Fix regex query to work with field alias

CHANGELOG.md

@@ -33,11 +33,11 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 ### Fixed
 - Add task cancellation checks in aggregators ([#18426](https://github.com/opensearch-project/OpenSearch/pull/18426))
 - Fix concurrent timings in profiler ([#18540](https://github.com/opensearch-project/OpenSearch/pull/18540))
+- Fix regex query from query string query to work with field alias ([#18215](https://github.com/opensearch-project/OpenSearch/issues/18215))
 - [Autotagging] Fix delete rule event consumption in InMemoryRuleProcessingService ([#18628](https://github.com/opensearch-project/OpenSearch/pull/18628))
 - Cannot communicate with HTTP/2 when reactor-netty is enabled ([#18599](https://github.com/opensearch-project/OpenSearch/pull/18599))
 - Fix the visit of sub queries for HasParentQuery and HasChildQuery ([#18621](https://github.com/opensearch-project/OpenSearch/pull/18621))
 
-
 ### Security
 
 [Unreleased 3.x]: https://github.com/opensearch-project/OpenSearch/compare/3.1...main

rest-api-spec/src/main/resources/rest-api-spec/test/search/61_query_string_field_alias.yml

@@ -0,0 +1,56 @@
+setup:
+  - skip:
+      version: " - 3.1.99"
+      reason: "regex query over field alias support starts 3.2"
+
+  - do:
+      indices.create:
+        index: test_index
+        body:
+          settings:
+            number_of_shards: 1
+            number_of_replicas: 0
+          mappings:
+            properties:
+              test:
+                type: text
+              test_alias:
+                type: alias
+                path: test
+
+  - do:
+      bulk:
+        refresh: true
+        body: |
+          {"index":{"_index":"test_index","_id":"1"}}
+          {"test":"hello"}
+          {"index":{"_index":"test_index","_id":"2"}}
+          {"test":"world"}
+
+---
+"regex search on normal field":
+  - do:
+      search:
+        rest_total_hits_as_int: true
+        index: test_index
+        body:
+          query:
+            query_string:
+              query: "test: /h[a-z].*/"
+
+  - match: {hits.total: 1}
+  - match: {hits.hits.0._id: "1"}
+
+---
+"regex search on alias field":
+  - do:
+      search:
+        rest_total_hits_as_int: true
+        index: test_index
+        body:
+          query:
+            query_string:
+              query: "test_alias: /h[a-z].*/"
+
+  - match: {hits.total: 1}
+  - match: {hits.hits.0._id: "1"}

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

@@ -56,6 +56,7 @@
 import org.apache.lucene.search.SynonymQuery;
 import org.apache.lucene.search.WildcardQuery;
 import org.apache.lucene.util.BytesRef;
+import org.apache.lucene.util.automaton.RegExp;
 import org.opensearch.common.lucene.search.Queries;
 import org.opensearch.common.regex.Regex;
 import org.opensearch.common.unit.Fuzziness;
@@ -787,8 +788,12 @@ private Query getRegexpQuerySingle(String field, String termStr) throws ParseExc
             if (currentFieldType == null) {
                 return newUnmappedFieldQuery(field);
             }
-            setAnalyzer(getSearchAnalyzer(currentFieldType));
-            return super.getRegexpQuery(field, termStr);
+            if (forceAnalyzer != null) {
+                setAnalyzer(forceAnalyzer);
+            }
+            // query string query normalizes search value
+            termStr = getAnalyzer().normalize(currentFieldType.name(), termStr).utf8ToString();
+            return currentFieldType.regexpQuery(termStr, RegExp.ALL, 0, getDeterminizeWorkLimit(), getMultiTermRewriteMethod(), context);
         } catch (RuntimeException e) {
             if (lenient) {
                 return newLenientFieldQuery(field, e);

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

@@ -786,6 +786,16 @@ public void testToQueryRegExpQuery() throws Exception {
         assertTrue(regexpQuery.toString().contains("/foo*bar/"));
     }
 
+    public void testRegexpQueryParserWithForceAnalyzer() throws Exception {
+        QueryStringQueryParser queryParser = new QueryStringQueryParser(createShardContext(), TEXT_FIELD_NAME);
+        queryParser.setForceAnalyzer(new org.apache.lucene.analysis.standard.StandardAnalyzer());
+        Query query = queryParser.parse("/aBc.*/");
+        assertThat(query, instanceOf(RegexpQuery.class));
+        RegexpQuery regexpQuery = (RegexpQuery) query;
+        // Standard analyzer normalizes to lowercase, verifying the normalization path with currentFieldType.name() is hit
+        assertTrue(regexpQuery.toString().contains("abc.*"));
+    }
+
     public void testToQueryRegExpQueryTooComplex() throws Exception {
         QueryStringQueryBuilder queryBuilder = queryStringQuery("/[ac]*a[ac]{50,200}/").defaultField(TEXT_FIELD_NAME);