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

@@ -19,6 +19,7 @@
 import co.elastic.clients.elasticsearch._types.analysis.Analyzer;
 import co.elastic.clients.elasticsearch._types.analysis.CharFilterDefinition;
 import co.elastic.clients.elasticsearch._types.analysis.CustomAnalyzer;
+import co.elastic.clients.elasticsearch._types.analysis.NGramTokenizer;
 import co.elastic.clients.elasticsearch._types.analysis.TokenFilterDefinition;
 import co.elastic.clients.elasticsearch._types.analysis.TokenizerDefinition;
 import co.elastic.clients.elasticsearch.indices.IndexSettingsAnalysis;
@@ -172,10 +173,28 @@ private static TokenizerDefinition loadTokenizer(NodeState state) {
             }
         }
         name = normalize(name);
+        if ("n_gram".equals(name)) {
+            // OAK-11568
+            // https://www.elastic.co/guide/en/elasticsearch/reference/current/analysis-ngram-tokenizer.html
+            Integer minGramSize = getIntegerSetting(args, "minGramSize", 2);
+            Integer maxGramSize = getIntegerSetting(args, "maxGramSize", 3);
+            TokenizerDefinition ngram = TokenizerDefinition.of(t -> t.ngram(
+                    NGramTokenizer.of(n -> n.minGram(minGramSize).maxGram(maxGramSize))));
+            return ngram;
+        }
         args.put(ANALYZER_TYPE, name);
         return new TokenizerDefinition(name, JsonData.of(args));
     }
 
+    private static Integer getIntegerSetting(Map<String, Object> args, String name, Integer defaultValue) {
+        Object value = args.getOrDefault(name, defaultValue);
+        if (!(value instanceof Integer)) {
+            LOG.warn("Setting {} value {} is not an integer; using default: {}", name, value, defaultValue);
+            return defaultValue;
+        }
+        return (Integer) value;
+    }
+
     private static <FD> LinkedHashMap<String, FD> loadFilters(NodeState state,
                                                               Function<String, Class<? extends AbstractAnalysisFactory>> lookup,
                                                               BiFunction<String, JsonData, FD> factory) {

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

@@ -1092,7 +1092,7 @@ public void analyzerWithEmptyCharFilterMapping() throws Exception {
 
             defaultAnalyzers.addChild(FulltextIndexConstants.ANL_TOKENIZER)
                 .setProperty(FulltextIndexConstants.ANL_NAME, "Standard");
-            Tree filters = analyzers.addChild(FulltextIndexConstants.ANL_FILTERS);
+            Tree filters = defaultAnalyzers.addChild(FulltextIndexConstants.ANL_FILTERS);
             filters.setOrderableChildren(true);
             filters.addChild("LowerCase");
         });
@@ -1106,6 +1106,32 @@ public void analyzerWithEmptyCharFilterMapping() throws Exception {
         });
     }
 
+    // OAK-11568
+    @Test
+    public void analyzerWithNGramTokenizer() throws Exception {
+        setup(List.of("foo"), idx -> {
+            Tree analyzers = idx.addChild(FulltextIndexConstants.ANALYZERS);
+            Tree defaultAnalyzers = analyzers.addChild(FulltextIndexConstants.ANL_DEFAULT);
+            Tree tokenizer = defaultAnalyzers.addChild(FulltextIndexConstants.ANL_TOKENIZER);
+            tokenizer.setProperty(FulltextIndexConstants.ANL_NAME, "NGram");
+            tokenizer.setProperty("maxGramSize", 2);
+            tokenizer.setProperty("minGramSize", 3);
+        });
+
+        Tree content = root.getTree("/").addChild("content");
+        content.addChild("bar").setProperty("foo", "foob bart");
+        root.commit();
+
+        assertEventually(() -> {
+            assertQuery("select * from [nt:base] where contains(*, 'fo')", List.of("/content/bar"));
+            assertQuery("select * from [nt:base] where contains(*, 'foo')", List.of("/content/bar"));
+            assertQuery("select * from [nt:base] where contains(*, 'oob')", List.of("/content/bar"));
+            assertQuery("select * from [nt:base] where contains(*, 'ba')", List.of("/content/bar"));
+            assertQuery("select * from [nt:base] where contains(*, 'bar')", List.of("/content/bar"));
+            assertQuery("select * from [nt:base] where contains(*, 'art')", List.of("/content/bar"));
+        });
+    }
+
     protected Tree addFilter(Tree analyzer, String filterName) {
         Tree filter = analyzer.addChild(filterName);
         // mimics nodes api