Search in sources :

Example 11 with SpanFocusQuery

use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.

the class TestMatchIndex method indexExampleExtend.

@Test
public void indexExampleExtend() throws IOException {
    KrillIndex ki = new KrillIndex();
    // abcabcabac
    FieldDocument fd = new FieldDocument();
    fd.addTV("base", "abcabcabac", "[(0-1)s:a|i:a|_0$<i>0<i>1|-:t$<i>10]" + "[(1-2)s:b|i:b|_1$<i>1<i>2]" + "[(2-3)s:c|i:c|_2$<i>2<i>3]" + "[(3-4)s:a|i:a|_3$<i>3<i>4]" + "[(4-5)s:b|i:b|_4$<i>4<i>5]" + "[(5-6)s:c|i:c|_5$<i>5<i>6]" + "[(6-7)s:a|i:a|_6$<i>6<i>7]" + "[(7-8)s:b|i:b|_7$<i>7<i>8]" + "[(8-9)s:a|i:a|_8$<i>8<i>9]" + "[(9-10)s:c|i:c|_9$<i>9<i>10]");
    ki.addDoc(fd);
    ki.commit();
    SpanQuery sq;
    Result kr;
    sq = new SpanFocusQuery(new SpanNextQuery(new SpanClassQuery(new SpanTermQuery(new Term("base", "s:a")), (byte) 2), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:b")), (byte) 3)), (byte) 3);
    kr = ki.search(sq, (short) 10);
    assertEquals("totalResults", kr.getTotalResults(), 3);
    Match km = kr.getMatch(0);
    assertEquals("StartPos (0)", 1, km.startPos);
    assertEquals("EndPos (0)", 2, km.endPos);
    assertEquals("SnippetBrackets (0)", "a[[{3:b}]]cabcab ...", km.getSnippetBrackets());
    sq = new SpanFocusQuery(new SpanFocusQuery(new SpanNextQuery(new SpanClassQuery(new SpanTermQuery(new Term("base", "s:a")), (byte) 2), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:b")), (byte) 3)), (byte) 3), (byte) 2);
    kr = ki.search(sq, (short) 10);
    km = kr.getMatch(0);
    assertEquals("StartPos (0)", 0, km.startPos);
    assertEquals("EndPos (0)", 1, km.endPos);
    assertEquals("SnippetBrackets (0)", "[[{2:a}]]bcabca ...", km.getSnippetBrackets());
// TODO: Check ID
}
Also used : SpanClassQuery(de.ids_mannheim.korap.query.SpanClassQuery) SpanTermQuery(org.apache.lucene.search.spans.SpanTermQuery) Term(org.apache.lucene.index.Term) KrillIndex(de.ids_mannheim.korap.KrillIndex) SpanFocusQuery(de.ids_mannheim.korap.query.SpanFocusQuery) SpanQuery(org.apache.lucene.search.spans.SpanQuery) Result(de.ids_mannheim.korap.response.Result) SpanNextQuery(de.ids_mannheim.korap.query.SpanNextQuery) Match(de.ids_mannheim.korap.response.Match) Test(org.junit.Test)

Example 12 with SpanFocusQuery

use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.

the class TestMatchIndex method indexExample1.

@Test
public void indexExample1() throws IOException {
    KrillIndex ki = new KrillIndex();
    // abcabcabac
    FieldDocument fd = new FieldDocument();
    fd.addTV("base", "abcabcabac", "[(0-1)s:a|i:a|_0$<i>0<i>1|-:t$<i>10]" + "[(1-2)s:b|i:b|_1$<i>1<i>2]" + "[(2-3)s:c|i:c|_2$<i>2<i>3]" + "[(3-4)s:a|i:a|_3$<i>3<i>4]" + "[(4-5)s:b|i:b|_4$<i>4<i>5]" + "[(5-6)s:c|i:c|_5$<i>5<i>6]" + "[(6-7)s:a|i:a|_6$<i>6<i>7]" + "[(7-8)s:b|i:b|_7$<i>7<i>8]" + "[(8-9)s:a|i:a|_8$<i>8<i>9]" + "[(9-10)s:c|i:c|_9$<i>9<i>10]");
    ki.addDoc(fd);
    ki.commit();
    SpanQuery sq;
    Result kr;
    sq = new SpanNextQuery(new SpanTermQuery(new Term("base", "s:b")), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:a"))));
    kr = ki.search(sq, (short) 10);
    assertEquals("totalResults", kr.getTotalResults(), 1);
    assertEquals("StartPos (0)", 7, kr.getMatch(0).startPos);
    assertEquals("EndPos (0)", 9, kr.getMatch(0).endPos);
    assertEquals("SnippetBrackets (0)", "... bcabca[[b{1:a}]]c", kr.getMatch(0).getSnippetBrackets());
    assertEquals("Test no 'more' context", "<span class=\"context-left\"><span class=\"more\"></span>bcabca</span><span class=\"match\"><mark>b<mark class=\"class-1 level-0\">a</mark></mark></span><span class=\"context-right\">c</span>", kr.getMatch(0).getSnippetHTML());
    sq = new SpanFocusQuery(new SpanNextQuery(new SpanTermQuery(new Term("base", "s:b")), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:a")))));
    kr = ki.search(sq, (short) 10);
    assertEquals("totalResults", kr.getTotalResults(), 1);
    assertEquals("StartPos (0)", 8, kr.getMatch(0).startPos);
    assertEquals("EndPos (0)", 9, kr.getMatch(0).endPos);
    assertEquals("SnippetBrackets (0)", "... cabcab[[{1:a}]]c", kr.getMatch(0).getSnippetBrackets());
    sq = new SpanFocusQuery(new SpanNextQuery(new SpanClassQuery(new SpanTermQuery(new Term("base", "s:a")), (byte) 2), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:b")), (byte) 3)), (byte) 3);
    kr = ki.search(sq, (short) 10);
    assertEquals("totalResults", kr.getTotalResults(), 3);
    assertEquals("StartPos (0)", 1, kr.getMatch(0).startPos);
    assertEquals("EndPos (0)", 2, kr.getMatch(0).endPos);
    assertEquals("SnippetBrackets (0)", "a[[{3:b}]]cabcab ...", kr.getMatch(0).getSnippetBrackets());
    assertEquals("<span class=\"context-left\">a</span><span class=\"match\"><mark><mark class=\"class-3 level-0\">b</mark></mark></span><span class=\"context-right\">cabcab<span class=\"more\"></span></span>", kr.getMatch(0).getSnippetHTML());
    assertEquals("StartPos (1)", 4, kr.getMatch(1).startPos);
    assertEquals("EndPos (1)", 5, kr.getMatch(1).endPos);
    assertEquals("SnippetBrackets (1)", "abca[[{3:b}]]cabac", kr.getMatch(1).getSnippetBrackets());
    assertEquals("<span class=\"context-left\">abca</span><span class=\"match\"><mark><mark class=\"class-3 level-0\">b</mark></mark></span><span class=\"context-right\">cabac</span>", kr.getMatch(1).getSnippetHTML());
    assertEquals("StartPos (2)", 7, kr.getMatch(2).startPos);
    assertEquals("EndPos (2)", 8, kr.getMatch(2).endPos);
    assertEquals("SnippetBrackets (2)", "... bcabca[[{3:b}]]ac", kr.getMatch(2).getSnippetBrackets());
    // abcabcabac
    sq = new SpanFocusQuery(new SpanNextQuery(new SpanTermQuery(new Term("base", "s:a")), new SpanClassQuery(new SpanNextQuery(new SpanTermQuery(new Term("base", "s:b")), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:a")))), (byte) 2)), (byte) 2);
    kr = ki.search(sq, (short) 10);
    assertEquals("totalResults", kr.getTotalResults(), 1);
    assertEquals("SnippetBrackets (0)", "... bcabca[[{2:b{1:a}}]]c", kr.getMatch(0).getSnippetBrackets());
    assertEquals("SnippetHTML (0) 1", "<span class=\"context-left\"><span class=\"more\"></span>bcabca</span><span class=\"match\"><mark><mark class=\"class-2 level-0\">b<mark class=\"class-1 level-1\">a</mark></mark></mark></span><span class=\"context-right\">c</span>", kr.getMatch(0).getSnippetHTML());
    // Offset tokens
    kr = ki.search(sq, 0, (short) 10, true, (short) 2, true, (short) 2);
    assertEquals("totalResults", kr.getTotalResults(), 1);
    assertEquals("SnippetBrackets (0)", "... ca[[{2:b{1:a}}]]c", kr.getMatch(0).getSnippetBrackets());
    // Offset Characters
    kr = ki.search(sq, 0, (short) 10, false, (short) 1, false, (short) 0);
    assertEquals("totalResults", kr.getTotalResults(), 1);
    assertEquals("SnippetBrackets (0)", "... a[[{2:b{1:a}}]] ...", kr.getMatch(0).getSnippetBrackets());
    assertEquals("SnippetHTML (0) 2", "<span class=\"context-left\"><span class=\"more\"></span>a</span><span class=\"match\"><mark><mark class=\"class-2 level-0\">b<mark class=\"class-1 level-1\">a</mark></mark></mark></span><span class=\"context-right\"><span class=\"more\"></span></span>", kr.getMatch(0).getSnippetHTML());
    // Don't match the expected class!
    sq = new SpanFocusQuery(new SpanNextQuery(new SpanClassQuery(new SpanTermQuery(new Term("base", "s:b")), (byte) 1), new SpanClassQuery(new SpanTermQuery(new Term("base", "s:c")), (byte) 2)), (byte) 3);
    kr = ki.search(sq, (short) 10);
    assertEquals("totalResults", kr.getTotalResults(), 0);
    sq = new SpanFocusQuery(new SpanNextQuery(new SpanTermQuery(new Term("base", "s:a")), new SpanClassQuery(new SpanNextQuery(new SpanTermQuery(new Term("base", "s:b")), new SpanTermQuery(new Term("base", "s:c"))))));
    kr = ki.search(sq, (short) 2);
    assertEquals("totalResults", kr.getTotalResults(), 2);
    assertEquals("StartPos (0)", 1, kr.getMatch(0).startPos);
    assertEquals("EndPos (0)", 3, kr.getMatch(0).endPos);
    assertEquals("SnippetBrackets (0)", "a[[{1:bc}]]abcaba ...", kr.getMatch(0).getSnippetBrackets());
    assertEquals("StartPos (1)", 4, kr.getMatch(1).startPos);
    assertEquals("EndPos (1)", 6, kr.getMatch(1).endPos);
    assertEquals("SnippetBrackets (1)", "abca[[{1:bc}]]abac", kr.getMatch(1).getSnippetBrackets());
    assertEquals(1, ki.numberOf("base", "documents"));
    assertEquals(10, ki.numberOf("base", "t"));
}
Also used : SpanTermQuery(org.apache.lucene.search.spans.SpanTermQuery) SpanClassQuery(de.ids_mannheim.korap.query.SpanClassQuery) Term(org.apache.lucene.index.Term) KrillIndex(de.ids_mannheim.korap.KrillIndex) SpanFocusQuery(de.ids_mannheim.korap.query.SpanFocusQuery) SpanQuery(org.apache.lucene.search.spans.SpanQuery) Result(de.ids_mannheim.korap.response.Result) SpanNextQuery(de.ids_mannheim.korap.query.SpanNextQuery) Test(org.junit.Test)

Example 13 with SpanFocusQuery

use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.

the class TestFocusIndex method testCase12.

/**
 * Check Skipto focus spans
 */
@Test
public void testCase12() throws IOException {
    ki.addDoc(TestRelationIndex.createFieldDoc0());
    ki.addDoc(TestRelationIndex.createFieldDoc1());
    ki.commit();
    SpanRelationQuery sq = new SpanRelationQuery(new SpanTermQuery(new Term("base", ">:xip/syntax-dep_rel")), true, RelationDirection.RIGHT);
    sq.setSourceClass((byte) 1);
    SpanFocusQuery sfq = new SpanFocusQuery(sq, (byte) 1);
    sfq.setSorted(false);
    SpanTermQuery stq = new SpanTermQuery(new Term("base", "s:c"));
    SpanNextQuery snq = new SpanNextQuery(stq, sfq);
    kr = ki.search(snq, (short) 20);
    assertEquals(0, kr.getMatch(0).getStartPos());
    assertEquals(2, kr.getMatch(0).getEndPos());
    assertEquals(5, kr.getMatch(1).getStartPos());
    assertEquals(9, kr.getMatch(1).getEndPos());
// for (Match m : kr.getMatches()) {
// System.out.println(m.getStartPos() + " " + m.getEndPos());
// }
}
Also used : SpanTermQuery(org.apache.lucene.search.spans.SpanTermQuery) SpanRelationQuery(de.ids_mannheim.korap.query.SpanRelationQuery) Term(org.apache.lucene.index.Term) SpanFocusQuery(de.ids_mannheim.korap.query.SpanFocusQuery) SpanNextQuery(de.ids_mannheim.korap.query.SpanNextQuery) Test(org.junit.Test)

Example 14 with SpanFocusQuery

use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.

the class SpanRelationWrapper method toFragmentQuery.

@Override
public SpanQuery toFragmentQuery() throws QueryException {
    if (this.isNull() || this.isEmpty()) {
        return null;
    }
    SpanQuery relationTermQuery = relationQuery.retrieveNode(this.retrieveNode).toFragmentQuery();
    // .retrieveNode(this.retrieveNode).toFragmentQuery();
    if (relationTermQuery == null)
        return null;
    SpanRelationQuery rq = new SpanRelationQuery(relationTermQuery, true, direction);
    SpanQuery subq1, subq2;
    if (subQuery1.isEmpty) {
        if (!subQuery2.isEmpty) {
            // match target
            subq2 = subQuery2.retrieveNode(this.retrieveNode).toFragmentQuery();
            if (subQuery1.hasClass) {
                rq.setSourceClass(subQuery1.getClassNumber());
            }
            return new SpanRelationMatchQuery(rq, subq2, true);
        }
    } else if (subQuery2.isEmpty) {
        if (!subQuery1.isEmpty) {
            // match source
            subq1 = subQuery1.retrieveNode(this.retrieveNode).toFragmentQuery();
            if (subQuery2.hasClass) {
                rq.setTargetClass(subQuery2.getClassNumber());
            }
            return new SpanRelationMatchQuery(rq, subq1, true);
        }
    } else {
        // match both
        subq1 = subQuery1.retrieveNode(this.retrieveNode).toFragmentQuery();
        subq2 = subQuery2.retrieveNode(this.retrieveNode).toFragmentQuery();
        return new SpanRelationMatchQuery(rq, subq1, subq2, true);
    }
    // both empty
    if (subQuery1.hasClass) {
        rq.setSourceClass(subQuery1.getClassNumber());
    }
    if (subQuery2.hasClass) {
        rq.setTargetClass(subQuery2.getClassNumber());
    }
    SpanFocusQuery fq = new SpanFocusQuery(rq, rq.getTempClassNumbers());
    fq.setMatchTemporaryClass(true);
    fq.setRemoveTemporaryClasses(true);
    fq.setSorted(false);
    return fq;
}
Also used : SpanRelationQuery(de.ids_mannheim.korap.query.SpanRelationQuery) SpanRelationMatchQuery(de.ids_mannheim.korap.query.SpanRelationMatchQuery) SpanFocusQuery(de.ids_mannheim.korap.query.SpanFocusQuery) SpanQuery(org.apache.lucene.search.spans.SpanQuery)

Example 15 with SpanFocusQuery

use of de.ids_mannheim.korap.query.SpanFocusQuery in project Krill by KorAP.

the class TestReferenceIndex method testCase1.

@Test
public void testCase1() throws IOException {
    ki = new KrillIndex();
    ki.addDoc(TestRelationIndex.createFieldDoc2());
    ki.commit();
    SpanTermQuery stq = new SpanTermQuery(new Term("base", "pos:V"));
    SpanElementQuery seq = new SpanElementQuery("base", "np");
    SpanClassQuery scq1 = new SpanClassQuery(stq, (byte) 1);
    SpanClassQuery scq2 = new SpanClassQuery(seq, (byte) 2);
    SpanNextQuery snq1 = new SpanNextQuery(scq1, scq2);
    SpanFocusQuery sfq1 = new SpanFocusQuery(snq1, (byte) 2);
    SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(new Term("base", "<:child-of")), true, RelationDirection.LEFT);
    SpanElementQuery seq2 = new SpanElementQuery("base", "pp");
    SpanClassQuery scq3 = new SpanClassQuery(seq2, (byte) 3);
    SpanRelationMatchQuery rq = new SpanRelationMatchQuery(srq, sfq1, scq3, true);
    // focus on np
    SpanFocusQuery sfq2 = new SpanFocusQuery(rq, (byte) 1);
    DistanceConstraint constraint = new DistanceConstraint(3, 3, true, false);
    SpanDistanceQuery sdq = new SpanDistanceQuery(sfq2, scq3, constraint, true);
    SpanReferenceQuery ref = new SpanReferenceQuery(sdq, (byte) 3, true);
    kr = ki.search(ref, (short) 10);
    /*
         * for (Match km : kr.getMatches()) {
         * 
         * System.out.println(km.getStartPos() + "," + km.getEndPos()
         * + " " + km.getSnippetBrackets()); }
         * System.out.println(kr.getTotalResults());
         */
    // cat=V & cat=np & cat=pp & #1 . #2 & #3 ->child-of #2 & #1 .{3,3} #3
    assertEquals("spanReference(spanDistance(focus(1: focus(#[1,2]spanSegment(" + "focus(#1: spanSegment(spanRelation(base:<:child-of), focus(2: spanNext(" + "{1: base:pos:V}, {2: <base:np />})))), {3: <base:pp />}))), " + "{3: <base:pp />}, [(w[3:3], ordered, notExcluded)]), 3)", ref.toString());
    assertEquals(1, kr.getMatch(0).getStartPos());
    assertEquals(7, kr.getMatch(0).getEndPos());
}
Also used : SpanTermQuery(org.apache.lucene.search.spans.SpanTermQuery) SpanClassQuery(de.ids_mannheim.korap.query.SpanClassQuery) SpanRelationQuery(de.ids_mannheim.korap.query.SpanRelationQuery) SpanDistanceQuery(de.ids_mannheim.korap.query.SpanDistanceQuery) SpanRelationMatchQuery(de.ids_mannheim.korap.query.SpanRelationMatchQuery) DistanceConstraint(de.ids_mannheim.korap.query.DistanceConstraint) Term(org.apache.lucene.index.Term) SpanReferenceQuery(de.ids_mannheim.korap.query.SpanReferenceQuery) KrillIndex(de.ids_mannheim.korap.KrillIndex) SpanElementQuery(de.ids_mannheim.korap.query.SpanElementQuery) SpanFocusQuery(de.ids_mannheim.korap.query.SpanFocusQuery) SpanNextQuery(de.ids_mannheim.korap.query.SpanNextQuery) Test(org.junit.Test)

Aggregations

SpanFocusQuery (de.ids_mannheim.korap.query.SpanFocusQuery)22 Term (org.apache.lucene.index.Term)20 SpanTermQuery (org.apache.lucene.search.spans.SpanTermQuery)20 Test (org.junit.Test)19 SpanRelationQuery (de.ids_mannheim.korap.query.SpanRelationQuery)15 SpanClassQuery (de.ids_mannheim.korap.query.SpanClassQuery)14 SpanElementQuery (de.ids_mannheim.korap.query.SpanElementQuery)10 SpanQuery (org.apache.lucene.search.spans.SpanQuery)10 KrillIndex (de.ids_mannheim.korap.KrillIndex)8 SpanRelationMatchQuery (de.ids_mannheim.korap.query.SpanRelationMatchQuery)8 SpanNextQuery (de.ids_mannheim.korap.query.SpanNextQuery)6 Result (de.ids_mannheim.korap.response.Result)6 SpanAttributeQuery (de.ids_mannheim.korap.query.SpanAttributeQuery)4 SpanWithAttributeQuery (de.ids_mannheim.korap.query.SpanWithAttributeQuery)4 SpanTermWithIdQuery (de.ids_mannheim.korap.query.SpanTermWithIdQuery)3 SpanWithinQuery (de.ids_mannheim.korap.query.SpanWithinQuery)3 DistanceConstraint (de.ids_mannheim.korap.query.DistanceConstraint)2 SpanDistanceQuery (de.ids_mannheim.korap.query.SpanDistanceQuery)2 SpanReferenceQuery (de.ids_mannheim.korap.query.SpanReferenceQuery)2 SpanSegmentQuery (de.ids_mannheim.korap.query.SpanSegmentQuery)2