Tweak copy_to handling in synthetic _source to account for nested objects

docs/changelog/120974.yaml

@@ -0,0 +1,6 @@
+pr: 120974
+summary: Tweak `copy_to` handling in synthetic `_source` to account for nested objects
+area: Mapping
+type: bug
+issues:
+ - 120831

rest-api-spec/src/yamlRestTest/resources/rest-api-spec/test/indices.create/20_synthetic_source.yml

@@ -1602,6 +1602,74 @@ synthetic_source with copy_to pointing inside object:
       hits.hits.2.fields:
           c.copy: [ "100", "hello", "zap" ]
 
+---
+synthetic_source with copy_to inside nested object:
+  - do:
+      indices.create:
+        index: test
+        body:
+          settings:
+            index:
+              mapping.source.mode: synthetic
+          mappings:
+            properties:
+              name:
+                type: keyword
+              my_values:
+                type: nested
+                properties:
+                  k:
+                    type: keyword
+                    copy_to: my_values.copy
+                  second_level:
+                    type: nested
+                    properties:
+                      k2:
+                        type: keyword
+                        copy_to: my_values.copy
+                  copy:
+                    type: keyword
+              dummy:
+                type: keyword
+
+  - do:
+      index:
+        index: test
+        id: 1
+        refresh: true
+        body:
+          name: "A"
+          my_values:
+            k: "hello"
+
+  - do:
+      index:
+        index: test
+        id: 2
+        refresh: true
+        body:
+          name: "B"
+          my_values:
+            second_level:
+              k2: "hello"
+
+  - do:
+      search:
+        index: test
+        sort: name
+
+  - match:
+      hits.hits.0._source:
+        name: "A"
+        my_values:
+          k: "hello"
+  - match:
+      hits.hits.1._source:
+        name: "B"
+        my_values:
+          second_level:
+            k2: "hello"
+
 ---
 synthetic_source with copy_to pointing to ambiguous field:
   - do:

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

@@ -367,19 +367,6 @@ public final DocumentParserContext maybeCloneForArray(Mapper mapper) throws IOEx
         return this;
     }
 
-    /**
-     * Creates a sub-context from the current {@link DocumentParserContext} to indicate that the source for the sub-context has been
-     * recorded and avoid duplicate recording for parts of the sub-context. Applies to synthetic source only.
-     */
-    public final DocumentParserContext cloneWithRecordedSource() throws IOException {
-        if (canAddIgnoredField()) {
-            DocumentParserContext subcontext = createChildContext(parent());
-            subcontext.setRecordedSource();  // Avoids double-storing parts of the source for the same parser subtree.
-            return subcontext;
-        }
-        return this;
-    }
-
     /**
      * Add the given {@code field} to the _field_names field
      *
@@ -466,10 +453,6 @@ public boolean isCopyToDestinationField(String name) {
         return copyToFields.contains(name);
     }
 
-    public Set<String> getCopyToFields() {
-        return copyToFields;
-    }
-
     /**
      * Add a new mapper dynamically created while parsing.
      *
@@ -706,6 +689,26 @@ public LuceneDocument doc() {
      * @param doc           the document to target
      */
     public final DocumentParserContext createCopyToContext(String copyToField, LuceneDocument doc) throws IOException {
+        /*
+            Mark field as containing copied data meaning it should not be present
+            in synthetic _source (to be consistent with stored _source).
+            Ignored source values take precedence over standard synthetic source implementation
+            so by adding the `XContentDataHelper.voidValue()` entry we disable the field in synthetic source.
+            Otherwise, it would be constructed f.e. from doc_values which leads to duplicate values
+            in copied field after reindexing.
+         */
+        if (mappingLookup.isSourceSynthetic() && indexSettings().getSkipIgnoredSourceWrite() == false) {
+            ObjectMapper parent = root().findParentMapper(copyToField);
+            // There are scenarios when this is false:
+            // 1. all values of the field that is the source of copy_to are null
+            // 2. copy_to points at a field inside a disabled object
+            // 3. copy_to points at dynamic field which is not yet applied to mapping, we will process it properly after the dynamic update
+            if (parent != null) {
+                int offset = parent.isRoot() ? 0 : parent.fullPath().length() + 1;
+                ignoredFieldValues.add(new IgnoredSourceFieldMapper.NameValue(copyToField, offset, XContentDataHelper.voidValue(), doc));
+            }
+        }
+
         ContentPath path = new ContentPath();
         XContentParser parser = DotExpandingXContentParser.expandDots(new CopyToParser(copyToField, parser()), path);
         return new Wrapper(root(), this) {

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

@@ -25,8 +25,6 @@
 
 import java.io.IOException;
 import java.nio.charset.StandardCharsets;
-import java.util.ArrayList;
-import java.util.Collection;
 import java.util.Collections;
 import java.util.Map;
 import java.util.Set;
@@ -161,33 +159,7 @@ public void postParse(DocumentParserContext context) {
             return;
         }
 
-        Collection<NameValue> ignoredValuesToWrite = context.getIgnoredFieldValues();
-        if (context.getCopyToFields().isEmpty() == false && indexSettings.getSkipIgnoredSourceWrite() == false) {
-            /*
-            Mark fields as containing copied data meaning they should not be present
-            in synthetic _source (to be consistent with stored _source).
-            Ignored source values take precedence over standard synthetic source implementation
-            so by adding the `XContentDataHelper.voidValue()` entry we disable the field in synthetic source.
-            Otherwise, it would be constructed f.e. from doc_values which leads to duplicate values
-            in copied field after reindexing.
-            */
-            var mutableList = new ArrayList<>(ignoredValuesToWrite);
-            for (String copyToField : context.getCopyToFields()) {
-                ObjectMapper parent = context.parent().findParentMapper(copyToField);
-                if (parent == null) {
-                    // There are scenarios when this can happen:
-                    // 1. all values of the field that is the source of copy_to are null
-                    // 2. copy_to points at a field inside a disabled object
-                    // 3. copy_to points at dynamic field which is not yet applied to mapping, we will process it properly on re-parse.
-                    continue;
-                }
-                int offset = parent.isRoot() ? 0 : parent.fullPath().length() + 1;
-                mutableList.add(new IgnoredSourceFieldMapper.NameValue(copyToField, offset, XContentDataHelper.voidValue(), context.doc()));
-            }
-            ignoredValuesToWrite = mutableList;
-        }
-
-        for (NameValue nameValue : ignoredValuesToWrite) {
+        for (NameValue nameValue : context.getIgnoredFieldValues()) {
             nameValue.doc().add(new StoredField(NAME, encode(nameValue)));
         }
     }