| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 1 | package de.ids_mannheim.korap.index; |
| 2 | |
| 3 | import static org.junit.Assert.assertEquals; |
| Eliza Margaretha | 4026054 | 2014-10-01 10:32:09 +0000 | [diff] [blame] | 4 | import static org.junit.Assert.fail; |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 5 | |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 6 | import java.util.*; |
| 7 | import java.io.*; |
| Akron | 67d2ff0 | 2018-06-19 10:51:16 +0200 | [diff] [blame] | 8 | import static de.ids_mannheim.korap.TestSimple.*; |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 9 | |
| 10 | import org.apache.lucene.index.Term; |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 11 | import org.apache.lucene.search.spans.SpanOrQuery; |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 12 | import org.apache.lucene.search.spans.SpanQuery; |
| 13 | import org.apache.lucene.search.spans.SpanTermQuery; |
| 14 | import org.junit.Test; |
| 15 | import org.junit.runner.RunWith; |
| 16 | import org.junit.runners.JUnit4; |
| 17 | |
| Nils Diewald | a14ecd6 | 2015-02-26 21:00:20 +0000 | [diff] [blame] | 18 | import de.ids_mannheim.korap.KrillIndex; |
| Nils Diewald | 0339d46 | 2015-02-26 14:53:56 +0000 | [diff] [blame] | 19 | import de.ids_mannheim.korap.KrillQuery; |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 20 | import de.ids_mannheim.korap.query.DistanceConstraint; |
| margaretha | 71c66ee | 2015-12-11 14:39:55 +0100 | [diff] [blame] | 21 | import de.ids_mannheim.korap.query.QueryBuilder; |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 22 | import de.ids_mannheim.korap.query.SpanDistanceQuery; |
| 23 | import de.ids_mannheim.korap.query.SpanElementQuery; |
| Eliza Margaretha | fead206 | 2014-01-28 16:34:09 +0000 | [diff] [blame] | 24 | import de.ids_mannheim.korap.query.SpanNextQuery; |
| Nils Diewald | 92729ce | 2014-10-06 16:00:17 +0000 | [diff] [blame] | 25 | import de.ids_mannheim.korap.query.wrap.SpanQueryWrapper; |
| margaretha | c66265c | 2016-12-14 13:48:45 +0100 | [diff] [blame] | 26 | import de.ids_mannheim.korap.response.Match; |
| margaretha | 71c66ee | 2015-12-11 14:39:55 +0100 | [diff] [blame] | 27 | import de.ids_mannheim.korap.response.Result; |
| Eliza Margaretha | 4026054 | 2014-10-01 10:32:09 +0000 | [diff] [blame] | 28 | import de.ids_mannheim.korap.util.QueryException; |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 29 | |
| margaretha | c66265c | 2016-12-14 13:48:45 +0100 | [diff] [blame] | 30 | /** |
| 31 | * @author margaretha |
| 32 | * |
| 33 | */ |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 34 | @RunWith(JUnit4.class) |
| Eliza Margaretha | 1751dbf | 2014-02-03 09:47:19 +0000 | [diff] [blame] | 35 | public class TestElementDistanceIndex { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 36 | |
| 37 | Result kr; |
| 38 | KrillIndex ki; |
| 39 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 40 | private FieldDocument createFieldDoc0 () { |
| 41 | FieldDocument fd = new FieldDocument(); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 42 | fd.addString("ID", "doc-0"); |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 43 | fd.addTV("tokens", "bbcbbb", |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 44 | "[(0-1)s:b|s:c|_1$<i>0<i>1|<>:s$<b>64<i>0<i>1<i>1<b>0]" |
| margaretha | 71c66ee | 2015-12-11 14:39:55 +0100 | [diff] [blame] | 45 | + "[(1-2)s:b|_2$<i>1<i>2]" |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 46 | + "[(2-3)s:c|_3$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>3<b>0]" |
| 47 | + "[(3-4)s:b|_4$<i>3<i>4|<>:s$<b>64<i>3<i>4<i>4<b>0]" |
| 48 | + "[(4-5)s:b|_5$<i>4<i>5|<>:s$<b>64<i>4<i>5<i>5<b>0]" |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 49 | + "[(5-6)s:b|_6$<i>5<i>6]" |
| 50 | + "[(6-7)s:c|_7$<i>6<i>7|<>:s$<b>64<i>6<i>7<i>7<b>0]"); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 51 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 52 | } |
| 53 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 54 | private FieldDocument createFieldDoc1 () { |
| 55 | FieldDocument fd = new FieldDocument(); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 56 | fd.addString("ID", "doc-1"); |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 57 | fd.addTV("tokens", "ecebdc", |
| Akron | 19b4be2 | 2016-06-03 14:27:24 +0200 | [diff] [blame] | 58 | "[(0-1)s:e|_1$<i>0<i>1|<>:base/s:t$<b>64<i>0<i>6<i>6<b>0|<>:s$<b>64<i>0<i>2<i>1<b>0]" |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 59 | + "[(1-2)s:c|s:b|_2$<i>1<i>2|<>:s$<b>64<i>1<i>2<i>2<b>0]" |
| 60 | + "[(2-3)s:e|_3$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>3<b>0]" |
| 61 | + "[(3-4)s:b|_4$<i>3<i>4|<>:s$<b>64<i>3<i>4<i>4<b>0]" |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 62 | + "[(4-5)s:c|_5$<i>4<i>5|<>:s$<b>64<i>4<i>5<i>5<b>0]" |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 63 | + "[(5-6)s:c|_6$<i>5<i>6|<>:s$<b>64<i>5<i>6<i>6<b>0]"); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 64 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 65 | } |
| 66 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 67 | private FieldDocument createFieldDoc2 () { |
| 68 | FieldDocument fd = new FieldDocument(); |
| Eliza Margaretha | 8f9934c | 2014-02-06 13:04:32 +0000 | [diff] [blame] | 69 | fd.addString("ID", "doc-2"); |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 70 | fd.addTV("tokens", "bbbddd", |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 71 | "[(0-1)s:b|_1$<i>0<i>1|<>:p$<b>64<i>0<i>2<i>1<b>0]" |
| margaretha | 71c66ee | 2015-12-11 14:39:55 +0100 | [diff] [blame] | 72 | + "[(1-2)s:b|_2$<i>1<i>2]" |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 73 | + "[(2-3)s:b|_3$<i>2<i>3|<>:p$<b>64<i>2<i>3<i>3<b>0]" |
| 74 | + "[(3-4)s:d|_4$<i>3<i>4|<>:p$<b>64<i>3<i>4<i>4<b>0]" |
| 75 | + "[(4-5)s:d|_5$<i>4<i>5|<>:p$<b>64<i>4<i>5<i>5<b>0]" |
| margaretha | 71c66ee | 2015-12-11 14:39:55 +0100 | [diff] [blame] | 76 | + "[(5-6)s:d|_6$<i>5<i>6]"); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 77 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 78 | } |
| 79 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 80 | private FieldDocument createFieldDoc3 () { |
| 81 | FieldDocument fd = new FieldDocument(); |
| Eliza Margaretha | 8f9934c | 2014-02-06 13:04:32 +0000 | [diff] [blame] | 82 | fd.addString("ID", "doc-3"); |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 83 | fd.addTV("tokens", "bdbcdd", |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 84 | "[(0-1)s:b|_1$<i>0<i>1|<>:s$<b>64<i>0<i>2<i>1<b>0]" |
| margaretha | 71c66ee | 2015-12-11 14:39:55 +0100 | [diff] [blame] | 85 | + "[(1-2)s:d|_2$<i>1<i>2]" |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 86 | + "[(2-3)s:b|_3$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>3<b>0]" |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 87 | + "[(3-4)s:c|_4$<i>3<i>4|<>:s$<b>64<i>3<i>5<i>5<b>0]" |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 88 | + "[(4-5)s:d|_5$<i>4<i>5|<>:s$<b>64<i>4<i>5<i>5<b>0]" |
| margaretha | 71c66ee | 2015-12-11 14:39:55 +0100 | [diff] [blame] | 89 | + "[(5-6)s:d|_6$<i>5<i>6]"); |
| Eliza Margaretha | fead206 | 2014-01-28 16:34:09 +0000 | [diff] [blame] | 90 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 91 | } |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 92 | |
| margaretha | 4ad135c | 2017-01-12 12:49:44 +0100 | [diff] [blame] | 93 | private FieldDocument createFieldDoc4 () { |
| 94 | FieldDocument fd = new FieldDocument(); |
| 95 | fd.addString("ID", "doc-4"); |
| 96 | fd.addTV("tokens", "bdbcdd", |
| 97 | "[(0-1)s:b|_1$<i>0<i>1|<>:s$<b>64<i>0<i>2<i>1<b>0]" |
| 98 | + "[(1-2)s:d|_2$<i>1<i>2]" |
| 99 | + "[(2-3)s:c|s:b|_3$<i>2<i>3|<>:s$<b>64<i>2<i>5<i>3<b>0]" |
| 100 | + "[(3-4)s:c|_4$<i>3<i>4|<>:s$<b>64<i>3<i>5<i>4<b>0]" |
| 101 | + "[(4-5)s:d|_5$<i>4<i>5|<>:s$<b>64<i>4<i>5<i>5<b>0]" |
| 102 | + "[(5-6)s:d|_6$<i>5<i>6]"); |
| 103 | return fd; |
| 104 | } |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 105 | |
| 106 | private FieldDocument createFieldDoc5 () { |
| 107 | FieldDocument fd = new FieldDocument(); |
| 108 | fd.addString("ID", "doc-5"); |
| 109 | fd.addTV("tokens", "edef", |
| 110 | "[(0-1)s:e|_1$<i>0<i>1]" |
| 111 | + "[(1-2)s:d|_2$<i>1<i>2]" |
| 112 | + "[(2-3)s:e|_3$<i>2<i>3]" |
| 113 | + "[(3-4)s:f|_4$<i>3<i>4]"); |
| 114 | return fd; |
| 115 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 116 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 117 | public SpanQuery createQuery (String elementType, String x, String y, |
| 118 | int min, int max, boolean isOrdered) { |
| 119 | |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 120 | SpanElementQuery e = new SpanElementQuery("tokens", elementType); |
| 121 | return new SpanDistanceQuery(new SpanTermQuery(new Term("tokens", x)), |
| 122 | new SpanTermQuery(new Term("tokens", y)), |
| 123 | new DistanceConstraint(e, min, max, isOrdered, false), true); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 124 | } |
| 125 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 126 | /** |
| 127 | * Multiple documents |
| 128 | * Ensure terms and elements are in the same doc |
| 129 | * Ensure terms are in elements |
| 130 | * Check filter candidate list |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 131 | */ |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 132 | @Test |
| 133 | public void testCase1 () throws IOException { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 134 | ki = new KrillIndex(); |
| 135 | ki.addDoc(createFieldDoc0()); |
| 136 | ki.addDoc(createFieldDoc1()); |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 137 | ki.addDoc(createFieldDoc3()); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 138 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 139 | |
| 140 | SpanQuery sq; |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 141 | sq = createQuery("s", "s:b", "s:c", 1, 1, true); |
| Akron | d292690 | 2016-02-13 18:37:36 +0100 | [diff] [blame] | 142 | |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 143 | kr = ki.search(sq, (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 144 | |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 145 | assertEquals(4, kr.getTotalResults()); |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 146 | assertEquals(0, kr.getMatch(0).startPos); |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 147 | assertEquals(3, kr.getMatch(0).endPos); |
| 148 | assertEquals(4, kr.getMatch(1).startPos); |
| 149 | assertEquals(7, kr.getMatch(1).endPos); |
| 150 | assertEquals(3, kr.getMatch(2).startPos); |
| 151 | assertEquals(5, kr.getMatch(2).endPos); |
| 152 | assertEquals(2, kr.getMatch(3).startPos); |
| 153 | assertEquals(4, kr.getMatch(3).endPos); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 154 | } |
| 155 | |
| margaretha | 4ad135c | 2017-01-12 12:49:44 +0100 | [diff] [blame] | 156 | /** |
| 157 | * Ignore nested element distance unit |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 158 | */ |
| margaretha | 4ad135c | 2017-01-12 12:49:44 +0100 | [diff] [blame] | 159 | @Test |
| 160 | public void testCase1b () throws IOException { |
| 161 | ki = new KrillIndex(); |
| 162 | ki.addDoc(createFieldDoc4()); |
| 163 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 164 | |
| margaretha | 4ad135c | 2017-01-12 12:49:44 +0100 | [diff] [blame] | 165 | SpanQuery sq; |
| 166 | sq = createQuery("s", "s:b", "s:c", 1, 1, true); |
| 167 | |
| 168 | kr = ki.search(sq, (short) 10); |
| 169 | |
| 170 | assertEquals(2, kr.getTotalResults()); |
| 171 | assertEquals(0, kr.getMatch(0).startPos); |
| 172 | assertEquals(3, kr.getMatch(0).endPos); |
| 173 | assertEquals(2, kr.getMatch(1).startPos); |
| 174 | assertEquals(4, kr.getMatch(1).endPos); |
| 175 | } |
| 176 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 177 | /** |
| 178 | * Ensure terms and elements are in the same doc |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 179 | */ |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 180 | @Test |
| 181 | public void testCase2 () throws IOException { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 182 | ki = new KrillIndex(); |
| 183 | ki.addDoc(createFieldDoc0()); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 184 | ki.addDoc(createFieldDoc1()); |
| 185 | ki.addDoc(createFieldDoc2()); |
| 186 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 187 | |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 188 | SpanQuery sq; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 189 | sq = createQuery("p", "s:b", "s:d", 1, 1, true); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 190 | kr = ki.search(sq, (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 191 | |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 192 | assertEquals(kr.getTotalResults(), 1); |
| 193 | assertEquals(2, kr.getMatch(0).getLocalDocID()); |
| 194 | assertEquals(2, kr.getMatch(0).startPos); |
| 195 | assertEquals(4, kr.getMatch(0).endPos); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 196 | |
| 197 | } |
| 198 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 199 | /** Skip to */ |
| 200 | @Test |
| 201 | public void testCase3 () throws IOException { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 202 | ki = new KrillIndex(); |
| 203 | ki.addDoc(createFieldDoc0()); |
| Eliza Margaretha | fead206 | 2014-01-28 16:34:09 +0000 | [diff] [blame] | 204 | ki.addDoc(createFieldDoc1()); |
| 205 | ki.addDoc(createFieldDoc3()); |
| 206 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 207 | |
| Eliza Margaretha | fead206 | 2014-01-28 16:34:09 +0000 | [diff] [blame] | 208 | SpanQuery sq, edq; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 209 | edq = createQuery("s", "s:b", "s:c", 1, 1, true); |
| 210 | |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 211 | sq = new SpanNextQuery(edq, |
| 212 | new SpanTermQuery(new Term("tokens", "s:d"))); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 213 | |
| Eliza Margaretha | fead206 | 2014-01-28 16:34:09 +0000 | [diff] [blame] | 214 | kr = ki.search(sq, (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 215 | |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 216 | assertEquals(1, kr.getTotalResults()); |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 217 | assertEquals(2, kr.getMatch(0).getLocalDocID()); |
| 218 | assertEquals(2, kr.getMatch(0).startPos); |
| 219 | assertEquals(5, kr.getMatch(0).endPos); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 220 | |
| 221 | } |
| 222 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 223 | /** Same tokens in different elements */ |
| 224 | @Test |
| 225 | public void testCase4 () throws IOException { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 226 | ki = new KrillIndex(); |
| 227 | ki.addDoc(createFieldDoc0()); |
| Eliza Margaretha | 45b5be1 | 2014-02-04 11:22:46 +0000 | [diff] [blame] | 228 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 229 | |
| Eliza Margaretha | 45b5be1 | 2014-02-04 11:22:46 +0000 | [diff] [blame] | 230 | SpanQuery sq; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 231 | sq = createQuery("s", "s:b", "s:b", 1, 2, true); |
| Eliza Margaretha | 45b5be1 | 2014-02-04 11:22:46 +0000 | [diff] [blame] | 232 | kr = ki.search(sq, (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 233 | |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 234 | assertEquals(kr.getTotalResults(), 2); |
| 235 | assertEquals(0, kr.getMatch(0).startPos); |
| 236 | assertEquals(4, kr.getMatch(0).endPos); |
| 237 | assertEquals(3, kr.getMatch(1).startPos); |
| 238 | assertEquals(5, kr.getMatch(1).endPos); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 239 | |
| Eliza Margaretha | 4026054 | 2014-10-01 10:32:09 +0000 | [diff] [blame] | 240 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 241 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 242 | /** Test query from json */ |
| 243 | @Test |
| 244 | public void testCase5 () throws Exception { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 245 | ki = new KrillIndex(); |
| 246 | ki.addDoc(getClass().getResourceAsStream("/wiki/00001.json.gz"), true); |
| Eliza Margaretha | 4026054 | 2014-10-01 10:32:09 +0000 | [diff] [blame] | 247 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 248 | |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 249 | String jsonPath = |
| 250 | getClass().getResource("/queries/cosmas1.json").getFile(); |
| Akron | d6f5f59 | 2018-06-19 15:58:16 +0200 | [diff] [blame] | 251 | SpanQueryWrapper sqwi = getJsonQuery(jsonPath); |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 252 | kr = ki.search(sqwi.toQuery(), (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 253 | |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 254 | assertEquals((long) 3, kr.getTotalResults()); |
| Eliza Margaretha | 42b407b | 2014-10-01 10:36:33 +0000 | [diff] [blame] | 255 | assertEquals(14, kr.getMatch(0).startPos); |
| 256 | assertEquals(19, kr.getMatch(0).endPos); |
| 257 | assertEquals(30, kr.getMatch(1).startPos); |
| 258 | assertEquals(33, kr.getMatch(1).endPos); |
| Eliza Margaretha | 4026054 | 2014-10-01 10:32:09 +0000 | [diff] [blame] | 259 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 260 | |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 261 | @Test |
| 262 | public void testCQLAnd () throws Exception { |
| 263 | ki = new KrillIndex(); |
| 264 | ki.addDoc(getClass().getResourceAsStream("/wiki/00001.json.gz"), true); |
| 265 | ki.commit(); |
| 266 | |
| 267 | String jsonPath = getClass() |
| 268 | .getResource("/queries/distances/cql-and.jsonld").getFile(); |
| 269 | SpanQueryWrapper sqwi = getJsonQuery(jsonPath); |
| 270 | SpanQuery query = sqwi.toQuery(); |
| 271 | assertEquals( |
| 272 | "spanElementDistance(tokens:s:Buchstaben, tokens:s:Alphabet, " |
| 273 | + "[(base/s:s[0:0], notOrdered, notExcluded)])", |
| 274 | query.toString()); |
| 275 | kr = ki.search(sqwi.toQuery(), (short) 10); |
| 276 | |
| 277 | assertEquals((long) 1, kr.getTotalResults()); |
| 278 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 279 | |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 280 | /** Test query from json (2) */ |
| 281 | @Test |
| 282 | public void testCase6 () throws Exception { |
| 283 | ki = new KrillIndex(); |
| 284 | ki.addDoc(createFieldDoc0()); |
| 285 | ki.addDoc(createFieldDoc1()); |
| 286 | ki.addDoc(createFieldDoc2()); |
| 287 | ki.commit(); |
| 288 | |
| 289 | SpanQueryWrapper sqwi; |
| 290 | sqwi = new QueryBuilder("tokens").tag("base/s:t"); |
| 291 | |
| 292 | kr = ki.search(sqwi.toQuery(), (short) 10); |
| 293 | assertEquals(kr.getTotalResults(), 1); |
| Akron | f05fde6 | 2016-08-03 23:46:17 +0200 | [diff] [blame] | 294 | assertEquals("[[ecebdc]]", kr.getMatch(0).getSnippetBrackets()); |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 295 | |
| Akron | 67d2ff0 | 2018-06-19 10:51:16 +0200 | [diff] [blame] | 296 | String jsonPath = getClass() |
| 297 | .getResource("/queries/distances/in-same-t.jsonld").getFile(); |
| Akron | d6f5f59 | 2018-06-19 15:58:16 +0200 | [diff] [blame] | 298 | sqwi = getJsonQuery(jsonPath); |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 299 | |
| 300 | assertEquals( |
| 301 | "spanElementDistance(tokens:s:c, tokens:s:e, [(base/s:t[0:0], ordered, notExcluded)])", |
| 302 | sqwi.toQuery().toString()); |
| 303 | |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 304 | kr = ki.search(sqwi.toQuery(), (short) 10); |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 305 | assertEquals(1, kr.getTotalResults()); // Is 1 correct or |
| 306 | // should it not be |
| 307 | // ordered? |
| Akron | f05fde6 | 2016-08-03 23:46:17 +0200 | [diff] [blame] | 308 | assertEquals("[[ec]]ebdc", kr.getMatch(0).getSnippetBrackets()); |
| margaretha | c66265c | 2016-12-14 13:48:45 +0100 | [diff] [blame] | 309 | } |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 310 | |
| 311 | @Test |
| 312 | public void testNoMoreFirstSpans () throws IOException { |
| 313 | ki = new KrillIndex(); |
| 314 | ki.addDoc(createFieldDoc3()); |
| 315 | ki.addDoc(createFieldDoc5()); |
| 316 | ki.commit(); |
| 317 | |
| 318 | // c or b |
| 319 | SpanOrQuery soq = new SpanOrQuery( |
| 320 | new SpanTermQuery(new Term("tokens", "s:c")), |
| 321 | new SpanTermQuery(new Term("tokens", "s:b"))); |
| 322 | |
| 323 | // (c or b) /s0 d |
| 324 | SpanElementQuery e = new SpanElementQuery("tokens", "s"); |
| 325 | SpanDistanceQuery sdq = new SpanDistanceQuery( |
| 326 | soq, |
| 327 | new SpanTermQuery(new Term("tokens", "s:d")), |
| 328 | new DistanceConstraint(e, 0, 0, true, false), true); |
| 329 | |
| 330 | kr = ki.search(sdq, (short) 10); |
| 331 | assertEquals(1, kr.getTotalResults()); |
| 332 | } |
| 333 | |
| 334 | @Test |
| 335 | public void testNoMoreSecondSpans () throws IOException { |
| 336 | ki = new KrillIndex(); |
| 337 | ki.addDoc(createFieldDoc3()); |
| 338 | ki.addDoc(createFieldDoc5()); |
| 339 | ki.commit(); |
| 340 | |
| 341 | // c or b |
| 342 | SpanOrQuery soq = new SpanOrQuery( |
| 343 | new SpanTermQuery(new Term("tokens", "s:c")), |
| 344 | new SpanTermQuery(new Term("tokens", "s:b"))); |
| 345 | |
| 346 | // d /s0(c or b) |
| 347 | SpanElementQuery e = new SpanElementQuery("tokens", "s"); |
| 348 | SpanDistanceQuery sdq = new SpanDistanceQuery( |
| 349 | new SpanTermQuery(new Term("tokens", "s:d")), |
| 350 | soq, |
| 351 | new DistanceConstraint(e, 0, 0, true, false), true); |
| 352 | |
| 353 | kr = ki.search(sdq, (short) 10); |
| 354 | assertEquals(0, kr.getTotalResults()); |
| 355 | } |
| 356 | |
| 357 | @Test |
| 358 | public void testNoElementSpans () throws IOException { |
| 359 | ki = new KrillIndex(); |
| 360 | ki.addDoc(createFieldDoc3()); |
| 361 | ki.addDoc(createFieldDoc5()); |
| 362 | ki.commit(); |
| 363 | |
| 364 | // c or e |
| 365 | SpanOrQuery soq = new SpanOrQuery( |
| 366 | new SpanTermQuery(new Term("tokens", "s:c")), |
| 367 | new SpanTermQuery(new Term("tokens", "s:e"))); |
| 368 | |
| 369 | // (c or e) /s0 d |
| 370 | SpanElementQuery e = new SpanElementQuery("tokens", "s"); |
| 371 | SpanDistanceQuery sdq = new SpanDistanceQuery( |
| 372 | soq, |
| 373 | new SpanTermQuery(new Term("tokens", "s:d")), |
| 374 | new DistanceConstraint(e, 0, 0, true, false), true); |
| 375 | |
| 376 | kr = ki.search(sdq, (short) 10); |
| 377 | } |
| margaretha | c66265c | 2016-12-14 13:48:45 +0100 | [diff] [blame] | 378 | } |