|
| 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.optimizer.rules.logical; |
| 9 | + |
| 10 | +import org.elasticsearch.xpack.esql.VerificationException; |
| 11 | +import org.elasticsearch.xpack.esql.expression.Foldables; |
| 12 | +import org.elasticsearch.xpack.esql.optimizer.AbstractLogicalPlanOptimizerTests; |
| 13 | +import org.elasticsearch.xpack.esql.plan.logical.Dissect; |
| 14 | +import org.elasticsearch.xpack.esql.plan.logical.Enrich; |
| 15 | +import org.elasticsearch.xpack.esql.plan.logical.Eval; |
| 16 | +import org.elasticsearch.xpack.esql.plan.logical.Limit; |
| 17 | +import org.elasticsearch.xpack.esql.plan.logical.Project; |
| 18 | + |
| 19 | +import static org.elasticsearch.xpack.esql.EsqlTestUtils.as; |
| 20 | +import static org.hamcrest.Matchers.containsString; |
| 21 | +import static org.hamcrest.Matchers.equalTo; |
| 22 | +import static org.hamcrest.Matchers.is; |
| 23 | +import static org.hamcrest.Matchers.not; |
| 24 | + |
| 25 | +public class HoistRemoteEnrichLimitTests extends AbstractLogicalPlanOptimizerTests { |
| 26 | + |
| 27 | + /** |
| 28 | + * <pre> |
| 29 | + * Limit[10[INTEGER],true,false] |
| 30 | + * \_Enrich[REMOTE,languages_remote[KEYWORD],id{r}#4,{"match":{"indices":[],"match_field":"id", |
| 31 | + * "enrich_fields":["language_code","language_name"]}},{=languages_idx},[language_code{r}#20, language_name{r}#21]] |
| 32 | + * \_Eval[[emp_no{f}#6 AS id#4]] |
| 33 | + * \_Limit[10[INTEGER],false,true] |
| 34 | + * \_EsRelation[test][_meta_field{f}#12, emp_no{f}#6, first_name{f}#7, ...] |
| 35 | + * </pre> |
| 36 | + */ |
| 37 | + public void testLimitWithinRemoteEnrich() { |
| 38 | + var plan = plan(randomFrom(""" |
| 39 | + from test |
| 40 | + | EVAL id = emp_no |
| 41 | + | LIMIT 10 |
| 42 | + | ENRICH _remote:languages_remote |
| 43 | + """, """ |
| 44 | + from test |
| 45 | + | LIMIT 10 |
| 46 | + | EVAL id = emp_no |
| 47 | + | ENRICH _remote:languages_remote |
| 48 | + """)); // it should be the same in any order |
| 49 | + |
| 50 | + var limit = as(plan, Limit.class); |
| 51 | + assertTrue(limit.duplicated()); |
| 52 | + assertFalse(limit.local()); |
| 53 | + var enrich = as(limit.child(), Enrich.class); |
| 54 | + assertThat(enrich.mode(), is(Enrich.Mode.REMOTE)); |
| 55 | + var eval = as(enrich.child(), Eval.class); |
| 56 | + var innerLimit = as(eval.child(), Limit.class); |
| 57 | + assertFalse(innerLimit.duplicated()); |
| 58 | + assertTrue(innerLimit.local()); |
| 59 | + } |
| 60 | + |
| 61 | + /** |
| 62 | + * <pre> |
| 63 | + * Project[[salary{f}#19 AS wage#10, emp_no{f}#14 AS id#4, first_name{r}#11, language_code{r}#28, language_name{r}#29]] |
| 64 | + * \_Limit[5[INTEGER],true,false] |
| 65 | + * \_Enrich[REMOTE,languages_remote[KEYWORD],emp_no{f}#14,{"match":{"indices":[],"match_field":"id","enrich_fields":["languag |
| 66 | + * e_code","language_name"]}},{=languages_idx},[language_code{r}#28, language_name{r}#29]] |
| 67 | + * \_Dissect[first_name{f}#15,Parser[pattern=%{first_name}s, appendSeparator=, parser=org.elasticsearch.dissect.DissectParse |
| 68 | + * r@7d5c7931],[first_name{r}#11]] |
| 69 | + * \_Limit[5[INTEGER],false,true] |
| 70 | + * \_EsRelation[test][_meta_field{f}#20, emp_no{f}#14, first_name{f}#15, ..] |
| 71 | + * </pre> |
| 72 | + */ |
| 73 | + public void testManyLimitsWithinRemoteEnrich() { |
| 74 | + var plan = plan(""" |
| 75 | + from test |
| 76 | + | LIMIT 10 |
| 77 | + | EVAL id = emp_no |
| 78 | + | KEEP first_name, salary, id |
| 79 | + | RENAME salary AS wage |
| 80 | + | DISSECT first_name "%{first_name}s" |
| 81 | + | LIMIT 5 |
| 82 | + | ENRICH _remote:languages_remote |
| 83 | + """); |
| 84 | + |
| 85 | + var project = as(plan, Project.class); |
| 86 | + var limit = as(project.child(), Limit.class); |
| 87 | + assertTrue(limit.duplicated()); |
| 88 | + assertFalse(limit.local()); |
| 89 | + assertThat(Foldables.limitValue(limit.limit(), limit.sourceText()), equalTo(5)); |
| 90 | + var enrich = as(limit.child(), Enrich.class); |
| 91 | + assertThat(enrich.mode(), is(Enrich.Mode.REMOTE)); |
| 92 | + var dissect = as(enrich.child(), Dissect.class); |
| 93 | + var innerLimit = as(dissect.child(), Limit.class); |
| 94 | + assertFalse(innerLimit.duplicated()); |
| 95 | + assertTrue(innerLimit.local()); |
| 96 | + assertThat(Foldables.limitValue(innerLimit.limit(), innerLimit.sourceText()), equalTo(5)); |
| 97 | + } |
| 98 | + |
| 99 | + /** |
| 100 | + * Project[[first_name{f}#14, salary{f}#18 AS wage#11, emp_no{f}#13 AS id#4, language_code{r}#32, language_name{r}#33]] |
| 101 | + * \_Limit[5[INTEGER],true,false] |
| 102 | + * \_Enrich[REMOTE,languages_remote[KEYWORD],emp_no{f}#13,{"match":{"indices":[],"match_field":"id","enrich_fields":["languag |
| 103 | + * e_code","language_name"]}},{=languages_idx},[language_code{r}#32, language_name{r}#33]] |
| 104 | + * \_Limit[5[INTEGER],true,true] |
| 105 | + * \_Enrich[REMOTE,languages_remote[KEYWORD],emp_no{f}#13,{"match":{"indices":[],"match_field":"id","enrich_fields":["languag |
| 106 | + * e_code","language_name"]}},{=languages_idx},[language_code{r}#27, language_name{r}#28]] |
| 107 | + * \_Limit[5[INTEGER],false,true] |
| 108 | + * \_EsRelation[test][_meta_field{f}#19, emp_no{f}#13, first_name{f}#14, ..] |
| 109 | + */ |
| 110 | + public void testLimitsWithinRemoteEnrichTwice() { |
| 111 | + var plan = plan(""" |
| 112 | + from test |
| 113 | + | LIMIT 10 |
| 114 | + | EVAL id = emp_no |
| 115 | + | KEEP first_name, salary, id |
| 116 | + | ENRICH _remote:languages_remote |
| 117 | + | RENAME salary AS wage |
| 118 | + | LIMIT 5 |
| 119 | + | ENRICH _remote:languages_remote |
| 120 | + """); |
| 121 | + var project = as(plan, Project.class); |
| 122 | + var limit = as(project.child(), Limit.class); |
| 123 | + assertTrue(limit.duplicated()); |
| 124 | + assertFalse(limit.local()); |
| 125 | + assertThat(Foldables.limitValue(limit.limit(), limit.sourceText()), equalTo(5)); |
| 126 | + var enrich = as(limit.child(), Enrich.class); |
| 127 | + assertThat(enrich.mode(), is(Enrich.Mode.REMOTE)); |
| 128 | + var innerLimit = as(enrich.child(), Limit.class); |
| 129 | + assertTrue(innerLimit.duplicated()); |
| 130 | + assertTrue(innerLimit.local()); |
| 131 | + assertThat(Foldables.limitValue(innerLimit.limit(), innerLimit.sourceText()), equalTo(5)); |
| 132 | + var secondEnrich = as(innerLimit.child(), Enrich.class); |
| 133 | + assertThat(secondEnrich.mode(), is(Enrich.Mode.REMOTE)); |
| 134 | + var innermostLimit = as(secondEnrich.child(), Limit.class); |
| 135 | + assertFalse(innermostLimit.duplicated()); |
| 136 | + assertTrue(innermostLimit.local()); |
| 137 | + assertThat(Foldables.limitValue(innermostLimit.limit(), innermostLimit.sourceText()), equalTo(5)); |
| 138 | + } |
| 139 | + |
| 140 | + // These cases do not get hoisting, and it's ok |
| 141 | + public void testLimitWithinOtherEnrich() { |
| 142 | + String enrichPolicy = randomFrom("languages_idx", "_any:languages_idx", "_coordinator:languages_coordinator"); |
| 143 | + var plan = plan(String.format(""" |
| 144 | + from test |
| 145 | + | EVAL id = emp_no |
| 146 | + | LIMIT 10 |
| 147 | + | ENRICH %s |
| 148 | + """, enrichPolicy)); |
| 149 | + // Here ENRICH is on top - no hoisting happens |
| 150 | + var enrich = as(plan, Enrich.class); |
| 151 | + assertThat(enrich.mode(), not(is(Enrich.Mode.REMOTE))); |
| 152 | + } |
| 153 | + |
| 154 | + // Non-cardinality preserving commands after limit |
| 155 | + public void testFilterLimitThenEnrich() { |
| 156 | + // Hoisting does not happen, so the verifier fails since LIMIT is before remote ENRICH |
| 157 | + failPlan(""" |
| 158 | + from test |
| 159 | + | EVAL id = emp_no |
| 160 | + | LIMIT 10 |
| 161 | + | WHERE first_name != "john" |
| 162 | + | ENRICH _remote:languages_remote |
| 163 | + """, "ENRICH with remote policy can't be executed after [LIMIT 10]"); |
| 164 | + } |
| 165 | + |
| 166 | + public void testMvExpandLimitThenEnrich() { |
| 167 | + // Hoisting does not happen, so the verifier fails since LIMIT is before remote ENRICH |
| 168 | + failPlan(""" |
| 169 | + from test |
| 170 | + | EVAL id = emp_no |
| 171 | + | LIMIT 10 |
| 172 | + | MV_EXPAND languages |
| 173 | + | ENRICH _remote:languages_remote |
| 174 | + """, "MV_EXPAND after LIMIT is incompatible with remote ENRICH"); |
| 175 | + } |
| 176 | + |
| 177 | + // Other cases where hoisting does not happen: |
| 178 | + // - ExecutesOn.COORDINATOR - this fails the verifier |
| 179 | + // - PipelineBreaker - all relevant ones are also ExecutesOn.COORDINATOR |
| 180 | + |
| 181 | + private void failPlan(String esql, String reason) { |
| 182 | + var e = expectThrows(VerificationException.class, () -> plan(esql)); |
| 183 | + assertThat(e.getMessage(), containsString(reason)); |
| 184 | + } |
| 185 | +} |
0 commit comments