Add serializer for pandas dataframe (apache#30390)

Add serializer for pandas dataframe

This adds an in-memory serializer for dataframes based on pyarrow. No
checking of size limits is being done and it is up to the user to
ensure the backend is configured to handle the potential size of
a dataframe.

airflow/serialization/serializers/pandas.py

@@ -0,0 +1,70 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements.  See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership.  The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License.  You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied.  See the License for the
+# specific language governing permissions and limitations
+# under the License.
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+from airflow.utils.module_loading import qualname
+
+# lazy loading for performance reasons
+serializers = [
+    "pandas.core.frame.DataFrame",
+]
+deserializers = serializers
+
+if TYPE_CHECKING:
+    from pandas import DataFrame
+
+    from airflow.serialization.serde import U
+
+__version__ = 1
+
+
+def serialize(o: object) -> tuple[U, str, int, bool]:
+    import pyarrow as pa
+    from pandas import DataFrame
+    from pyarrow import parquet as pq
+
+    if not isinstance(o, DataFrame):
+        return "", "", 0, False
+
+    # for now, we *always* serialize into in memory
+    # until we have a generic backend that manages
+    # sinks
+    table = pa.Table.from_pandas(o)
+    buf = pa.BufferOutputStream()
+    pq.write_table(table, buf, compression="snappy")
+
+    return buf.getvalue().hex().decode("utf-8"), qualname(o), __version__, True
+
+
+def deserialize(classname: str, version: int, data: object) -> DataFrame:
+    if version > __version__:
+        raise TypeError(f"serialized {version} of {classname} > {__version__}")
+
+    import io
+
+    from pyarrow import parquet as pq
+
+    if not isinstance(data, str):
+        raise TypeError(f"serialized {classname} has wrong data type {type(data)}")
+
+    buf = io.BytesIO(bytes.fromhex(data))
+    df = pq.read_table(buf).to_pandas()
+
+    return df

setup.py

@@ -301,9 +301,7 @@ def write_version(filename: str = str(AIRFLOW_SOURCES_ROOT / "airflow" / "git_ve
 ]
 leveldb = ["plyvel"]
 otel = ["opentelemetry-api==1.15.0", "opentelemetry-exporter-otlp", "opentelemetry-exporter-prometheus"]
-pandas = [
-    "pandas>=0.17.1",
-]
+pandas = ["pandas>=0.17.1", "pyarrow>=9.0.0"]
 password = [
     "bcrypt>=2.0.0",
     "flask-bcrypt>=0.7.1",

tests/serialization/serializers/test_serializers.py

@@ -20,6 +20,7 @@
 import decimal
 
 import numpy
+import pandas
 import pendulum.tz
 import pytest
 from pendulum import DateTime
@@ -91,3 +92,9 @@ def test_params(self):
         e = serialize(i)
         d = deserialize(e)
         assert i["x"] == d["x"]
+
+    def test_pandas(self):
+        i = pandas.DataFrame(data={"col1": [1, 2], "col2": [3, 4]})
+        e = serialize(i)
+        d = deserialize(e)
+        assert i.equals(d)