Call setReferences() on custom referring tokenfilters in _analyze (#3…

…2157)

When building custom tokenfilters without an index in the _analyze endpoint,
we need to ensure that referring filters are correctly built by calling
their #setReferences() method

Fixes #32154

modules/analysis-common/src/test/resources/rest-api-spec/test/analysis-common/40_token_filters.yml

@@ -1557,3 +1557,18 @@
             filter:    [my_bengali_stem]
     - length: { tokens: 1 }
     - match:  { tokens.0.token: কর }
+
+---
+"multiplexer":
+    - do:
+        indices.analyze:
+          body:
+            text: "The quick fox"
+            tokenizer: "standard"
+            filter:
+              - type: multiplexer
+                filters: [ lowercase, uppercase ]
+                preserve_original: false
+    - length: { tokens: 6 }
+    - match: { tokens.0.token: the }
+    - match: { tokens.1.token: THE }

server/src/main/java/org/elasticsearch/action/admin/indices/analyze/TransportAnalyzeAction.java

@@ -52,6 +52,7 @@
 import org.elasticsearch.index.analysis.IndexAnalyzers;
 import org.elasticsearch.index.analysis.MultiTermAwareComponent;
 import org.elasticsearch.index.analysis.NamedAnalyzer;
+import org.elasticsearch.index.analysis.ReferringFilterFactory;
 import org.elasticsearch.index.analysis.TokenFilterFactory;
 import org.elasticsearch.index.analysis.TokenizerFactory;
 import org.elasticsearch.index.mapper.KeywordFieldMapper;
@@ -574,6 +575,7 @@ private static List<TokenFilterFactory> parseTokenFilterFactories(AnalyzeRequest
                                                                 Environment environment, Tuple<String, TokenizerFactory> tokenizerFactory,
                                                                 List<CharFilterFactory> charFilterFactoryList, boolean normalizer) throws IOException {
         List<TokenFilterFactory> tokenFilterFactoryList = new ArrayList<>();
+        List<ReferringFilterFactory> referringFilters = new ArrayList<>();
         if (request.tokenFilters() != null && request.tokenFilters().size() > 0) {
             List<AnalyzeRequest.NameOrDefinition> tokenFilters = request.tokenFilters();
             for (AnalyzeRequest.NameOrDefinition tokenFilter : tokenFilters) {
@@ -594,7 +596,9 @@ private static List<TokenFilterFactory> parseTokenFilterFactories(AnalyzeRequest
                     tokenFilterFactory = tokenFilterFactoryFactory.get(getNaIndexSettings(settings), environment, "_anonymous_tokenfilter", settings);
                     tokenFilterFactory = CustomAnalyzerProvider.checkAndApplySynonymFilter(tokenFilterFactory, tokenizerFactory.v1(), tokenizerFactory.v2(), tokenFilterFactoryList,
                         charFilterFactoryList, environment);
-
+                    if (tokenFilterFactory instanceof ReferringFilterFactory) {
+                        referringFilters.add((ReferringFilterFactory)tokenFilterFactory);
+                    }
 
                 } else {
                     AnalysisModule.AnalysisProvider<TokenFilterFactory> tokenFilterFactoryFactory;
@@ -629,6 +633,26 @@ private static List<TokenFilterFactory> parseTokenFilterFactories(AnalyzeRequest
                 tokenFilterFactoryList.add(tokenFilterFactory);
             }
         }
+        if (referringFilters.isEmpty() == false) {
+            // The request included at least one custom referring tokenfilter that has not already been built by the
+            // analysis registry, so we need to set its references.  Note that this will only apply pre-built
+            // tokenfilters
+            if (indexSettings == null) {
+                Settings settings = Settings.builder()
+                    .put(IndexMetaData.SETTING_VERSION_CREATED, Version.CURRENT)
+                    .put(IndexMetaData.SETTING_NUMBER_OF_REPLICAS, 0)
+                    .put(IndexMetaData.SETTING_NUMBER_OF_SHARDS, 1)
+                    .put(IndexMetaData.SETTING_INDEX_UUID, UUIDs.randomBase64UUID())
+                    .build();
+                IndexMetaData metaData = IndexMetaData.builder(IndexMetaData.INDEX_UUID_NA_VALUE).settings(settings).build();
+                indexSettings = new IndexSettings(metaData, Settings.EMPTY);
+            }
+            Map<String, TokenFilterFactory> prebuiltFilters = analysisRegistry.buildTokenFilterFactories(indexSettings);
+            for (ReferringFilterFactory rff : referringFilters) {
+                rff.setReferences(prebuiltFilters);
+            }
+
+        }
         return tokenFilterFactoryList;
     }