(arrow-dotnet) 01/06: Expose IPC Message custom_metadata on ArrowStreamReader

[cu...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

CurtHagenlocher pushed a commit to branch ipc-message-custom-metadata
in repository https://gitbox.apache.org/repos/asf/arrow-dotnet.git

commit 7a5bacf9b6df0a5d1588e38a1bd9e1418d4328b1
Author: Christoph Mettler <c....@ethenea.com>
AuthorDate: Sun Mar 8 20:14:21 2026 +0100

    Expose IPC Message custom_metadata on ArrowStreamReader
    
    The Arrow IPC format supports custom_metadata on each Message
    (RecordBatch), but the C# implementation currently ignores it on read.
    This adds a LastBatchCustomMetadata property to ArrowStreamReader that
    exposes the key-value pairs from the most recently read batch's Message.
    
    This is the read-side counterpart to pyarrow's
    read_next_batch_with_custom_metadata() and enables use cases like
    RPC frameworks that embed method routing or log metadata in per-batch
    custom_metadata fields.
    
    Co-Authored-By: Claude Opus 4.6 <no...@anthropic.com>
---
 src/Apache.Arrow/Ipc/ArrowReaderImplementation.cs | 23 +++++++++++++++++++++++
 src/Apache.Arrow/Ipc/ArrowStreamReader.cs         |  8 ++++++++
 2 files changed, 31 insertions(+)

diff --git a/src/Apache.Arrow/Ipc/ArrowReaderImplementation.cs b/src/Apache.Arrow/Ipc/ArrowReaderImplementation.cs
index 2c380e5..acac52f 100644
--- a/src/Apache.Arrow/Ipc/ArrowReaderImplementation.cs
+++ b/src/Apache.Arrow/Ipc/ArrowReaderImplementation.cs
@@ -81,6 +81,11 @@ namespace Apache.Arrow.Ipc
         public abstract ValueTask<RecordBatch> ReadNextRecordBatchAsync(CancellationToken cancellationToken);
         public abstract RecordBatch ReadNextRecordBatch();
 
+        /// <summary>
+        /// Custom metadata from the most recently read RecordBatch Message, if any.
+        /// </summary>
+        internal IReadOnlyDictionary<string, string> LastBatchCustomMetadata { get; private protected set; }
+
         internal static T ReadMessage<T>(ByteBuffer bb)
             where T : struct, IFlatbufferObject
         {
@@ -148,6 +153,7 @@ namespace Apache.Arrow.Ipc
                     }
 
                     List<IArrowArray> arrays = BuildArrays(message.Version, Schema, bodyByteBuffer, rb);
+                    LastBatchCustomMetadata = ReadMessageCustomMetadata(message);
                     return new RecordBatch(Schema, memoryOwner, arrays, (int)rb.Length);
                 default:
                     // NOTE: Skip unsupported message type
@@ -158,6 +164,23 @@ namespace Apache.Arrow.Ipc
             return null;
         }
 
+        private static IReadOnlyDictionary<string, string> ReadMessageCustomMetadata(Flatbuf.Message message)
+        {
+            int count = message.CustomMetadataLength;
+            if (count == 0)
+                return null;
+
+            var result = new Dictionary<string, string>(count);
+            for (int i = 0; i < count; i++)
+            {
+                Flatbuf.KeyValue kv = message.CustomMetadata(i).GetValueOrDefault();
+                string key = kv.Key;
+                if (key != null)
+                    result[key] = kv.Value ?? "";
+            }
+            return result;
+        }
+
         internal static ByteBuffer CreateByteBuffer(ReadOnlyMemory<byte> buffer)
         {
             return new ByteBuffer(new ReadOnlyMemoryBufferAllocator(buffer), 0);
diff --git a/src/Apache.Arrow/Ipc/ArrowStreamReader.cs b/src/Apache.Arrow/Ipc/ArrowStreamReader.cs
index afa3713..8379b12 100644
--- a/src/Apache.Arrow/Ipc/ArrowStreamReader.cs
+++ b/src/Apache.Arrow/Ipc/ArrowStreamReader.cs
@@ -14,6 +14,7 @@
 // limitations under the License.
 
 using System;
+using System.Collections.Generic;
 using System.IO;
 using System.Threading;
 using System.Threading.Tasks;
@@ -151,5 +152,12 @@ namespace Apache.Arrow.Ipc
         {
             return _implementation.ReadNextRecordBatch();
         }
+
+        /// <summary>
+        /// Custom metadata from the most recently read RecordBatch Message.
+        /// Updated after each call to ReadNextRecordBatch/ReadNextRecordBatchAsync.
+        /// Returns null if the last batch had no custom metadata.
+        /// </summary>
+        public IReadOnlyDictionary<string, string> LastBatchCustomMetadata => _implementation.LastBatchCustomMetadata;
     }
 }