| 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 | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 43 | fd.addTV("tokens", "bbcbbbc", |
| 44 | "[(0-1)s:b|s:c|_0$<i>0<i>1|<>:s$<b>64<i>0<i>2<i>1<b>0]" |
| 45 | + "[(1-2)s:b|_1$<i>1<i>2]" |
| 46 | + "[(2-3)s:c|_2$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>3<b>0]" |
| 47 | + "[(3-4)s:b|_3$<i>3<i>4|<>:s$<b>64<i>3<i>4<i>4<b>0]" |
| 48 | + "[(4-5)s:b|_4$<i>4<i>5|<>:s$<b>64<i>4<i>5<i>5<b>0]" |
| 49 | + "[(5-6)s:b|_5$<i>5<i>6]" |
| 50 | + "[(6-7)s:c|_6$<i>6<i>7|<>:s$<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", |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 58 | "[(0-1)s:e|_0$<i>0<i>1|<>:base/s:t$<b>64<i>0<i>6<i>6<b>0|<>:s$<b>64<i>0<i>1<i>1<b>0]" |
| 59 | + "[(1-2)s:c|s:b|_1$<i>1<i>2|<>:s$<b>64<i>1<i>2<i>2<b>0]" |
| 60 | + "[(2-3)s:e|_2$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>3<b>0]" |
| 61 | + "[(3-4)s:b|_3$<i>3<i>4|<>:s$<b>64<i>3<i>4<i>4<b>0]" |
| 62 | + "[(4-5)s:c|_4$<i>4<i>5|<>:s$<b>64<i>4<i>5<i>5<b>0]" |
| 63 | + "[(5-6)s:c|_5$<i>5<i>6|<>:s$<b>64<i>5<i>6<i>6<b>0]"); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 64 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 65 | } |
| 66 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 67 | private FieldDocument createFieldDoc2 () { |
| 68 | FieldDocument fd = new FieldDocument(); |
| Eliza Margaretha | 8f9934c | 2014-02-06 13:04:32 +0000 | [diff] [blame] | 69 | fd.addString("ID", "doc-2"); |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 70 | fd.addTV("tokens", "bbbddd", |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 71 | "[(0-1)s:b|_0$<i>0<i>1|<>:p$<b>64<i>0<i>1<i>1<b>0]" |
| 72 | + "[(1-2)s:b|_1$<i>1<i>2]" |
| 73 | + "[(2-3)s:b|_2$<i>2<i>3|<>:p$<b>64<i>2<i>3<i>3<b>0]" |
| 74 | + "[(3-4)s:d|_3$<i>3<i>4|<>:p$<b>64<i>3<i>4<i>4<b>0]" |
| 75 | + "[(4-5)s:d|_4$<i>4<i>5|<>:p$<b>64<i>4<i>5<i>5<b>0]" |
| 76 | + "[(5-6)s:d|_5$<i>5<i>6]"); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 77 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 78 | } |
| 79 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 80 | private FieldDocument createFieldDoc3 () { |
| 81 | FieldDocument fd = new FieldDocument(); |
| Eliza Margaretha | 8f9934c | 2014-02-06 13:04:32 +0000 | [diff] [blame] | 82 | fd.addString("ID", "doc-3"); |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 83 | fd.addTV("tokens", "bdbcdd", |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 84 | "[(0-1)s:b|_0$<i>0<i>1|<>:s$<b>64<i>0<i>2<i>2<b>0]" |
| 85 | + "[(1-2)s:d|_1$<i>1<i>2]" |
| 86 | + "[(2-3)s:b|_2$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>3<b>0]" |
| 87 | + "[(3-4)s:c|_3$<i>3<i>4|<>:s$<b>64<i>3<i>5<i>5<b>0]" |
| 88 | + "[(4-5)s:d|_4$<i>4<i>5|<>:s$<b>64<i>4<i>5<i>5<b>0]" |
| 89 | + "[(5-6)s:d|_5$<i>5<i>6]"); |
| Eliza Margaretha | fead206 | 2014-01-28 16:34:09 +0000 | [diff] [blame] | 90 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 91 | } |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 92 | |
| margaretha | 4ad135c | 2017-01-12 12:49:44 +0100 | [diff] [blame] | 93 | private FieldDocument createFieldDoc4 () { |
| 94 | FieldDocument fd = new FieldDocument(); |
| 95 | fd.addString("ID", "doc-4"); |
| 96 | fd.addTV("tokens", "bdbcdd", |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 97 | "[(0-1)s:b|_0$<i>0<i>1|<>:s$<b>64<i>0<i>1<i>1<b>0]" |
| 98 | + "[(1-2)s:d|_1$<i>1<i>2]" |
| 99 | + "[(2-3)s:c|s:b|_2$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>3<b>0]" |
| 100 | + "[(3-4)s:c|_3$<i>3<i>4|<>:s$<b>64<i>3<i>5<i>5<b>0]" |
| 101 | + "[(4-5)s:d|_4$<i>4<i>5|<>:s$<b>64<i>4<i>5<i>5<b>0]" |
| 102 | + "[(5-6)s:d|_5$<i>5<i>6]"); |
| margaretha | 4ad135c | 2017-01-12 12:49:44 +0100 | [diff] [blame] | 103 | return fd; |
| 104 | } |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 105 | |
| 106 | private FieldDocument createFieldDoc5 () { |
| 107 | FieldDocument fd = new FieldDocument(); |
| 108 | fd.addString("ID", "doc-5"); |
| 109 | fd.addTV("tokens", "edef", |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 110 | "[(0-1)s:e|_0$<i>0<i>1]" |
| 111 | + "[(1-2)s:d|_1$<i>1<i>2]" |
| 112 | + "[(2-3)s:e|_2$<i>2<i>3]" |
| 113 | + "[(3-4)s:f|_3$<i>3<i>4]"); |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 114 | return fd; |
| 115 | } |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 116 | |
| 117 | private FieldDocument createFieldDoc6 () { |
| 118 | FieldDocument fd = new FieldDocument(); |
| 119 | fd.addString("ID", "doc-6"); |
| 120 | fd.addTV("tokens", "bebcee", |
| 121 | "[(0-1)s:b|_0$<i>0<i>1|<>:s$<b>64<i>0<i>4<i>4<b>0]" |
| 122 | + "[(1-2)s:e|_1$<i>1<i>2]" |
| 123 | + "[(2-3)s:b|_2$<i>2<i>3]" |
| 124 | + "[(3-4)s:c|_3$<i>3<i>4]" |
| 125 | + "[(4-5)s:e|_4$<i>4<i>5|<>:s$<b>64<i>4<i>6<i>6<b>0]" |
| 126 | + "[(5-6)s:e|_5$<i>5<i>6]"); |
| 127 | return fd; |
| 128 | } |
| 129 | |
| 130 | private FieldDocument createFieldDoc7 () { |
| 131 | FieldDocument fd = new FieldDocument(); |
| 132 | fd.addString("ID", "doc-5"); |
| 133 | fd.addTV("tokens", "beef", |
| 134 | "[(0-1)s:b|_0$<i>0<i>1|<>:s$<b>64<i>0<i>4<i>4<b>0]" |
| 135 | + "[(1-2)s:e|_1$<i>1<i>2]" |
| 136 | + "[(2-3)s:e|_2$<i>2<i>3]" |
| 137 | + "[(3-4)s:f|_3$<i>3<i>4]"); |
| 138 | return fd; |
| 139 | } |
| 140 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 141 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 142 | public SpanQuery createQuery (String elementType, String x, String y, |
| 143 | int min, int max, boolean isOrdered) { |
| 144 | |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 145 | SpanElementQuery e = new SpanElementQuery("tokens", elementType); |
| 146 | return new SpanDistanceQuery(new SpanTermQuery(new Term("tokens", x)), |
| 147 | new SpanTermQuery(new Term("tokens", y)), |
| 148 | new DistanceConstraint(e, min, max, isOrdered, false), true); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 149 | } |
| 150 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 151 | /** |
| 152 | * Multiple documents |
| 153 | * Ensure terms and elements are in the same doc |
| 154 | * Ensure terms are in elements |
| 155 | * Check filter candidate list |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 156 | */ |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 157 | @Test |
| 158 | public void testCase1 () throws IOException { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 159 | ki = new KrillIndex(); |
| 160 | ki.addDoc(createFieldDoc0()); |
| 161 | ki.addDoc(createFieldDoc1()); |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 162 | ki.addDoc(createFieldDoc3()); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 163 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 164 | |
| 165 | SpanQuery sq; |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 166 | sq = createQuery("s", "s:b", "s:c", 1, 1, true); |
| Akron | d292690 | 2016-02-13 18:37:36 +0100 | [diff] [blame] | 167 | |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 168 | kr = ki.search(sq, (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 169 | |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 170 | assertEquals(4, kr.getTotalResults()); |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 171 | assertEquals(0, kr.getMatch(0).startPos); |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 172 | assertEquals(3, kr.getMatch(0).endPos); |
| 173 | assertEquals(4, kr.getMatch(1).startPos); |
| 174 | assertEquals(7, kr.getMatch(1).endPos); |
| 175 | assertEquals(3, kr.getMatch(2).startPos); |
| 176 | assertEquals(5, kr.getMatch(2).endPos); |
| 177 | assertEquals(2, kr.getMatch(3).startPos); |
| 178 | assertEquals(4, kr.getMatch(3).endPos); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 179 | } |
| 180 | |
| margaretha | 4ad135c | 2017-01-12 12:49:44 +0100 | [diff] [blame] | 181 | /** |
| 182 | * Ignore nested element distance unit |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 183 | */ |
| margaretha | 4ad135c | 2017-01-12 12:49:44 +0100 | [diff] [blame] | 184 | @Test |
| 185 | public void testCase1b () throws IOException { |
| 186 | ki = new KrillIndex(); |
| 187 | ki.addDoc(createFieldDoc4()); |
| 188 | ki.commit(); |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 189 | // b~d~b/c~c~dd |
| margaretha | 4ad135c | 2017-01-12 12:49:44 +0100 | [diff] [blame] | 190 | SpanQuery sq; |
| 191 | sq = createQuery("s", "s:b", "s:c", 1, 1, true); |
| 192 | |
| 193 | kr = ki.search(sq, (short) 10); |
| 194 | |
| 195 | assertEquals(2, kr.getTotalResults()); |
| 196 | assertEquals(0, kr.getMatch(0).startPos); |
| 197 | assertEquals(3, kr.getMatch(0).endPos); |
| 198 | assertEquals(2, kr.getMatch(1).startPos); |
| 199 | assertEquals(4, kr.getMatch(1).endPos); |
| 200 | } |
| 201 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 202 | /** |
| 203 | * Ensure terms and elements are in the same doc |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 204 | */ |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 205 | @Test |
| 206 | public void testCase2 () throws IOException { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 207 | ki = new KrillIndex(); |
| 208 | ki.addDoc(createFieldDoc0()); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 209 | ki.addDoc(createFieldDoc1()); |
| 210 | ki.addDoc(createFieldDoc2()); |
| 211 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 212 | |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 213 | SpanQuery sq; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 214 | sq = createQuery("p", "s:b", "s:d", 1, 1, true); |
| Eliza Margaretha | b0e9143 | 2014-01-28 15:26:31 +0000 | [diff] [blame] | 215 | kr = ki.search(sq, (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 216 | |
| Marc Kupietz | 26e05d2 | 2025-09-24 13:59:26 +0200 | [diff] [blame] | 217 | assertEquals(1, kr.getTotalResults()); |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 218 | assertEquals(2, kr.getMatch(0).getLocalDocID()); |
| 219 | assertEquals(2, kr.getMatch(0).startPos); |
| 220 | assertEquals(4, kr.getMatch(0).endPos); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 221 | |
| 222 | } |
| 223 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 224 | /** Skip to */ |
| 225 | @Test |
| 226 | public void testCase3 () throws IOException { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 227 | ki = new KrillIndex(); |
| 228 | ki.addDoc(createFieldDoc0()); |
| Eliza Margaretha | fead206 | 2014-01-28 16:34:09 +0000 | [diff] [blame] | 229 | ki.addDoc(createFieldDoc1()); |
| 230 | ki.addDoc(createFieldDoc3()); |
| 231 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 232 | |
| Eliza Margaretha | fead206 | 2014-01-28 16:34:09 +0000 | [diff] [blame] | 233 | SpanQuery sq, edq; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 234 | edq = createQuery("s", "s:b", "s:c", 1, 1, true); |
| 235 | |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 236 | sq = new SpanNextQuery(edq, |
| 237 | new SpanTermQuery(new Term("tokens", "s:d"))); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 238 | |
| Eliza Margaretha | fead206 | 2014-01-28 16:34:09 +0000 | [diff] [blame] | 239 | kr = ki.search(sq, (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 240 | |
| margaretha | 93d0d62 | 2017-01-04 14:53:33 +0100 | [diff] [blame] | 241 | assertEquals(1, kr.getTotalResults()); |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 242 | assertEquals(2, kr.getMatch(0).getLocalDocID()); |
| 243 | assertEquals(2, kr.getMatch(0).startPos); |
| 244 | assertEquals(5, kr.getMatch(0).endPos); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 245 | |
| 246 | } |
| 247 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 248 | /** Same tokens in different elements */ |
| 249 | @Test |
| 250 | public void testCase4 () throws IOException { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 251 | ki = new KrillIndex(); |
| 252 | ki.addDoc(createFieldDoc0()); |
| Eliza Margaretha | 45b5be1 | 2014-02-04 11:22:46 +0000 | [diff] [blame] | 253 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 254 | |
| Eliza Margaretha | 45b5be1 | 2014-02-04 11:22:46 +0000 | [diff] [blame] | 255 | SpanQuery sq; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 256 | sq = createQuery("s", "s:b", "s:b", 1, 2, true); |
| Eliza Margaretha | 45b5be1 | 2014-02-04 11:22:46 +0000 | [diff] [blame] | 257 | kr = ki.search(sq, (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 258 | |
| Marc Kupietz | 26e05d2 | 2025-09-24 13:59:26 +0200 | [diff] [blame] | 259 | assertEquals(2, kr.getTotalResults()); |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 260 | assertEquals(0, kr.getMatch(0).startPos); |
| 261 | assertEquals(4, kr.getMatch(0).endPos); |
| 262 | assertEquals(3, kr.getMatch(1).startPos); |
| 263 | assertEquals(5, kr.getMatch(1).endPos); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 264 | |
| Eliza Margaretha | 4026054 | 2014-10-01 10:32:09 +0000 | [diff] [blame] | 265 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 266 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 267 | /** Test query from json */ |
| 268 | @Test |
| 269 | public void testCase5 () throws Exception { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 270 | ki = new KrillIndex(); |
| 271 | ki.addDoc(getClass().getResourceAsStream("/wiki/00001.json.gz"), true); |
| Eliza Margaretha | 4026054 | 2014-10-01 10:32:09 +0000 | [diff] [blame] | 272 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 273 | |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 274 | String jsonPath = |
| 275 | getClass().getResource("/queries/cosmas1.json").getFile(); |
| Akron | d6f5f59 | 2018-06-19 15:58:16 +0200 | [diff] [blame] | 276 | SpanQueryWrapper sqwi = getJsonQuery(jsonPath); |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 277 | kr = ki.search(sqwi.toQuery(), (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 278 | |
| Nils Diewald | e1ecd5e | 2014-11-27 02:17:24 +0000 | [diff] [blame] | 279 | assertEquals((long) 3, kr.getTotalResults()); |
| Eliza Margaretha | 42b407b | 2014-10-01 10:36:33 +0000 | [diff] [blame] | 280 | assertEquals(14, kr.getMatch(0).startPos); |
| 281 | assertEquals(19, kr.getMatch(0).endPos); |
| 282 | assertEquals(30, kr.getMatch(1).startPos); |
| 283 | assertEquals(33, kr.getMatch(1).endPos); |
| Eliza Margaretha | 4026054 | 2014-10-01 10:32:09 +0000 | [diff] [blame] | 284 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 285 | |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 286 | @Test |
| 287 | public void testCQLAnd () throws Exception { |
| 288 | ki = new KrillIndex(); |
| 289 | ki.addDoc(getClass().getResourceAsStream("/wiki/00001.json.gz"), true); |
| 290 | ki.commit(); |
| 291 | |
| 292 | String jsonPath = getClass() |
| 293 | .getResource("/queries/distances/cql-and.jsonld").getFile(); |
| 294 | SpanQueryWrapper sqwi = getJsonQuery(jsonPath); |
| 295 | SpanQuery query = sqwi.toQuery(); |
| 296 | assertEquals( |
| 297 | "spanElementDistance(tokens:s:Buchstaben, tokens:s:Alphabet, " |
| 298 | + "[(base/s:s[0:0], notOrdered, notExcluded)])", |
| 299 | query.toString()); |
| 300 | kr = ki.search(sqwi.toQuery(), (short) 10); |
| 301 | |
| 302 | assertEquals((long) 1, kr.getTotalResults()); |
| 303 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 304 | |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 305 | /** Test query from json (2) */ |
| 306 | @Test |
| 307 | public void testCase6 () throws Exception { |
| 308 | ki = new KrillIndex(); |
| 309 | ki.addDoc(createFieldDoc0()); |
| 310 | ki.addDoc(createFieldDoc1()); |
| 311 | ki.addDoc(createFieldDoc2()); |
| 312 | ki.commit(); |
| 313 | |
| 314 | SpanQueryWrapper sqwi; |
| 315 | sqwi = new QueryBuilder("tokens").tag("base/s:t"); |
| 316 | |
| 317 | kr = ki.search(sqwi.toQuery(), (short) 10); |
| Marc Kupietz | 26e05d2 | 2025-09-24 13:59:26 +0200 | [diff] [blame] | 318 | assertEquals(1, kr.getTotalResults()); |
| Akron | f05fde6 | 2016-08-03 23:46:17 +0200 | [diff] [blame] | 319 | assertEquals("[[ecebdc]]", kr.getMatch(0).getSnippetBrackets()); |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 320 | |
| Akron | 67d2ff0 | 2018-06-19 10:51:16 +0200 | [diff] [blame] | 321 | String jsonPath = getClass() |
| 322 | .getResource("/queries/distances/in-same-t.jsonld").getFile(); |
| Akron | d6f5f59 | 2018-06-19 15:58:16 +0200 | [diff] [blame] | 323 | sqwi = getJsonQuery(jsonPath); |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 324 | |
| 325 | assertEquals( |
| 326 | "spanElementDistance(tokens:s:c, tokens:s:e, [(base/s:t[0:0], ordered, notExcluded)])", |
| 327 | sqwi.toQuery().toString()); |
| 328 | |
| Akron | c12567c | 2016-06-03 00:40:52 +0200 | [diff] [blame] | 329 | kr = ki.search(sqwi.toQuery(), (short) 10); |
| margaretha | 9614d97 | 2019-07-17 15:05:32 +0200 | [diff] [blame] | 330 | assertEquals(1, kr.getTotalResults()); // Is 1 correct or |
| 331 | // should it not be |
| 332 | // ordered? |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 333 | assertEquals("e[[ce]]bdc", kr.getMatch(0).getSnippetBrackets()); |
| margaretha | c66265c | 2016-12-14 13:48:45 +0100 | [diff] [blame] | 334 | } |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 335 | |
| 336 | @Test |
| 337 | public void testNoMoreFirstSpans () throws IOException { |
| 338 | ki = new KrillIndex(); |
| 339 | ki.addDoc(createFieldDoc3()); |
| 340 | ki.addDoc(createFieldDoc5()); |
| 341 | ki.commit(); |
| 342 | |
| 343 | // c or b |
| 344 | SpanOrQuery soq = new SpanOrQuery( |
| 345 | new SpanTermQuery(new Term("tokens", "s:c")), |
| 346 | new SpanTermQuery(new Term("tokens", "s:b"))); |
| 347 | |
| 348 | // (c or b) /s0 d |
| 349 | SpanElementQuery e = new SpanElementQuery("tokens", "s"); |
| 350 | SpanDistanceQuery sdq = new SpanDistanceQuery( |
| 351 | soq, |
| 352 | new SpanTermQuery(new Term("tokens", "s:d")), |
| 353 | new DistanceConstraint(e, 0, 0, true, false), true); |
| 354 | |
| 355 | kr = ki.search(sdq, (short) 10); |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 356 | assertEquals(2, kr.getTotalResults()); |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 357 | } |
| 358 | |
| 359 | @Test |
| 360 | public void testNoMoreSecondSpans () throws IOException { |
| 361 | ki = new KrillIndex(); |
| 362 | ki.addDoc(createFieldDoc3()); |
| 363 | ki.addDoc(createFieldDoc5()); |
| 364 | ki.commit(); |
| 365 | |
| 366 | // c or b |
| 367 | SpanOrQuery soq = new SpanOrQuery( |
| 368 | new SpanTermQuery(new Term("tokens", "s:c")), |
| 369 | new SpanTermQuery(new Term("tokens", "s:b"))); |
| 370 | |
| 371 | // d /s0(c or b) |
| 372 | SpanElementQuery e = new SpanElementQuery("tokens", "s"); |
| 373 | SpanDistanceQuery sdq = new SpanDistanceQuery( |
| 374 | new SpanTermQuery(new Term("tokens", "s:d")), |
| 375 | soq, |
| 376 | new DistanceConstraint(e, 0, 0, true, false), true); |
| 377 | |
| 378 | kr = ki.search(sdq, (short) 10); |
| 379 | assertEquals(0, kr.getTotalResults()); |
| 380 | } |
| 381 | |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 382 | |
| 383 | |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 384 | @Test |
| 385 | public void testNoElementSpans () throws IOException { |
| 386 | ki = new KrillIndex(); |
| 387 | ki.addDoc(createFieldDoc3()); |
| 388 | ki.addDoc(createFieldDoc5()); |
| 389 | ki.commit(); |
| 390 | |
| 391 | // c or e |
| 392 | SpanOrQuery soq = new SpanOrQuery( |
| 393 | new SpanTermQuery(new Term("tokens", "s:c")), |
| 394 | new SpanTermQuery(new Term("tokens", "s:e"))); |
| 395 | |
| 396 | // (c or e) /s0 d |
| 397 | SpanElementQuery e = new SpanElementQuery("tokens", "s"); |
| 398 | SpanDistanceQuery sdq = new SpanDistanceQuery( |
| 399 | soq, |
| 400 | new SpanTermQuery(new Term("tokens", "s:d")), |
| 401 | new DistanceConstraint(e, 0, 0, true, false), true); |
| 402 | |
| 403 | kr = ki.search(sdq, (short) 10); |
| margaretha | f420cb3 | 2023-02-14 17:49:33 +0100 | [diff] [blame] | 404 | assertEquals(1, kr.getTotalResults()); |
| 405 | } |
| 406 | |
| 407 | @Test |
| 408 | public void testNoMoreFirstSpanWithSpanOrQuery () throws IOException { |
| 409 | ki = new KrillIndex(); |
| 410 | ki.addDoc(createFieldDoc3()); |
| 411 | ki.addDoc(createFieldDoc5()); |
| 412 | ki.commit(); |
| 413 | |
| 414 | // c or d |
| 415 | SpanOrQuery soq = new SpanOrQuery( |
| 416 | new SpanTermQuery(new Term("tokens", "s:c")), |
| 417 | new SpanTermQuery(new Term("tokens", "s:d"))); |
| 418 | |
| 419 | // b /s0 (c or d) |
| 420 | SpanElementQuery e = new SpanElementQuery("tokens", "s"); |
| 421 | SpanDistanceQuery sdq = new SpanDistanceQuery( |
| 422 | new SpanTermQuery(new Term("tokens", "s:b")), |
| 423 | soq, |
| 424 | new DistanceConstraint(e, 0, 0, true, false), true); |
| 425 | |
| 426 | kr = ki.search(sdq, (short) 10); |
| 427 | assertEquals(1, kr.getTotalResults()); |
| 428 | } |
| 429 | |
| 430 | @Test |
| 431 | public void testNoMoreSecondSpansOrQuery () throws IOException { |
| 432 | ki = new KrillIndex(); |
| 433 | ki.addDoc(createFieldDoc6()); |
| 434 | ki.addDoc(createFieldDoc5()); |
| 435 | ki.addDoc(createFieldDoc7()); |
| 436 | ki.commit(); |
| 437 | |
| 438 | // c or d |
| 439 | SpanOrQuery soq = new SpanOrQuery( |
| 440 | new SpanTermQuery(new Term("tokens", "s:c")), |
| 441 | new SpanTermQuery(new Term("tokens", "s:d"))); |
| 442 | |
| 443 | // b /s0(c or d) |
| 444 | SpanElementQuery e = new SpanElementQuery("tokens", "s"); |
| 445 | SpanDistanceQuery sdq = new SpanDistanceQuery( |
| 446 | new SpanTermQuery(new Term("tokens", "s:b")), |
| 447 | soq, |
| 448 | new DistanceConstraint(e, 0, 0, true, false), true); |
| 449 | |
| 450 | kr = ki.search(sdq, (short) 10); |
| 451 | |
| 452 | assertEquals(2, kr.getTotalResults()); |
| 453 | |
| 454 | // System.out.println(kr.getTotalResults()); |
| 455 | // for (Match m : kr.getMatches()) { |
| 456 | // System.out.println(m.getSnippetBrackets()); |
| 457 | // } |
| margaretha | e3a166c | 2021-09-30 16:57:24 +0200 | [diff] [blame] | 458 | } |
| margaretha | c66265c | 2016-12-14 13:48:45 +0100 | [diff] [blame] | 459 | } |