OAK-11568 Elastic: improved compatibility for aggregation definitions

Author: thomasmueller

oak-search-elastic/src/main/java/org/apache/jackrabbit/oak/plugins/index/elastic/index/ElasticCustomAnalyzer.java

@@ -40,6 +40,7 @@
 import org.apache.lucene.analysis.AbstractAnalysisFactory;
 import org.apache.lucene.analysis.CharFilterFactory;
 import org.apache.lucene.analysis.TokenFilterFactory;
+import org.apache.lucene.analysis.charfilter.MappingCharFilterFactory;
 import org.apache.lucene.analysis.en.AbstractWordsFileFilterFactory;
 import org.apache.lucene.util.ResourceLoader;
 import org.jetbrains.annotations.NotNull;
@@ -188,6 +189,7 @@ private static <FD> LinkedHashMap<String, FD> loadFilters(NodeState state,
             String name;
             List<String> content = null;
             List<ParameterTransformer> transformers;
+            boolean skipEntry = false;
             try {
                 Class<? extends AbstractAnalysisFactory> tff = lookup.apply(t.getName());
 
@@ -208,6 +210,13 @@ private static <FD> LinkedHashMap<String, FD> loadFilters(NodeState state,
                     wordsFF.inform(new NodeStateResourceLoader(child));
                     content = wordsFF.getWords().stream().map(w -> new String(((char[]) w))).collect(Collectors.toList());
                 }
+                if (luceneFactory instanceof MappingCharFilterFactory) {
+                    MappingCharFilterFactory map = (MappingCharFilterFactory) luceneFactory;
+                    if (map.getOriginalArgs().isEmpty()) {
+                        skipEntry = true;
+                        LOG.warn("Empty CharFilter mapping: ignoring");
+                    }
+                }
 
                 name = normalize((String) tff.getField("NAME").get(null));
                 transformers = LUCENE_ELASTIC_TRANSFORMERS.entrySet().stream()
@@ -245,6 +254,9 @@ private static <FD> LinkedHashMap<String, FD> loadFilters(NodeState state,
             }
             args.put(ANALYZER_TYPE, name);
 
+            if (skipEntry) {
+                continue;
+            }
             filters.put(name + "_" + i, factory.apply(name, JsonData.of(args)));
             i++;
         }

oak-search-elastic/src/test/java/org/apache/jackrabbit/oak/plugins/index/elastic/index/ElasticIndexHelperTest.java

@@ -105,7 +105,7 @@ public void analyzerWithEmptyTokenizer() {
         Tree defaultAnalyzer = analyzer.addChild("default");
         defaultAnalyzer.setProperty(FulltextIndexConstants.ANL_CLASS, "org.apache.lucene.analysis.en.EnglishAnalyzer");
         defaultAnalyzer.addChild("tokenizer");
-        defaultAnalyzer.addChild("filter");
+        defaultAnalyzer.addChild("filters");
 
         NodeState nodeState = builder.build();
         ElasticIndexDefinition definition =
@@ -128,6 +128,34 @@ public void analyzerWithEmptyDefault() {
         ElasticIndexHelper.createIndexRequest("prefix.path", definition);
     }
 
+    @Test
+    public void analyzerWithWordDelimiter() {
+        IndexDefinitionBuilder builder = new ElasticIndexDefinitionBuilder();
+        IndexDefinitionBuilder.IndexRule indexRule = builder.indexRule("idxRule");
+        indexRule.property("foo").type("String").useInSimilarity();
+
+        Tree analyzer = builder.getBuilderTree().addChild("analyzers");
+        Tree defaultAnalyzer = analyzer.addChild("default");
+        Tree tokenizer = defaultAnalyzer.addChild("tokenizer");
+        tokenizer.setProperty("name", "Standard");
+        Tree filters = defaultAnalyzer.addChild("filters");
+        filters.addChild("LowerCase");
+        filters.addChild("WordDelimiter");
+        Tree synonym = filters.addChild("Synonym");
+        synonym.setProperty("format", "solr");
+        synonym.setProperty("ignoreCase", true);
+        synonym.setProperty("synonyms", "synonyms.txt");
+        Tree synonymsText = filters.addChild("synonyms.txt");
+        Tree synonymsContent = synonymsText.addChild("jcr:content");
+        synonymsContent.setProperty("jcr:data", "test");
+        filters.addChild("PorterStem");
+
+        NodeState nodeState = builder.build();
+        ElasticIndexDefinition definition =
+                new ElasticIndexDefinition(nodeState, nodeState, "path", "prefix");
+        ElasticIndexHelper.createIndexRequest("prefix.path", definition);
+    }
+
     @Test()
     public void indexSettingsAreCorrectlySet() {
         IndexDefinitionBuilder builder = new ElasticIndexDefinitionBuilder();

oak-search/src/test/java/org/apache/jackrabbit/oak/plugins/index/FullTextAnalyzerCommonTest.java

@@ -1075,6 +1075,37 @@ public void wildcardQueryToLookupUnanalyzedText() throws Exception {
         });
     }
 
+    // OAK-11568
+    @Test
+    public void analyzerWithEmptyCharFilterMapping() throws Exception {
+        setup(List.of("foo"), idx -> {
+            Tree analyzers = idx.addChild(FulltextIndexConstants.ANALYZERS);
+            Tree defaultAnalyzers = analyzers.addChild(FulltextIndexConstants.ANL_DEFAULT);
+            Tree charFilters = defaultAnalyzers.addChild(FulltextIndexConstants.ANL_CHAR_FILTERS);
+            charFilters.addChild("HTMLStrip");
+
+            // having the mappings, but not having any content, resulted in:
+            // co.elastic.clients.elasticsearch._types.ElasticsearchException:
+            // [es/indices.create] failed: [illegal_argument_exception]
+            // mapping requires either `mappings` or `mappings_path` to be configured
+            charFilters.addChild("Mapping");
+
+            defaultAnalyzers.addChild(FulltextIndexConstants.ANL_TOKENIZER)
+                .setProperty(FulltextIndexConstants.ANL_NAME, "Standard");
+            Tree filters = analyzers.addChild(FulltextIndexConstants.ANL_FILTERS);
+            filters.setOrderableChildren(true);
+            filters.addChild("LowerCase");
+        });
+
+        Tree content = root.getTree("/").addChild("content");
+        content.addChild("bar").setProperty("foo", "foo bar");
+        root.commit();
+
+        assertEventually(() -> {
+            assertQuery("select * from [nt:base] where CONTAINS(*, 'foo')", List.of("/content/bar"));
+        });
+    }
+
     protected Tree addFilter(Tree analyzer, String filterName) {
         Tree filter = analyzer.addChild(filterName);
         // mimics nodes api