HSEARCH-2219 Translate Lucene analyzer definitions to their Elasticse…

…arch equivalent automatically

There are a few gotchas, most notably:

 * parameters expecting a list of file paths only accept *one* file
   in Elasticsearch
 * the files targeted by file path parameters must be in the config
   directory of Elasticsearch, instead of being in the classpath
 * TypeTokenFilterFactory is not supported, because the "keep_types"
   parameter is a file path in Lucene, but an array of elements in
   Elasticsearch
 * HunspellStemFilterFactory is not supported, because the "dictionary"
   parameter is a list of files in Lucene, but does not exist in
   Elasticsearch (a "locale" parameter is expected instead)
 * A few parameters that are supported in Lucene but not in
   Elasticsearch will trigger an exception when used
 * For now, extra parameters that are not even supported by Lucene will
   not trigger any exception. This may change in the future.

elasticsearch/src/main/java/org/hibernate/search/elasticsearch/analyzer/impl/CustomElasticsearchAnalyzerImpl.java

@@ -0,0 +1,53 @@
+/*
+ * Hibernate Search, full-text search for your domain model
+ *
+ * License: GNU Lesser General Public License (LGPL), version 2.1 or later
+ * See the lgpl.txt file in the root directory or <http://www.gnu.org/licenses/lgpl-2.1.html>.
+ */
+package org.hibernate.search.elasticsearch.analyzer.impl;
+
+import org.hibernate.search.annotations.AnalyzerDef;
+
+/**
+ * A description of an Elasticsearch analyzer built through an analyzer definition.
+ * <p>
+ * This implementation is used whenever {@code @Analyzer(definition = "foo")} is encountered
+ * and an {@code @AnalyzerDefinition} exists with the given name ("foo" in this example).
+ *
+ * @author Guillaume Smet
+ * @author Yoann Rodiere
+ */
+public class CustomElasticsearchAnalyzerImpl implements ElasticsearchAnalyzer {
+
+	private final AnalyzerDef definition;
+
+	public CustomElasticsearchAnalyzerImpl(AnalyzerDef definition) {
+		this.definition = definition;
+	}
+
+	@Override
+	public String getName(String fieldName) {
+		return definition.name();
+	}
+
+	@Override
+	public AnalyzerDef getDefinition(String fieldName) {
+		return definition;
+	}
+
+	@Override
+	public void close() {
+		// nothing to close
+	}
+
+	@Override
+	public String toString() {
+		StringBuilder sb = new StringBuilder();
+		sb.append( getClass().getSimpleName() );
+		sb.append( "<" );
+		sb.append( definition );
+		sb.append( ">" );
+		return sb.toString();
+	}
+
+}

elasticsearch/src/main/java/org/hibernate/search/elasticsearch/analyzer/impl/ElasticsearchAnalyzer.java

@@ -7,6 +7,7 @@
 package org.hibernate.search.elasticsearch.analyzer.impl;
 
 import org.hibernate.search.analyzer.impl.RemoteAnalyzer;
+import org.hibernate.search.annotations.AnalyzerDef;
 
 /**
  * A description of an Elasticsearch analyzer.
@@ -15,4 +16,6 @@
  */
 public interface ElasticsearchAnalyzer extends RemoteAnalyzer {
 
+	AnalyzerDef getDefinition(String fieldName);
+
 }

elasticsearch/src/main/java/org/hibernate/search/elasticsearch/analyzer/impl/ElasticsearchAnalyzerStrategy.java

@@ -21,12 +21,12 @@ public class ElasticsearchAnalyzerStrategy implements AnalyzerStrategy<Elasticse
 
 	@Override
 	public ElasticsearchAnalyzerReference createDefaultAnalyzerReference() {
-		return new ElasticsearchAnalyzerReference( new ElasticsearchAnalyzerImpl( "default" ) );
+		return new ElasticsearchAnalyzerReference( new UndefinedElasticsearchAnalyzerImpl( "default" ) );
 	}
 
 	@Override
 	public ElasticsearchAnalyzerReference createPassThroughAnalyzerReference() {
-		return new ElasticsearchAnalyzerReference( new ElasticsearchAnalyzerImpl( "keyword" ) );
+		return new ElasticsearchAnalyzerReference( new UndefinedElasticsearchAnalyzerImpl( "keyword" ) );
 	}
 
 	@Override
@@ -57,18 +57,19 @@ private void initializeReference(Map<String, ElasticsearchAnalyzer> initializedA
 		ElasticsearchAnalyzer analyzer = initializedAnalyzers.get( name );
 
 		if ( analyzer == null ) {
-			// TODO HSEARCH-2219 Actually use the definition
-			analyzer = buildAnalyzer( name );
+			AnalyzerDef analyzerDefinition = analyzerDefinitions.get( name );
+			if ( analyzerDefinition == null ) {
+				analyzer = new UndefinedElasticsearchAnalyzerImpl( name );
+			}
+			else {
+				analyzer = new CustomElasticsearchAnalyzerImpl( analyzerDefinition );
+			}
 			initializedAnalyzers.put( name, analyzer );
 		}
 
 		analyzerReference.initialize( analyzer );
 	}
 
-	private ElasticsearchAnalyzer buildAnalyzer(String name) {
-		return new ElasticsearchAnalyzerImpl( name );
-	}
-
 	@Override
 	public ScopedElasticsearchAnalyzer.Builder buildScopedAnalyzer(ElasticsearchAnalyzerReference initialGlobalAnalyzerReference) {
 		return new ScopedElasticsearchAnalyzer.Builder(

elasticsearch/src/main/java/org/hibernate/search/elasticsearch/analyzer/impl/ScopedElasticsearchAnalyzer.java

@@ -11,6 +11,7 @@
 
 import org.hibernate.search.analyzer.spi.AnalyzerReference;
 import org.hibernate.search.analyzer.spi.ScopedAnalyzer;
+import org.hibernate.search.annotations.AnalyzerDef;
 import org.hibernate.search.util.logging.impl.Log;
 import org.hibernate.search.util.logging.impl.LoggerFactory;
 
@@ -38,13 +39,22 @@ public ScopedElasticsearchAnalyzer(Builder builder) {
 		this.scopedAnalyzers.putAll( builder.scopedAnalyzers );
 	}
 
-	@Override
-	public String getName(String fieldName) {
+	private ElasticsearchAnalyzerReference getDelegate(String fieldName) {
 		ElasticsearchAnalyzerReference analyzerReference = scopedAnalyzers.get( fieldName );
 		if ( analyzerReference == null ) {
 			analyzerReference = globalAnalyzerReference;
 		}
-		return analyzerReference.getAnalyzer().getName( fieldName );
+		return analyzerReference;
+	}
+
+	@Override
+	public String getName(String fieldName) {
+		return getDelegate( fieldName ).getAnalyzer().getName( fieldName );
+	}
+
+	@Override
+	public AnalyzerDef getDefinition(String fieldName) {
+		return getDelegate( fieldName ).getAnalyzer().getDefinition( fieldName );
 	}
 
 	@Override

elasticsearch/src/main/java/org/hibernate/search/elasticsearch/analyzer/impl/UndefinedElasticsearchAnalyzerImpl.java

@@ -0,0 +1,46 @@
+/*
+ * Hibernate Search, full-text search for your domain model
+ *
+ * License: GNU Lesser General Public License (LGPL), version 2.1 or later
+ * See the lgpl.txt file in the root directory or <http://www.gnu.org/licenses/lgpl-2.1.html>.
+ */
+package org.hibernate.search.elasticsearch.analyzer.impl;
+
+import org.hibernate.search.annotations.AnalyzerDef;
+
+/**
+ * An Elasticsearch analyzer for which no definition was found in the Hibernate Search mapping.
+ * <p>
+ * Such an analyzer is expected to be defined separately on Elasticsearch.
+ * <p>
+ * This implementation is used whenever {@code @Analyzer(definition = "foo")} is encountered
+ * and <strong>no</strong> {@code @AnalyzerDefinition} exists with the given name
+ * ("foo" in this example).
+ *
+ * @author Yoann Rodiere
+ */
+public class UndefinedElasticsearchAnalyzerImpl implements ElasticsearchAnalyzer {
+
+	private final String remoteName;
+
+	public UndefinedElasticsearchAnalyzerImpl(String remoteName) {
+		this.remoteName = remoteName;
+	}
+
+	@Override
+	public void close() {
+		// nothing to do
+	}
+
+	@Override
+	public String getName(String fieldName) {
+		return remoteName;
+	}
+
+	@Override
+	public AnalyzerDef getDefinition(String fieldName) {
+		// No definition
+		return null;
+	}
+
+}

elasticsearch/src/main/java/org/hibernate/search/elasticsearch/impl/ElasticsearchIndexManager.java

@@ -330,14 +330,12 @@ private void reinitializeIndex(Set<Class<?>> entityTypesToInitialize) {
 	}
 
 	private IndexMetadata createIndexMetadata(Collection<Class<?>> classes) {
-		IndexMetadata index = new IndexMetadata();
-		index.setName( actualIndexName );
+		List<EntityIndexBinding> descriptors = new ArrayList<>();
 		for ( Class<?> entityType : classes ) {
-			String entityName = entityType.getName();
 			EntityIndexBinding descriptor = searchIntegrator.getIndexBinding( entityType );
-			index.putMapping( entityName, schemaTranslator.translate( descriptor, schemaManagementExecutionOptions ) );
+			descriptors.add( descriptor );
 		}
-		return index;
+		return schemaTranslator.translate( actualIndexName, descriptors, schemaManagementExecutionOptions );
 	}
 
 	@Override

elasticsearch/src/main/java/org/hibernate/search/elasticsearch/logging/impl/Log.java

@@ -306,4 +306,47 @@ BulkRequestFailedException elasticsearchBulkRequestFailed(String request, String
 	)
 	SearchException indexedEmbeddedPrefixBypass(Class<?> entityType, String fieldPath, String expectedParent);
 
+	@Message(id = ES_BACKEND_MESSAGES_START_ID + 55,
+			value = "The same tokenizer name '%1$s' is used multiple times. The tokenizer names must be unique."
+					+ " If this name was automatically generated, you may override it by using @TokenizerDef.name." )
+	SearchException tokenizerNamingConflict(String remoteName);
+
+	@Message(id = ES_BACKEND_MESSAGES_START_ID + 56,
+			value = "The same char filter name '%1$s' is used multiple times. The char filter names must be unique."
+					+ " If this name was automatically generated, you may override it by using @CharFilterDef.name." )
+	SearchException charFilterNamingConflict(String remoteName);
+
+	@Message(id = ES_BACKEND_MESSAGES_START_ID + 57,
+			value = "The same token filter name '%1$s' is used multiple times. The token filter names must be unique."
+					+ " If this name was automatically generated, you may override it by using @TokenFilterDef.name." )
+	SearchException tokenFilterNamingConflict(String remoteName);
+
+	@Message(id = ES_BACKEND_MESSAGES_START_ID + 58,
+			value = "The char filter factory '%1$s' is not supported with Elasticsearch."
+					+ " Please only use builtin Lucene factories that have a builtin equivalent in Elasticsearch." )
+	SearchException unsupportedCharFilterFactory(@FormatWith(ClassFormatter.class) Class<?> factoryType);
+
+	@Message(id = ES_BACKEND_MESSAGES_START_ID + 59,
+			value = "The tokenizer factory '%1$s' is not supported with Elasticsearch."
+					+ " Please only use builtin Lucene factories that have a builtin equivalent in Elasticsearch." )
+	SearchException unsupportedTokenizerFactory(@FormatWith(ClassFormatter.class) Class<?> factoryType);
+
+	@Message(id = ES_BACKEND_MESSAGES_START_ID + 60,
+			value = "The token filter factory '%1$s' is not supported with Elasticsearch."
+					+ " Please only use builtin Lucene factories that have a builtin equivalent in Elasticsearch." )
+	SearchException unsupportedTokenFilterFactory(@FormatWith(ClassFormatter.class) Class<?> factoryType);
+
+	@Message(id = ES_BACKEND_MESSAGES_START_ID + 61,
+			value = "The parameter '%2$s' is not supported for the factory '%1$s' with Elasticsearch." )
+	SearchException unsupportedAnalysisFactoryParameter(@FormatWith(ClassFormatter.class) Class<?> factoryType, String parameter);
+
+	@Message(id = ES_BACKEND_MESSAGES_START_ID + 62,
+			value = "The parameter '%2$s' for the factory '%1$s' refers to the class '%3$s',"
+					+ " which cannot be converted to a builtin Elasticsearch tokenizer type." )
+	SearchException unsupportedAnalysisFactoryTokenizerClassNameParameter(@FormatWith(ClassFormatter.class) Class<?> factoryClass, String parameterName, String tokenizerClass);
+
+	@Message(id = ES_BACKEND_MESSAGES_START_ID + 63,
+			value = "The parameter '%2$s' for the factory '%1$s' has an unsupported value: '%3$s' is unsupported with Elasticsearch." )
+	SearchException unsupportedAnalysisDefinitionParameterValue(@FormatWith(ClassFormatter.class) Class<?> factoryClass, String parameterName, String parameterValue);
+
 }