Files
superset2/superset/distributed_lock/__init__.py
Claude Code dfd3f7b316 ci(lint): enforce no function-body imports (PLC0415) with targeted ignores
Follow-up to #40231 (merged), where a reviewer flagged a function-body
`from datetime import datetime, timedelta` instead of a top-of-file
import. Adds a `ruff-import-placement` pre-commit hook running
`ruff check --select PLC0415 --preview --no-fix`.

Per @rusackas's pushback on the first cut of this PR — which spammed
2,657 `# noqa: PLC0415` annotations across ~410 files without fixing
anything — this revision is a much smaller surface area:

1. **Per-file-ignores** for whole directories where function-body
   imports are a deliberate pattern, not an oversight:
   - `superset/cli/**` and `scripts/**`: subcommand-deferred imports
     keep heavy modules out of the CLI startup path.
   - `superset/tasks/**`: Celery task bodies defer imports of the
     modules they orchestrate.
   - `superset/migrations/versions/**`: Alembic migrations interact
     with model state at runtime, not at module load.
   - `superset/mcp_service/**`: MCP tools lazy-load resources on
     invocation so the server can register many tools without paying
     their import cost at startup.
   - `superset/db_engine_specs/**`: engine specs defer driver imports
     so optional DB drivers don't have to be installed.
   - `superset/initialization/__init__.py`, `superset/extensions/__init__.py`,
     `superset/app.py`: the app-factory and extension wiring are
     intentionally full of circular-import workarounds.
   - `tests/**`: test files routinely defer imports for fixture
     isolation; the rule still applies to production code.

2. **Per-line `# noqa: PLC0415`** on the 259 remaining genuine
   circular-import sites (security/manager.py, sql/execution/executor.py,
   semantic_layers/labels.py, tags/core.py, core_api_injection.py, etc.).
   These are foundational modules where moving the imports up would
   actually break things.

Net result: ~410 files / 2,657 grandfathered → ~73 files / 259 actual
noqa annotations. The rule still catches every new function-body
import outside the explicitly-allowed directories.

Also: silences a pre-existing C901 on `mcp_service/sql_lab/tool/execute_sql.py`
that fires under newer local ruff but not CI's pinned ruff 0.9.7 — blocks
the local pre-commit run otherwise.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-20 13:55:14 -07:00

67 lines
2.5 KiB
Python

# 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
import uuid
from collections.abc import Iterator
from contextlib import contextmanager
from typing import Any
from superset.distributed_lock.utils import get_key
@contextmanager
def DistributedLock( # noqa: N802
namespace: str,
ttl_seconds: int | None = None,
**kwargs: Any,
) -> Iterator[uuid.UUID]:
"""
Distributed lock for coordinating operations across workers.
Automatically uses Redis-based locking when DISTRIBUTED_COORDINATION_CONFIG is
configured, falling back to database-backed locking otherwise.
Redis locking uses SET NX EX for atomic acquisition with automatic expiration.
Database locking uses the KeyValue table with manual expiration cleanup.
:param namespace: Lock namespace for grouping related locks
:param ttl_seconds: Lock TTL in seconds. Defaults to 30 seconds.
After expiration, the lock is automatically released
to prevent deadlocks from crashed processes.
:param kwargs: Additional key parameters to differentiate locks
:yields: UUID identifying this lock acquisition
:raises AcquireDistributedLockFailedException: If lock is already held
or Redis connection fails
"""
# pylint: disable=import-outside-toplevel
from superset.commands.distributed_lock.acquire import ( # noqa: PLC0415
AcquireDistributedLock,
)
from superset.commands.distributed_lock.release import ( # noqa: PLC0415
ReleaseDistributedLock,
)
key = get_key(namespace, **kwargs)
AcquireDistributedLock(namespace, kwargs, ttl_seconds).run()
try:
yield key
finally:
ReleaseDistributedLock(namespace, kwargs).run()