|
| 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.hamcrest.Matchers; |
| 17 | +import org.junit.Before; |
| 18 | + |
| 19 | +import java.util.List; |
| 20 | + |
| 21 | +import static org.elasticsearch.test.hamcrest.ElasticsearchAssertions.assertAcked; |
| 22 | +import static org.elasticsearch.xpack.esql.EsqlTestUtils.getValuesList; |
| 23 | +import static org.hamcrest.CoreMatchers.containsString; |
| 24 | + |
| 25 | +//@TestLogging(value = "org.elasticsearch.xpack.esql:TRACE,org.elasticsearch.compute:TRACE", reason = "debug") |
| 26 | +public class MatchPhraseFunctionIT extends AbstractEsqlIntegTestCase { |
| 27 | + |
| 28 | + @Before |
| 29 | + public void setupIndex() { |
| 30 | + createAndPopulateIndex(); |
| 31 | + } |
| 32 | + |
| 33 | + public void testSimpleWhereMatchPhrase() { |
| 34 | + var query = """ |
| 35 | + FROM test |
| 36 | + | WHERE match_phrase(content, "brown fox") |
| 37 | + | KEEP id |
| 38 | + | SORT id |
| 39 | + """; |
| 40 | + |
| 41 | + try (var resp = run(query)) { |
| 42 | + assertColumnNames(resp.columns(), List.of("id")); |
| 43 | + assertColumnTypes(resp.columns(), List.of("integer")); |
| 44 | + assertValues(resp.values(), List.of(List.of(1), List.of(6))); |
| 45 | + } |
| 46 | + } |
| 47 | + |
| 48 | + public void testCombinedWhereMatchPhrase() { |
| 49 | + var query = """ |
| 50 | + FROM test |
| 51 | + | WHERE match_phrase(content, "brown fox") AND id > 5 |
| 52 | + | KEEP id |
| 53 | + | SORT id |
| 54 | + """; |
| 55 | + |
| 56 | + try (var resp = run(query)) { |
| 57 | + assertColumnNames(resp.columns(), List.of("id")); |
| 58 | + assertColumnTypes(resp.columns(), List.of("integer")); |
| 59 | + assertValues(resp.values(), List.of(List.of(6))); |
| 60 | + } |
| 61 | + } |
| 62 | + |
| 63 | + public void testMultipleMatchPhrase() { |
| 64 | + var query = """ |
| 65 | + FROM test |
| 66 | + | WHERE match_phrase(content, "the quick") AND match_phrase(content, "brown fox") |
| 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(1), List.of(6))); |
| 75 | + } |
| 76 | + } |
| 77 | + |
| 78 | + public void testMultipleWhereMatchPhrase() { |
| 79 | + var query = """ |
| 80 | + FROM test |
| 81 | + | WHERE match_phrase(content, "the quick") AND match_phrase(content, "brown fox") |
| 82 | + | EVAL summary = CONCAT("document with id: ", to_str(id), "and content: ", content) |
| 83 | + | SORT summary |
| 84 | + | LIMIT 4 |
| 85 | + | WHERE match_phrase(content, "lazy dog") |
| 86 | + | KEEP id |
| 87 | + """; |
| 88 | + |
| 89 | + var error = expectThrows(ElasticsearchException.class, () -> run(query)); |
| 90 | + assertThat(error.getMessage(), containsString("[MATCH_PHRASE] function cannot be used after LIMIT")); |
| 91 | + } |
| 92 | + |
| 93 | + public void testNotWhereMatchPhrase() { |
| 94 | + var query = """ |
| 95 | + FROM test |
| 96 | + | WHERE NOT match_phrase(content, "brown fox") |
| 97 | + | KEEP id |
| 98 | + | SORT id |
| 99 | + """; |
| 100 | + |
| 101 | + try (var resp = run(query)) { |
| 102 | + assertColumnNames(resp.columns(), List.of("id")); |
| 103 | + assertColumnTypes(resp.columns(), List.of("integer")); |
| 104 | + assertValues(resp.values(), List.of(List.of(5))); |
| 105 | + } |
| 106 | + } |
| 107 | + |
| 108 | + public void testWhereMatchPhraseWithScoring() { |
| 109 | + var query = """ |
| 110 | + FROM test |
| 111 | + METADATA _score |
| 112 | + | WHERE match_phrase(content, "brown fox") |
| 113 | + | KEEP id, _score |
| 114 | + | SORT id ASC |
| 115 | + """; |
| 116 | + |
| 117 | + try (var resp = run(query)) { |
| 118 | + assertColumnNames(resp.columns(), List.of("id", "_score")); |
| 119 | + assertColumnTypes(resp.columns(), List.of("integer", "double")); |
| 120 | + assertValues(resp.values(), List.of(List.of(1, 1.156558871269226), List.of(6, 0.9114001989364624))); |
| 121 | + } |
| 122 | + } |
| 123 | + |
| 124 | + public void testWhereMatchPhraseWithScoringDifferentSort() { |
| 125 | + |
| 126 | + var query = """ |
| 127 | + FROM test |
| 128 | + METADATA _score |
| 129 | + | WHERE match_phrase(content, "brown fox") |
| 130 | + | KEEP id, _score |
| 131 | + | SORT id DESC |
| 132 | + """; |
| 133 | + |
| 134 | + try (var resp = run(query)) { |
| 135 | + assertColumnNames(resp.columns(), List.of("id", "_score")); |
| 136 | + assertColumnTypes(resp.columns(), List.of("integer", "double")); |
| 137 | + assertValues(resp.values(), List.of(List.of(6, 0.9114001989364624), List.of(1, 1.156558871269226))); |
| 138 | + } |
| 139 | + } |
| 140 | + |
| 141 | + public void testWhereMatchPhraseWithScoringSortScore() { |
| 142 | + var query = """ |
| 143 | + FROM test |
| 144 | + METADATA _score |
| 145 | + | WHERE match_phrase(content, "brown fox") |
| 146 | + | KEEP id, _score |
| 147 | + | SORT _score DESC |
| 148 | + """; |
| 149 | + |
| 150 | + try (var resp = run(query)) { |
| 151 | + assertColumnNames(resp.columns(), List.of("id", "_score")); |
| 152 | + assertColumnTypes(resp.columns(), List.of("integer", "double")); |
| 153 | + assertValues(resp.values(), List.of(List.of(1, 1.156558871269226), List.of(6, 0.9114001989364624))); |
| 154 | + } |
| 155 | + } |
| 156 | + |
| 157 | + public void testWhereMatchPhraseWithScoringNoSort() { |
| 158 | + var query = """ |
| 159 | + FROM test |
| 160 | + METADATA _score |
| 161 | + | WHERE match_phrase(content, "brown fox") |
| 162 | + | KEEP id, _score |
| 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 | + assertValuesInAnyOrder(resp.values(), List.of(List.of(1, 1.156558871269226), List.of(6, 0.9114001989364624))); |
| 169 | + } |
| 170 | + } |
| 171 | + |
| 172 | + public void testNonExistingColumn() { |
| 173 | + var query = """ |
| 174 | + FROM test |
| 175 | + | WHERE match_phrase(something, "brown fox") |
| 176 | + """; |
| 177 | + |
| 178 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 179 | + assertThat(error.getMessage(), containsString("Unknown column [something]")); |
| 180 | + } |
| 181 | + |
| 182 | + public void testWhereMatchPhraseEvalColumn() { |
| 183 | + var query = """ |
| 184 | + FROM test |
| 185 | + | EVAL upper_content = to_upper(content) |
| 186 | + | WHERE match_phrase(upper_content, "BROWN FOX") |
| 187 | + | KEEP id |
| 188 | + """; |
| 189 | + |
| 190 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 191 | + assertThat( |
| 192 | + error.getMessage(), |
| 193 | + containsString("[MATCH_PHRASE] function cannot operate on [upper_content], which is not a field from an index mapping") |
| 194 | + ); |
| 195 | + } |
| 196 | + |
| 197 | + public void testWhereMatchPhraseOverWrittenColumn() { |
| 198 | + var query = """ |
| 199 | + FROM test |
| 200 | + | DROP content |
| 201 | + | EVAL content = CONCAT("document with ID ", to_str(id)) |
| 202 | + | WHERE match_phrase(content, "document content") |
| 203 | + """; |
| 204 | + |
| 205 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 206 | + assertThat( |
| 207 | + error.getMessage(), |
| 208 | + containsString("[MATCH_PHRASE] function cannot operate on [content], which is not a field from an index mapping") |
| 209 | + ); |
| 210 | + } |
| 211 | + |
| 212 | + public void testWhereMatchPhraseAfterStats() { |
| 213 | + var query = """ |
| 214 | + FROM test |
| 215 | + | STATS count(*) |
| 216 | + | WHERE match_phrase(content, "brown fox") |
| 217 | + """; |
| 218 | + |
| 219 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 220 | + assertThat(error.getMessage(), containsString("Unknown column [content]")); |
| 221 | + } |
| 222 | + |
| 223 | + public void testWhereMatchPhraseNotPushedDown() { |
| 224 | + var query = """ |
| 225 | + FROM test |
| 226 | + | WHERE match(content, "brown fox") OR length(content) < 20 |
| 227 | + | KEEP id |
| 228 | + | SORT id |
| 229 | + """; |
| 230 | + |
| 231 | + try (var resp = run(query)) { |
| 232 | + assertColumnNames(resp.columns(), List.of("id")); |
| 233 | + assertColumnTypes(resp.columns(), List.of("integer")); |
| 234 | + assertValues(resp.values(), List.of(List.of(1), List.of(2), List.of(6))); |
| 235 | + } |
| 236 | + } |
| 237 | + |
| 238 | + public void testWhereMatchPhraseWithRow() { |
| 239 | + var query = """ |
| 240 | + ROW content = "a brown fox" |
| 241 | + | WHERE match(content, "brown fox") |
| 242 | + """; |
| 243 | + |
| 244 | + var error = expectThrows(ElasticsearchException.class, () -> run(query)); |
| 245 | + assertThat( |
| 246 | + error.getMessage(), |
| 247 | + containsString("line 2:15: [MATCH_PHRASE] function cannot operate on [content], which is not a field from an index mapping") |
| 248 | + ); |
| 249 | + } |
| 250 | + |
| 251 | + public void testMatchPhraseithStats() { |
| 252 | + var errorQuery = """ |
| 253 | + FROM test |
| 254 | + | STATS c = count(*) BY match_phrase(content, "brown fox") |
| 255 | + """; |
| 256 | + |
| 257 | + var error = expectThrows(ElasticsearchException.class, () -> run(errorQuery)); |
| 258 | + assertThat(error.getMessage(), containsString("[MATCH_PHRASE] function is only supported in WHERE and STATS commands")); |
| 259 | + |
| 260 | + var query = """ |
| 261 | + FROM test |
| 262 | + | STATS c = count(*) WHERE match_phrase(content, "brown fox"), d = count(*) WHERE match_phrase(content, "lazy dog") |
| 263 | + """; |
| 264 | + |
| 265 | + try (var resp = run(query)) { |
| 266 | + assertColumnNames(resp.columns(), List.of("c", "d")); |
| 267 | + assertColumnTypes(resp.columns(), List.of("long", "long")); |
| 268 | + assertValues(resp.values(), List.of(List.of(2L, 4L))); |
| 269 | + } |
| 270 | + |
| 271 | + query = """ |
| 272 | + FROM test METADATA _score |
| 273 | + | WHERE match_phrase(content, "brown fox") |
| 274 | + | STATS m = max(_score), n = min(_score) |
| 275 | + """; |
| 276 | + |
| 277 | + try (var resp = run(query)) { |
| 278 | + assertColumnNames(resp.columns(), List.of("m", "n")); |
| 279 | + assertColumnTypes(resp.columns(), List.of("double", "double")); |
| 280 | + List<List<Object>> valuesList = getValuesList(resp.values()); |
| 281 | + assertEquals(1, valuesList.size()); |
| 282 | + assertThat((double) valuesList.get(0).get(0), Matchers.greaterThan(1.0)); |
| 283 | + assertThat((double) valuesList.get(0).get(1), Matchers.greaterThan(0.0)); |
| 284 | + } |
| 285 | + } |
| 286 | + |
| 287 | + public void testMatchPhraseWithinEval() { |
| 288 | + var query = """ |
| 289 | + FROM test |
| 290 | + | EVAL matches_query = match_phrase(content, "brown fox") |
| 291 | + """; |
| 292 | + |
| 293 | + var error = expectThrows(VerificationException.class, () -> run(query)); |
| 294 | + assertThat(error.getMessage(), containsString("[MATCH_PHRASE] function is only supported in WHERE and STATS commands")); |
| 295 | + } |
| 296 | + |
| 297 | + private void createAndPopulateIndex() { |
| 298 | + var indexName = "test"; |
| 299 | + var client = client().admin().indices(); |
| 300 | + var CreateRequest = client.prepareCreate(indexName) |
| 301 | + .setSettings(Settings.builder().put("index.number_of_shards", 1)) |
| 302 | + .setMapping("id", "type=integer", "content", "type=text"); |
| 303 | + assertAcked(CreateRequest); |
| 304 | + client().prepareBulk() |
| 305 | + .add(new IndexRequest(indexName).id("1").source("id", 1, "content", "This is a brown fox")) |
| 306 | + .add(new IndexRequest(indexName).id("2").source("id", 2, "content", "This is a brown dog")) |
| 307 | + .add(new IndexRequest(indexName).id("3").source("id", 3, "content", "This dog is really brown")) |
| 308 | + .add(new IndexRequest(indexName).id("4").source("id", 4, "content", "The dog is brown but this document is very very long")) |
| 309 | + .add(new IndexRequest(indexName).id("5").source("id", 5, "content", "There is also a white cat")) |
| 310 | + .add(new IndexRequest(indexName).id("6").source("id", 6, "content", "The quick brown fox jumps over the lazy dog")) |
| 311 | + .setRefreshPolicy(WriteRequest.RefreshPolicy.IMMEDIATE) |
| 312 | + .get(); |
| 313 | + ensureYellow(indexName); |
| 314 | + } |
| 315 | +} |
0 commit comments