def refactor_amazon_package(self):
"""
Fixes to "amazon" providers package.
Copies some of the classes used from core Airflow to "common.utils" package of
the provider and renames imports to use them from there.
We copy typing_compat.py and change import as in example diff:
.. code-block:: diff
--- ./airflow/providers/amazon/aws/operators/ecs.py
+++ ./airflow/providers/amazon/aws/operators/ecs.py
@@ -24,7 +24,7 @@
from airflow.models import BaseOperator
from airflow.providers.amazon.aws.hooks.base_aws import AwsBaseHook
from airflow.providers.amazon.aws.hooks.logs import AwsLogsHook
-from airflow.typing_compat import Protocol, runtime_checkable
+from airflow.providers.amazon.common.utils.typing_compat import Protocol, runtime_checkable
from airflow.utils.decorators import apply_defaults
"""
def amazon_package_filter(node: LN, capture: Capture,
filename: Filename) -> bool:
return filename.startswith("./airflow/providers/amazon/")
os.makedirs(os.path.join(get_target_providers_package_folder("amazon"),
"common", "utils"),
exist_ok=True)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"__init__.py"),
os.path.join(get_target_providers_package_folder("amazon"),
"common", "__init__.py"),
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"__init__.py"),
os.path.join(get_target_providers_package_folder("amazon"),
"common", "utils", "__init__.py"),
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow",
"typing_compat.py"),
os.path.join(get_target_providers_package_folder("amazon"),
"common", "utils", "typing_compat.py"),
)
(self.qry.select_module("airflow.typing_compat").filter(
callback=amazon_package_filter).rename(
"airflow.providers.amazon.common.utils.typing_compat"))
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"email.py"),
os.path.join(get_target_providers_package_folder("amazon"),
"common", "utils", "email.py"),
)
(self.qry.select_module("airflow.utils.email").filter(
callback=amazon_package_filter).rename(
"airflow.providers.amazon.common.utils.email"))
def refactor_apache_beam_package(self):
r"""
Fixes to "apache_beam" providers package.
Copies some of the classes used from core Airflow to "common.utils" package of the
the provider and renames imports to use them from there. Note that in this case we also rename
the imports in the copied files.
For example we copy python_virtualenv.py, process_utils.py and change import as in example diff:
.. code-block:: diff
--- ./airflow/providers/apache/beam/common/utils/python_virtualenv.py
+++ ./airflow/providers/apache/beam/common/utils/python_virtualenv.py
@@ -21,7 +21,7 @@
\"\"\"
from typing import List, Optional
-from airflow.utils.process_utils import execute_in_subprocess
+from airflow.providers.apache.beam.common.utils.process_utils import execute_in_subprocess
def _generate_virtualenv_cmd(tmp_dir: str, python_bin: str, system_site_packages: bool)
"""
def apache_beam_package_filter(node: LN, capture: Capture, filename: Filename) -> bool:
return filename.startswith("./airflow/providers/apache/beam")
os.makedirs(
os.path.join(get_target_providers_package_folder("apache.beam"), "common", "utils"), exist_ok=True
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils", "__init__.py"),
os.path.join(
get_target_providers_package_folder("apache.beam"), "common", "utils", "__init__.py"
),
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils", "python_virtualenv.py"),
os.path.join(
get_target_providers_package_folder("apache.beam"), "common", "utils", "python_virtualenv.py"
),
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils", "process_utils.py"),
os.path.join(
get_target_providers_package_folder("apache.beam"), "common", "utils", "process_utils.py"
),
)
(
self.qry.select_module("airflow.utils.python_virtualenv")
.filter(callback=apache_beam_package_filter)
.rename("airflow.providers.apache.beam.common.utils.python_virtualenv")
)
(
self.qry.select_module("airflow.utils.process_utils")
.filter(callback=apache_beam_package_filter)
.rename("airflow.providers.apache.beam.common.utils.process_utils")
)
def copy_helper_py_file(target_file_path: str) -> None:
"""
Copies. airflow/utils/helper.py to a new location within provider package
The helper has two methods (chain, cross_downstream) that are moved from the original helper to
'airflow.models.baseoperator'. so in 1.10 they should reimport the original 'airflow.utils.helper'
methods. Those deprecated methods use import with import_string("<IMPORT>") so it is easier to
replace them as strings rather than with Bowler
:param target_file_path: target path name for the helpers.py
"""
source_helper_file_path = os.path.join(get_source_airflow_folder(), "airflow", "utils", "helpers.py")
with open(source_helper_file_path) as in_file:
with open(target_file_path, "wt") as out_file:
for line in in_file:
out_file.write(line.replace('airflow.models.baseoperator', 'airflow.utils.helpers'))
def refactor_odbc_package(self):
"""
Fixes to "odbc" providers package.
Copies some of the classes used from core Airflow to "common.utils" package of the
the provider and renames imports to use them from there.
We copy helpers.py and change import as in example diff:
.. code-block:: diff
--- ./airflow/providers/google/cloud/example_dags/example_datacatalog.py
+++ ./airflow/providers/google/cloud/example_dags/example_datacatalog.py
@@ -37,7 +37,7 @@
CloudDataCatalogUpdateTagTemplateOperator,
)
from airflow.utils.dates import days_ago
-from airflow.utils.helpers import chain
+from airflow.providers.odbc.utils.helpers import chain
default_args = {"start_date": days_ago(1)}
"""
def odbc_package_filter(node: LN, capture: Capture,
filename: Filename) -> bool:
return filename.startswith("./airflow/providers/odbc/")
os.makedirs(os.path.join(get_target_providers_folder(), "odbc",
"utils"),
exist_ok=True)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"__init__.py"),
os.path.join(get_target_providers_package_folder("odbc"), "utils",
"__init__.py"),
)
copy_helper_py_file(
os.path.join(get_target_providers_package_folder("odbc"), "utils",
"helpers.py"))
(self.qry.select_module("airflow.utils.helpers").filter(
callback=odbc_package_filter).rename(
"airflow.providers.odbc.utils.helpers"))
def refactor_google_package(self):
r"""
Fixes to "google" providers package.
Copies some of the classes used from core Airflow to "common.utils" package of the
the provider and renames imports to use them from there. Note that in this case we also rename
the imports in the copied files.
For example we copy python_virtualenv.py, process_utils.py and change import as in example diff:
.. code-block:: diff
--- ./airflow/providers/google/cloud/operators/kubernetes_engine.py
+++ ./airflow/providers/google/cloud/operators/kubernetes_engine.py
@@ -28,11 +28,11 @@
from airflow.exceptions import AirflowException
from airflow.models import BaseOperator
-from airflow.providers.cncf.kubernetes.operators.kubernetes_pod import KubernetesPodOperator
+from airflow.contrib.operators.kubernetes_pod_operator import KubernetesPodOperator
from airflow.providers.google.cloud.hooks.kubernetes_engine import GKEHook
from airflow.providers.google.common.hooks.base_google import GoogleBaseHook
from airflow.utils.decorators import apply_defaults
-from airflow.utils.process_utils import execute_in_subprocess, patch_environ
+from airflow.providers.google.common.utils.process_utils import execute_in_subprocess
And in the copied python_virtualenv.py we also change import to process_utils.py. This happens
automatically and is solved by Pybowler.
.. code-block:: diff
--- ./airflow/providers/google/common/utils/python_virtualenv.py
+++ ./airflow/providers/google/common/utils/python_virtualenv.py
@@ -21,7 +21,7 @@
\"\"\"
from typing import List, Optional
-from airflow.utils.process_utils import execute_in_subprocess
+from airflow.providers.google.common.utils.process_utils import execute_in_subprocess
def _generate_virtualenv_cmd(tmp_dir: str, python_bin: str, system_site_packages: bool)
We also rename Base operator links to deprecated names:
.. code-block:: diff
--- ./airflow/providers/google/cloud/operators/mlengine.py
+++ ./airflow/providers/google/cloud/operators/mlengine.py
@@ -24,7 +24,7 @@
from typing import List, Optional
from airflow.exceptions import AirflowException
-from airflow.models import BaseOperator, BaseOperatorLink
+from airflow.models.baseoperator import BaseOperator, BaseOperatorLink
from airflow.models.taskinstance import TaskInstance
from airflow.providers.google.cloud.hooks.mlengine import MLEngineHook
from airflow.utils.decorators import apply_defaults
We also copy (google.common.utils) and rename imports to the helpers.
.. code-block:: diff
--- ./airflow/providers/google/cloud/example_dags/example_datacatalog.py
+++ ./airflow/providers/google/cloud/example_dags/example_datacatalog.py
@@ -37,7 +37,7 @@
CloudDataCatalogUpdateTagTemplateOperator,
)
from airflow.utils.dates import days_ago
-from airflow.utils.helpers import chain
+from airflow.providers.google.common.utils.helpers import chain
default_args = {"start_date": days_ago(1)}
And also module_loading which is used by helpers
.. code-block:: diff
--- ./airflow/providers/google/common/utils/helpers.py
+++ ./airflow/providers/google/common/utils/helpers.py
@@ -26,7 +26,7 @@
from jinja2 import Template
from airflow.exceptions import AirflowException
-from airflow.utils.module_loading import import_string
+from airflow.providers.google.common.utils.module_loading import import_string
KEY_REGEX = re.compile(r'^[\\w.-]+$')
"""
def google_package_filter(node: LN, capture: Capture,
filename: Filename) -> bool:
return filename.startswith("./airflow/providers/google/")
def pure_airflow_models_filter(node: LN, capture: Capture,
filename: Filename) -> bool:
"""Check if select is exactly [airflow, . , models]"""
return len(list(node.children[1].leaves())) == 3
def _contains_chain_in_import_filter(node: LN, capture: Capture,
filename: Filename) -> bool:
if "module_import" in capture:
return bool(
"chain" in capture["module_import"].value
) and filename.startswith("./airflow/providers/google/")
return False
os.makedirs(os.path.join(get_target_providers_package_folder("google"),
"common", "utils"),
exist_ok=True)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"__init__.py"),
os.path.join(get_target_providers_package_folder("google"),
"common", "utils", "__init__.py"),
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"python_virtualenv.py"),
os.path.join(get_target_providers_package_folder("google"),
"common", "utils", "python_virtualenv.py"),
)
copy_helper_py_file(
os.path.join(get_target_providers_package_folder("google"),
"common", "utils", "helpers.py"))
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"module_loading.py"),
os.path.join(get_target_providers_package_folder("google"),
"common", "utils", "module_loading.py"),
)
(self.qry.select_module("airflow.utils.python_virtualenv").filter(
callback=google_package_filter).rename(
"airflow.providers.google.common.utils.python_virtualenv"))
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"process_utils.py"),
os.path.join(get_target_providers_package_folder("google"),
"common", "utils", "process_utils.py"),
)
(self.qry.select_module("airflow.utils.process_utils").filter(
callback=google_package_filter).rename(
"airflow.providers.google.common.utils.process_utils"))
(self.qry.select_module("airflow.models.baseoperator").filter(
callback=_contains_chain_in_import_filter).rename(
"airflow.providers.google.common.utils.helpers"))
(self.qry.select_module("airflow.utils.helpers").filter(
callback=google_package_filter).rename(
"airflow.providers.google.common.utils.helpers"))
(self.qry.select_module("airflow.utils.module_loading").filter(
callback=google_package_filter).rename(
"airflow.providers.google.common.utils.module_loading"))
(
# Fix BaseOperatorLinks imports
self.qry.select_module("airflow.models").is_filename(
include=r"bigquery\.py|mlengine\.py"
).filter(callback=google_package_filter
).filter(pure_airflow_models_filter).rename(
"airflow.models.baseoperator"))
def refactor_elasticsearch_package(self):
"""
Fixes to "elasticsearch" providers package.
Copies some of the classes used from core Airflow to "common.utils" package of
the provider and renames imports to use them from there.
We copy file_task_handler.py and change import as in example diff:
.. code-block:: diff
--- ./airflow/providers/elasticsearch/log/es_task_handler.py
+++ ./airflow/providers/elasticsearch/log/es_task_handler.py
@@ -24,7 +24,7 @@
from airflow.configuration import conf
from airflow.models import TaskInstance
from airflow.utils import timezone
from airflow.utils.helpers import parse_template_string
-from airflow.utils.log.file_task_handler import FileTaskHandler
+from airflow.providers.elasticsearch.common.utils.log.file_task_handler import FileTaskHandler
from airflow.utils.log.json_formatter import JSONFormatter
from airflow.utils.log.logging_mixin import LoggingMixin
"""
def elasticsearch_package_filter(node: LN, capture: Capture,
filename: Filename) -> bool:
return filename.startswith("./airflow/providers/elasticsearch/")
os.makedirs(
os.path.join(get_target_providers_package_folder("elasticsearch"),
"common", "utils", "log"),
exist_ok=True,
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"__init__.py"),
os.path.join(get_target_providers_package_folder("elasticsearch"),
"common", "__init__.py"),
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"__init__.py"),
os.path.join(get_target_providers_package_folder("elasticsearch"),
"common", "utils", "__init__.py"),
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"log", "__init__.py"),
os.path.join(get_target_providers_package_folder("elasticsearch"),
"common", "utils", "log", "__init__.py"),
)
copyfile(
os.path.join(get_source_airflow_folder(), "airflow", "utils",
"log", "file_task_handler.py"),
os.path.join(
get_target_providers_package_folder("elasticsearch"),
"common",
"utils",
"log",
"file_task_handler.py",
),
)
(self.qry.select_module("airflow.utils.log.file_task_handler").filter(
callback=elasticsearch_package_filter
).rename(
"airflow.providers.elasticsearch.common.utils.log.file_task_handler"
))
