| Eliza Margaretha | 67a8857 | 2014-11-04 14:38:56 +0000 | [diff] [blame] | 1 | package de.ids_mannheim.korap.highlight; |
| 2 | |
| 3 | import static org.junit.Assert.assertEquals; |
| 4 | import static org.junit.Assert.fail; |
| 5 | |
| 6 | import java.io.BufferedReader; |
| 7 | import java.io.FileReader; |
| 8 | import java.io.IOException; |
| 9 | |
| Eliza Margaretha | b698f82 | 2014-11-12 10:14:21 +0000 | [diff] [blame] | 10 | import org.apache.lucene.search.spans.SpanQuery; |
| Eliza Margaretha | 67a8857 | 2014-11-04 14:38:56 +0000 | [diff] [blame] | 11 | import org.junit.Test; |
| 12 | |
| Nils Diewald | a14ecd6 | 2015-02-26 21:00:20 +0000 | [diff] [blame] | 13 | import de.ids_mannheim.korap.KrillIndex; |
| Nils Diewald | 392bcf3 | 2015-02-26 20:01:17 +0000 | [diff] [blame] | 14 | import de.ids_mannheim.korap.response.Match; |
| Nils Diewald | 0339d46 | 2015-02-26 14:53:56 +0000 | [diff] [blame] | 15 | import de.ids_mannheim.korap.KrillQuery; |
| Nils Diewald | 884dbcf | 2015-02-27 17:02:28 +0000 | [diff] [blame] | 16 | import de.ids_mannheim.korap.response.Result; |
| Nils Diewald | bbd39a5 | 2015-02-23 19:56:57 +0000 | [diff] [blame] | 17 | import de.ids_mannheim.korap.Krill; |
| Eliza Margaretha | 67a8857 | 2014-11-04 14:38:56 +0000 | [diff] [blame] | 18 | import de.ids_mannheim.korap.query.SpanNextQuery; |
| 19 | import de.ids_mannheim.korap.query.wrap.SpanQueryWrapper; |
| 20 | import de.ids_mannheim.korap.util.QueryException; |
| 21 | |
| 22 | |
| 23 | public class TestClass { |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 24 | KrillIndex ki; |
| 25 | Result kr; |
| 26 | Krill ks; |
| Eliza Margaretha | 67a8857 | 2014-11-04 14:38:56 +0000 | [diff] [blame] | 27 | |
| Nils Diewald | 11e9186 | 2014-11-12 16:29:18 +0000 | [diff] [blame] | 28 | |
| Nils Diewald | 7d32064 | 2014-11-12 17:39:42 +0000 | [diff] [blame] | 29 | @Test |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 30 | public void queryJSONpoly1 () throws QueryException, IOException { |
| Nils Diewald | 7d32064 | 2014-11-12 17:39:42 +0000 | [diff] [blame] | 31 | |
| Nils Diewald | bb33da2 | 2015-03-04 16:24:25 +0000 | [diff] [blame] | 32 | String jsonPath = getClass().getResource("/queries/poly1.json") |
| 33 | .getFile(); |
| 34 | String jsonQuery = readFile(jsonPath); |
| 35 | SpanQueryWrapper sqwi = new KrillQuery("tokens").fromJson(jsonQuery); |
| 36 | |
| 37 | SpanNextQuery sq = (SpanNextQuery) sqwi.toQuery(); |
| 38 | //System.out.println(sq.toString()); |
| 39 | |
| 40 | ki = new KrillIndex(); |
| 41 | ki.addDoc(getClass().getResourceAsStream("/wiki/JJJ-00785.json.gz"), |
| 42 | true); |
| 43 | ki.addDoc(getClass().getResourceAsStream("/wiki/DDD-01402.json.gz"), |
| 44 | true); |
| 45 | ki.commit(); |
| 46 | kr = ki.search(sq, (short) 10); |
| 47 | |
| 48 | assertEquals(61, kr.getMatch(0).getStartPos()); |
| 49 | assertEquals(64, kr.getMatch(0).getEndPos()); |
| 50 | assertEquals( |
| 51 | "... Bruckner (Wien) und Mathis Lussy (Paris). [{1:Inspiriert} " |
| 52 | + "{2:durch die}] additiven Modelle arabischer Rhythmik (er half ...", |
| 53 | kr.getMatch(0).getSnippetBrackets()); |
| 54 | |
| 55 | assertEquals(31, kr.getMatch(1).getStartPos()); |
| 56 | assertEquals(34, kr.getMatch(1).getEndPos()); |
| 57 | assertEquals( |
| 58 | "... des Sendens wird ein unhörbarer Unterton [{1:mitgesendet}, " |
| 59 | + "{2:auf den}] das angesprochene Funkgerät reagiert. Die Abkürzung ...", |
| 60 | kr.getMatch(1).getSnippetBrackets()); |
| 61 | } |
| 62 | |
| 63 | |
| 64 | @Test |
| 65 | public void queryJSONpoly4 () throws QueryException, IOException { |
| 66 | |
| 67 | String jsonPath = getClass().getResource("/queries/poly4.json") |
| 68 | .getFile(); |
| 69 | String jsonQuery = readFile(jsonPath); |
| 70 | SpanQueryWrapper sqwi = new KrillQuery("tokens").fromJson(jsonQuery); |
| 71 | SpanQuery sq = sqwi.toQuery(); |
| 72 | |
| 73 | // System.out.println(sq.toString()); |
| 74 | |
| 75 | |
| 76 | ki = new KrillIndex(); |
| 77 | ki.addDoc(getClass().getResourceAsStream("/wiki/SSS-09803.json.gz"), |
| 78 | true); |
| 79 | |
| 80 | ki.commit(); |
| 81 | kr = ki.search(sq, (short) 10); |
| 82 | |
| 83 | /* |
| 84 | for (Match km : kr.getMatches()){ |
| 85 | System.out.println(km.getStartPos() +","+km.getEndPos()+" " |
| 86 | +km.getSnippetBrackets() |
| 87 | ); |
| 88 | } |
| 89 | */ |
| 90 | assertEquals((long) 495, kr.getTotalResults()); |
| 91 | assertEquals(3, kr.getMatch(0).getStartPos()); |
| 92 | assertEquals(5, kr.getMatch(0).getEndPos()); |
| 93 | |
| 94 | //fail("Tests have to be updated"); |
| 95 | } |
| 96 | |
| 97 | |
| 98 | private String readFile (String path) { |
| 99 | StringBuilder sb = new StringBuilder(); |
| 100 | try { |
| 101 | BufferedReader in = new BufferedReader(new FileReader(path)); |
| 102 | String str; |
| 103 | while ((str = in.readLine()) != null) { |
| 104 | sb.append(str); |
| 105 | }; |
| 106 | in.close(); |
| 107 | } |
| 108 | catch (IOException e) { |
| 109 | fail(e.getMessage()); |
| 110 | } |
| 111 | return sb.toString(); |
| 112 | } |
| Eliza Margaretha | 67a8857 | 2014-11-04 14:38:56 +0000 | [diff] [blame] | 113 | } |