| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 1 | package de.ids_mannheim.korap.index; |
| 2 | |
| Akron | d6f5f59 | 2018-06-19 15:58:16 +0200 | [diff] [blame] | 3 | import static de.ids_mannheim.korap.TestSimple.*; |
| margaretha | 71c66ee | 2015-12-11 14:39:55 +0100 | [diff] [blame] | 4 | import static org.junit.Assert.assertEquals; |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 5 | |
| 6 | import java.io.IOException; |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 7 | import java.nio.file.Paths; |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 8 | import java.util.ArrayList; |
| 9 | import java.util.List; |
| 10 | |
| 11 | import org.apache.lucene.index.Term; |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 12 | import org.apache.lucene.search.WildcardQuery; |
| 13 | import org.apache.lucene.search.spans.SpanMultiTermQueryWrapper; |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 14 | import org.apache.lucene.search.spans.SpanQuery; |
| 15 | import org.apache.lucene.search.spans.SpanTermQuery; |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 16 | import org.apache.lucene.store.MMapDirectory; |
| 17 | import org.junit.Ignore; |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 18 | import org.junit.Test; |
| 19 | import org.junit.runner.RunWith; |
| 20 | import org.junit.runners.JUnit4; |
| 21 | |
| margaretha | 953fd01 | 2017-09-04 16:33:39 +0200 | [diff] [blame] | 22 | import com.fasterxml.jackson.databind.JsonNode; |
| 23 | import com.fasterxml.jackson.databind.ObjectMapper; |
| 24 | |
| 25 | import de.ids_mannheim.korap.Krill; |
| 26 | import de.ids_mannheim.korap.KrillCollection; |
| Nils Diewald | a14ecd6 | 2015-02-26 21:00:20 +0000 | [diff] [blame] | 27 | import de.ids_mannheim.korap.KrillIndex; |
| margaretha | 953fd01 | 2017-09-04 16:33:39 +0200 | [diff] [blame] | 28 | import de.ids_mannheim.korap.KrillQuery; |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 29 | import de.ids_mannheim.korap.query.DistanceConstraint; |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 30 | import de.ids_mannheim.korap.query.SpanClassQuery; |
| Eliza Margaretha | 5f60692 | 2014-02-18 15:33:49 +0000 | [diff] [blame] | 31 | import de.ids_mannheim.korap.query.SpanDistanceQuery; |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 32 | import de.ids_mannheim.korap.query.SpanElementQuery; |
| 33 | import de.ids_mannheim.korap.query.SpanMultipleDistanceQuery; |
| 34 | import de.ids_mannheim.korap.query.SpanNextQuery; |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 35 | import de.ids_mannheim.korap.query.wrap.SpanQueryWrapper; |
| margaretha | 59b905f | 2017-09-05 11:15:48 +0200 | [diff] [blame] | 36 | import de.ids_mannheim.korap.response.Match; |
| margaretha | 71c66ee | 2015-12-11 14:39:55 +0100 | [diff] [blame] | 37 | import de.ids_mannheim.korap.response.Result; |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 38 | import de.ids_mannheim.korap.util.QueryException; |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 39 | |
| 40 | @RunWith(JUnit4.class) |
| 41 | public class TestMultipleDistanceIndex { |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 42 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 43 | private KrillIndex ki; |
| 44 | private Result kr; |
| 45 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 46 | public SpanQuery createQuery (String x, String y, |
| 47 | List<DistanceConstraint> constraints, boolean isOrdered) { |
| 48 | |
| 49 | SpanQuery sx = new SpanTermQuery(new Term("base", x)); |
| 50 | SpanQuery sy = new SpanTermQuery(new Term("base", y)); |
| 51 | |
| 52 | return new SpanMultipleDistanceQuery(sx, sy, constraints, isOrdered, |
| 53 | true); |
| 54 | } |
| 55 | |
| 56 | |
| margaretha | 3d81110 | 2017-09-06 14:20:42 +0200 | [diff] [blame] | 57 | public static DistanceConstraint createConstraint (String unit, int min, int max, |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 58 | boolean isOrdered, boolean exclusion) { |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 59 | return createConstraint("base", unit, min, max, isOrdered, exclusion); |
| 60 | } |
| 61 | |
| 62 | |
| margaretha | 3d81110 | 2017-09-06 14:20:42 +0200 | [diff] [blame] | 63 | public static DistanceConstraint createConstraint (String field, String unit, |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 64 | int min, int max, boolean isOrdered, boolean exclusion) { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 65 | |
| 66 | if (unit.equals("w")) { |
| 67 | return new DistanceConstraint(min, max, isOrdered, exclusion); |
| 68 | } |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 69 | return new DistanceConstraint(new SpanElementQuery(field, unit), min, |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 70 | max, isOrdered, exclusion); |
| 71 | } |
| 72 | |
| 73 | |
| 74 | private FieldDocument createFieldDoc0 () { |
| 75 | FieldDocument fd = new FieldDocument(); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 76 | fd.addString("ID", "doc-0"); |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 77 | fd.addTV("base", "text", |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 78 | "[(0-1)s:b|_1$<i>0<i>1|<>:s$<b>64<i>0<i>2<i>2<b>0|<>:p$<b>64<i>0<i>4<i>4<b>0]" |
| 79 | + "[(1-2)s:b|s:c|_2$<i>1<i>2]" |
| 80 | + "[(2-3)s:c|_3$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>4<b>0]" |
| 81 | + "[(3-4)s:b|_4$<i>3<i>4]" |
| 82 | + "[(4-5)s:c|_5$<i>4<i>5|<>:s$<b>64<i>4<i>6<i>6<b>0|<>:p$<b>64<i>4<i>6<i>6<b>0]" |
| 83 | + "[(5-6)s:e|_6$<i>5<i>6]"); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 84 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 85 | } |
| 86 | |
| 87 | |
| 88 | private FieldDocument createFieldDoc1 () { |
| 89 | FieldDocument fd = new FieldDocument(); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 90 | fd.addString("ID", "doc-1"); |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 91 | fd.addTV("base", "text", |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 92 | "[(0-1)s:c|_1$<i>0<i>1|<>:s$<b>64<i>0<i>2<i>2<b>0|<>:p$<b>64<i>0<i>4<i>4<b>0]" |
| 93 | + "[(1-2)s:c|s:e|_2$<i>1<i>2]" |
| 94 | + "[(2-3)s:e|_3$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>4<b>0]" |
| 95 | + "[(3-4)s:c|_4$<i>3<i>4]" |
| 96 | + "[(4-5)s:e|_5$<i>4<i>5|<>:s$<b>64<i>4<i>6<i>6<b>0|<>:p$<b>64<i>4<i>6<i>6<b>0]" |
| 97 | + "[(5-6)s:c|_6$<i>5<i>6]"); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 98 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 99 | } |
| 100 | |
| 101 | |
| 102 | private FieldDocument createFieldDoc2 () { |
| 103 | FieldDocument fd = new FieldDocument(); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 104 | fd.addString("ID", "doc-2"); |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 105 | fd.addTV("base", "text", |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 106 | "[(0-1)s:b|_1$<i>0<i>1|<>:s$<b>64<i>0<i>2<i>2<b>0|<>:p$<b>64<i>0<i>4<i>4<b>0]" |
| 107 | + "[(1-2)s:b|s:e|_2$<i>1<i>2]" |
| 108 | + "[(2-3)s:e|_3$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>4<b>0]" |
| 109 | + "[(3-4)s:b|s:c|_4$<i>3<i>4]" |
| 110 | + "[(4-5)s:e|_5$<i>4<i>5|<>:s$<b>64<i>4<i>6<i>6<b>0|<>:p$<b>64<i>4<i>6<i>6<b>0]" |
| 111 | + "[(5-6)s:d|_6$<i>5<i>6]" |
| 112 | + "[(6-7)s:b|_7$<i>6<i>7|<>:s$<b>64<i>6<i>7<i>7<b>0|<>:p$<b>64<i>6<i>7<i>7<b>0]"); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 113 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 114 | } |
| 115 | |
| 116 | |
| 117 | private FieldDocument createFieldDoc3 () { |
| 118 | FieldDocument fd = new FieldDocument(); |
| margaretha | 329e167 | 2017-06-20 15:04:24 +0200 | [diff] [blame] | 119 | fd.addString("ID", "doc-3"); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 120 | fd.addTV("base", "text", |
| margaretha | 4f99558 | 2015-12-14 14:14:34 +0100 | [diff] [blame] | 121 | "[(0-1)s:b|_1$<i>0<i>1|<>:s$<b>64<i>0<i>2<i>2<b>0|<>:p$<b>64<i>0<i>4<i>4<b>0]" |
| 122 | + "[(1-2)s:b|s:c|_2$<i>1<i>2]" |
| 123 | + "[(2-3)s:c|_3$<i>2<i>3|<>:s$<b>64<i>2<i>3<i>5<b>0]" |
| 124 | + "[(3-4)s:b|_4$<i>3<i>4]" + "[(4-5)s:b|_5$<i>4<i>5]" |
| 125 | + "[(5-6)s:b|_6$<i>5<i>6]" + // gap |
| 126 | "[(6-7)s:c|_7$<i>6<i>7|<>:s$<b>64<i>6<i>7<i>7<b>0|<>:p$<b>64<i>6<i>7<i>7<b>0]"); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 127 | return fd; |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 128 | } |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 129 | |
| 130 | |
| margaretha | 329e167 | 2017-06-20 15:04:24 +0200 | [diff] [blame] | 131 | private FieldDocument createFieldDoc4 () { |
| 132 | FieldDocument fd = new FieldDocument(); |
| 133 | fd.addString("ID", "doc-4"); |
| 134 | fd.addTV("base", "text", |
| 135 | "[(0-1)s:Zum|_1$<i>0<i>1|<>:s$<b>64<i>0<i>9<i>9<b>0]" |
| 136 | + "[(1-2)s:Begin|_2$<i>1<i>2]" |
| 137 | + "[(2-3)s:der|_3$<i>2<i>3]" |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 138 | + "[(3-4)s:Veranstaltung|_4$<i>3<i>4]" |
| margaretha | 329e167 | 2017-06-20 15:04:24 +0200 | [diff] [blame] | 139 | + "[(4-5)s:ruft|_5$<i>4<i>5]" |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 140 | + "[(5-6)s:der|_6$<i>5<i>6]" |
| margaretha | 329e167 | 2017-06-20 15:04:24 +0200 | [diff] [blame] | 141 | + "[(6-7)s:Moderator|_7$<i>6<i>7]" |
| 142 | + "[(7-8)s:die|_8$<i>7<i>8]" |
| 143 | + "[(8-9)s:Gäste|_9$<i>8<i>9]"); |
| 144 | return fd; |
| 145 | } |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 146 | |
| 147 | |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 148 | private FieldDocument createFieldDoc5 () { |
| 149 | FieldDocument fd = new FieldDocument(); |
| 150 | fd.addString("ID", "doc-5"); |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 151 | fd.addTV("tokens", "text", |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 152 | "[(0-1)s:Meine|_1$<i>0<i>1|<>:s$<b>64<i>0<i>9<i>9<b>0]" |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 153 | + "[(1-2)l:Erfahrung|_2$<i>1<i>2]" |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 154 | + "[(2-3)s:Meiner|_3$<i>2<i>3]" |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 155 | + "[(3-4)l:Erfahrung|_4$<i>3<i>4]" |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 156 | + "[(4-5)s:Mein|_5$<i>4<i>5]" |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 157 | + "[(5-6)l:Erfahrung|_6$<i>5<i>6]" |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 158 | + "[(6-7)s:Meinem|_7$<i>6<i>7]" |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 159 | + "[(7-8)l:Erfahrung|_8$<i>7<i>8]" |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 160 | + "[(8-9)s:Meinen|_9$<i>8<i>9]"); |
| 161 | return fd; |
| 162 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 163 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 164 | |
| 165 | private FieldDocument createFieldDoc6 () { |
| 166 | FieldDocument fd = new FieldDocument(); |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 167 | fd.addString("ID", "doc-6"); |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 168 | fd.addTV("tokens", "text", |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 169 | "[(0-1)s:Meine|_1$<i>0<i>1|<>:s$<b>64<i>0<i>5<i>5<b>0]" |
| 170 | + "[(1-2)s:Meiner|_2$<i>1<i>2]" |
| 171 | + "[(2-3)s:Mein|_3$<i>2<i>3]" |
| 172 | + "[(3-4)s:Meinem|_4$<i>3<i>4]" |
| 173 | + "[(4-5)s:Meinen|_5$<i>4<i>5]"); |
| 174 | return fd; |
| 175 | } |
| 176 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 177 | |
| 178 | private FieldDocument createFieldDoc7 () { |
| 179 | FieldDocument fd = new FieldDocument(); |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 180 | fd.addString("ID", "doc-7"); |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 181 | fd.addTV("tokens", "text", |
| 182 | "[(0-1)l:Erfahrung|_1$<i>0<i>1|<>:s$<b>64<i>0<i>4<i>4<b>0]" |
| 183 | + "[(1-2)l:Erfahrung|_2$<i>1<i>2]" |
| 184 | + "[(2-3)l:Erfahrung|_3$<i>2<i>3]" |
| 185 | + "[(3-4)l:Erfahrung|_4$<i>3<i>4]"); |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 186 | return fd; |
| 187 | } |
| 188 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 189 | |
| 190 | private FieldDocument createFieldDoc8 () { |
| 191 | FieldDocument fd = new FieldDocument(); |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 192 | fd.addString("ID", "doc-8"); |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 193 | fd.addTV("tokens", "text", |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 194 | "[(0-1)s:Meine|_1$<i>0<i>1|<>:s$<b>64<i>0<i>9<i>9<b>0]" |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 195 | + "[(1-2)l:Erfahrung|_2$<i>1<i>2]" |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 196 | + "[(2-3)s:Meiner|_3$<i>2<i>3]" |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 197 | + "[(3-4)l:Erfahrung|_4$<i>3<i>4]" |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 198 | + "[(4-5)s:Mein|_5$<i>4<i>5]" |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 199 | + "[(5-6)l:Erfahrung|_6$<i>5<i>6]" |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 200 | + "[(6-7)s:Meinem|_7$<i>6<i>7]" |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 201 | + "[(7-8)l:Erfahrung|_8$<i>7<i>8]" |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 202 | + "[(8-9)s:Meinen|_9$<i>8<i>9]"); |
| 203 | return fd; |
| 204 | } |
| 205 | |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 206 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 207 | @Test |
| 208 | public void testQueryWithWildCard () throws IOException { |
| 209 | // meine* /+w1:2,s0 &Erfahrung |
| 210 | ki = new KrillIndex(); |
| 211 | ki.addDoc(createFieldDoc5()); |
| 212 | ki.commit(); |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 213 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 214 | // Check simple rewriting |
| 215 | WildcardQuery wcquery = |
| 216 | new WildcardQuery(new Term("tokens", "s:Meine*")); |
| 217 | SpanMultiTermQueryWrapper<WildcardQuery> mtq = |
| 218 | new SpanMultiTermQueryWrapper<WildcardQuery>(wcquery); |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 219 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 220 | assertEquals(wcquery.toString(), "tokens:s:Meine*"); |
| Akron | b4ba430 | 2017-08-30 17:25:27 +0200 | [diff] [blame] | 221 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 222 | kr = ki.search(mtq, (short) 10); |
| 223 | assertEquals(4, kr.getMatches().size()); |
| 224 | assertEquals(0, kr.getMatch(0).getStartPos()); |
| 225 | assertEquals(1, kr.getMatch(0).getEndPos()); |
| Akron | 5782e59 | 2017-08-30 19:58:47 +0200 | [diff] [blame] | 226 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 227 | // Check rewriting in multidistance query |
| 228 | SpanQuery sq = new SpanTermQuery(new Term("tokens", "l:Erfahrung")); |
| 229 | kr = ki.search(sq, (short) 10); |
| 230 | assertEquals(4, kr.getMatches().size()); |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 231 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 232 | List<DistanceConstraint> constraints = |
| 233 | new ArrayList<DistanceConstraint>(); |
| Akron | 5782e59 | 2017-08-30 19:58:47 +0200 | [diff] [blame] | 234 | constraints.add(createConstraint("w", 1, 2, true, false)); |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 235 | constraints.add(createConstraint("tokens", "s", 0, 0, true, false)); |
| Akron | 5782e59 | 2017-08-30 19:58:47 +0200 | [diff] [blame] | 236 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 237 | SpanQuery mdsq = |
| 238 | new SpanMultipleDistanceQuery(mtq, sq, constraints, true, true); |
| 239 | assertEquals(mdsq.toString(), |
| 240 | "spanMultipleDistance(SpanMultiTermQueryWrapper(tokens:s:Meine*), " |
| 241 | + "tokens:l:Erfahrung, [(w[1:2], ordered, notExcluded), (s[0:0], " |
| 242 | + "ordered, notExcluded)])"); |
| Akron | e68b9a1 | 2017-08-30 20:07:16 +0200 | [diff] [blame] | 243 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 244 | kr = ki.search(mdsq, (short) 10); |
| 245 | assertEquals(3, kr.getMatches().size()); |
| 246 | assertEquals(0, kr.getMatch(0).getStartPos()); |
| 247 | assertEquals(2, kr.getMatch(0).getEndPos()); |
| 248 | |
| 249 | // Check skipping with multiple documents |
| 250 | ki.addDoc(createFieldDoc6()); |
| 251 | ki.addDoc(createFieldDoc7()); |
| 252 | ki.addDoc(createFieldDoc8()); |
| 253 | ki.commit(); |
| 254 | kr = ki.search(mdsq, (short) 10); |
| 255 | assertEquals(6, kr.getMatches().size()); |
| 256 | } |
| 257 | |
| Akron | d3e077f | 2017-09-04 18:58:12 +0200 | [diff] [blame] | 258 | |
| 259 | @Test |
| 260 | public void queryJSONwildcardNoFoundry () throws QueryException, IOException { |
| 261 | // meine* |
| 262 | ki = new KrillIndex(); |
| 263 | ki.addDoc(createFieldDoc5()); |
| 264 | ki.commit(); |
| 265 | |
| 266 | // treat merging gracefully |
| Akron | d6f5f59 | 2018-06-19 15:58:16 +0200 | [diff] [blame] | 267 | SpanQueryWrapper sqw = getJsonQuery( |
| Akron | d3e077f | 2017-09-04 18:58:12 +0200 | [diff] [blame] | 268 | getClass().getResource("/queries/bugs/cosmas_wildcards_missingfoundry.jsonld") |
| 269 | .getFile()); |
| 270 | SpanQuery sq = sqw.toQuery(); |
| 271 | assertEquals(sq.toString(),"SpanMultiTermQueryWrapper(tokens:l:Erfahr*)"); |
| 272 | |
| 273 | kr = ki.search(sq, (short) 10); |
| 274 | assertEquals(4, kr.getMatches().size()); |
| 275 | assertEquals(1, kr.getMatch(0).getStartPos()); |
| 276 | assertEquals(2, kr.getMatch(0).getEndPos()); |
| 277 | }; |
| 278 | |
| margaretha | 9880aec | 2017-08-31 16:58:07 +0200 | [diff] [blame] | 279 | |
| 280 | @Test |
| margaretha | 329e167 | 2017-06-20 15:04:24 +0200 | [diff] [blame] | 281 | public void testUnorderedTokenDistance () throws IOException { |
| 282 | ki = new KrillIndex(); |
| 283 | ki.addDoc(createFieldDoc4()); |
| 284 | ki.commit(); |
| 285 | |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 286 | List<DistanceConstraint> constraints = |
| 287 | new ArrayList<DistanceConstraint>(); |
| margaretha | 329e167 | 2017-06-20 15:04:24 +0200 | [diff] [blame] | 288 | constraints.add(createConstraint("w", 0, 5, true, false)); |
| 289 | constraints.add(createConstraint("s", 0, 0, true, false)); |
| 290 | |
| 291 | SpanQuery mdq; |
| 292 | mdq = createQuery("s:Begin", "s:Moderator", constraints, false); |
| 293 | kr = ki.search(mdq, (short) 10); |
| 294 | assertEquals(1, kr.getMatch(0).getStartPos()); |
| 295 | assertEquals(7, kr.getMatch(0).getEndPos()); |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 296 | |
| 297 | SpanQuery sq = new SpanDistanceQuery(mdq, |
| margaretha | 329e167 | 2017-06-20 15:04:24 +0200 | [diff] [blame] | 298 | new SpanTermQuery(new Term("base", "s:ruft")), |
| 299 | new DistanceConstraint(0, 0, false, false), true); |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 300 | |
| margaretha | 329e167 | 2017-06-20 15:04:24 +0200 | [diff] [blame] | 301 | kr = ki.search(sq, (short) 10); |
| 302 | assertEquals(1, kr.getMatch(0).getStartPos()); |
| 303 | assertEquals(7, kr.getMatch(0).getEndPos()); |
| 304 | } |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 305 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 306 | |
| 307 | /** |
| 308 | * Unordered, same sentence |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 309 | */ |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 310 | @Test |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 311 | public void testCase1 () throws IOException { |
| 312 | ki = new KrillIndex(); |
| 313 | ki.addDoc(createFieldDoc0()); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 314 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 315 | |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 316 | List<DistanceConstraint> constraints = |
| 317 | new ArrayList<DistanceConstraint>(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 318 | constraints.add(createConstraint("w", 0, 2, false, false)); |
| 319 | constraints.add(createConstraint("s", 0, 0, false, false)); |
| 320 | |
| 321 | SpanQuery mdq; |
| 322 | mdq = createQuery("s:b", "s:c", constraints, false); |
| 323 | kr = ki.search(mdq, (short) 10); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 324 | |
| 325 | assertEquals((long) 3, kr.getTotalResults()); |
| 326 | assertEquals(0, kr.getMatch(0).getStartPos()); |
| 327 | assertEquals(2, kr.getMatch(0).getEndPos()); |
| 328 | assertEquals(1, kr.getMatch(1).getStartPos()); |
| 329 | assertEquals(2, kr.getMatch(1).getEndPos()); |
| 330 | assertEquals(2, kr.getMatch(2).getStartPos()); |
| 331 | assertEquals(4, kr.getMatch(2).getEndPos()); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 332 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 333 | |
| 334 | |
| 335 | /** |
| 336 | * Ordered |
| 337 | * Unordered |
| 338 | * Two constraints |
| 339 | * Three constraints |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 340 | */ |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 341 | @Test |
| 342 | public void testCase2 () throws IOException { |
| 343 | ki = new KrillIndex(); |
| 344 | ki.addDoc(createFieldDoc0()); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 345 | ki.commit(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 346 | |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 347 | // Ordered - two constraints |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 348 | List<DistanceConstraint> constraints = |
| 349 | new ArrayList<DistanceConstraint>(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 350 | constraints.add(createConstraint("w", 0, 2, true, false)); |
| 351 | constraints.add(createConstraint("s", 1, 1, true, false)); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 352 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 353 | SpanQuery mdq; |
| 354 | mdq = createQuery("s:b", "s:c", constraints, true); |
| 355 | kr = ki.search(mdq, (short) 10); |
| 356 | assertEquals((long) 3, kr.getTotalResults()); |
| 357 | assertEquals(0, kr.getMatch(0).getStartPos()); |
| 358 | assertEquals(3, kr.getMatch(0).getEndPos()); |
| 359 | assertEquals(1, kr.getMatch(1).getStartPos()); |
| 360 | assertEquals(3, kr.getMatch(1).getEndPos()); |
| 361 | assertEquals(3, kr.getMatch(2).getStartPos()); |
| 362 | assertEquals(5, kr.getMatch(2).getEndPos()); |
| Eliza Margaretha | 5f60692 | 2014-02-18 15:33:49 +0000 | [diff] [blame] | 363 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 364 | // Three constraints |
| 365 | constraints.add(createConstraint("p", 0, 0, true, false)); |
| 366 | mdq = createQuery("s:b", "s:c", constraints, true); |
| 367 | kr = ki.search(mdq, (short) 10); |
| 368 | assertEquals((long) 2, kr.getTotalResults()); |
| 369 | |
| 370 | |
| 371 | // Unordered - two constraints |
| 372 | constraints.clear(); |
| 373 | constraints.add(createConstraint("w", 0, 2, false, false)); |
| 374 | constraints.add(createConstraint("s", 1, 1, false, false)); |
| 375 | |
| 376 | mdq = createQuery("s:c", "s:b", constraints, false); |
| 377 | kr = ki.search(mdq, (short) 10); |
| 378 | assertEquals((long) 4, kr.getTotalResults()); |
| 379 | assertEquals(1, kr.getMatch(2).getStartPos()); |
| 380 | assertEquals(4, kr.getMatch(2).getEndPos()); |
| 381 | |
| 382 | // Three constraints |
| 383 | constraints.add(createConstraint("p", 0, 0, false, false)); |
| 384 | mdq = createQuery("s:b", "s:c", constraints, false); |
| 385 | kr = ki.search(mdq, (short) 10); |
| 386 | assertEquals((long) 3, kr.getTotalResults()); |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 387 | |
| 388 | } |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 389 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 390 | |
| 391 | /** |
| 392 | * Multiple documents |
| 393 | * Ensure same doc (inner term span) |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 394 | */ |
| Eliza Margaretha | cdb769b | 2014-02-11 17:24:13 +0000 | [diff] [blame] | 395 | @Test |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 396 | public void testCase3 () throws IOException { |
| 397 | ki = new KrillIndex(); |
| 398 | ki.addDoc(createFieldDoc0()); |
| 399 | ki.addDoc(createFieldDoc1()); |
| 400 | ki.addDoc(createFieldDoc2()); |
| 401 | ki.commit(); |
| 402 | |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 403 | List<DistanceConstraint> constraints = |
| 404 | new ArrayList<DistanceConstraint>(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 405 | constraints.add(createConstraint("w", 1, 2, false, false)); |
| 406 | constraints.add(createConstraint("s", 1, 2, false, false)); |
| 407 | |
| 408 | SpanQuery mdq; |
| 409 | mdq = createQuery("s:b", "s:e", constraints, false); |
| 410 | kr = ki.search(mdq, (short) 10); |
| 411 | |
| 412 | assertEquals((long) 5, kr.getTotalResults()); |
| 413 | assertEquals(3, kr.getMatch(0).getStartPos()); |
| 414 | assertEquals(6, kr.getMatch(0).getEndPos()); |
| 415 | assertEquals(2, kr.getMatch(1).getLocalDocID()); |
| 416 | assertEquals(1, kr.getMatch(2).getStartPos()); |
| 417 | assertEquals(4, kr.getMatch(2).getEndPos()); |
| 418 | assertEquals(3, kr.getMatch(3).getStartPos()); |
| 419 | assertEquals(5, kr.getMatch(3).getEndPos()); |
| 420 | assertEquals(4, kr.getMatch(4).getStartPos()); |
| 421 | assertEquals(7, kr.getMatch(4).getEndPos()); |
| 422 | |
| margaretha | 59b905f | 2017-09-05 11:15:48 +0200 | [diff] [blame] | 423 | // System.out.print(kr.getTotalResults()+"\n"); |
| 424 | // for (int i=0; i< kr.getTotalResults(); i++){ |
| 425 | // System.out.println( |
| 426 | // kr.match(i).getLocalDocID()+" "+ |
| 427 | // kr.match(i).startPos + " " + |
| 428 | // kr.match(i).endPos |
| 429 | // ); |
| 430 | // } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 431 | |
| Eliza Margaretha | 0192918 | 2014-02-19 11:48:59 +0000 | [diff] [blame] | 432 | } |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 433 | |
| 434 | |
| 435 | /** |
| 436 | * Skip to |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 437 | */ |
| Eliza Margaretha | e335beb | 2014-02-27 12:56:14 +0000 | [diff] [blame] | 438 | @Test |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 439 | public void testCase4 () throws IOException { |
| 440 | ki = new KrillIndex(); |
| 441 | ki.addDoc(createFieldDoc0()); |
| 442 | ki.addDoc(createFieldDoc3()); |
| 443 | ki.addDoc(createFieldDoc1()); |
| 444 | ki.addDoc(createFieldDoc2()); |
| 445 | ki.commit(); |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 446 | |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 447 | List<DistanceConstraint> constraints = |
| 448 | new ArrayList<DistanceConstraint>(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 449 | constraints.add(createConstraint("w", 1, 2, false, false)); |
| 450 | constraints.add(createConstraint("s", 1, 2, false, false)); |
| Eliza Margaretha | e335beb | 2014-02-27 12:56:14 +0000 | [diff] [blame] | 451 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 452 | SpanQuery mdq; |
| 453 | mdq = createQuery("s:b", "s:c", constraints, false); |
| 454 | |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 455 | SpanQuery sq = new SpanNextQuery(mdq, |
| 456 | new SpanTermQuery(new Term("base", "s:e"))); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 457 | kr = ki.search(sq, (short) 10); |
| 458 | |
| 459 | assertEquals((long) 2, kr.getTotalResults()); |
| 460 | assertEquals(3, kr.getMatch(0).getStartPos()); |
| 461 | assertEquals(6, kr.getMatch(0).getEndPos()); |
| 462 | assertEquals(3, kr.getMatch(1).getLocalDocID()); |
| 463 | assertEquals(1, kr.getMatch(1).getStartPos()); |
| 464 | assertEquals(5, kr.getMatch(1).getEndPos()); |
| 465 | |
| 466 | } |
| 467 | |
| 468 | |
| 469 | /** |
| margaretha | 3512087 | 2016-12-19 18:24:22 +0100 | [diff] [blame] | 470 | * Same tokens: unordered yields twice the same results as ordered |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 471 | */ |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 472 | @Test |
| 473 | public void testCase5 () throws IOException { |
| 474 | ki = new KrillIndex(); |
| 475 | ki.addDoc(createFieldDoc0()); |
| 476 | ki.addDoc(createFieldDoc1()); |
| 477 | ki.commit(); |
| 478 | |
| margaretha | 3512087 | 2016-12-19 18:24:22 +0100 | [diff] [blame] | 479 | // ordered |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 480 | List<DistanceConstraint> constraints = |
| 481 | new ArrayList<DistanceConstraint>(); |
| margaretha | 3512087 | 2016-12-19 18:24:22 +0100 | [diff] [blame] | 482 | constraints.add(createConstraint("w", 1, 2, true, false)); |
| 483 | constraints.add(createConstraint("s", 1, 2, true, false)); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 484 | |
| 485 | SpanQuery mdq; |
| 486 | mdq = createQuery("s:c", "s:c", constraints, false); |
| 487 | kr = ki.search(mdq, (short) 10); |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 488 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 489 | assertEquals((long) 4, kr.getTotalResults()); |
| 490 | assertEquals(1, kr.getMatch(0).getStartPos()); |
| 491 | assertEquals(3, kr.getMatch(0).getEndPos()); |
| 492 | assertEquals(2, kr.getMatch(1).getStartPos()); |
| 493 | assertEquals(5, kr.getMatch(1).getEndPos()); |
| 494 | assertEquals(1, kr.getMatch(2).getLocalDocID()); |
| 495 | assertEquals(1, kr.getMatch(2).getStartPos()); |
| 496 | assertEquals(4, kr.getMatch(2).getEndPos()); |
| 497 | assertEquals(3, kr.getMatch(3).getStartPos()); |
| 498 | assertEquals(6, kr.getMatch(3).getEndPos()); |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 499 | |
| margaretha | 3512087 | 2016-12-19 18:24:22 +0100 | [diff] [blame] | 500 | //unordered |
| 501 | constraints = new ArrayList<DistanceConstraint>(); |
| 502 | constraints.add(createConstraint("w", 1, 2, false, false)); |
| 503 | constraints.add(createConstraint("s", 1, 2, false, false)); |
| 504 | |
| 505 | mdq = createQuery("s:c", "s:c", constraints, false); |
| 506 | kr = ki.search(mdq, (short) 10); |
| 507 | assertEquals((long) 8, kr.getTotalResults()); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 508 | |
| 509 | } |
| 510 | |
| 511 | |
| 512 | /** |
| 513 | * Exclusion |
| 514 | * Gaps |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 515 | */ |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 516 | @Test |
| 517 | public void testCase6 () throws IOException { |
| 518 | ki = new KrillIndex(); |
| 519 | ki.addDoc(createFieldDoc3()); |
| 520 | ki.commit(); |
| 521 | |
| 522 | // First constraint - token exclusion |
| 523 | SpanQuery sx = new SpanTermQuery(new Term("base", "s:b")); |
| 524 | SpanQuery sy = new SpanTermQuery(new Term("base", "s:c")); |
| 525 | |
| 526 | DistanceConstraint dc1 = createConstraint("w", 0, 1, false, true); |
| 527 | SpanDistanceQuery sq = new SpanDistanceQuery(sx, sy, dc1, true); |
| 528 | |
| 529 | kr = ki.search(sq, (short) 10); |
| 530 | assertEquals((long) 1, kr.getTotalResults()); |
| 531 | // 4-5 |
| 532 | |
| 533 | // Second constraint - element distance |
| 534 | DistanceConstraint dc2 = createConstraint("s", 1, 1, false, false); |
| 535 | sq = new SpanDistanceQuery(sx, sy, dc2, true); |
| 536 | kr = ki.search(sq, (short) 10); |
| 537 | // 0-3, 1-3, 1-4, 1-5, 3-7, 4-7 |
| 538 | assertEquals((long) 6, kr.getTotalResults()); |
| 539 | |
| 540 | |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 541 | List<DistanceConstraint> constraints = |
| 542 | new ArrayList<DistanceConstraint>(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 543 | constraints.add(dc1); |
| 544 | constraints.add(dc2); |
| 545 | |
| 546 | SpanQuery mdq; |
| 547 | mdq = createQuery("s:b", "s:c", constraints, false); |
| 548 | kr = ki.search(mdq, (short) 10); |
| 549 | |
| 550 | assertEquals((long) 2, kr.getTotalResults()); |
| 551 | assertEquals(1, kr.getMatch(0).getStartPos()); |
| 552 | assertEquals(5, kr.getMatch(0).getEndPos()); |
| 553 | assertEquals(4, kr.getMatch(1).getStartPos()); |
| 554 | assertEquals(7, kr.getMatch(1).getEndPos()); |
| 555 | } |
| 556 | |
| 557 | |
| 558 | /** |
| 559 | * Exclusion, multiple documents |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 560 | */ |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 561 | @Test |
| 562 | public void testCase7 () throws IOException { |
| 563 | ki = new KrillIndex(); |
| 564 | ki.addDoc(createFieldDoc2()); |
| 565 | ki.commit(); |
| 566 | |
| 567 | SpanQuery sx = new SpanTermQuery(new Term("base", "s:b")); |
| 568 | SpanQuery sy = new SpanTermQuery(new Term("base", "s:c")); |
| 569 | // Second constraint |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 570 | SpanDistanceQuery sq = new SpanDistanceQuery(sx, sy, |
| 571 | createConstraint("s", 0, 0, false, true), true); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 572 | kr = ki.search(sq, (short) 10); |
| 573 | assertEquals((long) 3, kr.getTotalResults()); |
| 574 | // 0-1, 1-2, 6-7 |
| 575 | |
| 576 | // Exclusion within the same sentence |
| margaretha | f14f380 | 2017-08-30 16:34:01 +0200 | [diff] [blame] | 577 | List<DistanceConstraint> constraints = |
| 578 | new ArrayList<DistanceConstraint>(); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 579 | constraints.add(createConstraint("w", 0, 2, false, true)); |
| 580 | constraints.add(createConstraint("s", 0, 0, false, true)); |
| 581 | |
| 582 | SpanQuery mdq; |
| 583 | mdq = createQuery("s:b", "s:c", constraints, false); |
| 584 | kr = ki.search(mdq, (short) 10); |
| 585 | assertEquals((long) 2, kr.getTotalResults()); |
| 586 | assertEquals(0, kr.getMatch(0).getStartPos()); |
| 587 | assertEquals(1, kr.getMatch(0).getEndPos()); |
| 588 | assertEquals(6, kr.getMatch(1).getStartPos()); |
| 589 | assertEquals(7, kr.getMatch(1).getEndPos()); |
| 590 | |
| 591 | |
| 592 | // Third constraint |
| Eliza Margaretha | 6f98920 | 2016-10-14 21:48:29 +0200 | [diff] [blame] | 593 | sq = new SpanDistanceQuery(sx, sy, |
| 594 | createConstraint("p", 0, 0, false, true), true); |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 595 | kr = ki.search(sq, (short) 10); |
| 596 | assertEquals((long) 1, kr.getTotalResults()); |
| 597 | // 6-7 |
| 598 | |
| 599 | constraints.add(createConstraint("p", 0, 0, false, true)); |
| 600 | mdq = createQuery("s:b", "s:c", constraints, false); |
| 601 | kr = ki.search(mdq, (short) 10); |
| 602 | |
| 603 | assertEquals((long) 1, kr.getTotalResults()); |
| 604 | assertEquals(6, kr.getMatch(0).getStartPos()); |
| 605 | assertEquals(7, kr.getMatch(0).getEndPos()); |
| 606 | |
| 607 | } |
| Eliza Margaretha | e18d62e | 2014-02-11 11:30:48 +0000 | [diff] [blame] | 608 | } |