|
| 1 | +/* |
| 2 | + * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one |
| 3 | + * or more contributor license agreements. Licensed under the Elastic License |
| 4 | + * 2.0; you may not use this file except in compliance with the Elastic License |
| 5 | + * 2.0. |
| 6 | + */ |
| 7 | + |
| 8 | +package org.elasticsearch.xpack.esql.plugin; |
| 9 | + |
| 10 | +import org.elasticsearch.ElasticsearchException; |
| 11 | +import org.elasticsearch.action.index.IndexRequest; |
| 12 | +import org.elasticsearch.action.support.WriteRequest; |
| 13 | +import org.elasticsearch.common.settings.Settings; |
| 14 | +import org.elasticsearch.xpack.esql.VerificationException; |
| 15 | +import org.elasticsearch.xpack.esql.action.AbstractEsqlIntegTestCase; |
| 16 | +import org.junit.Before; |
| 17 | + |
| 18 | +import java.util.List; |
| 19 | + |
| 20 | +import static org.elasticsearch.test.hamcrest.ElasticsearchAssertions.assertAcked; |
| 21 | +import static org.hamcrest.CoreMatchers.containsString; |
| 22 | + |
| 23 | +//@TestLogging(value = "org.elasticsearch.xpack.esql:TRACE,org.elasticsearch.compute:TRACE", reason = "debug") |
| 24 | +public class ChickenScoreFunctionIT extends AbstractEsqlIntegTestCase { |
| 25 | + |
| 26 | + @Before |
| 27 | + public void setupIndex() { |
| 28 | + createAndPopulateIndex(); |
| 29 | + } |
| 30 | + |
| 31 | + public void testSimpleChickenScoreWithWhereMatch() { |
| 32 | + var query = """ |
| 33 | + FROM test METADATA _score |
| 34 | + | EVAL first_score = chicken_score(match(content, "brown")) |
| 35 | + | WHERE match(content, "fox") |
| 36 | + | KEEP id, _score, first_score |
| 37 | + | SORT id |
| 38 | + """; |
| 39 | + |
| 40 | + try (var resp = run(query)) { |
| 41 | + assertColumnNames(resp.columns(), List.of("id", "_score", "first_score")); |
| 42 | + assertColumnTypes(resp.columns(), List.of("integer", "double", "double")); |
| 43 | + assertValues(resp.values(), List.of(List.of(1, 1.156558871269226, 0.2708943784236908), List.of(6, 0.9114001989364624, 0.21347221732139587))); |
| 44 | + } |
| 45 | + } |
| 46 | + |
| 47 | + public void testSimpleEvalScoreWithWhereMatch() { |
| 48 | + var query = """ |
| 49 | + FROM test METADATA _score |
| 50 | + | EVAL first_matching = match(content, "brown") |
| 51 | + | WHERE match(content, "fox") |
| 52 | + | KEEP id, _score, first_score |
| 53 | + | SORT id |
| 54 | + """; |
| 55 | + |
| 56 | + try (var resp = run(query)) { |
| 57 | + assertColumnNames(resp.columns(), List.of("id", "_score", "first_matching")); |
| 58 | + assertColumnTypes(resp.columns(), List.of("integer", "double", "boolean")); |
| 59 | + assertValues(resp.values(), List.of(List.of(1, 1.156558871269226, true), List.of(6, 0.9114001989364624, true))); |
| 60 | + } |
| 61 | + } |
| 62 | + |
| 63 | + /*public void testCombinedWhereMatch() { |
| 64 | + var query = """ |
| 65 | + FROM test |
| 66 | + | WHERE match(content, "fox") AND id > 5 |
| 67 | + | KEEP id |
| 68 | + | SORT id |
| 69 | + """; |
| 70 | +
|
| 71 | + try (var resp = run(query)) { |
| 72 | + assertColumnNames(resp.columns(), List.of("id")); |
| 73 | + assertColumnTypes(resp.columns(), List.of("integer")); |
| 74 | + assertValues(resp.values(), List.of(List.of(6))); |
| 75 | + } |
| 76 | + } |
| 77 | +
|
| 78 | + public void testMultipleMatch() { |
| 79 | + var query = """ |
| 80 | + FROM test |
| 81 | + | WHERE match(content, "fox") AND match(content, "brown") |
| 82 | + | KEEP id |
| 83 | + | SORT id |
| 84 | + """; |
| 85 | +
|
| 86 | + try (var resp = run(query)) { |
| 87 | + assertColumnNames(resp.columns(), List.of("id")); |
| 88 | + assertColumnTypes(resp.columns(), List.of("integer")); |
| 89 | + assertValues(resp.values(), List.of(List.of(1), List.of(6))); |
| 90 | + } |
| 91 | + } |
| 92 | +
|
| 93 | + public void testMultipleWhereMatch() { |
| 94 | + var query = """ |
| 95 | + FROM test |
| 96 | + | WHERE match(content, "fox") AND match(content, "brown") |
| 97 | + | EVAL summary = CONCAT("document with id: ", to_str(id), "and content: ", content) |
| 98 | + | SORT summary |
| 99 | + | LIMIT 4 |
| 100 | + | WHERE match(content, "brown fox") |
| 101 | + | KEEP id |
| 102 | + """; |
| 103 | +
|
| 104 | + var error = expectThrows(ElasticsearchException.class, () -> run(query)); |
| 105 | + assertThat(error.getMessage(), containsString("[MATCH] function cannot be used after LIMIT")); |
| 106 | + } |
| 107 | +
|
| 108 | + public void testNotWhereMatch() { |
| 109 | + var query = """ |
| 110 | + FROM test |
| 111 | + | WHERE NOT match(content, "brown fox") |
| 112 | + | KEEP id |
| 113 | + | SORT id |
| 114 | + """; |
| 115 | +
|
| 116 | + try (var resp = run(query)) { |
| 117 | + assertColumnNames(resp.columns(), List.of("id")); |
| 118 | + assertColumnTypes(resp.columns(), List.of("integer")); |
| 119 | + assertValues(resp.values(), List.of(List.of(5))); |
| 120 | + } |
| 121 | + } |
| 122 | +
|
| 123 | + public void testWhereMatchWithScoring() { |
| 124 | + var query = """ |
| 125 | + FROM test |
| 126 | + METADATA _score |
| 127 | + | WHERE match(content, "fox") |
| 128 | + | KEEP id, _score |
| 129 | + | SORT id ASC |
| 130 | + """; |
| 131 | +
|
| 132 | + try (var resp = run(query)) { |
| 133 | + assertColumnNames(resp.columns(), List.of("id", "_score")); |
| 134 | + assertColumnTypes(resp.columns(), List.of("integer", "double")); |
| 135 | + assertValues(resp.values(), List.of(List.of(1, 1.156558871269226), List.of(6, 0.9114001989364624))); |
| 136 | + } |
| 137 | + } |
| 138 | +
|
| 139 | + public void testWhereMatchWithScoringDifferentSort() { |
| 140 | +
|
| 141 | + var query = """ |
| 142 | + FROM test |
| 143 | + METADATA _score |
| 144 | + | WHERE match(content, "fox") |
| 145 | + | KEEP id, _score |
| 146 | + | SORT id DESC |
| 147 | + """; |
| 148 | +
|
| 149 | + try (var resp = run(query)) { |
| 150 | + assertColumnNames(resp.columns(), List.of("id", "_score")); |
| 151 | + assertColumnTypes(resp.columns(), List.of("integer", "double")); |
| 152 | + assertValues(resp.values(), List.of(List.of(6, 0.9114001989364624), List.of(1, 1.156558871269226))); |
| 153 | + } |
| 154 | + } |
| 155 | +
|
| 156 | + public void testWhereMatchWithScoringSortScore() { |
| 157 | + var query = """ |
| 158 | + FROM test |
| 159 | + METADATA _score |
| 160 | + | WHERE match(content, "fox") |
| 161 | + | KEEP id, _score |
| 162 | + | SORT _score DESC |
| 163 | + """; |
| 164 | +
|
| 165 | + try (var resp = run(query)) { |
| 166 | + assertColumnNames(resp.columns(), List.of("id", "_score")); |
| 167 | + assertColumnTypes(resp.columns(), List.of("integer", "double")); |
| 168 | + assertValues(resp.values(), List.of(List.of(1, 1.156558871269226), List.of(6, 0.9114001989364624))); |
| 169 | + } |
| 170 | + } |
| 171 | +
|
| 172 | + public void testWhereMatchWithScoringNoSort() { |
| 173 | + var query = """ |
| 174 | + FROM test |
| 175 | + METADATA _score |
| 176 | + | WHERE match(content, "fox") |
| 177 | + | KEEP id, _score |
| 178 | + """; |
| 179 | +
|
| 180 | + try (var resp = run(query)) { |
| 181 | + assertColumnNames(resp.columns(), List.of("id", "_score")); |
| 182 | + assertColumnTypes(resp.columns(), List.of("integer", "double")); |
| 183 | + assertValuesInAnyOrder(resp.values(), List.of(List.of(1, 1.156558871269226), List.of(6, 0.9114001989364624))); |
| 184 | + } |
| 185 | + } |
| 186 | +
|
| 187 | + public void testNonExistingColumn() { |
| 188 | + var query = """ |
| 189 | + FROM test |
| 190 | + | WHERE match(something, "fox") |
| 191 | + """; |
| 192 | +
|
| 193 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 194 | + assertThat(error.getMessage(), containsString("Unknown column [something]")); |
| 195 | + } |
| 196 | +
|
| 197 | + public void testWhereMatchEvalColumn() { |
| 198 | + var query = """ |
| 199 | + FROM test |
| 200 | + | EVAL upper_content = to_upper(content) |
| 201 | + | WHERE match(upper_content, "FOX") |
| 202 | + | KEEP id |
| 203 | + """; |
| 204 | +
|
| 205 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 206 | + assertThat( |
| 207 | + error.getMessage(), |
| 208 | + containsString("[MATCH] function cannot operate on [upper_content], which is not a field from an index mapping") |
| 209 | + ); |
| 210 | + } |
| 211 | +
|
| 212 | + public void testWhereMatchOverWrittenColumn() { |
| 213 | + var query = """ |
| 214 | + FROM test |
| 215 | + | DROP content |
| 216 | + | EVAL content = CONCAT("document with ID ", to_str(id)) |
| 217 | + | WHERE match(content, "document") |
| 218 | + """; |
| 219 | +
|
| 220 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 221 | + assertThat( |
| 222 | + error.getMessage(), |
| 223 | + containsString("[MATCH] function cannot operate on [content], which is not a field from an index mapping") |
| 224 | + ); |
| 225 | + } |
| 226 | +
|
| 227 | + public void testWhereMatchAfterStats() { |
| 228 | + var query = """ |
| 229 | + FROM test |
| 230 | + | STATS count(*) |
| 231 | + | WHERE match(content, "fox") |
| 232 | + """; |
| 233 | +
|
| 234 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 235 | + assertThat(error.getMessage(), containsString("Unknown column [content]")); |
| 236 | + } |
| 237 | +
|
| 238 | + public void testWhereMatchNotPushedDown() { |
| 239 | + var query = """ |
| 240 | + FROM test |
| 241 | + | WHERE match(content, "fox") OR length(content) < 20 |
| 242 | + | KEEP id |
| 243 | + | SORT id |
| 244 | + """; |
| 245 | +
|
| 246 | + try (var resp = run(query)) { |
| 247 | + assertColumnNames(resp.columns(), List.of("id")); |
| 248 | + assertColumnTypes(resp.columns(), List.of("integer")); |
| 249 | + assertValues(resp.values(), List.of(List.of(1), List.of(2), List.of(6))); |
| 250 | + } |
| 251 | + } |
| 252 | +
|
| 253 | + public void testWhereMatchWithRow() { |
| 254 | + var query = """ |
| 255 | + ROW content = "a brown fox" |
| 256 | + | WHERE match(content, "fox") |
| 257 | + """; |
| 258 | +
|
| 259 | + var error = expectThrows(ElasticsearchException.class, () -> run(query)); |
| 260 | + assertThat( |
| 261 | + error.getMessage(), |
| 262 | + containsString("line 2:15: [MATCH] function cannot operate on [content], which is not a field from an index mapping") |
| 263 | + ); |
| 264 | + } |
| 265 | +
|
| 266 | + public void testMatchWithinEval() { |
| 267 | + var query = """ |
| 268 | + FROM test |
| 269 | + | EVAL matches_query = match(content, "fox") |
| 270 | + """; |
| 271 | +
|
| 272 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 273 | + assertThat(error.getMessage(), containsString("[MATCH] function is only supported in WHERE commands")); |
| 274 | + }*/ |
| 275 | + |
| 276 | + private void createAndPopulateIndex() { |
| 277 | + var indexName = "test"; |
| 278 | + var client = client().admin().indices(); |
| 279 | + var CreateRequest = client.prepareCreate(indexName) |
| 280 | + .setSettings(Settings.builder().put("index.number_of_shards", 1)) |
| 281 | + .setMapping("id", "type=integer", "content", "type=text"); |
| 282 | + assertAcked(CreateRequest); |
| 283 | + client().prepareBulk() |
| 284 | + .add(new IndexRequest(indexName).id("1").source("id", 1, "content", "This is a brown fox")) |
| 285 | + .add(new IndexRequest(indexName).id("2").source("id", 2, "content", "This is a brown dog")) |
| 286 | + .add(new IndexRequest(indexName).id("3").source("id", 3, "content", "This dog is really brown")) |
| 287 | + .add(new IndexRequest(indexName).id("4").source("id", 4, "content", "The dog is brown but this document is very very long")) |
| 288 | + .add(new IndexRequest(indexName).id("5").source("id", 5, "content", "There is also a white cat")) |
| 289 | + .add(new IndexRequest(indexName).id("6").source("id", 6, "content", "The quick brown fox jumps over the lazy dog")) |
| 290 | + .setRefreshPolicy(WriteRequest.RefreshPolicy.IMMEDIATE) |
| 291 | + .get(); |
| 292 | + ensureYellow(indexName); |
| 293 | + } |
| 294 | +} |
0 commit comments