanQuery.setMaxClauseCount(originalMaxClauses);
  
public voidtestBooleanQuery()

    BooleanQuery.setMaxClauseCount(2);
    try {
      QueryParser.parse("one two three", "field", new WhitespaceAnalyzer());
      fail("ParseException expected due to too many boolean clauses");
    } catch (ParseException expected) {
      // too many boolean clauses, so ParseException is expected
    }
  
public voidtestBoost()

    StandardAnalyzer oneStopAnalyzer = new StandardAnalyzer(new String[]{"on"});
    QueryParser qp = new QueryParser("field", oneStopAnalyzer);
    Query q = qp.parse("on^1.0");
    assertNotNull(q);
    q = qp.parse("\"hello\"^2.0");
    assertNotNull(q);
    assertEquals(q.getBoost(), (float) 2.0, (float) 0.5);
    q = qp.parse("hello^2.0");
    assertNotNull(q);
    assertEquals(q.getBoost(), (float) 2.0, (float) 0.5);
    q = qp.parse("\"on\"^1.0");
    assertNotNull(q);

    q = QueryParser.parse("the^3", "field", new StandardAnalyzer());
    assertNotNull(q);
  
public voidtestCustomQueryParserFuzzy()

    try {
      new QPTestParser("contents", new WhitespaceAnalyzer()).parse("xunit~");
    } catch (ParseException expected) {
      return;
    }
    fail("Fuzzy queries should not be allowed");
  
public voidtestCustomQueryParserWildcard()

    try {
      new QPTestParser("contents", new WhitespaceAnalyzer()).parse("a?t");
    } catch (ParseException expected) {
      return;
    }
    fail("Wildcard queries should not be allowed");
  
public voidtestDateRange()

    String startDate = getLocalizedDate(2002, 1, 1);
    String endDate = getLocalizedDate(2002, 1, 4);
    assertQueryEquals("[ " + startDate + " TO " + endDate + "]", null,
                      "[" + getDate(startDate) + " TO " + getDate(endDate) + "]");
    assertQueryEquals("{  " + startDate + "    " + endDate + "   }", null,
                      "{" + getDate(startDate) + " TO " + getDate(endDate) + "}");
  
public voidtestEscaped()

    Analyzer a = new WhitespaceAnalyzer();
    
    /*assertQueryEquals("\\[brackets", a, "\\[brackets");
    assertQueryEquals("\\[brackets", null, "brackets");
    assertQueryEquals("\\\\", a, "\\\\");
    assertQueryEquals("\\+blah", a, "\\+blah");
    assertQueryEquals("\\(blah", a, "\\(blah");

    assertQueryEquals("\\-blah", a, "\\-blah");
    assertQueryEquals("\\!blah", a, "\\!blah");
    assertQueryEquals("\\{blah", a, "\\{blah");
    assertQueryEquals("\\}blah", a, "\\}blah");
    assertQueryEquals("\\:blah", a, "\\:blah");
    assertQueryEquals("\\^blah", a, "\\^blah");
    assertQueryEquals("\\[blah", a, "\\[blah");
    assertQueryEquals("\\]blah", a, "\\]blah");
    assertQueryEquals("\\\"blah", a, "\\\"blah");
    assertQueryEquals("\\(blah", a, "\\(blah");
    assertQueryEquals("\\)blah", a, "\\)blah");
    assertQueryEquals("\\~blah", a, "\\~blah");
    assertQueryEquals("\\*blah", a, "\\*blah");
    assertQueryEquals("\\?blah", a, "\\?blah");
    //assertQueryEquals("foo \\&\\& bar", a, "foo \\&\\& bar");
    //assertQueryEquals("foo \\|| bar", a, "foo \\|| bar");
    //assertQueryEquals("foo \\AND bar", a, "foo \\AND bar");*/

    assertQueryEquals("a\\-b:c", a, "a-b:c");
    assertQueryEquals("a\\+b:c", a, "a+b:c");
    assertQueryEquals("a\\:b:c", a, "a:b:c");
    assertQueryEquals("a\\\\b:c", a, "a\\b:c");

    assertQueryEquals("a:b\\-c", a, "a:b-c");
    assertQueryEquals("a:b\\+c", a, "a:b+c");
    assertQueryEquals("a:b\\:c", a, "a:b:c");
    assertQueryEquals("a:b\\\\c", a, "a:b\\c");

    assertQueryEquals("a:b\\-c*", a, "a:b-c*");
    assertQueryEquals("a:b\\+c*", a, "a:b+c*");
    assertQueryEquals("a:b\\:c*", a, "a:b:c*");

    assertQueryEquals("a:b\\\\c*", a, "a:b\\c*");

    assertQueryEquals("a:b\\-?c", a, "a:b-?c");
    assertQueryEquals("a:b\\+?c", a, "a:b+?c");
    assertQueryEquals("a:b\\:?c", a, "a:b:?c");

    assertQueryEquals("a:b\\\\?c", a, "a:b\\?c");

    assertQueryEquals("a:b\\-c~", a, "a:b-c~0.5");
    assertQueryEquals("a:b\\+c~", a, "a:b+c~0.5");
    assertQueryEquals("a:b\\:c~", a, "a:b:c~0.5");
    assertQueryEquals("a:b\\\\c~", a, "a:b\\c~0.5");

    assertQueryEquals("[ a\\- TO a\\+ ]", null, "[a- TO a+]");
    assertQueryEquals("[ a\\: TO a\\~ ]", null, "[a: TO a~]");
    assertQueryEquals("[ a\\\\ TO a\\* ]", null, "[a\\ TO a*]");
  
public voidtestException()

    try {
      assertQueryEquals("\"some phrase", null, "abc");
      fail("ParseException expected, not thrown");
    } catch (ParseException expected) {
    }
  
public voidtestNumber()

// The numbers go away because SimpleAnalzyer ignores them
    assertQueryEquals("3", null, "");
    assertQueryEquals("term 1.0 1 2", null, "term");
    assertQueryEquals("term term1 term2", null, "term term term");

    Analyzer a = new StandardAnalyzer();
    assertQueryEquals("3", a, "3");
    assertQueryEquals("term 1.0 1 2", a, "term 1.0 1 2");
    assertQueryEquals("term term1 term2", a, "term term1 term2");
  
public voidtestPunct()

    Analyzer a = new WhitespaceAnalyzer();
    assertQueryEquals("a&b", a, "a&b");
    assertQueryEquals("a&&b", a, "a&&b");
    assertQueryEquals(".NET", a, ".NET");
  
public voidtestQPA()

    assertQueryEquals("term term term", qpAnalyzer, "term term term");
    assertQueryEquals("term +stop term", qpAnalyzer, "term term");
    assertQueryEquals("term -stop term", qpAnalyzer, "term term");
    assertQueryEquals("drop AND stop AND roll", qpAnalyzer, "+drop +roll");
    assertQueryEquals("term phrase term", qpAnalyzer,
                      "term \"phrase1 phrase2\" term");
    assertQueryEquals("term AND NOT phrase term", qpAnalyzer,
                      "+term -\"phrase1 phrase2\" term");
    assertQueryEquals("stop", qpAnalyzer, "");
    assertTrue(getQuery("term term term", qpAnalyzer) instanceof BooleanQuery);
    assertTrue(getQuery("term +stop", qpAnalyzer) instanceof TermQuery);
  
public voidtestRange()

    assertQueryEquals("[ a TO z]", null, "[a TO z]");
    assertTrue(getQuery("[ a TO z]", null) instanceof RangeQuery);
    assertQueryEquals("[ a TO z ]", null, "[a TO z]");
    assertQueryEquals("{ a TO z}", null, "{a TO z}");
    assertQueryEquals("{ a TO z }", null, "{a TO z}");
    assertQueryEquals("{ a TO z }^2.0", null, "{a TO z}^2.0");
    assertQueryEquals("[ a TO z] OR bar", null, "[a TO z] bar");
    assertQueryEquals("[ a TO z] AND bar", null, "+[a TO z] +bar");
    assertQueryEquals("( bar blar { a TO z}) ", null, "bar blar {a TO z}");
    assertQueryEquals("gack ( bar blar { a TO z}) ", null, "gack (bar blar {a TO z})");
  
public voidtestSimple()

    assertQueryEquals("term term term", null, "term term term");
    assertQueryEquals("türm term term", null, "türm term term");
    assertQueryEquals("ümlaut", null, "ümlaut");

    assertQueryEquals("a AND b", null, "+a +b");
    assertQueryEquals("(a AND b)", null, "+a +b");
    assertQueryEquals("c OR (a AND b)", null, "c (+a +b)");
    assertQueryEquals("a AND NOT b", null, "+a -b");
    assertQueryEquals("a AND -b", null, "+a -b");
    assertQueryEquals("a AND !b", null, "+a -b");
    assertQueryEquals("a && b", null, "+a +b");
    assertQueryEquals("a && ! b", null, "+a -b");

    assertQueryEquals("a OR b", null, "a b");
    assertQueryEquals("a || b", null, "a b");
    assertQueryEquals("a OR !b", null, "a -b");
    assertQueryEquals("a OR ! b", null, "a -b");
    assertQueryEquals("a OR -b", null, "a -b");

    assertQueryEquals("+term -term term", null, "+term -term term");
    assertQueryEquals("foo:term AND field:anotherTerm", null,
                      "+foo:term +anotherterm");
    assertQueryEquals("term AND \"phrase phrase\"", null,
                      "+term +\"phrase phrase\"");
    assertQueryEquals("\"hello there\"", null, "\"hello there\"");
    assertTrue(getQuery("a AND b", null) instanceof BooleanQuery);
    assertTrue(getQuery("hello", null) instanceof TermQuery);
    assertTrue(getQuery("\"hello there\"", null) instanceof PhraseQuery);

    assertQueryEquals("germ term^2.0", null, "germ term^2.0");
    assertQueryEquals("(term)^2.0", null, "term^2.0");
    assertQueryEquals("(germ term)^2.0", null, "(germ term)^2.0");
    assertQueryEquals("term^2.0", null, "term^2.0");
    assertQueryEquals("term^2", null, "term^2.0");
    assertQueryEquals("\"germ term\"^2.0", null, "\"germ term\"^2.0");
    assertQueryEquals("\"term germ\"^2", null, "\"term germ\"^2.0");

    assertQueryEquals("(foo OR bar) AND (baz OR boo)", null,
                      "+(foo bar) +(baz boo)");
    assertQueryEquals("((a OR b) AND NOT c) OR d", null,
                      "(+(a b) -c) d");
    assertQueryEquals("+(apple \"steve jobs\") -(foo bar baz)", null,
                      "+(apple \"steve jobs\") -(foo bar baz)");
    assertQueryEquals("+title:(dog OR cat) -author:\"bob dole\"", null,
                      "+(title:dog title:cat) -author:\"bob dole\"");
  
public voidtestSimpleDAO()

    assertQueryEqualsDOA("term term term", null, "+term +term +term");
    assertQueryEqualsDOA("term +term term", null, "+term +term +term");
    assertQueryEqualsDOA("term term +term", null, "+term +term +term");
    assertQueryEqualsDOA("term +term +term", null, "+term +term +term");
    assertQueryEqualsDOA("-term term term", null, "-term +term +term");
  
public voidtestSlop()

    assertQueryEquals("\"term germ\"~2", null, "\"term germ\"~2");
    assertQueryEquals("\"term germ\"~2 flork", null, "\"term germ\"~2 flork");
    assertQueryEquals("\"term\"~2", null, "term");
    assertQueryEquals("\" \"~2 germ", null, "germ");
    assertQueryEquals("\"term germ\"~2^2", null, "\"term germ\"~2^2.0");
  
public voidtestTabNewlineCarriageReturn()

    assertQueryEqualsDOA("+weltbank +worlbank", null,
      "+weltbank +worlbank");

    assertQueryEqualsDOA("+weltbank\n+worlbank", null,
      "+weltbank +worlbank");
    assertQueryEqualsDOA("weltbank \n+worlbank", null,
      "+weltbank +worlbank");
    assertQueryEqualsDOA("weltbank \n +worlbank", null,
      "+weltbank +worlbank");

    assertQueryEqualsDOA("+weltbank\r+worlbank", null,
      "+weltbank +worlbank");
    assertQueryEqualsDOA("weltbank \r+worlbank", null,
      "+weltbank +worlbank");
    assertQueryEqualsDOA("weltbank \r +worlbank", null,
      "+weltbank +worlbank");

    assertQueryEqualsDOA("+weltbank\r\n+worlbank", null,
      "+weltbank +worlbank");
    assertQueryEqualsDOA("weltbank \r\n+worlbank", null,
      "+weltbank +worlbank");
    assertQueryEqualsDOA("weltbank \r\n +worlbank", null,
      "+weltbank +worlbank");
    assertQueryEqualsDOA("weltbank \r \n +worlbank", null,
      "+weltbank +worlbank");

    assertQueryEqualsDOA("+weltbank\t+worlbank", null,
      "+weltbank +worlbank");
    assertQueryEqualsDOA("weltbank \t+worlbank", null,
      "+weltbank +worlbank");
    assertQueryEqualsDOA("weltbank \t +worlbank", null,
      "+weltbank +worlbank");
  
public voidtestWildcard()

    assertQueryEquals("term*", null, "term*");
    assertQueryEquals("term*^2", null, "term*^2.0");
    assertQueryEquals("term~", null, "term~0.5");
    assertQueryEquals("term~0.7", null, "term~0.7");
    assertQueryEquals("term~^2", null, "term^2.0~0.5");
    assertQueryEquals("term^2~", null, "term^2.0~0.5");
    assertQueryEquals("term*germ", null, "term*germ");
    assertQueryEquals("term*germ^3", null, "term*germ^3.0");

    assertTrue(getQuery("term*", null) instanceof PrefixQuery);
    assertTrue(getQuery("term*^2", null) instanceof PrefixQuery);
    assertTrue(getQuery("term~", null) instanceof FuzzyQuery);
    assertTrue(getQuery("term~0.7", null) instanceof FuzzyQuery);
    FuzzyQuery fq = (FuzzyQuery)getQuery("term~0.7", null);
    assertEquals(0.7f, fq.getMinSimilarity(), 0.1f);
    assertEquals(0, fq.getPrefixLength());
    fq = (FuzzyQuery)getQuery("term~", null);
    assertEquals(0.5f, fq.getMinSimilarity(), 0.1f);
    assertEquals(0, fq.getPrefixLength());
    try {
      getQuery("term~1.1", null);   // value > 1, throws exception
      fail();
    } catch(ParseException pe) {
      // expected exception
    }
    assertTrue(getQuery("term*germ", null) instanceof WildcardQuery);

/* Tests to see that wild card terms are (or are not) properly
	 * lower-cased with propery parser configuration
	 */
// First prefix queries:
    assertWildcardQueryEquals("term*", true, "term*");
    assertWildcardQueryEquals("Term*", true, "term*");
    assertWildcardQueryEquals("TERM*", true, "term*");
    assertWildcardQueryEquals("term*", false, "term*");
    assertWildcardQueryEquals("Term*", false, "Term*");
    assertWildcardQueryEquals("TERM*", false, "TERM*");
// Then 'full' wildcard queries:
    assertWildcardQueryEquals("te?m", true, "te?m");
    assertWildcardQueryEquals("Te?m", true, "te?m");
    assertWildcardQueryEquals("TE?M", true, "te?m");
    assertWildcardQueryEquals("Te?m*gerM", true, "te?m*germ");
    assertWildcardQueryEquals("te?m", false, "te?m");
    assertWildcardQueryEquals("Te?m", false, "Te?m");
    assertWildcardQueryEquals("TE?M", false, "TE?M");
    assertWildcardQueryEquals("Te?m*gerM", false, "Te?m*gerM");