Stop Needlessly Copying Bytes in XContent Parsing (#61447) (#61469)

Wrapping a `BytesArray` in a `StreamInput` for deserialization is inefficient.
This forces Jackson to internally buffer (i.e. copy) all bytes from the `BytesArray`
before deserializing, adding overhead for copying the bytes and managing the buffers.

This commit fixes a number of spots where `BytesArray` is the most common type of
`BytesReference` to special case this type and parse it more efficiently.
Also improves parsing `String`s to use the more efficient direct `String` parsing APIs.

libs/x-content/src/main/java/org/elasticsearch/common/xcontent/cbor/CborXContent.java

@@ -36,7 +36,6 @@ import java.io.IOException;
 import java.io.InputStream;
 import java.io.OutputStream;
 import java.io.Reader;
-import java.io.StringReader;
 import java.util.Set;
 
 /**
@@ -81,7 +80,7 @@ public class CborXContent implements XContent {
     @Override
     public XContentParser createParser(NamedXContentRegistry xContentRegistry,
             DeprecationHandler deprecationHandler, String content) throws IOException {
-        return new CborXContentParser(xContentRegistry, deprecationHandler, cborFactory.createParser(new StringReader(content)));
+        return new CborXContentParser(xContentRegistry, deprecationHandler, cborFactory.createParser(content));
     }
 
     @Override

libs/x-content/src/main/java/org/elasticsearch/common/xcontent/json/JsonXContent.java

@@ -35,7 +35,6 @@ import java.io.IOException;
 import java.io.InputStream;
 import java.io.OutputStream;
 import java.io.Reader;
-import java.io.StringReader;
 import java.util.Set;
 
 /**
@@ -82,7 +81,7 @@ public class JsonXContent implements XContent {
     @Override
     public XContentParser createParser(NamedXContentRegistry xContentRegistry,
             DeprecationHandler deprecationHandler, String content) throws IOException {
-        return new JsonXContentParser(xContentRegistry, deprecationHandler, jsonFactory.createParser(new StringReader(content)));
+        return new JsonXContentParser(xContentRegistry, deprecationHandler, jsonFactory.createParser(content));
     }
 
     @Override

libs/x-content/src/main/java/org/elasticsearch/common/xcontent/smile/SmileXContent.java

@@ -36,7 +36,6 @@ import java.io.IOException;
 import java.io.InputStream;
 import java.io.OutputStream;
 import java.io.Reader;
-import java.io.StringReader;
 import java.util.Set;
 
 /**
@@ -83,7 +82,7 @@ public class SmileXContent implements XContent {
     @Override
     public XContentParser createParser(NamedXContentRegistry xContentRegistry,
             DeprecationHandler deprecationHandler, String content) throws IOException {
-        return new SmileXContentParser(xContentRegistry, deprecationHandler, smileFactory.createParser(new StringReader(content)));
+        return new SmileXContentParser(xContentRegistry, deprecationHandler, smileFactory.createParser(content));
     }
 
     @Override

libs/x-content/src/main/java/org/elasticsearch/common/xcontent/yaml/YamlXContent.java

@@ -34,7 +34,6 @@ import java.io.IOException;
 import java.io.InputStream;
 import java.io.OutputStream;
 import java.io.Reader;
-import java.io.StringReader;
 import java.util.Set;
 
 /**
@@ -76,7 +75,7 @@ public class YamlXContent implements XContent {
     @Override
     public XContentParser createParser(NamedXContentRegistry xContentRegistry,
             DeprecationHandler deprecationHandler, String content) throws IOException {
-        return new YamlXContentParser(xContentRegistry, deprecationHandler, yamlFactory.createParser(new StringReader(content)));
+        return new YamlXContentParser(xContentRegistry, deprecationHandler, yamlFactory.createParser(content));
     }
 
     @Override

server/src/main/java/org/elasticsearch/action/bulk/BulkRequestParser.java

@@ -26,6 +26,7 @@ import org.elasticsearch.action.index.IndexRequest;
 import org.elasticsearch.action.update.UpdateRequest;
 import org.elasticsearch.common.Nullable;
 import org.elasticsearch.common.ParseField;
+import org.elasticsearch.common.bytes.BytesArray;
 import org.elasticsearch.common.bytes.BytesReference;
 import org.elasticsearch.common.logging.DeprecationLogger;
 import org.elasticsearch.common.lucene.uid.Versions;
@@ -40,7 +41,6 @@ import org.elasticsearch.rest.action.document.RestBulkAction;
 import org.elasticsearch.search.fetch.subphase.FetchSourceContext;
 
 import java.io.IOException;
-import java.io.InputStream;
 import java.util.HashMap;
 import java.util.Map;
 import java.util.function.Consumer;
@@ -156,10 +156,7 @@ public final class BulkRequestParser {
             line++;
 
             // now parse the action
-            // EMPTY is safe here because we never call namedObject
+            try (XContentParser parser = createParser(data, xContent, from, nextMarker)) {
-            try (InputStream stream = data.slice(from, nextMarker - from).streamInput();
-                    XContentParser parser = xContent
-                            .createParser(NamedXContentRegistry.EMPTY, LoggingDeprecationHandler.INSTANCE, stream)) {
                 // move pointers
                 from = nextMarker + 1;
 
@@ -204,7 +201,7 @@ public final class BulkRequestParser {
                         if (token == XContentParser.Token.FIELD_NAME) {
                             currentFieldName = parser.currentName();
                         } else if (token.isValue()) {
-                            if (INDEX.match(currentFieldName, parser.getDeprecationHandler())){
+                            if (INDEX.match(currentFieldName, parser.getDeprecationHandler())) {
                                 if (!allowExplicitIndex) {
                                     throw new IllegalArgumentException("explicit index in bulk is not allowed");
                                 }
@@ -299,10 +296,8 @@ public final class BulkRequestParser {
                                 .setIfSeqNo(ifSeqNo).setIfPrimaryTerm(ifPrimaryTerm)
                                 .setRequireAlias(requireAlias)
                                 .routing(routing);
-                        // EMPTY is safe here because we never call namedObject
+                        try (XContentParser sliceParser = createParser(
-                        try (InputStream dataStream = sliceTrimmingCarriageReturn(data, from, nextMarker, xContentType).streamInput();
+                                sliceTrimmingCarriageReturn(data, from, nextMarker, xContentType), xContent)) {
-                                XContentParser sliceParser = xContent.createParser(NamedXContentRegistry.EMPTY,
-                                        LoggingDeprecationHandler.INSTANCE, dataStream)) {
                             updateRequest.fromXContent(sliceParser);
                         }
                         if (fetchSourceContext != null) {
@@ -322,4 +317,35 @@ public final class BulkRequestParser {
         }
     }
 
+    private static XContentParser createParser(BytesReference data, XContent xContent) throws IOException {
+        if (data instanceof BytesArray) {
+            return parseBytesArray(xContent, (BytesArray) data, 0, data.length());
+        } else {
+            return xContent.createParser(NamedXContentRegistry.EMPTY, LoggingDeprecationHandler.INSTANCE, data.streamInput());
+        }
+    }
+
+    // Create an efficient parser of the given bytes, trying to directly parse a byte array if possible and falling back to stream wrapping
+    // otherwise.
+    private static XContentParser createParser(BytesReference data, XContent xContent, int from, int nextMarker) throws IOException {
+        if (data instanceof BytesArray) {
+            return parseBytesArray(xContent, (BytesArray) data, from, nextMarker);
+        } else {
+            final int length = nextMarker - from;
+            final BytesReference slice = data.slice(from, length);
+            if (slice instanceof BytesArray) {
+                return parseBytesArray(xContent, (BytesArray) slice, 0, length);
+            } else {
+                // EMPTY is safe here because we never call namedObject
+                return xContent.createParser(NamedXContentRegistry.EMPTY, LoggingDeprecationHandler.INSTANCE, slice.streamInput());
+            }
+        }
+    }
+
+    private static XContentParser parseBytesArray(XContent xContent, BytesArray array, int from, int nextMarker) throws IOException {
+        final int offset = array.offset();
+        // EMPTY is safe here because we never call namedObject
+        return xContent.createParser(NamedXContentRegistry.EMPTY, LoggingDeprecationHandler.INSTANCE, array.array(),
+                offset + from, nextMarker - from);
+    }
 }

server/src/main/java/org/elasticsearch/common/xcontent/XContentHelper.java

@@ -76,6 +76,11 @@ public class XContentHelper {
             }
             return XContentFactory.xContent(xContentType).createParser(xContentRegistry, deprecationHandler, compressedInput);
         } else {
+            if (bytes instanceof BytesArray) {
+                final BytesArray array = (BytesArray) bytes;
+                return xContentType.xContent().createParser(
+                        xContentRegistry, deprecationHandler, array.array(), array.offset(), array.length());
+            }
             return xContentType.xContent().createParser(xContentRegistry, deprecationHandler, bytes.streamInput());
         }
     }
@@ -106,10 +111,19 @@ public class XContentHelper {
                     compressedStreamInput = new BufferedInputStream(compressedStreamInput);
                 }
                 input = compressedStreamInput;
+                contentType = xContentType != null ? xContentType : XContentFactory.xContentType(input);
+            } else if (bytes instanceof BytesArray) {
+                final BytesArray arr = (BytesArray) bytes;
+                final byte[] raw = arr.array();
+                final int offset = arr.offset();
+                final int length = arr.length();
+                contentType = xContentType != null ? xContentType : XContentFactory.xContentType(raw, offset, length);
+                return new Tuple<>(Objects.requireNonNull(contentType),
+                        convertToMap(XContentFactory.xContent(contentType), raw, offset, length, ordered));
             } else {
                 input = bytes.streamInput();
-            }
                 contentType = xContentType != null ? xContentType : XContentFactory.xContentType(input);
+            }
             try (InputStream stream = input) {
                 return new Tuple<>(Objects.requireNonNull(contentType),
                     convertToMap(XContentFactory.xContent(contentType), stream, ordered));
@@ -148,6 +162,21 @@ public class XContentHelper {
         }
     }
 
+    /**
+     * Convert a byte array in some {@link XContent} format to a {@link Map}. Throws an {@link ElasticsearchParseException} if there is any
+     * error. Note that unlike {@link #convertToMap(BytesReference, boolean)}, this doesn't automatically uncompress the input.
+     */
+    public static Map<String, Object> convertToMap(XContent xContent, byte[] bytes, int offset, int length, boolean ordered)
+            throws ElasticsearchParseException {
+        // It is safe to use EMPTY here because this never uses namedObject
+        try (XContentParser parser = xContent.createParser(NamedXContentRegistry.EMPTY,
+                DeprecationHandler.THROW_UNSUPPORTED_OPERATION, bytes, offset, length)) {
+            return ordered ? parser.mapOrdered() : parser.map();
+        } catch (IOException e) {
+            throw new ElasticsearchParseException("Failed to parse content to map", e);
+        }
+    }
+
     @Deprecated
     public static String convertToJson(BytesReference bytes, boolean reformatJson) throws IOException {
         return convertToJson(bytes, reformatJson, false);
@@ -183,9 +212,22 @@ public class XContentHelper {
         }
 
         // It is safe to use EMPTY here because this never uses namedObject
+        if (bytes instanceof BytesArray) {
+            final BytesArray array = (BytesArray) bytes;
+            try (XContentParser parser = XContentFactory.xContent(xContentType).createParser(NamedXContentRegistry.EMPTY,
+                         DeprecationHandler.THROW_UNSUPPORTED_OPERATION, array.array(), array.offset(), array.length())) {
+                return toJsonString(prettyPrint, parser);
+            }
+        } else {
             try (InputStream stream = bytes.streamInput();
                  XContentParser parser = XContentFactory.xContent(xContentType).createParser(NamedXContentRegistry.EMPTY,
                          DeprecationHandler.THROW_UNSUPPORTED_OPERATION, stream)) {
+                return toJsonString(prettyPrint, parser);
+            }
+        }
+    }
+
+    private static String toJsonString(boolean prettyPrint, XContentParser parser) throws IOException {
         parser.nextToken();
         XContentBuilder builder = XContentFactory.jsonBuilder();
         if (prettyPrint) {
@@ -194,7 +236,6 @@ public class XContentHelper {
         builder.copyCurrentStructure(parser);
         return Strings.toString(builder);
     }
-    }
 
     /**
      * Updates the provided changes into the source. If the key exists in the changes, it overrides the one in source

server/src/test/java/org/elasticsearch/index/IndexingSlowLogTests.java

@@ -254,12 +254,12 @@ public class IndexingSlowLogTests extends ESTestCase {
             () -> new IndexingSlowLogMessage(index, doc, 10, true, 3));
         assertThat(e, hasToString(containsString("_failed_to_convert_[Unrecognized token 'invalid':"
             + " was expecting (JSON String, Number, Array, Object or token 'null', 'true' or 'false')\\n"
-            + " at [Source: (org.elasticsearch.common.bytes.AbstractBytesReference$MarkSupportingStreamInputWrapper)")));
+            + " at [Source: ")));
         assertNotNull(e.getCause());
         assertThat(e.getCause(), instanceOf(JsonParseException.class));
         assertThat(e.getCause(), hasToString(containsString("Unrecognized token 'invalid':"
                 + " was expecting (JSON String, Number, Array, Object or token 'null', 'true' or 'false')\n"
-            + " at [Source: (org.elasticsearch.common.bytes.AbstractBytesReference$MarkSupportingStreamInputWrapper)")));
+                + " at [Source: ")));
     }
 
     public void testReformatSetting() {

test/framework/src/main/java/org/elasticsearch/test/ESTestCase.java

@@ -55,6 +55,7 @@ import org.elasticsearch.cluster.metadata.IndexMetadata;
 import org.elasticsearch.cluster.node.DiscoveryNode;
 import org.elasticsearch.common.CheckedRunnable;
 import org.elasticsearch.common.SuppressForbidden;
+import org.elasticsearch.common.bytes.BytesArray;
 import org.elasticsearch.common.bytes.BytesReference;
 import org.elasticsearch.common.io.PathUtils;
 import org.elasticsearch.common.io.PathUtilsForTesting;
@@ -1283,8 +1284,7 @@ public abstract class ESTestCase extends LuceneTestCase {
      * Create a new {@link XContentParser}.
      */
     protected final XContentParser createParser(XContentBuilder builder) throws IOException {
-        return builder.generator().contentType().xContent()
+        return createParser(builder.contentType().xContent(), BytesReference.bytes(builder));
-            .createParser(xContentRegistry(), LoggingDeprecationHandler.INSTANCE, BytesReference.bytes(builder).streamInput());
     }
 
     /**
@@ -1312,7 +1312,7 @@ public abstract class ESTestCase extends LuceneTestCase {
      * Create a new {@link XContentParser}.
      */
     protected final XContentParser createParser(XContent xContent, BytesReference data) throws IOException {
-        return xContent.createParser(xContentRegistry(), LoggingDeprecationHandler.INSTANCE, data.streamInput());
+        return createParser(xContentRegistry(), xContent, data);
     }
 
     /**
@@ -1320,6 +1320,11 @@ public abstract class ESTestCase extends LuceneTestCase {
      */
     protected final XContentParser createParser(NamedXContentRegistry namedXContentRegistry, XContent xContent,
                                                 BytesReference data) throws IOException {
+        if (data instanceof BytesArray) {
+            final BytesArray array = (BytesArray) data;
+            return xContent.createParser(
+                    namedXContentRegistry, LoggingDeprecationHandler.INSTANCE, array.array(), array.offset(), array.length());
+        }
         return xContent.createParser(namedXContentRegistry, LoggingDeprecationHandler.INSTANCE, data.streamInput());
     }