Propagate TooComplexToDeterminizeException in query_string regex queries (opensearch-project#18883)

Signed-off-by: Peter Alfonsi <[email protected]>
Co-authored-by: Peter Alfonsi <[email protected]>

Author: 2 people

CHANGELOG.md

@@ -97,6 +97,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Ignore awareness attributes when a custom preference string is included with a search request ([#18848](https://github.com/opensearch-project/OpenSearch/pull/18848))
 - Use ScoreDoc instead of FieldDoc when creating TopScoreDocCollectorManager to avoid unnecessary conversion ([#18802](https://github.com/opensearch-project/OpenSearch/pull/18802))
 - Fix leafSorter optimization for ReadOnlyEngine and NRTReplicationEngine ([#18639](https://github.com/opensearch-project/OpenSearch/pull/18639))
+- Fix query string regex queries incorrectly swallowing TooComplexToDeterminizeException ([#18883](https://github.com/opensearch-project/OpenSearch/pull/18883))
 
 ### Security
 

server/src/main/java/org/opensearch/index/search/QueryStringQueryParser.java

@@ -57,6 +57,7 @@
 import org.apache.lucene.search.WildcardQuery;
 import org.apache.lucene.util.BytesRef;
 import org.apache.lucene.util.automaton.RegExp;
+import org.apache.lucene.util.automaton.TooComplexToDeterminizeException;
 import org.opensearch.common.lucene.search.Queries;
 import org.opensearch.common.regex.Regex;
 import org.opensearch.common.unit.Fuzziness;
@@ -795,7 +796,9 @@ private Query getRegexpQuerySingle(String field, String termStr) throws ParseExc
             termStr = getAnalyzer().normalize(currentFieldType.name(), termStr).utf8ToString();
             return currentFieldType.regexpQuery(termStr, RegExp.ALL, 0, getDeterminizeWorkLimit(), getMultiTermRewriteMethod(), context);
         } catch (RuntimeException e) {
-            if (lenient) {
+            // Lenient queries are intended for data type mismatches, but TooComplexToDeterminizeException
+            // comes up from the same place in the code. Don't create a lenient query in this case.
+            if (lenient && !(e instanceof TooComplexToDeterminizeException)) {
                 return newLenientFieldQuery(field, e);
             }
             throw e;

server/src/test/java/org/opensearch/index/query/QueryStringQueryBuilderTests.java

@@ -803,8 +803,14 @@ public void testToQueryRegExpQueryTooComplex() throws Exception {
             TooComplexToDeterminizeException.class,
             () -> queryBuilder.toQuery(createShardContext())
         );
-        assertThat(e.getMessage(), containsString("Determinizing automaton"));
-        assertThat(e.getMessage(), containsString("would require more than 10000 effort"));
+        assertTrue(e.getMessage().contains("Determinizing automaton"));
+        assertTrue(e.getMessage().contains("would require more than 10000 effort"));
+
+        // TooComplexToDeterminizeException should be thrown even if lenient is true
+        QueryStringQueryBuilder lenientQueryBuilder = queryStringQuery("/[ac]*a[ac]{50,200}/").defaultField(TEXT_FIELD_NAME).lenient(true);
+        e = expectThrows(TooComplexToDeterminizeException.class, () -> lenientQueryBuilder.toQuery(createShardContext()));
+        assertTrue(e.getMessage().contains("Determinizing automaton"));
+        assertTrue(e.getMessage().contains("would require more than 10000 effort"));
     }
 
     /**