GH-46222: [Python] Allow to specify footer when opening IPC file for writing (#46222)

Author: chaubold

python/pyarrow/includes/libarrow.pxd

@@ -1976,13 +1976,15 @@ cdef extern from "arrow/ipc/api.h" namespace "arrow::ipc" nogil:
 
         CIpcReadStats stats()
 
+        shared_ptr[const CKeyValueMetadata] metadata()
+
     CResult[shared_ptr[CRecordBatchWriter]] MakeStreamWriter(
         shared_ptr[COutputStream] sink, const shared_ptr[CSchema]& schema,
         CIpcWriteOptions& options)
 
     CResult[shared_ptr[CRecordBatchWriter]] MakeFileWriter(
         shared_ptr[COutputStream] sink, const shared_ptr[CSchema]& schema,
-        CIpcWriteOptions& options)
+        CIpcWriteOptions& options, shared_ptr[const CKeyValueMetadata] metadata)
 
     CResult[unique_ptr[CMessage]] ReadMessage(CInputStream* stream,
                                               CMemoryPool* pool)

python/pyarrow/ipc.pxi

@@ -1020,15 +1020,21 @@ cdef class _RecordBatchStreamReader(RecordBatchReader):
 cdef class _RecordBatchFileWriter(_RecordBatchStreamWriter):
 
     def _open(self, sink, Schema schema not None,
-              IpcWriteOptions options=IpcWriteOptions()):
+              IpcWriteOptions options=IpcWriteOptions(),
+              metadata=None):
         cdef:
             shared_ptr[COutputStream] c_sink
+            shared_ptr[const CKeyValueMetadata] c_meta
 
         self.options = options.c_options
         get_writer(sink, &c_sink)
+
+        metadata = ensure_metadata(metadata, allow_none=True)
+        c_meta = pyarrow_unwrap_metadata(metadata)
+
         with nogil:
             self.writer = GetResultValue(
-                MakeFileWriter(c_sink, schema.sp_schema, self.options))
+                MakeFileWriter(c_sink, schema.sp_schema, self.options, c_meta))
 
 _RecordBatchWithMetadata = namedtuple(
     'RecordBatchWithMetadata',
@@ -1192,6 +1198,13 @@ cdef class _RecordBatchFileReader(_Weakrefable):
             raise ValueError("Operation on closed reader")
         return _wrap_read_stats(self.reader.get().stats())
 
+    @property
+    def metadata(self):
+        """
+        File-level custom KeyValueMetadata written via ``ipc.new_file(..., metadata=)``.
+        """
+        return pyarrow_wrap_metadata(self.reader.get().metadata())
+
 
 def get_tensor_size(Tensor tensor):
     """

python/pyarrow/ipc.py

@@ -109,11 +109,16 @@ class RecordBatchFileWriter(lib._RecordBatchFileWriter):
 
     __doc__ = """Writer to create the Arrow binary file format
 
-{}""".format(_ipc_writer_class_doc)
+{}
+metadata : dict | pyarrow.KeyValueMetadata, optional
+        Key/value pairs (both **must** be bytes-like) that will be stored
+        in the file footer and are retrievable with
+        ``ipc.open_file(...).metadata``.
+""".format(_ipc_writer_class_doc)
 
-    def __init__(self, sink, schema, *, options=None):
+    def __init__(self, sink, schema, *, options=None, metadata=None):
         options = _get_legacy_format_default(options)
-        self._open(sink, schema, options=options)
+        self._open(sink, schema, options=options, metadata=metadata)
 
 
 def _get_legacy_format_default(options):
@@ -180,9 +185,8 @@ def open_stream(source, *, options=None, memory_pool=None):
                                    memory_pool=memory_pool)
 
 
-def new_file(sink, schema, *, options=None):
-    return RecordBatchFileWriter(sink, schema,
-                                 options=options)
+def new_file(sink, schema, *, metadata=None, options=None):
+    return RecordBatchFileWriter(sink, schema, options=options, metadata=metadata)
 
 
 new_file.__doc__ = """\

python/pyarrow/tests/test_ipc.py

@@ -1304,3 +1304,17 @@ def test_record_batch_reader_cast_nulls():
     casted_reader = reader.cast(schema_dst)
     with pytest.raises(pa.lib.ArrowInvalid, match="Can't cast array"):
         casted_reader.read_all()
+
+
+def test_record_batch_file_writer_with_metadata():
+    # https://github.com/apache/arrow/issues/46222
+    tbl = pa.table({"a": [1, 2, 3]})
+    meta = {b"creator": b"test", b"version": b"0.1.0"}
+    sink = pa.BufferOutputStream()
+
+    with pa.ipc.new_file(sink, tbl.schema, metadata=meta) as w:
+        w.write_table(tbl)
+
+    buffer = sink.getvalue()
+    with pa.ipc.open_file(buffer) as r:
+        assert dict(r.metadata) == meta