(arrow-rs) branch master updated: Align buffers from Python (FFI) (#6472)

[tu...@apache.org]

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

tustvold pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/arrow-rs.git


The following commit(s) were added to refs/heads/master by this push:
     new 2c524f16a Align buffers from Python (FFI) (#6472)
2c524f16a is described below

commit 2c524f16a7ed484456700d69dc5ccfb9b3af3058
Author: Enrico Minack <gi...@enrico.minack.dev>
AuthorDate: Wed Oct 2 14:22:44 2024 +0200

    Align buffers from Python (FFI) (#6472)
    
    * Align buffers in RecordBatch.from_pyarrow_bound
    
    * Update arrow/src/pyarrow.rs
    
    * cargo fmt
    
    ---------
    
    Co-authored-by: Andrew Lamb <an...@nerdnetworks.org>
---
 arrow/src/pyarrow.rs | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/arrow/src/pyarrow.rs b/arrow/src/pyarrow.rs
index b05c967d7..6effe1c03 100644
--- a/arrow/src/pyarrow.rs
+++ b/arrow/src/pyarrow.rs
@@ -363,13 +363,19 @@ impl FromPyArrow for RecordBatch {
 
             let schema_ptr = unsafe { schema_capsule.reference::<FFI_ArrowSchema>() };
             let ffi_array = unsafe { FFI_ArrowArray::from_raw(array_capsule.pointer().cast()) };
-            let array_data = unsafe { ffi::from_ffi(ffi_array, schema_ptr) }.map_err(to_py_err)?;
+            let mut array_data =
+                unsafe { ffi::from_ffi(ffi_array, schema_ptr) }.map_err(to_py_err)?;
             if !matches!(array_data.data_type(), DataType::Struct(_)) {
                 return Err(PyTypeError::new_err(
                     "Expected Struct type from __arrow_c_array.",
                 ));
             }
             let options = RecordBatchOptions::default().with_row_count(Some(array_data.len()));
+            // Ensure data is aligned (by potentially copying the buffers).
+            // This is needed because some python code (for example the
+            // python flight client) produces unaligned buffers
+            // See https://github.com/apache/arrow/issues/43552 for details
+            array_data.align_buffers();
             let array = StructArray::from(array_data);
             // StructArray does not embed metadata from schema. We need to override
             // the output schema with the schema from the capsule.