Pass through preserve_index to RecordBatch.from_pandas in serialize_pandas

Change-Id: I21268556e18b739db3de8601cce83388227dec0d

python/pyarrow/ipc.py

@@ -142,7 +142,7 @@ def open_file(source, footer_offset=None):
     return RecordBatchFileReader(source, footer_offset=footer_offset)
 
 
-def serialize_pandas(df, nthreads=None):
+def serialize_pandas(df, nthreads=None, preserve_index=True):
     """Serialize a pandas DataFrame into a buffer protocol compatible object.
 
     Parameters
@@ -150,13 +150,17 @@ def serialize_pandas(df, nthreads=None):
     df : pandas.DataFrame
     nthreads : int, default None
         Number of threads to use for conversion to Arrow, default all CPUs
+    preserve_index : boolean, default True
+        If True, preserve the pandas index data, otherwise the result will have
+        a default RangeIndex
 
     Returns
     -------
     buf : buffer
         An object compatible with the buffer protocol
     """
-    batch = pa.RecordBatch.from_pandas(df, nthreads=nthreads)
+    batch = pa.RecordBatch.from_pandas(df, nthreads=nthreads,
+                                       preserve_index=preserve_index)
     sink = pa.BufferOutputStream()
     writer = pa.RecordBatchStreamWriter(sink, batch.schema)
     writer.write_batch(batch)

python/pyarrow/pandas_compat.py

@@ -485,8 +485,8 @@ def table_to_blockmanager(options, table, memory_pool, nthreads=1):
             )
         ]
         new_levels = [
-            level if level.dtype == dtype else level.astype(dtype)
-            for level, dtype in levels_dtypes
+            _level if _level.dtype == _dtype else _level.astype(_dtype)
+            for _level, _dtype in levels_dtypes
         ]
         columns = pd.MultiIndex(
             levels=new_levels,

python/pyarrow/tests/test_ipc.py

@@ -416,6 +416,15 @@ def test_pandas_serialize_round_trip_not_string_columns():
     assert_frame_equal(result, df)
 
 
+def test_serialize_pandas_no_preserve_index():
+    df = pd.DataFrame({'a': [1, 2, 3]}, index=[1, 2, 3])
+    expected = pd.DataFrame({'a': [1, 2, 3]})
+
+    buf = pa.serialize_pandas(df, preserve_index=False)
+    result = pa.deserialize_pandas(buf)
+    assert_frame_equal(result, expected)
+
+
 def test_schema_batch_serialize_methods():
     nrows = 5
     df = pd.DataFrame({