| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 1 | package de.ids_mannheim.korap.index; |
| 2 | |
| 3 | import static org.junit.Assert.*; |
| 4 | |
| 5 | import java.io.IOException; |
| 6 | |
| 7 | import org.apache.lucene.index.Term; |
| 8 | import org.apache.lucene.search.spans.SpanOrQuery; |
| 9 | import org.apache.lucene.search.spans.SpanQuery; |
| 10 | import org.apache.lucene.search.spans.SpanTermQuery; |
| 11 | import org.junit.Test; |
| 12 | |
| 13 | import de.ids_mannheim.korap.KorapIndex; |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 14 | import de.ids_mannheim.korap.KorapMatch; |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 15 | import de.ids_mannheim.korap.KorapResult; |
| 16 | import de.ids_mannheim.korap.query.SpanNextQuery; |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 17 | import de.ids_mannheim.korap.query.SpanRepetitionQuery; |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 18 | |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 19 | public class TestRepetitionIndex { |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 20 | |
| 21 | private KorapIndex ki; |
| 22 | private KorapResult kr; |
| 23 | |
| 24 | private FieldDocument createFieldDoc0(){ |
| 25 | FieldDocument fd = new FieldDocument(); |
| 26 | fd.addString("ID", "doc-0"); |
| 27 | fd.addTV("base", |
| 28 | "text", |
| 29 | "[(0-1)s:c|_1#0-1]" + |
| 30 | "[(1-2)s:e|_2#1-2]" + |
| 31 | "[(2-3)s:c|_3#2-3|<>:y#2-4$<i>4]" + |
| 32 | "[(3-4)s:c|s:b|_4#3-4|<>:x#3-7$<i>7]" + |
| 33 | "[(4-5)s:e|s:d|_5#4-5|<>:y#4-6$<i>6]" + |
| 34 | "[(5-6)s:c|_6#5-6|<>:y#5-8$<i>8]" + |
| 35 | "[(6-7)s:d|_7#6-7]" + |
| 36 | "[(7-8)s:e|_8#7-8|<>:x#7-9$<i>9]" + |
| 37 | "[(8-9)s:e|s:b|_9#8-9|<>:x#8-10$<i>10]" + |
| 38 | "[(9-10)s:d|_10#9-10]"); |
| 39 | return fd; |
| 40 | } |
| 41 | |
| 42 | private FieldDocument createFieldDoc1() { |
| 43 | FieldDocument fd = new FieldDocument(); |
| 44 | fd.addString("ID", "doc-1"); |
| 45 | fd.addTV("base", |
| 46 | "text", |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 47 | "[(0-1)s:b|_1#0-1]" + |
| 48 | "[(1-2)s:e|_2#1-2]" + |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 49 | "[(2-3)s:c|_3#2-3]" + |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 50 | "[(3-4)s:c|s:d]" + |
| 51 | "[(4-5)s:d|s:c|_5#4-5]" + |
| 52 | "[(5-6)s:e|s:c|_6#5-6]" + |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 53 | "[(6-7)s:e|_7#6-7]" + |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 54 | "[(7-8)s:c|_8#7-8]" + |
| 55 | "[(8-9)s:d|_9#8-9]" + |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 56 | "[(9-10)s:d|_10#9-10]"); |
| 57 | return fd; |
| 58 | } |
| 59 | |
| 60 | private FieldDocument createFieldDoc2() { |
| 61 | FieldDocument fd = new FieldDocument(); |
| 62 | fd.addString("ID", "doc-2"); |
| 63 | fd.addTV("base", |
| 64 | "text", |
| 65 | "[(0-1)s:b|s:c|_1#0-1|<>:s#0-2$<i>1]" + |
| 66 | "[(1-2)s:c|_2#1-2]" + |
| 67 | "[(2-3)s:b|_3#2-3|<>:s#2-3$<i>3]" + |
| 68 | "[(3-4)s:c|_4#3-4|<>:s#3-4$<i>4]" + |
| 69 | "[(4-5)s:c|_5#4-5|<>:s#4-5$<i>5]" + |
| 70 | "[(5-6)s:b|_6#5-6]" + |
| 71 | "[(6-7)s:c|_7#6-7|<>:s#6-7$<i>7]"); |
| 72 | return fd; |
| 73 | } |
| 74 | |
| 75 | private FieldDocument createFieldDoc3() { |
| 76 | FieldDocument fd = new FieldDocument(); |
| 77 | fd.addString("ID", "doc-3"); |
| 78 | fd.addTV("base", |
| 79 | "text", |
| 80 | "[(0-1)s:a|_1#0-1|<>:s#0-2$<i>1]" + |
| 81 | "[(1-2)s:d|_2#1-2|<>:s#1-2$<i>3]" + |
| 82 | "[(2-3)s:e|_3#2-3]"); |
| 83 | return fd; |
| 84 | } |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 85 | |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 86 | @Test |
| 87 | public void testCase1() throws IOException{ |
| 88 | ki = new KorapIndex(); |
| 89 | ki.addDoc(createFieldDoc0()); |
| 90 | ki.commit(); |
| 91 | |
| 92 | SpanQuery sq, sq2; |
| 93 | // Quantifier only |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 94 | // c{1,2} |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 95 | sq = new SpanRepetitionQuery(new SpanTermQuery(new Term("base","s:c")),1,2, true); |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 96 | kr = ki.search(sq, (short) 10); |
| 97 | // 0-1, 2-3, 2-4, 3-4, 5-6 |
| 98 | assertEquals(5,kr.getTotalResults()); |
| 99 | |
| 100 | // ec{1,2} |
| 101 | sq = new SpanNextQuery( |
| 102 | new SpanTermQuery(new Term("base", "s:e")), |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 103 | new SpanRepetitionQuery(new SpanTermQuery(new Term("base","s:c")),1,2, true) |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 104 | ); |
| 105 | |
| 106 | kr = ki.search(sq, (short) 10); |
| 107 | // 1-3, 1-4, 4-6 |
| 108 | assertEquals(3,kr.getTotalResults()); |
| 109 | |
| 110 | // ec{1,2}d |
| 111 | sq2 = new SpanNextQuery(sq, new SpanTermQuery(new Term("base", "s:d"))); |
| 112 | kr = ki.search(sq2, (short) 10); |
| 113 | assertEquals(2,kr.getTotalResults()); |
| 114 | assertEquals(1, kr.getMatch(0).startPos); |
| 115 | assertEquals(5, kr.getMatch(0).endPos); |
| 116 | assertEquals(4, kr.getMatch(1).startPos); |
| 117 | assertEquals(7, kr.getMatch(1).endPos); |
| 118 | |
| 119 | // Multiple documents |
| 120 | ki.addDoc(createFieldDoc1()); |
| 121 | ki.commit(); |
| 122 | kr = ki.search(sq2, (short) 10); |
| 123 | assertEquals(5,kr.getTotalResults()); |
| 124 | } |
| 125 | |
| 126 | /** Skip to */ |
| 127 | @Test |
| 128 | public void testCase2() throws IOException{ |
| 129 | ki = new KorapIndex(); |
| 130 | ki.addDoc(createFieldDoc0()); |
| 131 | ki.addDoc(createFieldDoc3()); |
| 132 | ki.addDoc(createFieldDoc2()); |
| 133 | ki.addDoc(createFieldDoc1()); |
| 134 | ki.commit(); |
| 135 | |
| 136 | SpanQuery sq; |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 137 | // c{2,2} |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 138 | sq = new SpanRepetitionQuery(new SpanTermQuery(new Term("base","s:c")),2,2, true); |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 139 | kr = ki.search(sq, (short) 10); |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 140 | // doc1 2-4, 3-5, 4-6 |
| 141 | assertEquals(6,kr.getTotalResults()); |
| 142 | |
| 143 | // ec{2,2} |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 144 | kr = ki.search(sq, (short) 10); |
| 145 | sq = new SpanNextQuery( |
| 146 | new SpanTermQuery(new Term("base", "s:e")), |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 147 | new SpanRepetitionQuery(new SpanTermQuery(new Term("base","s:c")),2,2, true) |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 148 | ); |
| 149 | |
| 150 | kr = ki.search(sq, (short) 10); |
| 151 | assertEquals(2,kr.getTotalResults()); |
| 152 | assertEquals(3,kr.getMatch(1).getLocalDocID()); |
| 153 | |
| 154 | } |
| 155 | |
| 156 | /** OR */ |
| 157 | @Test |
| 158 | public void testCase3() throws IOException{ |
| 159 | ki = new KorapIndex(); |
| 160 | ki.addDoc(createFieldDoc0()); |
| 161 | ki.commit(); |
| 162 | |
| 163 | SpanQuery sq,sq2; |
| 164 | // ec{1,2} |
| 165 | sq = new SpanNextQuery( |
| 166 | new SpanTermQuery(new Term("base", "s:e")), |
| 167 | new SpanOrQuery( |
| Eliza Margaretha | d469346 | 2014-03-17 13:16:18 +0000 | [diff] [blame] | 168 | new SpanRepetitionQuery(new SpanTermQuery(new Term("base","s:c")),1,1, true), |
| 169 | new SpanRepetitionQuery(new SpanTermQuery(new Term("base","s:b")),1,1, true) |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 170 | ) |
| 171 | ); |
| 172 | kr = ki.search(sq, (short) 10); |
| 173 | assertEquals(3,kr.getTotalResults()); |
| 174 | assertEquals(1, kr.getMatch(0).startPos); |
| 175 | assertEquals(3, kr.getMatch(0).endPos); |
| 176 | assertEquals(4, kr.getMatch(1).startPos); |
| 177 | assertEquals(6, kr.getMatch(1).endPos); |
| 178 | assertEquals(7, kr.getMatch(2).startPos); |
| 179 | assertEquals(9, kr.getMatch(2).endPos); |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 180 | |
| 181 | } |
| 182 | |
| 183 | @Test |
| 184 | public void testCase4() throws IOException { |
| 185 | ki = new KorapIndex(); |
| 186 | ki.addDoc(createFieldDoc1()); |
| 187 | ki.commit(); |
| 188 | |
| 189 | SpanQuery sq; |
| 190 | // c{2,2} |
| 191 | sq = new SpanRepetitionQuery(new SpanTermQuery(new Term("base","s:c")),1,3, true); |
| 192 | kr = ki.search(sq, (short) 10); |
| 193 | // 2-3, 2-4, 2-5, 3-4, 3-5, 3-6, 4-5, 4-6, 5-6, 7-8 |
| 194 | assertEquals(10,kr.getTotalResults()); |
| 195 | |
| 196 | sq = new SpanRepetitionQuery(new SpanTermQuery(new Term("base","s:c")),2,3, true); |
| 197 | kr = ki.search(sq, (short) 10); |
| 198 | // 2-4, 2-5, 3-5, 3-6, 4-6 |
| 199 | assertEquals(5,kr.getTotalResults()); |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 200 | |
| 201 | // System.out.print(kr.getTotalResults()+"\n"); |
| 202 | // for (int i=0; i< kr.getTotalResults(); i++){ |
| 203 | // System.out.println( |
| 204 | // kr.match(i).getLocalDocID()+" "+ |
| 205 | // kr.match(i).startPos + " " + |
| 206 | // kr.match(i).endPos |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 207 | // ); |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 208 | // } |
| 209 | } |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 210 | |
| 211 | @Test |
| 212 | public void testCase5() throws IOException { |
| 213 | ki = new KorapIndex(); |
| Eliza Margaretha | faa548f | 2014-09-30 17:22:11 +0000 | [diff] [blame] | 214 | ki.addDocFile( |
| 215 | getClass().getResource("/wiki/00001.json.gz").getFile(), true |
| 216 | ); |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 217 | ki.commit(); |
| 218 | |
| 219 | SpanQuery sq0, sq1, sq2; |
| Eliza Margaretha | faa548f | 2014-09-30 17:22:11 +0000 | [diff] [blame] | 220 | sq0 = new SpanTermQuery(new Term("tokens", "tt/p:NN")); |
| 221 | sq1 = new SpanRepetitionQuery(new SpanTermQuery(new Term("tokens","tt/p:ADJA")),2,3, true); |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 222 | sq2 = new SpanNextQuery(sq1,sq0); |
| 223 | kr = ki.search(sq2, (short) 10); |
| 224 | |
| 225 | assertEquals(2,kr.getTotalResults()); |
| Eliza Margaretha | faa548f | 2014-09-30 17:22:11 +0000 | [diff] [blame] | 226 | assertEquals(73, kr.getMatch(0).getStartPos()); |
| 227 | assertEquals(77, kr.getMatch(0).getEndPos()); |
| 228 | assertEquals(74, kr.getMatch(1).getStartPos()); |
| 229 | assertEquals(77, kr.getMatch(1).getEndPos()); |
| 230 | /* for (KorapMatch km : kr.getMatches()){ |
| 231 | System.out.println(km.getSnippetBrackets()); |
| 232 | System.out.println(km.getStartPos() +","+km.getEndPos()); |
| 233 | }*/ |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 234 | |
| 235 | sq2 = new SpanNextQuery( |
| Eliza Margaretha | faa548f | 2014-09-30 17:22:11 +0000 | [diff] [blame] | 236 | new SpanTermQuery(new Term("tokens", "s:offenen")), |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 237 | sq2); |
| 238 | kr = ki.search(sq2, (short) 10); |
| 239 | |
| 240 | assertEquals(1,kr.getTotalResults()); |
| Eliza Margaretha | faa548f | 2014-09-30 17:22:11 +0000 | [diff] [blame] | 241 | assertEquals(73, kr.getMatch(0).getStartPos()); |
| 242 | assertEquals(77, kr.getMatch(0).getEndPos()); |
| 243 | /* |
| 244 | for (KorapMatch km : kr.getMatches()){ |
| Eliza Margaretha | 7788a98 | 2014-08-29 16:10:52 +0000 | [diff] [blame] | 245 | System.out.println(km.getSnippetBrackets()); |
| 246 | System.out.println(km.getStartPos() +","+km.getEndPos()); |
| 247 | }*/ |
| 248 | } |
| Eliza Margaretha | d28469f | 2014-03-10 12:42:21 +0000 | [diff] [blame] | 249 | } |