Declare pyarrow for Parquet output of SqlToS3Operator (#74691)

Writing Parquet files and reading with the pyarrow dtype backend both need pyarrow, but the provider never declared it. It only worked when another installed package happened to bring pyarrow in, and otherwise failed at runtime with pandas' "Unable to find a usable engine" error. A separate extra keeps the roughly 40 MB dependency out of installs that only write CSV or JSON.

Author: Dev-iL

providers/amazon/docs/index.rst

@@ -174,9 +174,9 @@ Install them when installing from PyPI. For example:
     pip install apache-airflow-providers-amazon[aiobotocore]
 
 
-====================  ============================================================================================================================================================
+====================  ======================================================================================================================================================================
 Extra                 Dependencies
-====================  ============================================================================================================================================================
+====================  ======================================================================================================================================================================
 ``aiobotocore``       ``aiobotocore>=3.0.0``
 ``cncf.kubernetes``   ``apache-airflow-providers-cncf-kubernetes>=7.2.0``
 ``s3fs``              ``s3fs>=2023.10.0``
@@ -191,6 +191,7 @@ Extra                 Dependencies
 ``microsoft.azure``   ``apache-airflow-providers-microsoft-azure``
 ``mongo``             ``apache-airflow-providers-mongo``
 ``pandas``            ``pandas>=2.1.2; python_version <"3.13"``, ``pandas>=2.2.3; python_version >="3.13" and python_version <"3.14"``, ``pandas>=2.3.3; python_version >="3.14"``
+``parquet``           ``pyarrow>=16.1.0; python_version < '3.13'``, ``pyarrow>=18.0.0; python_version >= '3.13' and python_version < '3.14'``, ``pyarrow>=22.0.0; python_version >= '3.14'``
 ``openlineage``       ``apache-airflow-providers-openlineage>=2.3.0``
 ``salesforce``        ``apache-airflow-providers-salesforce``
 ``ssh``               ``apache-airflow-providers-ssh``
@@ -199,7 +200,7 @@ Extra                 Dependencies
 ``common.messaging``  ``apache-airflow-providers-common-messaging>=2.0.0``
 ``duckdb``            ``apache-airflow-providers-duckdb>=0.1.0``
 ``sqlalchemy``        ``sqlalchemy>=1.4.54,!=2.1.0``
-====================  ============================================================================================================================================================
+====================  ======================================================================================================================================================================
 
 Downloading official packages
 -----------------------------

providers/amazon/docs/transfer/sql_to_s3.rst

@@ -24,6 +24,10 @@ Use ``SqlToS3Operator`` to copy data from a SQL server to an Amazon Simple Stora
 converts the SQL result to `pandas dataframe <https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.html>`__
 (e.g. MySQL, Hive, ...).
 
+With the default ``df_type="pandas"`` the operator needs the ``pandas`` extra. Writing Parquet files
+(``file_format="parquet"``) or reading with ``read_kwargs={"dtype_backend": "pyarrow"}`` also needs
+pyarrow, which the ``parquet`` extra installs: ``apache-airflow-providers-amazon[pandas,parquet]``.
+
 Prerequisite Tasks
 ------------------
 

providers/amazon/pyproject.toml

@@ -136,6 +136,11 @@ dependencies = [
     'pandas>=2.2.3; python_version >="3.13" and python_version <"3.14"',
     'pandas>=2.3.3; python_version >="3.14"',
 ]
+"parquet" = [
+    "pyarrow>=16.1.0; python_version < '3.13'",
+    "pyarrow>=18.0.0; python_version >= '3.13' and python_version < '3.14'",
+    "pyarrow>=22.0.0; python_version >= '3.14'",
+]
 "openlineage" = [
     "apache-airflow-providers-openlineage>=2.3.0"
 ]
@@ -201,6 +206,7 @@ dev = [
     "responses>=0.25.0",
     "apache-airflow-providers-common-sql[pandas,polars]",
     "apache-airflow-providers-amazon[sqlalchemy]",
+    "apache-airflow-providers-amazon[parquet]",
     "apache-airflow-providers-duckdb"
 ]