iter

server/src/main/java/org/elasticsearch/action/update/TransportUpdateAction.java

@@ -412,7 +412,7 @@ private static UpdateHelper.Result deleteInferenceResults(
                         // This has two important side effects:
                         // - The inference field value will remain parsable by its mapper
                         // - The inference results will be removed, forcing them to be re-generated downstream
-                        updatedSource.put(inferenceFieldName, inferenceFieldMapper.getOriginalValue(updatedSource));
+                        updatedSource.put(inferenceFieldName, getOriginalValueLegacy(inferenceFieldName, updatedSource));
                         updatedSourceModified = true;
                         break;
                     }
@@ -435,4 +435,23 @@ private static UpdateHelper.Result deleteInferenceResults(
 
         return returnedResult;
     }
+
+    /**
+     * Get the field's original value (i.e. the value the user specified) from the provided source.
+     *
+     * @param sourceAsMap The source as a map
+     * @return The field's original value, or {@code null} if none was provided
+     */
+    private static Object getOriginalValueLegacy(String fullPath, Map<String, Object> sourceAsMap) {
+        Object fieldValue = sourceAsMap.get(fullPath);
+        if (fieldValue == null) {
+            return null;
+        } else if (fieldValue instanceof Map<?, ?> == false) {
+            // Don't try to further validate the non-map value, that will be handled when the source is fully parsed
+            return fieldValue;
+        }
+
+        Map<String, Object> fieldValueMap = XContentMapValues.nodeMapValue(fieldValue, "Field [" + fullPath + "]");
+        return XContentMapValues.extractValue("text", fieldValueMap);
+    }
 }

server/src/main/java/org/elasticsearch/index/mapper/DocumentParserContext.java

@@ -66,8 +66,8 @@ public boolean isWithinInferenceMetadata() {
         }
 
         @Override
-        public void markInferenceMetadata() {
-            in.markInferenceMetadata();
+        public void markInferenceMetadataField() {
+            in.markInferenceMetadataField();
         }
 
         @Override
@@ -155,7 +155,7 @@ private enum Scope {
     // Indicates if the source for this context has been marked to be recorded. Applies to synthetic source only.
     private boolean recordedSource;
 
-    private boolean inferenceMetadata;
+    private boolean hasInferenceMetadata;
 
     private DocumentParserContext(
         MappingLookup mappingLookup,
@@ -349,12 +349,19 @@ public final DocumentParserContext addIgnoredFieldFromContext(IgnoredSourceField
         return this;
     }
 
-    public void markInferenceMetadata() {
-        this.inferenceMetadata = true;
+    /**
+     * Called by {@link InferenceMetadataFieldsMapper} to indicate whether the metadata field is present
+     * in _source.
+     */
+    public void markInferenceMetadataField() {
+        this.hasInferenceMetadata = true;
     }
 
-    public final boolean hasInferenceMetadata() {
-        return false;// TODO: inferenceMetadata;
+    /**
+     * Returns whether the _source contains an inference metadata field.
+     */
+    public final boolean hasInferenceMetadataField() {
+        return hasInferenceMetadata;
     }
 
     /**

server/src/main/java/org/elasticsearch/index/mapper/InferenceFieldMapper.java

@@ -12,7 +12,6 @@
 import org.elasticsearch.cluster.metadata.InferenceFieldMetadata;
 import org.elasticsearch.inference.InferenceService;
 
-import java.util.Map;
 import java.util.Set;
 
 /**
@@ -26,12 +25,4 @@ public interface InferenceFieldMapper {
      * @param sourcePaths The source path that populates the input for the field (before inference)
      */
     InferenceFieldMetadata getMetadata(Set<String> sourcePaths);
-
-    /**
-     * Get the field's original value (i.e. the value the user specified) from the provided source.
-     *
-     * @param sourceAsMap The source as a map
-     * @return The field's original value, or {@code null} if none was provided
-     */
-    Object getOriginalValue(Map<String, Object> sourceAsMap);
 }

server/src/main/java/org/elasticsearch/index/mapper/InferenceMetadataFieldsMapper.java

@@ -71,10 +71,11 @@ protected boolean supportsParsingObject() {
     protected void parseCreateField(DocumentParserContext context) throws IOException {
         XContentParser parser = context.parser();
         XContentParserUtils.ensureExpectedToken(XContentParser.Token.START_OBJECT, parser.currentToken(), parser);
-        context.markInferenceMetadata();
+        context.markInferenceMetadataField();
         while (parser.nextToken() != XContentParser.Token.END_OBJECT) {
             XContentParserUtils.ensureExpectedToken(XContentParser.Token.FIELD_NAME, parser.currentToken(), parser);
             String fieldName = parser.currentName();
+            // TODO: Find the leaf field under objects
             Mapper mapper = context.mappingLookup().getMapper(fieldName);
             if (mapper != null && mapper instanceof InferenceFieldMapper && mapper instanceof FieldMapper fieldMapper) {
                 fieldMapper.parseCreateField(new DocumentParserContext.Wrapper(context.parent(), context) {

server/src/main/java/org/elasticsearch/index/mapper/SourceFieldMapper.java

@@ -428,7 +428,7 @@ public BytesReference applyFilters(
             return null;
         }
         var modSourceFilter = sourceFilter;
-        if (context != null && context.hasInferenceMetadata()) {
+        if (context != null && context.hasInferenceMetadataField()) {
             String[] modExcludes = new String[excludes != null ? excludes.length + 1 : 1];
             if (excludes != null) {
                 System.arraycopy(excludes, 0, modExcludes, 0, excludes.length);

test/framework/src/main/java/org/elasticsearch/index/mapper/MapperTestCase.java

@@ -1166,8 +1166,10 @@ public void testSupportsParsingObject() throws IOException {
             Object sampleValueForDocument = getSampleObjectForDocument();
             assertThat(sampleValueForDocument, instanceOf(Map.class));
             SourceToParse source = source(builder -> {
+                builder.startObject(InferenceMetadataFieldsMapper.NAME);
                 builder.field("field");
                 builder.value(sampleValueForDocument);
+                builder.endObject();
             });
             ParsedDocument doc = mapper.parse(source);
             assertNotNull(doc);

x-pack/plugin/core/build.gradle

@@ -68,6 +68,7 @@ dependencies {
   testImplementation project(path: ':modules:analysis-common')
   testImplementation project(path: ':modules:rest-root')
   testImplementation project(path: ':modules:health-shards-availability')
+  testImplementation project(path: ':modules:mapper-extras')
   // Needed for Fips140ProviderVerificationTests
   testCompileOnly('org.bouncycastle:bc-fips:1.0.2.5')
 

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/queries/SparseVectorQuery.java → x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/ml/search/SparseVectorQuery.java

@@ -5,7 +5,7 @@
  * 2.0.
  */
 
-package org.elasticsearch.xpack.inference.queries;
+package org.elasticsearch.xpack.core.ml.search;
 
 import org.apache.lucene.search.BooleanClause;
 import org.apache.lucene.search.IndexSearcher;

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/queries/SparseVectorQueryBuilder.java → x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/ml/search/SparseVectorQueryBuilder.java

@@ -5,7 +5,7 @@
  * 2.0.
  */
 
-package org.elasticsearch.xpack.inference.queries;
+package org.elasticsearch.xpack.core.ml.search;
 
 import org.apache.lucene.search.MatchNoDocsQuery;
 import org.apache.lucene.search.Query;
@@ -33,7 +33,6 @@
 import org.elasticsearch.xpack.core.ml.inference.results.TextExpansionResults;
 import org.elasticsearch.xpack.core.ml.inference.results.WarningInferenceResults;
 import org.elasticsearch.xpack.core.ml.inference.trainedmodel.TextExpansionConfigUpdate;
-import org.elasticsearch.xpack.core.ml.search.WeightedToken;
 
 import java.io.IOException;
 import java.util.ArrayList;

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/queries/TextExpansionQueryBuilder.java → x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/ml/search/TextExpansionQueryBuilder.java

@@ -5,7 +5,7 @@
  * 2.0.
  */
 
-package org.elasticsearch.xpack.inference.queries;
+package org.elasticsearch.xpack.core.ml.search;
 
 import org.apache.lucene.search.Query;
 import org.apache.lucene.util.SetOnce;
@@ -39,7 +39,7 @@
 
 import static org.elasticsearch.xpack.core.ClientHelper.ML_ORIGIN;
 import static org.elasticsearch.xpack.core.ClientHelper.executeAsyncWithOrigin;
-import static org.elasticsearch.xpack.inference.queries.WeightedTokensQueryBuilder.PRUNING_CONFIG;
+import static org.elasticsearch.xpack.core.ml.search.WeightedTokensQueryBuilder.PRUNING_CONFIG;
 
 /**
  * @deprecated Replaced by sparse_vector query

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/queries/TokenPruningConfig.java → x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/ml/search/TokenPruningConfig.java

@@ -5,7 +5,7 @@
  * 2.0.
  */
 
-package org.elasticsearch.xpack.inference.queries;
+package org.elasticsearch.xpack.core.ml.search;
 
 import org.elasticsearch.common.ParsingException;
 import org.elasticsearch.common.io.stream.StreamInput;
@@ -22,7 +22,7 @@
 import java.util.Objects;
 import java.util.Set;
 
-import static org.elasticsearch.xpack.inference.queries.WeightedTokensQueryBuilder.PRUNING_CONFIG;
+import static org.elasticsearch.xpack.core.ml.search.WeightedTokensQueryBuilder.PRUNING_CONFIG;
 
 public class TokenPruningConfig implements Writeable, ToXContentObject {
     public static final ParseField TOKENS_FREQ_RATIO_THRESHOLD = new ParseField("tokens_freq_ratio_threshold");

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/queries/WeightedTokensQueryBuilder.java → x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/ml/search/WeightedTokensQueryBuilder.java

@@ -5,7 +5,7 @@
  * 2.0.
  */
 
-package org.elasticsearch.xpack.inference.queries;
+package org.elasticsearch.xpack.core.ml.search;
 
 import org.apache.lucene.search.MatchNoDocsQuery;
 import org.apache.lucene.search.Query;
@@ -24,7 +24,6 @@
 import org.elasticsearch.xcontent.ParseField;
 import org.elasticsearch.xcontent.XContentBuilder;
 import org.elasticsearch.xcontent.XContentParser;
-import org.elasticsearch.xpack.core.ml.search.WeightedToken;
 
 import java.io.IOException;
 import java.util.ArrayList;

x-pack/plugin/inference/src/main/java/org/elasticsearch/xpack/inference/queries/WeightedTokensUtils.java → x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/ml/search/WeightedTokensUtils.java

@@ -5,7 +5,7 @@
  * 2.0.
  */
 
-package org.elasticsearch.xpack.inference.queries;
+package org.elasticsearch.xpack.core.ml.search;
 
 import org.apache.lucene.index.IndexReader;
 import org.apache.lucene.index.Term;
@@ -16,7 +16,6 @@
 import org.apache.lucene.search.Query;
 import org.elasticsearch.index.mapper.MappedFieldType;
 import org.elasticsearch.index.query.SearchExecutionContext;
-import org.elasticsearch.xpack.core.ml.search.WeightedToken;
 
 import java.io.IOException;
 import java.util.List;