Skip to content

Commit fa3fde2

Browse files
committed
Make simple_query_string leniency more fine-grained
Previously, the leniency was on a per-query basis, with each query being parsed into multiple queries, one for each field. If any one of these queries failed, the entire query was discarded in the name of being lenient. Now query parts will only be discarded if they fail for a particular field, the entire query is not discarded. This helps when performing a query over a numeric and string field, as only the sub-queries that are invalid due to format exceptions will be discarded. Also moves the `simple_query_string` queries out of SimpleQueryTests and into a dedicated SimpleQueryStringTests class. Fixes #7967
1 parent e3f6c31 commit fa3fde2

File tree

3 files changed

+317
-226
lines changed

3 files changed

+317
-226
lines changed

src/main/java/org/elasticsearch/index/query/SimpleQueryParser.java

+48-17
Original file line numberDiff line numberDiff line change
@@ -19,7 +19,8 @@
1919
package org.elasticsearch.index.query;
2020

2121
import org.apache.lucene.analysis.Analyzer;
22-
import org.apache.lucene.search.Query;
22+
import org.apache.lucene.index.Term;
23+
import org.apache.lucene.search.*;
2324

2425
import java.util.Locale;
2526
import java.util.Map;
@@ -50,11 +51,19 @@ private Query rethrowUnlessLenient(RuntimeException e) {
5051

5152
@Override
5253
public Query newDefaultQuery(String text) {
53-
try {
54-
return super.newDefaultQuery(text);
55-
} catch (RuntimeException e) {
56-
return rethrowUnlessLenient(e);
54+
BooleanQuery bq = new BooleanQuery(true);
55+
for (Map.Entry<String,Float> entry : weights.entrySet()) {
56+
try {
57+
Query q = createBooleanQuery(entry.getKey(), text, super.getDefaultOperator());
58+
if (q != null) {
59+
q.setBoost(entry.getValue());
60+
bq.add(q, BooleanClause.Occur.SHOULD);
61+
}
62+
} catch (RuntimeException e) {
63+
rethrowUnlessLenient(e);
64+
}
5765
}
66+
return super.simplify(bq);
5867
}
5968

6069
/**
@@ -66,20 +75,36 @@ public Query newFuzzyQuery(String text, int fuzziness) {
6675
if (settings.lowercaseExpandedTerms()) {
6776
text = text.toLowerCase(settings.locale());
6877
}
69-
try {
70-
return super.newFuzzyQuery(text, fuzziness);
71-
} catch (RuntimeException e) {
72-
return rethrowUnlessLenient(e);
78+
BooleanQuery bq = new BooleanQuery(true);
79+
for (Map.Entry<String,Float> entry : weights.entrySet()) {
80+
try {
81+
Query q = new FuzzyQuery(new Term(entry.getKey(), text), fuzziness);
82+
if (q != null) {
83+
q.setBoost(entry.getValue());
84+
bq.add(q, BooleanClause.Occur.SHOULD);
85+
}
86+
} catch (RuntimeException e) {
87+
rethrowUnlessLenient(e);
88+
}
7389
}
90+
return super.simplify(bq);
7491
}
7592

7693
@Override
7794
public Query newPhraseQuery(String text, int slop) {
78-
try {
79-
return super.newPhraseQuery(text, slop);
80-
} catch (RuntimeException e) {
81-
return rethrowUnlessLenient(e);
95+
BooleanQuery bq = new BooleanQuery(true);
96+
for (Map.Entry<String,Float> entry : weights.entrySet()) {
97+
try {
98+
Query q = createPhraseQuery(entry.getKey(), text, slop);
99+
if (q != null) {
100+
q.setBoost(entry.getValue());
101+
bq.add(q, BooleanClause.Occur.SHOULD);
102+
}
103+
} catch (RuntimeException e) {
104+
rethrowUnlessLenient(e);
105+
}
82106
}
107+
return super.simplify(bq);
83108
}
84109

85110
/**
@@ -91,11 +116,17 @@ public Query newPrefixQuery(String text) {
91116
if (settings.lowercaseExpandedTerms()) {
92117
text = text.toLowerCase(settings.locale());
93118
}
94-
try {
95-
return super.newPrefixQuery(text);
96-
} catch (RuntimeException e) {
97-
return rethrowUnlessLenient(e);
119+
BooleanQuery bq = new BooleanQuery(true);
120+
for (Map.Entry<String,Float> entry : weights.entrySet()) {
121+
try {
122+
PrefixQuery prefix = new PrefixQuery(new Term(entry.getKey(), text));
123+
prefix.setBoost(entry.getValue());
124+
bq.add(prefix, BooleanClause.Occur.SHOULD);
125+
} catch (RuntimeException e) {
126+
return rethrowUnlessLenient(e);
127+
}
98128
}
129+
return super.simplify(bq);
99130
}
100131

101132
/**
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,269 @@
1+
/*
2+
* Licensed to Elasticsearch under one or more contributor
3+
* license agreements. See the NOTICE file distributed with
4+
* this work for additional information regarding copyright
5+
* ownership. Elasticsearch licenses this file to you under
6+
* the Apache License, Version 2.0 (the "License"); you may
7+
* not use this file except in compliance with the License.
8+
* You may obtain a copy of the License at
9+
*
10+
* http://www.apache.org/licenses/LICENSE-2.0
11+
*
12+
* Unless required by applicable law or agreed to in writing,
13+
* software distributed under the License is distributed on an
14+
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
15+
* KIND, either express or implied. See the License for the
16+
* specific language governing permissions and limitations
17+
* under the License.
18+
*/
19+
20+
package org.elasticsearch.search.query;
21+
22+
import org.elasticsearch.action.search.SearchResponse;
23+
import org.elasticsearch.index.query.BoolQueryBuilder;
24+
import org.elasticsearch.index.query.SimpleQueryStringBuilder;
25+
import org.elasticsearch.index.query.SimpleQueryStringFlag;
26+
import org.elasticsearch.test.ElasticsearchIntegrationTest;
27+
import org.junit.Test;
28+
29+
import java.io.IOException;
30+
import java.util.Locale;
31+
import java.util.concurrent.ExecutionException;
32+
33+
import static org.elasticsearch.common.xcontent.XContentFactory.jsonBuilder;
34+
import static org.elasticsearch.index.query.QueryBuilders.boolQuery;
35+
import static org.elasticsearch.index.query.QueryBuilders.queryString;
36+
import static org.elasticsearch.index.query.QueryBuilders.simpleQueryString;
37+
import static org.elasticsearch.test.hamcrest.ElasticsearchAssertions.*;
38+
import static org.elasticsearch.test.hamcrest.ElasticsearchAssertions.assertSearchHits;
39+
import static org.hamcrest.Matchers.equalTo;
40+
41+
/**
42+
* Tests for the {@code simple_query_string} query
43+
*/
44+
public class SimpleQueryStringTests extends ElasticsearchIntegrationTest {
45+
46+
@Test
47+
public void testSimpleQueryString() throws ExecutionException, InterruptedException {
48+
createIndex("test");
49+
indexRandom(true, false,
50+
client().prepareIndex("test", "type1", "1").setSource("body", "foo"),
51+
client().prepareIndex("test", "type1", "2").setSource("body", "bar"),
52+
client().prepareIndex("test", "type1", "3").setSource("body", "foo bar"),
53+
client().prepareIndex("test", "type1", "4").setSource("body", "quux baz eggplant"),
54+
client().prepareIndex("test", "type1", "5").setSource("body", "quux baz spaghetti"),
55+
client().prepareIndex("test", "type1", "6").setSource("otherbody", "spaghetti"));
56+
57+
SearchResponse searchResponse = client().prepareSearch().setQuery(simpleQueryString("foo bar")).get();
58+
assertHitCount(searchResponse, 3l);
59+
assertSearchHits(searchResponse, "1", "2", "3");
60+
61+
searchResponse = client().prepareSearch().setQuery(
62+
simpleQueryString("foo bar").defaultOperator(SimpleQueryStringBuilder.Operator.AND)).get();
63+
assertHitCount(searchResponse, 1l);
64+
assertFirstHit(searchResponse, hasId("3"));
65+
66+
searchResponse = client().prepareSearch().setQuery(simpleQueryString("\"quux baz\" +(eggplant | spaghetti)")).get();
67+
assertHitCount(searchResponse, 2l);
68+
assertSearchHits(searchResponse, "4", "5");
69+
70+
searchResponse = client().prepareSearch().setQuery(
71+
simpleQueryString("eggplants").analyzer("snowball")).get();
72+
assertHitCount(searchResponse, 1l);
73+
assertFirstHit(searchResponse, hasId("4"));
74+
75+
searchResponse = client().prepareSearch().setQuery(
76+
simpleQueryString("spaghetti").field("body", 10.0f).field("otherbody", 2.0f).queryName("myquery")).get();
77+
assertHitCount(searchResponse, 2l);
78+
assertFirstHit(searchResponse, hasId("5"));
79+
assertSearchHits(searchResponse, "5", "6");
80+
assertThat(searchResponse.getHits().getAt(0).getMatchedQueries()[0], equalTo("myquery"));
81+
82+
searchResponse = client().prepareSearch().setQuery(simpleQueryString("spaghetti").field("*body")).get();
83+
assertHitCount(searchResponse, 2l);
84+
assertSearchHits(searchResponse, "5", "6");
85+
86+
// Have to bypass the builder here because the builder always uses "fields" instead of "field"
87+
searchResponse = client().prepareSearch().setQuery("{\"simple_query_string\": {\"query\": \"spaghetti\", \"field\": \"_all\"}}").get();
88+
assertHitCount(searchResponse, 2l);
89+
assertSearchHits(searchResponse, "5", "6");
90+
}
91+
92+
@Test
93+
public void testSimpleQueryStringLowercasing() {
94+
createIndex("test");
95+
client().prepareIndex("test", "type1", "1").setSource("body", "Professional").get();
96+
refresh();
97+
98+
SearchResponse searchResponse = client().prepareSearch().setQuery(simpleQueryString("Professio*")).get();
99+
assertHitCount(searchResponse, 1l);
100+
assertSearchHits(searchResponse, "1");
101+
102+
searchResponse = client().prepareSearch().setQuery(
103+
simpleQueryString("Professio*").lowercaseExpandedTerms(false)).get();
104+
assertHitCount(searchResponse, 0l);
105+
106+
searchResponse = client().prepareSearch().setQuery(
107+
simpleQueryString("Professionan~1")).get();
108+
assertHitCount(searchResponse, 1l);
109+
assertSearchHits(searchResponse, "1");
110+
111+
searchResponse = client().prepareSearch().setQuery(
112+
simpleQueryString("Professionan~1").lowercaseExpandedTerms(false)).get();
113+
assertHitCount(searchResponse, 0l);
114+
}
115+
116+
@Test
117+
public void testQueryStringLocale() {
118+
createIndex("test");
119+
client().prepareIndex("test", "type1", "1").setSource("body", "bılly").get();
120+
refresh();
121+
122+
SearchResponse searchResponse = client().prepareSearch().setQuery(simpleQueryString("BILL*")).get();
123+
assertHitCount(searchResponse, 0l);
124+
searchResponse = client().prepareSearch().setQuery(queryString("body:BILL*")).get();
125+
assertHitCount(searchResponse, 0l);
126+
127+
searchResponse = client().prepareSearch().setQuery(
128+
simpleQueryString("BILL*").locale(new Locale("tr", "TR"))).get();
129+
assertHitCount(searchResponse, 1l);
130+
assertSearchHits(searchResponse, "1");
131+
searchResponse = client().prepareSearch().setQuery(
132+
queryString("body:BILL*").locale(new Locale("tr", "TR"))).get();
133+
assertHitCount(searchResponse, 1l);
134+
assertSearchHits(searchResponse, "1");
135+
}
136+
137+
@Test
138+
public void testNestedFieldSimpleQueryString() throws IOException {
139+
assertAcked(prepareCreate("test")
140+
.addMapping("type1", jsonBuilder()
141+
.startObject()
142+
.startObject("type1")
143+
.startObject("properties")
144+
.startObject("body").field("type", "string")
145+
.startObject("fields")
146+
.startObject("sub").field("type", "string")
147+
.endObject() // sub
148+
.endObject() // fields
149+
.endObject() // body
150+
.endObject() // properties
151+
.endObject() // type1
152+
.endObject()));
153+
client().prepareIndex("test", "type1", "1").setSource("body", "foo bar baz").get();
154+
refresh();
155+
156+
SearchResponse searchResponse = client().prepareSearch().setQuery(
157+
simpleQueryString("foo bar baz").field("body")).get();
158+
assertHitCount(searchResponse, 1l);
159+
assertSearchHits(searchResponse, "1");
160+
161+
searchResponse = client().prepareSearch().setQuery(
162+
simpleQueryString("foo bar baz").field("type1.body")).get();
163+
assertHitCount(searchResponse, 1l);
164+
assertSearchHits(searchResponse, "1");
165+
166+
searchResponse = client().prepareSearch().setQuery(
167+
simpleQueryString("foo bar baz").field("body.sub")).get();
168+
assertHitCount(searchResponse, 1l);
169+
assertSearchHits(searchResponse, "1");
170+
171+
searchResponse = client().prepareSearch().setQuery(
172+
simpleQueryString("foo bar baz").field("type1.body.sub")).get();
173+
assertHitCount(searchResponse, 1l);
174+
assertSearchHits(searchResponse, "1");
175+
}
176+
177+
@Test
178+
public void testSimpleQueryStringFlags() throws ExecutionException, InterruptedException {
179+
createIndex("test");
180+
indexRandom(true,
181+
client().prepareIndex("test", "type1", "1").setSource("body", "foo"),
182+
client().prepareIndex("test", "type1", "2").setSource("body", "bar"),
183+
client().prepareIndex("test", "type1", "3").setSource("body", "foo bar"),
184+
client().prepareIndex("test", "type1", "4").setSource("body", "quux baz eggplant"),
185+
client().prepareIndex("test", "type1", "5").setSource("body", "quux baz spaghetti"),
186+
client().prepareIndex("test", "type1", "6").setSource("otherbody", "spaghetti"));
187+
188+
SearchResponse searchResponse = client().prepareSearch().setQuery(
189+
simpleQueryString("foo bar").flags(SimpleQueryStringFlag.ALL)).get();
190+
assertHitCount(searchResponse, 3l);
191+
assertSearchHits(searchResponse, "1", "2", "3");
192+
193+
// Sending a negative 'flags' value is the same as SimpleQueryStringFlag.ALL
194+
searchResponse = client().prepareSearch().setQuery("{\"simple_query_string\": {\"query\": \"foo bar\", \"flags\": -1}}").get();
195+
assertHitCount(searchResponse, 3l);
196+
assertSearchHits(searchResponse, "1", "2", "3");
197+
198+
searchResponse = client().prepareSearch().setQuery(
199+
simpleQueryString("foo | bar")
200+
.defaultOperator(SimpleQueryStringBuilder.Operator.AND)
201+
.flags(SimpleQueryStringFlag.OR)).get();
202+
assertHitCount(searchResponse, 3l);
203+
assertSearchHits(searchResponse, "1", "2", "3");
204+
205+
searchResponse = client().prepareSearch().setQuery(
206+
simpleQueryString("foo | bar")
207+
.defaultOperator(SimpleQueryStringBuilder.Operator.AND)
208+
.flags(SimpleQueryStringFlag.NONE)).get();
209+
assertHitCount(searchResponse, 1l);
210+
assertFirstHit(searchResponse, hasId("3"));
211+
212+
searchResponse = client().prepareSearch().setQuery(
213+
simpleQueryString("baz | egg*")
214+
.defaultOperator(SimpleQueryStringBuilder.Operator.AND)
215+
.flags(SimpleQueryStringFlag.NONE)).get();
216+
assertHitCount(searchResponse, 0l);
217+
218+
searchResponse = client().prepareSearch().setSource("{\n" +
219+
" \"query\": {\n" +
220+
" \"simple_query_string\": {\n" +
221+
" \"query\": \"foo|bar\",\n" +
222+
" \"default_operator\": \"AND\"," +
223+
" \"flags\": \"NONE\"\n" +
224+
" }\n" +
225+
" }\n" +
226+
"}").get();
227+
assertHitCount(searchResponse, 1l);
228+
229+
searchResponse = client().prepareSearch().setQuery(
230+
simpleQueryString("baz | egg*")
231+
.defaultOperator(SimpleQueryStringBuilder.Operator.AND)
232+
.flags(SimpleQueryStringFlag.WHITESPACE, SimpleQueryStringFlag.PREFIX)).get();
233+
assertHitCount(searchResponse, 1l);
234+
assertFirstHit(searchResponse, hasId("4"));
235+
}
236+
237+
@Test
238+
public void testSimpleQueryStringLenient() throws ExecutionException, InterruptedException {
239+
createIndex("test1", "test2");
240+
indexRandom(true, client().prepareIndex("test1", "type1", "1").setSource("field", "foo"),
241+
client().prepareIndex("test2", "type1", "10").setSource("field", 5));
242+
refresh();
243+
244+
SearchResponse searchResponse = client().prepareSearch().setQuery(simpleQueryString("foo").field("field")).get();
245+
assertFailures(searchResponse);
246+
assertHitCount(searchResponse, 1l);
247+
assertSearchHits(searchResponse, "1");
248+
249+
searchResponse = client().prepareSearch().setQuery(simpleQueryString("foo").field("field").lenient(true)).get();
250+
assertNoFailures(searchResponse);
251+
assertHitCount(searchResponse, 1l);
252+
assertSearchHits(searchResponse, "1");
253+
}
254+
255+
@Test // see: https://github.com/elasticsearch/elasticsearch/issues/7967
256+
public void testLenientFlagBeingTooLenient() throws Exception {
257+
indexRandom(true,
258+
client().prepareIndex("test", "doc", "1").setSource("num", 1, "body", "foo bar baz"),
259+
client().prepareIndex("test", "doc", "2").setSource("num", 2, "body", "eggplant spaghetti lasagna"));
260+
261+
BoolQueryBuilder q = boolQuery().should(simpleQueryString("bar").field("num").field("body").lenient(true));
262+
SearchResponse resp = client().prepareSearch("test").setQuery(q).get();
263+
assertNoFailures(resp);
264+
// the bug is that this would be parsed into basically a match_all
265+
// query and this would match both documents
266+
assertHitCount(resp, 1);
267+
assertSearchHits(resp, "1");
268+
}
269+
}

0 commit comments

Comments
 (0)