| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 1 | package de.ids_mannheim.korap.index; |
| 2 | |
| 3 | import static org.junit.Assert.assertEquals; |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 4 | import static org.junit.Assert.assertNotEquals; |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 5 | import static org.junit.Assert.assertTrue; |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 6 | |
| Akron | e370fd8 | 2021-10-18 19:13:30 +0200 | [diff] [blame] | 7 | import java.util.List; |
| 8 | |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 9 | import java.io.IOException; |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 10 | import java.nio.file.Files; |
| 11 | import java.nio.file.Path; |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 12 | import java.nio.file.Paths; |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 13 | |
| Akron | e370fd8 | 2021-10-18 19:13:30 +0200 | [diff] [blame] | 14 | import org.apache.lucene.index.LeafReaderContext; |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 15 | import org.apache.lucene.store.MMapDirectory; |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 16 | import org.junit.Ignore; |
| 17 | import org.junit.Test; |
| 18 | |
| 19 | import de.ids_mannheim.korap.KrillIndex; |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 20 | |
| 21 | public class TestIndexRevision { |
| 22 | |
| 23 | @Test |
| 24 | public void testIndexRevisionAdd () throws IOException { |
| 25 | KrillIndex ki = new KrillIndex(); |
| 26 | |
| 27 | assertEquals(ki.getFingerprint(),"null"); |
| 28 | |
| 29 | ki.addDoc(getClass().getResourceAsStream("/wiki/00001.json.gz"), true); |
| 30 | ki.commit(); |
| 31 | |
| 32 | String x1 = ki.getFingerprint(); |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 33 | assertEquals("ibtSULzKIMrfGAtES3GXRA==",x1); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 34 | |
| 35 | ki.addDoc(getClass().getResourceAsStream("/wiki/00002.json.gz"), true); |
| 36 | ki.addDoc(getClass().getResourceAsStream("/wiki/00003.json.gz"), true); |
| 37 | ki.addDoc(getClass().getResourceAsStream("/wiki/00004.json.gz"), true); |
| 38 | ki.commit(); |
| 39 | |
| 40 | String x2 = ki.getFingerprint(); |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 41 | assertEquals("0UIQZpZVfiGDD2leAq6YQA==",x2); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 42 | |
| 43 | ki.addDoc(getClass().getResourceAsStream("/wiki/00006.json.gz"), true); |
| 44 | ki.commit(); |
| 45 | |
| 46 | String x3 = ki.getFingerprint(); |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 47 | assertEquals("fS3GqnKynhPQ5wFyC9/XWw==",x3); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 48 | |
| 49 | |
| 50 | // Check if the same changes will have the same effect |
| 51 | KrillIndex ki2 = new KrillIndex(); |
| 52 | |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 53 | assertEquals("null",ki2.getFingerprint()); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 54 | |
| 55 | ki2.addDoc(getClass().getResourceAsStream("/wiki/00001.json.gz"), true); |
| 56 | ki2.commit(); |
| 57 | |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 58 | assertEquals(x1, ki2.getFingerprint()); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 59 | |
| 60 | ki2.addDoc(getClass().getResourceAsStream("/wiki/00002.json.gz"), true); |
| 61 | ki2.addDoc(getClass().getResourceAsStream("/wiki/00003.json.gz"), true); |
| 62 | ki2.addDoc(getClass().getResourceAsStream("/wiki/00004.json.gz"), true); |
| 63 | ki2.commit(); |
| 64 | |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 65 | assertEquals(x2, ki2.getFingerprint()); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 66 | |
| 67 | ki2.addDoc(getClass().getResourceAsStream("/wiki/00006.json.gz"), true); |
| 68 | ki2.commit(); |
| 69 | |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 70 | assertEquals(x3, ki2.getFingerprint()); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 71 | }; |
| 72 | |
| 73 | @Test |
| 74 | public void testIndexRevisionDel () throws IOException { |
| 75 | KrillIndex ki = new KrillIndex(); |
| 76 | |
| 77 | assertEquals(ki.getFingerprint(),"null"); |
| 78 | |
| 79 | ki.addDoc(getClass().getResourceAsStream("/wiki/00001.json.gz"), true); |
| 80 | ki.commit(); |
| 81 | |
| 82 | String x1 = ki.getFingerprint(); |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 83 | assertEquals("ibtSULzKIMrfGAtES3GXRA==",x1); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 84 | |
| 85 | assertTrue(ki.delDocs("title", "A")); |
| 86 | ki.commit(); |
| 87 | |
| 88 | String x2 = ki.getFingerprint(); |
| 89 | assertNotEquals(x1, x2); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 90 | }; |
| 91 | |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 92 | @Test |
| 93 | public void testIndexRevisionTempFile () throws IOException { |
| 94 | |
| 95 | Path tmpdir = Files.createTempDirectory("wiki"); |
| 96 | KrillIndex ki = new KrillIndex(new MMapDirectory(tmpdir)); |
| 97 | |
| 98 | assertEquals("null", ki.getFingerprint()); |
| 99 | |
| 100 | ki.addDoc(getClass().getResourceAsStream("/wiki/00001.json.gz"), true); |
| 101 | ki.commit(); |
| 102 | |
| 103 | ki.addDoc(getClass().getResourceAsStream("/wiki/00002.json.gz"), true); |
| 104 | ki.addDoc(getClass().getResourceAsStream("/wiki/00003.json.gz"), true); |
| 105 | ki.addDoc(getClass().getResourceAsStream("/wiki/00004.json.gz"), true); |
| 106 | ki.commit(); |
| 107 | |
| 108 | ki.addDoc(getClass().getResourceAsStream("/wiki/00006.json.gz"), true); |
| 109 | ki.commit(); |
| 110 | |
| 111 | assertTrue(ki.delDocs("title", "A")); |
| 112 | ki.commit(); |
| 113 | |
| 114 | assertEquals(false, ki.isReaderOpen()); |
| 115 | |
| 116 | String fingerp = "aoD2zQvZKa8oQPjFJlji1g=="; |
| 117 | assertEquals(fingerp, ki.getFingerprint()); |
| 118 | |
| 119 | assertEquals(true, ki.isReaderOpen()); |
| 120 | assertEquals(4, ki.numberOf("base", "documents")); |
| 121 | |
| 122 | assertEquals(fingerp, ki.getFingerprint()); |
| 123 | |
| 124 | ki.close(); |
| 125 | |
| 126 | // Reload index |
| 127 | ki = new KrillIndex(new MMapDirectory(tmpdir)); |
| 128 | |
| 129 | assertEquals(fingerp, ki.getFingerprint()); |
| 130 | |
| 131 | ki.close(); |
| 132 | }; |
| 133 | |
| Akron | e370fd8 | 2021-10-18 19:13:30 +0200 | [diff] [blame] | 134 | public void testIndexRevisionLeafTempFile () throws IOException { |
| 135 | |
| 136 | String x1, x2, x3; |
| 137 | |
| 138 | Path tmpdir = Files.createTempDirectory("wikileaf"); |
| 139 | KrillIndex ki = new KrillIndex(new MMapDirectory(tmpdir)); |
| 140 | |
| 141 | assertEquals("null", ki.getFingerprint()); |
| 142 | |
| 143 | ki.addDoc(getClass().getResourceAsStream("/wiki/00001.json.gz"), true); |
| 144 | ki.commit(); |
| 145 | |
| 146 | List<LeafReaderContext> contexts = ki.reader().leaves(); |
| 147 | assertEquals(1, contexts.size()); |
| 148 | x1 = contexts.get(0).reader().getCombinedCoreAndDeletesKey().toString(); |
| 149 | assertEquals("_0(5.0.0):c1", x1); |
| 150 | |
| 151 | ki.addDoc(getClass().getResourceAsStream("/wiki/00002.json.gz"), true); |
| 152 | ki.addDoc(getClass().getResourceAsStream("/wiki/00003.json.gz"), true); |
| 153 | ki.addDoc(getClass().getResourceAsStream("/wiki/00004.json.gz"), true); |
| 154 | ki.commit(); |
| 155 | |
| 156 | ki.addDoc(getClass().getResourceAsStream("/wiki/00006.json.gz"), true); |
| 157 | ki.commit(); |
| 158 | |
| 159 | contexts = ki.reader().leaves(); |
| 160 | assertEquals(3, contexts.size()); |
| 161 | x1 = contexts.get(0).reader().getCombinedCoreAndDeletesKey().toString(); |
| 162 | assertEquals("_0(5.0.0):c1", x1); |
| 163 | x2 = contexts.get(1).reader().getCombinedCoreAndDeletesKey().toString(); |
| 164 | assertEquals("_1(5.0.0):c3", x2); |
| 165 | x3 = contexts.get(2).reader().getCombinedCoreAndDeletesKey().toString(); |
| 166 | assertEquals("_2(5.0.0):c1", x3); |
| 167 | |
| 168 | assertTrue(ki.delDocs("title", "A (Band)")); |
| 169 | ki.commit(); |
| 170 | |
| 171 | assertEquals(false, ki.isReaderOpen()); |
| 172 | |
| 173 | contexts = ki.reader().leaves(); |
| 174 | assertEquals(3, contexts.size()); |
| 175 | x1 = contexts.get(0).reader().getCombinedCoreAndDeletesKey().toString(); |
| 176 | assertEquals("_0(5.0.0):c1", x1); |
| 177 | x2 = contexts.get(1).reader().getCombinedCoreAndDeletesKey().toString(); |
| 178 | assertEquals("_1(5.0.0):c3/1:delGen=1", x2); |
| 179 | x3 = contexts.get(2).reader().getCombinedCoreAndDeletesKey().toString(); |
| 180 | assertEquals("_2(5.0.0):c1", x3); |
| 181 | |
| 182 | String fingerp = "241/XHj/9ZxeO5Lm3zZ+iw=="; |
| 183 | assertEquals(fingerp, ki.getFingerprint()); |
| 184 | |
| 185 | ki.close(); |
| 186 | |
| 187 | // Reload index |
| 188 | ki = new KrillIndex(new MMapDirectory(tmpdir)); |
| 189 | |
| 190 | assertEquals(fingerp, ki.getFingerprint()); |
| 191 | |
| 192 | contexts = ki.reader().leaves(); |
| 193 | assertEquals(3, contexts.size()); |
| 194 | x1 = contexts.get(0).reader().getCombinedCoreAndDeletesKey().toString(); |
| 195 | assertEquals("_0(5.0.0):c1", x1); |
| 196 | x2 = contexts.get(1).reader().getCombinedCoreAndDeletesKey().toString(); |
| 197 | assertEquals("_1(5.0.0):c3/1:delGen=1", x2); |
| 198 | x3 = contexts.get(2).reader().getCombinedCoreAndDeletesKey().toString(); |
| 199 | assertEquals("_2(5.0.0):c1", x3); |
| 200 | |
| 201 | ki.close(); |
| 202 | }; |
| 203 | |
| 204 | |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 205 | @Ignore |
| 206 | public void testIndexRevisionSample () throws IOException { |
| 207 | KrillIndex ki = new KrillIndex(new MMapDirectory( |
| 208 | Paths.get(getClass().getResource("/sample-index").getFile()))); |
| 209 | |
| Akron | 412727b | 2021-10-18 18:46:41 +0200 | [diff] [blame] | 210 | assertEquals("Wes8Bd4h1OypPqbWF5njeQ==",ki.getFingerprint()); |
| Akron | d999162 | 2020-06-02 11:45:33 +0200 | [diff] [blame] | 211 | }; |
| 212 | }; |