GH-41302: [C#][Integration] Fix writing list and binary arrays with zero length offsets to IPC format (#41303)

adamreeve · web-flow · commit bfc0dcbe1baf · 2024-04-18T19:41:56.000-07:00
### Rationale for this change Fixes the integration test failures caused by #41230 ### What changes are included in this PR? Only try to access the offset values if the array length is non-zero when writing list and binary arrays to IPC format. ### Are these changes tested? Yes, I've manually run the integration tests with C# and Java to verify they pass (when also including the changes from #41264), and also added new unit tests for this. ### Are there any user-facing changes? This may also be a bug that affects users but it isn't in a released version. * GitHub Issue: #41302 Authored-by: Adam Reeve <adreeve@gmail.com> Signed-off-by: Curt Hagenlocher <curt@hagenlocher.org>
diff --git a/csharp/src/Apache.Arrow/Ipc/ArrowStreamWriter.cs b/csharp/src/Apache.Arrow/Ipc/ArrowStreamWriter.cs
@@ -165,8 +165,13 @@ public void Visit(ListArray array)
                 _buffers.Add(CreateBitmapBuffer(array.NullBitmapBuffer, array.Offset, array.Length));
                 _buffers.Add(CreateBuffer(GetZeroBasedValueOffsets(array.ValueOffsetsBuffer, array.Offset, array.Length)));
 
-                int valuesOffset = array.ValueOffsets[0];
-                int valuesLength = array.ValueOffsets[array.Length] - valuesOffset;
+                int valuesOffset = 0;
+                int valuesLength = 0;
+                if (array.Length > 0)
+                {
+                    valuesOffset = array.ValueOffsets[0];
+                    valuesLength = array.ValueOffsets[array.Length] - valuesOffset;
+                }
 
                 var values = array.Values;
                 if (valuesOffset > 0 || valuesLength < values.Length)
@@ -206,8 +211,13 @@ public void Visit(BinaryArray array)
                 _buffers.Add(CreateBitmapBuffer(array.NullBitmapBuffer, array.Offset, array.Length));
                 _buffers.Add(CreateBuffer(GetZeroBasedValueOffsets(array.ValueOffsetsBuffer, array.Offset, array.Length)));
 
-                int valuesOffset = array.ValueOffsets[0];
-                int valuesLength = array.ValueOffsets[array.Length] - valuesOffset;
+                int valuesOffset = 0;
+                int valuesLength = 0;
+                if (array.Length > 0)
+                {
+                    valuesOffset = array.ValueOffsets[0];
+                    valuesLength = array.ValueOffsets[array.Length] - valuesOffset;
+                }
 
                 _buffers.Add(CreateSlicedBuffer<byte>(array.ValueBuffer, valuesOffset, valuesLength));
             }
diff --git a/csharp/test/Apache.Arrow.Tests/ArrowFileWriterTests.cs b/csharp/test/Apache.Arrow.Tests/ArrowFileWriterTests.cs
@@ -193,5 +193,57 @@ public async Task WritesEmptyFileAsync()
             Assert.Null(readBatch);
             SchemaComparer.Compare(originalBatch.Schema, reader.Schema);
         }
+
+        [Fact]
+        public async Task WriteBinaryArrayWithEmptyOffsets()
+        {
+            // Empty binary arrays generated by the C# builder have a single offset,
+            // but some implementations may produce an empty offsets buffer.
+
+            var array = new BinaryArray(
+                new BinaryType(),
+                length: 0,
+                valueOffsetsBuffer: ArrowBuffer.Empty,
+                dataBuffer: ArrowBuffer.Empty,
+                nullBitmapBuffer: ArrowBuffer.Empty,
+                nullCount: 0);
+
+            var recordBatch = new RecordBatch.Builder().Append("x", true, array).Build();
+
+            var stream = new MemoryStream();
+            var writer = new ArrowFileWriter(stream, recordBatch.Schema, leaveOpen: true);
+
+            await writer.WriteRecordBatchAsync(recordBatch);
+            await writer.WriteEndAsync();
+
+            stream.Position = 0;
+
+            await ValidateRecordBatchFile(stream, recordBatch, strictCompare: false);
+        }
+
+        [Fact]
+        public async Task WriteListArrayWithEmptyOffsets()
+        {
+            var values = new Int32Array.Builder().Build();
+            var array = new ListArray(
+                new ListType(new Int32Type()),
+                length: 0,
+                valueOffsetsBuffer: ArrowBuffer.Empty,
+                values: values,
+                nullBitmapBuffer: ArrowBuffer.Empty,
+                nullCount: 0);
+
+            var recordBatch = new RecordBatch.Builder().Append("x", true, array).Build();
+
+            var stream = new MemoryStream();
+            var writer = new ArrowFileWriter(stream, recordBatch.Schema, leaveOpen: true);
+
+            await writer.WriteRecordBatchAsync(recordBatch);
+            await writer.WriteEndAsync();
+
+            stream.Position = 0;
+
+            await ValidateRecordBatchFile(stream, recordBatch, strictCompare: false);
+        }
     }
 }
