| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 1 | package de.ids_mannheim.korap.search; |
| 2 | |
| 3 | import static org.junit.Assert.assertEquals; |
| 4 | |
| 5 | import java.io.IOException; |
| 6 | import java.io.InputStream; |
| 7 | |
| 8 | import org.apache.commons.io.IOUtils; |
| 9 | import org.junit.Test; |
| 10 | |
| 11 | import com.fasterxml.jackson.databind.JsonNode; |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 12 | import com.fasterxml.jackson.databind.ObjectMapper; |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 13 | |
| 14 | import de.ids_mannheim.korap.Krill; |
| 15 | import de.ids_mannheim.korap.KrillIndex; |
| 16 | import de.ids_mannheim.korap.index.FieldDocument; |
| 17 | import de.ids_mannheim.korap.response.Response; |
| 18 | |
| 19 | public class TestVcField { |
| 20 | private KrillIndex ki; |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 21 | private ObjectMapper mapper = new ObjectMapper(); |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 22 | |
| 23 | private FieldDocument createFieldDoc (int uid, String textSigle) { |
| 24 | FieldDocument fd = new FieldDocument(); |
| 25 | fd.addString("textSigle", textSigle); |
| 26 | fd.setUID(uid); |
| 27 | return fd; |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 28 | } |
| 29 | |
| 30 | |
| 31 | public TestVcField () throws IOException { |
| 32 | ki = new KrillIndex(); |
| 33 | ki.addDoc(createFieldDoc(1, "WPD/AAA/00001")); |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 34 | |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 35 | FieldDocument fd = createFieldDoc(2, "WPD/AAA/00002"); |
| 36 | fd.addString("author", "Frank"); |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 37 | fd.addString("quote", "The \"quoted\" example"); |
| 38 | fd.addString("empty", ""); |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 39 | ki.addDoc(fd); |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 40 | |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 41 | fd = createFieldDoc(3, "WPD/AAA/00003"); |
| 42 | fd.addTV("tokens", "a b c", "[(0-1)s:a|i:a|_0$<i>0<i>1|-:t$<i>3]" |
| 43 | + "[(2-3)s:b|i:b|_1$<i>2<i>3]" + "[(4-5)s:c|i:c|_2$<i>4<i>5]"); |
| 44 | ki.addDoc(fd); |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 45 | |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 46 | ki.addDoc(createFieldDoc(4, "WPD/AAA/00004")); |
| 47 | ki.commit(); |
| 48 | ki.addDoc(createFieldDoc(5, "WPD/AAA/00005")); |
| 49 | ki.addDoc(createFieldDoc(6, "WPD/AAA/00006")); |
| 50 | ki.commit(); |
| 51 | } |
| 52 | |
| 53 | |
| 54 | private String getJsonResource (String file) throws IOException { |
| 55 | InputStream is = TestVcField.class.getResourceAsStream(file); |
| 56 | return IOUtils.toString(is, "utf-8"); |
| 57 | } |
| 58 | |
| 59 | |
| 60 | @Test |
| 61 | public void testRetrieveTextSiglesOfVc1 () throws IOException { |
| 62 | |
| 63 | String file = "/queries/collections/named-vcs/named-vc1.jsonld"; |
| 64 | String json = getJsonResource(file); |
| 65 | JsonNode n = new Krill().retrieveFieldValues(json, ki, "textSigle"); |
| 66 | |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 67 | assertEquals(Response.KORAL_VERSION, n.at("/@context").asText()); |
| 68 | assertEquals("koral:doc", n.at("/corpus/@type").asText()); |
| 69 | assertEquals("textSigle", n.at("/corpus/key").asText()); |
| 70 | assertEquals("type:string", n.at("/corpus/type").asText()); |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 71 | |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 72 | assertEquals("[\"WPD/AAA/00002\",\"WPD/AAA/00003\"]", |
| 73 | n.at("/corpus/value").toString()); |
| 74 | |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 75 | testRetrieveAuthorOfVc1(json); |
| 76 | testRetrieveTokensOfVc1(json); |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 77 | testRetrieveNullOfVc1(json); |
| 78 | testRetrieveQuoteOfVc1(json); |
| 79 | testRetrieveEmptyOfVc1(json); |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 80 | } |
| 81 | |
| 82 | |
| 83 | private void testRetrieveAuthorOfVc1 (String json) { |
| 84 | JsonNode n = new Krill().retrieveFieldValues(json, ki, "author"); |
| 85 | assertEquals("author", n.at("/corpus/key").asText()); |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 86 | assertEquals("[\"Frank\"]", n.at("/corpus/value").toString()); |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 87 | } |
| 88 | |
| 89 | |
| 90 | private void testRetrieveTokensOfVc1 (String json) { |
| 91 | JsonNode n = new Krill().retrieveFieldValues(json, ki, "tokens"); |
| 92 | assertEquals("tokens", n.at("/corpus/key").asText()); |
| margaretha | da237c9 | 2022-02-07 11:07:37 +0100 | [diff] [blame] | 93 | assertEquals("[]", n.at("/corpus/value").toString()); |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 94 | } |
| 95 | |
| 96 | private void testRetrieveNullOfVc1 (String json) { |
| 97 | JsonNode n = new Krill().retrieveFieldValues(json, ki, "hello"); |
| 98 | assertEquals("hello", n.at("/corpus/key").asText()); |
| 99 | assertEquals("[]", n.at("/corpus/value").toString()); |
| 100 | } |
| 101 | |
| 102 | private void testRetrieveQuoteOfVc1 (String json) { |
| 103 | JsonNode n = new Krill().retrieveFieldValues(json, ki, "quote"); |
| 104 | assertEquals("quote", n.at("/corpus/key").asText()); |
| 105 | assertEquals("[\"The \\\"quoted\\\" example\"]", n.at("/corpus/value").toString()); |
| 106 | } |
| 107 | |
| 108 | private void testRetrieveEmptyOfVc1 (String json) { |
| 109 | JsonNode n = new Krill().retrieveFieldValues(json, ki, "empty"); |
| 110 | assertEquals("empty", n.at("/corpus/key").asText()); |
| 111 | assertEquals("[]", n.at("/corpus/value").toString()); |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 112 | } |
| 113 | |
| 114 | |
| 115 | @Test |
| 116 | public void testRetrieveTextSiglesOfVc3 () throws IOException { |
| 117 | // uid 5000 is not in the index |
| 118 | String file = "/queries/collections/named-vcs/named-vc3.jsonld"; |
| 119 | JsonNode n = new Krill().retrieveFieldValues(getJsonResource(file), ki, |
| 120 | "textSigle"); |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 121 | |
| 122 | n = mapper.readTree(n.at("/corpus/value").toString()); |
| 123 | assertEquals(2, n.size()); |
| 124 | assertEquals("WPD/AAA/00002", n.at("/0").asText()); |
| 125 | assertEquals("WPD/AAA/00003", n.at("/1").asText()); |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 126 | } |
| 127 | |
| 128 | |
| 129 | @Test |
| 130 | public void testRetrieveTextSiglesOfVc4 () throws IOException { |
| 131 | |
| 132 | String file = "/queries/collections/named-vcs/named-vc4.jsonld"; |
| 133 | JsonNode n = new Krill().retrieveFieldValues(getJsonResource(file), ki, |
| 134 | "textSigle"); |
| margaretha | a4d543d | 2022-02-04 09:27:35 +0100 | [diff] [blame] | 135 | n = mapper.readTree(n.at("/corpus/value").toString()); |
| 136 | assertEquals(3, n.size()); |
| 137 | assertEquals("WPD/AAA/00001", n.at("/0").asText()); |
| 138 | assertEquals("WPD/AAA/00002", n.at("/1").asText()); |
| 139 | assertEquals("WPD/AAA/00003", n.at("/2").asText()); } |
| margaretha | 3827d53 | 2022-01-31 14:41:55 +0100 | [diff] [blame] | 140 | } |