Switch PythonVirtualenvOperator to venv from virtualenv package

The PythonVirtualenvOperator has been using virtualenv in order
to support Python 2.7 and pre Python 3.5, however we do not need
those any more - and we can use built-in venv instead.

Fixes: apache#40420

INSTALL

@@ -259,8 +259,7 @@ Those extras are available as regular core airflow extras - they install optiona
 # START CORE EXTRAS HERE
 
 aiobotocore, apache-atlas, apache-webhdfs, async, cgroups, cloudpickle, github-enterprise, google-
-auth, graphviz, kerberos, ldap, leveldb, otel, pandas, password, rabbitmq, s3fs, sentry, statsd, uv,
-virtualenv
+auth, graphviz, kerberos, ldap, leveldb, otel, pandas, password, rabbitmq, s3fs, sentry, statsd, uv
 
 # END CORE EXTRAS HERE
 

contributing-docs/12_airflow_dependencies_and_extras.rst

@@ -165,8 +165,7 @@ Those extras are available as regular core airflow extras - they install optiona
   .. START CORE EXTRAS HERE
 
 aiobotocore, apache-atlas, apache-webhdfs, async, cgroups, cloudpickle, github-enterprise, google-
-auth, graphviz, kerberos, ldap, leveldb, otel, pandas, password, rabbitmq, s3fs, sentry, statsd, uv,
-virtualenv
+auth, graphviz, kerberos, ldap, leveldb, otel, pandas, password, rabbitmq, s3fs, sentry, statsd, uv
 
   .. END CORE EXTRAS HERE
 

docs/apache-airflow/extra-packages-ref.rst

@@ -87,8 +87,6 @@ python dependencies for the provided package.
 +---------------------+-----------------------------------------------------+----------------------------------------------------------------------------+
 | uv                  | ``pip install 'apache-airflow[uv]'``                | Install uv - fast, Rust-based package installer (experimental)             |
 +---------------------+-----------------------------------------------------+----------------------------------------------------------------------------+
-| virtualenv          | ``pip install 'apache-airflow[virtualenv]'``        | Running python tasks in local virtualenv                                   |
-+---------------------+-----------------------------------------------------+----------------------------------------------------------------------------+
 | cloudpickle         | pip install apache-airflow[cloudpickle]             | Cloudpickle hooks and operators                                            |
 +---------------------+-----------------------------------------------------+----------------------------------------------------------------------------+
 

generated/provider_dependencies.json

@@ -1286,8 +1286,7 @@
   "standard": {
     "deps": [
       "apache-airflow-providers-common-sql>=1.18.0",
-      "apache-airflow>=2.8.0",
-      "virtualenv>=20.26.0"
+      "apache-airflow>=2.8.0"
     ],
     "devel-deps": [],
     "plugins": [],

hatch_build.py

@@ -137,9 +137,6 @@
     "uv": [
         "uv>=0.1.32",
     ],
-    "virtualenv": [
-        "virtualenv>=20.26.0",
-    ],
 }
 
 DOC_EXTRAS: dict[str, list[str]] = {

newsfragments/43568.significant.rst

@@ -0,0 +1 @@
+Remove ``virtualenv`` extra as PythonVirtualenvOperator has been moved to standard provider and switched to use built-in ven package.

providers/src/airflow/providers/standard/CHANGELOG.rst

@@ -42,3 +42,4 @@ Breaking changes
 ~~~~~~~~~~~~~~~~
 
 * ``In BranchDayOfWeekOperator, DayOfWeekSensor, BranchDateTimeOperator parameter use_task_execution_date has been removed. Please use use_task_logical_date.``
+* ``PythonVirtualenvOperator uses built-in venv instead of virtualenv package.``

providers/src/airflow/providers/standard/operators/python.py

@@ -17,7 +17,6 @@
 # under the License.
 from __future__ import annotations
 
-import importlib
 import inspect
 import json
 import logging
@@ -73,17 +72,6 @@
     from airflow.utils.context import Context
 
 
-def is_venv_installed() -> bool:
-    """
-    Check if the virtualenv package is installed via checking if it is on the path or installed as package.
-
-    :return: True if it is. Whichever way of checking it works, is fine.
-    """
-    if shutil.which("virtualenv") or importlib.util.find_spec("virtualenv"):
-        return True
-    return False
-
-
 @cache
 def _parse_version_info(text: str) -> tuple[int, int, int, str, int]:
     """Parse python version info from a text."""
@@ -710,8 +698,6 @@ def __init__(
                 "Passing non-string types (e.g. int or float) as python_version not supported"
             )
 
-        if not is_venv_installed():
-            raise AirflowException("PythonVirtualenvOperator requires virtualenv, please install it.")
         if use_airflow_context and (not expect_airflow and not system_site_packages):
             error_msg = "use_airflow_context is set to True, but expect_airflow and system_site_packages are set to False."
             raise AirflowException(error_msg)

providers/src/airflow/providers/standard/provider.yaml

@@ -30,7 +30,6 @@ versions:
 dependencies:
   - apache-airflow>=2.8.0
   - apache-airflow-providers-common-sql>=1.18.0
-  - virtualenv>=20.26.0
 
 integrations:
   - integration-name: Standard

providers/src/airflow/providers/standard/utils/python_virtualenv.py

@@ -29,12 +29,13 @@
 from airflow.utils.process_utils import execute_in_subprocess
 
 
-def _generate_virtualenv_cmd(tmp_dir: str, python_bin: str, system_site_packages: bool) -> list[str]:
-    cmd = [sys.executable, "-m", "virtualenv", tmp_dir]
+def _generate_venv_cmd(tmp_dir: str, python_bin: str, system_site_packages: bool) -> list[str]:
+    """We are using venv command instead of venv module to allow creation of venv for different python versions."""
+    if python_bin is None:
+        python_bin = sys.executable
+    cmd = [python_bin, "-m", "venv", tmp_dir]
     if system_site_packages:
         cmd.append("--system-site-packages")
-    if python_bin is not None:
-        cmd.append(f"--python={python_bin}")
     return cmd
 
 
@@ -90,8 +91,8 @@ def prepare_virtualenv(
     if index_urls is not None:
         _generate_pip_conf(Path(venv_directory) / "pip.conf", index_urls)
 
-    virtualenv_cmd = _generate_virtualenv_cmd(venv_directory, python_bin, system_site_packages)
-    execute_in_subprocess(virtualenv_cmd)
+    venv_cmd = _generate_venv_cmd(venv_directory, python_bin, system_site_packages)
+    execute_in_subprocess(venv_cmd)
 
     if requirements is not None and requirements_file_path is not None:
         raise ValueError("Either requirements OR requirements_file_path has to be passed, but not both")

pyproject.toml

@@ -78,8 +78,7 @@ dynamic = ["version", "optional-dependencies", "dependencies"]
 # START CORE EXTRAS HERE
 #
 # aiobotocore, apache-atlas, apache-webhdfs, async, cgroups, cloudpickle, github-enterprise, google-
-# auth, graphviz, kerberos, ldap, leveldb, otel, pandas, password, rabbitmq, s3fs, sentry, statsd, uv,
-# virtualenv
+# auth, graphviz, kerberos, ldap, leveldb, otel, pandas, password, rabbitmq, s3fs, sentry, statsd, uv
 #
 # END CORE EXTRAS HERE
 #