Use preconfigured filters correctly in Analyze API (#43568)

When a named token filter or char filter is passed as part of an Analyze API
request with no index, we currently try and build the relevant filter using no
index settings. However, this can miss cases where there is a pre-configured
filter defined in the analysis registry. One example here is the elision filter, which
has a pre-configured version built with the french elision set; when used as part
of normal analysis, this preconfigured set is used, but when used as part of the
Analyze API we end up with NPEs because it tries to instantiate the filter with
no index settings.

This commit changes the Analyze API to check for pre-configured filters in the case
that the request has no index defined, and is using a name rather than a custom
definition for a filter.

It also changes the pre-configured `word_delimiter_graph` filter and `edge_ngram`
tokenizer to make their settings consistent with the defaults used when creating
them with no settings

Closes #43002
Closes #43621
Closes #43582

modules/analysis-common/src/main/java/org/elasticsearch/analysis/common/CommonAnalysisPlugin.java

@@ -83,6 +83,7 @@ import org.apache.lucene.analysis.miscellaneous.TrimFilter;
 import org.apache.lucene.analysis.miscellaneous.TruncateTokenFilter;
 import org.apache.lucene.analysis.miscellaneous.WordDelimiterFilter;
 import org.apache.lucene.analysis.miscellaneous.WordDelimiterGraphFilter;
+import org.apache.lucene.analysis.miscellaneous.WordDelimiterIterator;
 import org.apache.lucene.analysis.ngram.EdgeNGramTokenFilter;
 import org.apache.lucene.analysis.ngram.EdgeNGramTokenizer;
 import org.apache.lucene.analysis.ngram.NGramTokenFilter;
@@ -110,6 +111,7 @@ import org.apache.lucene.analysis.tr.ApostropheFilter;
 import org.apache.lucene.analysis.tr.TurkishAnalyzer;
 import org.apache.lucene.analysis.util.ElisionFilter;
 import org.apache.lucene.util.SetOnce;
+import org.elasticsearch.Version;
 import org.elasticsearch.client.Client;
 import org.elasticsearch.cluster.service.ClusterService;
 import org.elasticsearch.common.io.stream.NamedWriteableRegistry;
@@ -488,13 +490,15 @@ public class CommonAnalysisPlugin extends Plugin implements AnalysisPlugin, Scri
                       | WordDelimiterFilter.SPLIT_ON_CASE_CHANGE
                       | WordDelimiterFilter.SPLIT_ON_NUMERICS
                       | WordDelimiterFilter.STEM_ENGLISH_POSSESSIVE, null)));
-        filters.add(PreConfiguredTokenFilter.singleton("word_delimiter_graph", false, input ->
+        filters.add(PreConfiguredTokenFilter.singletonWithVersion("word_delimiter_graph", false, (input, version) -> {
-                new WordDelimiterGraphFilter(input,
+            boolean adjustOffsets = version.onOrAfter(Version.V_7_3_0);
+            return new WordDelimiterGraphFilter(input, adjustOffsets, WordDelimiterIterator.DEFAULT_WORD_DELIM_TABLE,
                         WordDelimiterGraphFilter.GENERATE_WORD_PARTS
                       | WordDelimiterGraphFilter.GENERATE_NUMBER_PARTS
                       | WordDelimiterGraphFilter.SPLIT_ON_CASE_CHANGE
                       | WordDelimiterGraphFilter.SPLIT_ON_NUMERICS
-                      | WordDelimiterGraphFilter.STEM_ENGLISH_POSSESSIVE, null)));
+                      | WordDelimiterGraphFilter.STEM_ENGLISH_POSSESSIVE, null);
+        }));
         return filters;
     }
 
@@ -508,8 +512,12 @@ public class CommonAnalysisPlugin extends Plugin implements AnalysisPlugin, Scri
         tokenizers.add(PreConfiguredTokenizer.singleton("letter", LetterTokenizer::new));
         tokenizers.add(PreConfiguredTokenizer.singleton("whitespace", WhitespaceTokenizer::new));
         tokenizers.add(PreConfiguredTokenizer.singleton("ngram", NGramTokenizer::new));
-        tokenizers.add(PreConfiguredTokenizer.singleton("edge_ngram",
+        tokenizers.add(PreConfiguredTokenizer.elasticsearchVersion("edge_ngram", (version) -> {
-            () -> new EdgeNGramTokenizer(EdgeNGramTokenizer.DEFAULT_MIN_GRAM_SIZE, EdgeNGramTokenizer.DEFAULT_MAX_GRAM_SIZE)));
+            if (version.onOrAfter(Version.V_7_3_0)) {
+                return new EdgeNGramTokenizer(NGramTokenizer.DEFAULT_MIN_NGRAM_SIZE, NGramTokenizer.DEFAULT_MAX_NGRAM_SIZE);
+            }
+            return new EdgeNGramTokenizer(EdgeNGramTokenizer.DEFAULT_MIN_GRAM_SIZE, EdgeNGramTokenizer.DEFAULT_MAX_GRAM_SIZE);
+        }));
         tokenizers.add(PreConfiguredTokenizer.singleton("pattern", () -> new PatternTokenizer(Regex.compile("\\W+", null), -1)));
         tokenizers.add(PreConfiguredTokenizer.singleton("thai", ThaiTokenizer::new));
         // TODO deprecate and remove in API
@@ -518,8 +526,12 @@ public class CommonAnalysisPlugin extends Plugin implements AnalysisPlugin, Scri
 
         // Temporary shim for aliases. TODO deprecate after they are moved
         tokenizers.add(PreConfiguredTokenizer.singleton("nGram", NGramTokenizer::new));
-        tokenizers.add(PreConfiguredTokenizer.singleton("edgeNGram",
+        tokenizers.add(PreConfiguredTokenizer.elasticsearchVersion("edgeNGram", (version) -> {
-            () -> new EdgeNGramTokenizer(EdgeNGramTokenizer.DEFAULT_MIN_GRAM_SIZE, EdgeNGramTokenizer.DEFAULT_MAX_GRAM_SIZE)));
+            if (version.onOrAfter(Version.V_7_3_0)) {
+                return new EdgeNGramTokenizer(NGramTokenizer.DEFAULT_MIN_NGRAM_SIZE, NGramTokenizer.DEFAULT_MAX_NGRAM_SIZE);
+            }
+            return new EdgeNGramTokenizer(EdgeNGramTokenizer.DEFAULT_MIN_GRAM_SIZE, EdgeNGramTokenizer.DEFAULT_MAX_GRAM_SIZE);
+        }));
         tokenizers.add(PreConfiguredTokenizer.singleton("PathHierarchy", PathHierarchyTokenizer::new));
 
         return tokenizers;

modules/analysis-common/src/test/java/org/elasticsearch/analysis/common/EdgeNGramTokenizerTests.java

@@ -0,0 +1,98 @@
+/*
+ * Licensed to Elasticsearch under one or more contributor
+ * license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright
+ * ownership. Elasticsearch licenses this file to you under
+ * the Apache License, Version 2.0 (the "License"); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+package org.elasticsearch.analysis.common;
+
+import org.elasticsearch.Version;
+import org.elasticsearch.cluster.metadata.IndexMetaData;
+import org.elasticsearch.common.settings.Settings;
+import org.elasticsearch.env.Environment;
+import org.elasticsearch.env.TestEnvironment;
+import org.elasticsearch.index.IndexSettings;
+import org.elasticsearch.index.analysis.IndexAnalyzers;
+import org.elasticsearch.index.analysis.NamedAnalyzer;
+import org.elasticsearch.indices.analysis.AnalysisModule;
+import org.elasticsearch.test.ESTokenStreamTestCase;
+import org.elasticsearch.test.IndexSettingsModule;
+import org.elasticsearch.test.VersionUtils;
+
+import java.io.IOException;
+import java.util.Collections;
+
+public class EdgeNGramTokenizerTests extends ESTokenStreamTestCase {
+
+    private IndexAnalyzers buildAnalyzers(Version version, String tokenizer) throws IOException {
+        Settings settings = Settings.builder()
+            .put(Environment.PATH_HOME_SETTING.getKey(), createTempDir().toString())
+            .build();
+        Settings indexSettings = Settings.builder()
+            .put(IndexMetaData.SETTING_VERSION_CREATED, version)
+            .put("index.analysis.analyzer.my_analyzer.tokenizer", tokenizer)
+            .build();
+        IndexSettings idxSettings = IndexSettingsModule.newIndexSettings("index", indexSettings);
+        return new AnalysisModule(TestEnvironment.newEnvironment(settings),
+            Collections.singletonList(new CommonAnalysisPlugin())).getAnalysisRegistry().build(idxSettings);
+    }
+
+    public void testPreConfiguredTokenizer() throws IOException {
+
+        // Before 7.3 we return ngrams of length 1 only
+        {
+            Version version = VersionUtils.randomVersionBetween(random(), Version.V_7_0_0,
+                VersionUtils.getPreviousVersion(Version.V_7_3_0));
+            try (IndexAnalyzers indexAnalyzers = buildAnalyzers(version, "edge_ngram")) {
+                NamedAnalyzer analyzer = indexAnalyzers.get("my_analyzer");
+                assertNotNull(analyzer);
+                assertAnalyzesTo(analyzer, "test", new String[]{"t"});
+            }
+        }
+
+        // Check deprecated name as well
+        {
+            Version version = VersionUtils.randomVersionBetween(random(), Version.V_7_0_0,
+                VersionUtils.getPreviousVersion(Version.V_7_3_0));
+            try (IndexAnalyzers indexAnalyzers = buildAnalyzers(version, "edgeNGram")) {
+                NamedAnalyzer analyzer = indexAnalyzers.get("my_analyzer");
+                assertNotNull(analyzer);
+                assertAnalyzesTo(analyzer, "test", new String[]{"t"});
+            }
+        }
+
+        // Afterwards, we return ngrams of length 1 and 2, to match the default factory settings
+        {
+            try (IndexAnalyzers indexAnalyzers = buildAnalyzers(Version.CURRENT, "edge_ngram")) {
+                NamedAnalyzer analyzer = indexAnalyzers.get("my_analyzer");
+                assertNotNull(analyzer);
+                assertAnalyzesTo(analyzer, "test", new String[]{"t", "te"});
+            }
+        }
+
+        // Check deprecated name as well
+        {
+            try (IndexAnalyzers indexAnalyzers = buildAnalyzers(Version.CURRENT, "edgeNGram")) {
+                NamedAnalyzer analyzer = indexAnalyzers.get("my_analyzer");
+                assertNotNull(analyzer);
+                assertAnalyzesTo(analyzer, "test", new String[]{"t", "te"});
+
+            }
+        }
+
+    }
+
+}

modules/analysis-common/src/test/java/org/elasticsearch/analysis/common/WordDelimiterGraphTokenFilterFactoryTests.java

@@ -20,14 +20,24 @@ package org.elasticsearch.analysis.common;
 
 import org.apache.lucene.analysis.Tokenizer;
 import org.apache.lucene.analysis.core.WhitespaceTokenizer;
+import org.elasticsearch.Version;
+import org.elasticsearch.cluster.metadata.IndexMetaData;
 import org.elasticsearch.common.settings.Settings;
 import org.elasticsearch.env.Environment;
+import org.elasticsearch.env.TestEnvironment;
+import org.elasticsearch.index.IndexSettings;
 import org.elasticsearch.index.analysis.AnalysisTestsHelper;
+import org.elasticsearch.index.analysis.IndexAnalyzers;
+import org.elasticsearch.index.analysis.NamedAnalyzer;
 import org.elasticsearch.index.analysis.TokenFilterFactory;
+import org.elasticsearch.indices.analysis.AnalysisModule;
 import org.elasticsearch.test.ESTestCase;
+import org.elasticsearch.test.IndexSettingsModule;
+import org.elasticsearch.test.VersionUtils;
 
 import java.io.IOException;
 import java.io.StringReader;
+import java.util.Collections;
 
 public class WordDelimiterGraphTokenFilterFactoryTests
         extends BaseWordDelimiterTokenFilterFactoryTestCase {
@@ -107,4 +117,51 @@ public class WordDelimiterGraphTokenFilterFactoryTests
         assertTokenStreamContents(tokenFilter.create(tokenizer), expected, expectedStartOffsets, expectedEndOffsets, null,
             expectedIncr, expectedPosLen, null);
     }
+
+    public void testPreconfiguredFilter() throws IOException {
+        // Before 7.3 we don't adjust offsets
+        {
+            Settings settings = Settings.builder()
+                .put(Environment.PATH_HOME_SETTING.getKey(), createTempDir().toString())
+                .build();
+            Settings indexSettings = Settings.builder()
+                .put(IndexMetaData.SETTING_VERSION_CREATED,
+                    VersionUtils.randomVersionBetween(random(), Version.V_7_0_0, VersionUtils.getPreviousVersion(Version.V_7_3_0)))
+                .put("index.analysis.analyzer.my_analyzer.tokenizer", "standard")
+                .putList("index.analysis.analyzer.my_analyzer.filter", "word_delimiter_graph")
+                .build();
+            IndexSettings idxSettings = IndexSettingsModule.newIndexSettings("index", indexSettings);
+
+            try (IndexAnalyzers indexAnalyzers = new AnalysisModule(TestEnvironment.newEnvironment(settings),
+                Collections.singletonList(new CommonAnalysisPlugin())).getAnalysisRegistry().build(idxSettings)) {
+
+                NamedAnalyzer analyzer = indexAnalyzers.get("my_analyzer");
+                assertNotNull(analyzer);
+                assertAnalyzesTo(analyzer, "h100", new String[]{"h", "100"}, new int[]{ 0, 0 }, new int[]{ 4, 4 });
+
+            }
+        }
+
+        // Afger 7.3 we do adjust offsets
+        {
+            Settings settings = Settings.builder()
+                .put(Environment.PATH_HOME_SETTING.getKey(), createTempDir().toString())
+                .build();
+            Settings indexSettings = Settings.builder()
+                .put(IndexMetaData.SETTING_VERSION_CREATED, Version.CURRENT)
+                .put("index.analysis.analyzer.my_analyzer.tokenizer", "standard")
+                .putList("index.analysis.analyzer.my_analyzer.filter", "word_delimiter_graph")
+                .build();
+            IndexSettings idxSettings = IndexSettingsModule.newIndexSettings("index", indexSettings);
+
+            try (IndexAnalyzers indexAnalyzers = new AnalysisModule(TestEnvironment.newEnvironment(settings),
+                Collections.singletonList(new CommonAnalysisPlugin())).getAnalysisRegistry().build(idxSettings)) {
+
+                NamedAnalyzer analyzer = indexAnalyzers.get("my_analyzer");
+                assertNotNull(analyzer);
+                assertAnalyzesTo(analyzer, "h100", new String[]{"h", "100"}, new int[]{ 0, 1 }, new int[]{ 1, 4 });
+
+            }
+        }
+    }
 }

server/src/main/java/org/elasticsearch/index/analysis/AnalysisRegistry.java

@@ -111,6 +111,7 @@ public final class AnalysisRegistry implements Closeable {
     private <T> T getComponentFactory(IndexSettings settings, NameOrDefinition nod,
                                       String componentType,
                                       Function<String, AnalysisProvider<T>> globalComponentProvider,
+                                      Function<String, AnalysisProvider<T>> prebuiltComponentProvider,
                                       BiFunction<String, IndexSettings, AnalysisProvider<T>> indexComponentProvider) throws IOException {
         if (nod.definition != null) {
             // custom component, so we build it from scratch
@@ -128,10 +129,14 @@ public final class AnalysisRegistry implements Closeable {
             return factory.get(settings, environment, "__anonymous__" + type, nod.definition);
         }
         if (settings == null) {
-            // no index provided, so we use global analysis components only
+            // no index provided, so we use prebuilt analysis components
-            AnalysisProvider<T> factory = globalComponentProvider.apply(nod.name);
+            AnalysisProvider<T> factory = prebuiltComponentProvider.apply(nod.name);
             if (factory == null) {
-                throw new IllegalArgumentException("failed to find global " + componentType + " under [" + nod.name + "]");
+                // if there's no prebuilt component, try loading a global one to build with no settings
+                factory = globalComponentProvider.apply(nod.name);
+                if (factory == null) {
+                    throw new IllegalArgumentException("failed to find global " + componentType + " under [" + nod.name + "]");
+                }
             }
             return factory.get(environment, nod.name);
         } else {
@@ -219,25 +224,26 @@ public final class AnalysisRegistry implements Closeable {
     public NamedAnalyzer buildCustomAnalyzer(IndexSettings indexSettings, boolean normalizer, NameOrDefinition tokenizer,
                                              List<NameOrDefinition> charFilters, List<NameOrDefinition> tokenFilters) throws IOException {
         TokenizerFactory tokenizerFactory
-            = getComponentFactory(indexSettings, tokenizer, "tokenizer", this::getTokenizerProvider, this::getTokenizerProvider);
+            = getComponentFactory(indexSettings, tokenizer, "tokenizer",
+            this::getTokenizerProvider, prebuiltAnalysis::getTokenizerFactory, this::getTokenizerProvider);
 
         List<CharFilterFactory> charFilterFactories = new ArrayList<>();
         for (NameOrDefinition nod : charFilters) {
             charFilterFactories.add(getComponentFactory(indexSettings, nod, "char_filter",
-                this::getCharFilterProvider, this::getCharFilterProvider));
+                this::getCharFilterProvider, prebuiltAnalysis::getCharFilterFactory, this::getCharFilterProvider));
         }
 
         List<TokenFilterFactory> tokenFilterFactories = new ArrayList<>();
         for (NameOrDefinition nod : tokenFilters) {
             TokenFilterFactory tff = getComponentFactory(indexSettings, nod, "filter",
-                this::getTokenFilterProvider, this::getTokenFilterProvider);
+                this::getTokenFilterProvider, prebuiltAnalysis::getTokenFilterFactory, this::getTokenFilterProvider);
             if (normalizer && tff instanceof NormalizingTokenFilterFactory == false) {
                 throw new IllegalArgumentException("Custom normalizer may not use filter [" + tff.name() + "]");
             }
             tff = tff.getChainAwareTokenFilterFactory(tokenizerFactory, charFilterFactories, tokenFilterFactories, name -> {
                 try {
                     return getComponentFactory(indexSettings, new NameOrDefinition(name), "filter",
-                        this::getTokenFilterProvider, this::getTokenFilterProvider);
+                        this::getTokenFilterProvider, prebuiltAnalysis::getTokenFilterFactory, this::getTokenFilterProvider);
                 } catch (IOException e) {
                     throw new UncheckedIOException(e);
                 }

server/src/test/java/org/elasticsearch/action/admin/indices/TransportAnalyzeActionTests.java

@@ -142,7 +142,7 @@ public class TransportAnalyzeActionTests extends ESTestCase {
 
             @Override
             public List<PreConfiguredCharFilter> getPreConfiguredCharFilters() {
-                return singletonList(PreConfiguredCharFilter.singleton("append_foo", false, reader -> new AppendCharFilter(reader, "foo")));
+                return singletonList(PreConfiguredCharFilter.singleton("append", false, reader -> new AppendCharFilter(reader, "foo")));
             }
         };
         registry = new AnalysisModule(environment, singletonList(plugin)).getAnalysisRegistry();
@@ -171,24 +171,11 @@ public class TransportAnalyzeActionTests extends ESTestCase {
         List<AnalyzeAction.AnalyzeToken> tokens = analyze.getTokens();
         assertEquals(4, tokens.size());
 
-        // Refer to a token filter by its type so we get its default configuration
-        request = new AnalyzeAction.Request();
-        request.text("the qu1ck brown fox");
-        request.tokenizer("standard");
-        request.addTokenFilter("mock");
-        analyze
-            = TransportAnalyzeAction.analyze(request, registry, null, maxTokenCount);
-        tokens = analyze.getTokens();
-        assertEquals(3, tokens.size());
-        assertEquals("qu1ck", tokens.get(0).getTerm());
-        assertEquals("brown", tokens.get(1).getTerm());
-        assertEquals("fox", tokens.get(2).getTerm());
-
         // We can refer to a pre-configured token filter by its name to get it
         request = new AnalyzeAction.Request();
         request.text("the qu1ck brown fox");
         request.tokenizer("standard");
-        request.addCharFilter("append_foo");
+        request.addCharFilter("append");        // <-- no config, so use preconfigured filter
         analyze
             = TransportAnalyzeAction.analyze(request, registry, null, maxTokenCount);
         tokens = analyze.getTokens();
@@ -198,31 +185,32 @@ public class TransportAnalyzeActionTests extends ESTestCase {
         assertEquals("brown", tokens.get(2).getTerm());
         assertEquals("foxfoo", tokens.get(3).getTerm());
 
-        // We can refer to a token filter by its type to get its default configuration
+        // If the preconfigured filter doesn't exist, we use a global filter with no settings
         request = new AnalyzeAction.Request();
         request.text("the qu1ck brown fox");
         request.tokenizer("standard");
-        request.addCharFilter("append");
+        request.addTokenFilter("mock");     // <-- not preconfigured, but a global one available
-        request.text("the qu1ck brown fox");
         analyze
             = TransportAnalyzeAction.analyze(request, registry, null, maxTokenCount);
         tokens = analyze.getTokens();
-        assertEquals(4, tokens.size());
+        assertEquals(3, tokens.size());
-        assertEquals("the", tokens.get(0).getTerm());
+        assertEquals("qu1ck", tokens.get(0).getTerm());
-        assertEquals("qu1ck", tokens.get(1).getTerm());
+        assertEquals("brown", tokens.get(1).getTerm());
-        assertEquals("brown", tokens.get(2).getTerm());
+        assertEquals("fox", tokens.get(2).getTerm());
-        assertEquals("foxbar", tokens.get(3).getTerm());
 
-        // We can pass a new configuration
-        request = new AnalyzeAction.Request();
-        request.text("the qu1ck brown fox");
-        request.tokenizer("standard");
         Map<String, Object> tokenFilterConfig = new HashMap<>();
         tokenFilterConfig.put("type", "mock");
         tokenFilterConfig.put("stopword", "brown");
-        request.addTokenFilter(tokenFilterConfig);
+
-        request.addCharFilter("append");
+        Map<String, Object> charFilterConfig = new HashMap<>();
+        charFilterConfig.put("type", "append");
+
+        // We can build a new char filter to get default values
+        request = new AnalyzeAction.Request();
         request.text("the qu1ck brown fox");
+        request.tokenizer("standard");
+        request.addTokenFilter(tokenFilterConfig);
+        request.addCharFilter(charFilterConfig);    // <-- basic config, uses defaults
         analyze
             = TransportAnalyzeAction.analyze(request, registry, null, maxTokenCount);
         tokens = analyze.getTokens();
@@ -230,6 +218,21 @@ public class TransportAnalyzeActionTests extends ESTestCase {
         assertEquals("the", tokens.get(0).getTerm());
         assertEquals("qu1ck", tokens.get(1).getTerm());
         assertEquals("foxbar", tokens.get(2).getTerm());
+
+        // We can pass a new configuration
+        request = new AnalyzeAction.Request();
+        request.text("the qu1ck brown fox");
+        request.tokenizer("standard");
+        request.addTokenFilter(tokenFilterConfig);
+        charFilterConfig.put("suffix", "baz");
+        request.addCharFilter(charFilterConfig);
+        analyze
+            = TransportAnalyzeAction.analyze(request, registry, null, maxTokenCount);
+        tokens = analyze.getTokens();
+        assertEquals(3, tokens.size());
+        assertEquals("the", tokens.get(0).getTerm());
+        assertEquals("qu1ck", tokens.get(1).getTerm());
+        assertEquals("foxbaz", tokens.get(2).getTerm());
     }
 
     public void testFillsAttributes() throws IOException {