Merge remote-tracking branch 'upstream/master' into is8102/add-search-api-in-storage

Author: giancarloromeo

packages/aws-library/src/aws_library/s3/__init__.py

@@ -22,10 +22,10 @@
 )
 
 __all__: tuple[str, ...] = (
-    "CopiedBytesTransferredCallback",
-    "MultiPartUploadLinks",
     "PRESIGNED_LINK_MAX_SIZE",
     "S3_MAX_FILE_SIZE",
+    "CopiedBytesTransferredCallback",
+    "MultiPartUploadLinks",
     "S3AccessError",
     "S3BucketInvalidError",
     "S3DestinationNotEmptyError",
@@ -37,8 +37,8 @@
     "S3RuntimeError",
     "S3UploadNotFoundError",
     "SimcoreS3API",
-    "UploadedBytesTransferredCallback",
     "UploadID",
+    "UploadedBytesTransferredCallback",
 )
 
 # nopycln: file

packages/celery-library/src/celery_library/backends/_redis.py renamed to packages/celery-library/src/celery_library/backends/redis.py

@@ -3,24 +3,22 @@
 import logging
 from collections.abc import AsyncIterable
 from datetime import timedelta
-from typing import Final
+from typing import TYPE_CHECKING, Final
 
 from models_library.progress_bar import ProgressReport
 from pydantic import ValidationError
 from servicelib.celery.models import (
     Task,
     TaskFilter,
     TaskID,
+    TaskInfoStore,
     TaskMetadata,
-    TaskUUID,
+    Wildcard,
 )
-from servicelib.redis import RedisClientSDK
-
-from ..utils import build_task_id_prefix
+from servicelib.redis import RedisClientSDK, handle_redis_returns_union_types
 
 _CELERY_TASK_INFO_PREFIX: Final[str] = "celery-task-info-"
 _CELERY_TASK_ID_KEY_ENCODING = "utf-8"
-_CELERY_TASK_ID_KEY_SEPARATOR: Final[str] = ":"
 _CELERY_TASK_SCAN_COUNT_PER_BATCH: Final[int] = 1000
 _CELERY_TASK_METADATA_KEY: Final[str] = "metadata"
 _CELERY_TASK_PROGRESS_KEY: Final[str] = "progress"
@@ -43,18 +41,24 @@ async def create_task(
         expiry: timedelta,
     ) -> None:
         task_key = _build_key(task_id)
-        await self._redis_client_sdk.redis.hset(
-            name=task_key,
-            key=_CELERY_TASK_METADATA_KEY,
-            value=task_metadata.model_dump_json(),
-        )  # type: ignore
+        await handle_redis_returns_union_types(
+            self._redis_client_sdk.redis.hset(
+                name=task_key,
+                key=_CELERY_TASK_METADATA_KEY,
+                value=task_metadata.model_dump_json(),
+            )
+        )
         await self._redis_client_sdk.redis.expire(
             task_key,
             expiry,
         )
 
     async def get_task_metadata(self, task_id: TaskID) -> TaskMetadata | None:
-        raw_result = await self._redis_client_sdk.redis.hget(_build_key(task_id), _CELERY_TASK_METADATA_KEY)  # type: ignore
+        raw_result = await handle_redis_returns_union_types(
+            self._redis_client_sdk.redis.hget(
+                _build_key(task_id), _CELERY_TASK_METADATA_KEY
+            )
+        )
         if not raw_result:
             return None
 
@@ -67,7 +71,11 @@ async def get_task_metadata(self, task_id: TaskID) -> TaskMetadata | None:
             return None
 
     async def get_task_progress(self, task_id: TaskID) -> ProgressReport | None:
-        raw_result = await self._redis_client_sdk.redis.hget(_build_key(task_id), _CELERY_TASK_PROGRESS_KEY)  # type: ignore
+        raw_result = await handle_redis_returns_union_types(
+            self._redis_client_sdk.redis.hget(
+                _build_key(task_id), _CELERY_TASK_PROGRESS_KEY
+            )
+        )
         if not raw_result:
             return None
 
@@ -80,17 +88,14 @@ async def get_task_progress(self, task_id: TaskID) -> ProgressReport | None:
             return None
 
     async def list_tasks(self, task_filter: TaskFilter) -> list[Task]:
-        search_key = (
-            _CELERY_TASK_INFO_PREFIX
-            + build_task_id_prefix(task_filter)
-            + _CELERY_TASK_ID_KEY_SEPARATOR
+        search_key = _CELERY_TASK_INFO_PREFIX + task_filter.create_task_id(
+            task_uuid=Wildcard()
         )
-        search_key_len = len(search_key)
 
         keys: list[str] = []
         pipeline = self._redis_client_sdk.redis.pipeline()
         async for key in self._redis_client_sdk.redis.scan_iter(
-            match=search_key + "*", count=_CELERY_TASK_SCAN_COUNT_PER_BATCH
+            match=search_key, count=_CELERY_TASK_SCAN_COUNT_PER_BATCH
         ):
             # fake redis (tests) returns bytes, real redis returns str
             _key = (
@@ -112,7 +117,7 @@ async def list_tasks(self, task_filter: TaskFilter) -> list[Task]:
                 task_metadata = TaskMetadata.model_validate_json(raw_metadata)
                 tasks.append(
                     Task(
-                        uuid=TaskUUID(key[search_key_len:]),
+                        uuid=TaskFilter.get_task_uuid(key),
                         metadata=task_metadata,
                     )
                 )
@@ -153,13 +158,19 @@ async def stream_task_result(
                         return
 
     async def set_task_progress(self, task_id: TaskID, report: ProgressReport) -> None:
-        await self._redis_client_sdk.redis.hset(
-            name=_build_key(task_id),
-            key=_CELERY_TASK_PROGRESS_KEY,
-            value=report.model_dump_json(),
-        )  # type: ignore
+        await handle_redis_returns_union_types(
+            self._redis_client_sdk.redis.hset(
+                name=_build_key(task_id),
+                key=_CELERY_TASK_PROGRESS_KEY,
+                value=report.model_dump_json(),
+            )
+        )
 
     async def task_exists(self, task_id: TaskID) -> bool:
         n = await self._redis_client_sdk.redis.exists(_build_key(task_id))
         assert isinstance(n, int)  # nosec
         return n > 0
+
+
+if TYPE_CHECKING:
+    _: type[TaskInfoStore] = RedisTaskInfoStore

packages/celery-library/src/celery_library/common.py

@@ -2,13 +2,9 @@
 from typing import Any
 
 from celery import Celery  # type: ignore[import-untyped]
-from servicelib.redis import RedisClientSDK
 from settings_library.celery import CelerySettings
 from settings_library.redis import RedisDatabase
 
-from .backends._redis import RedisTaskInfoStore
-from .task_manager import CeleryTaskManager
-
 
 def _celery_configure(celery_settings: CelerySettings) -> dict[str, Any]:
     base_config = {
@@ -36,22 +32,3 @@ def create_app(settings: CelerySettings) -> Celery:
         ),
         **_celery_configure(settings),
     )
-
-
-async def create_task_manager(
-    app: Celery, settings: CelerySettings
-) -> CeleryTaskManager:
-    redis_client_sdk = RedisClientSDK(
-        settings.CELERY_REDIS_RESULT_BACKEND.build_redis_dsn(
-            RedisDatabase.CELERY_TASKS
-        ),
-        client_name="celery_tasks",
-    )
-    await redis_client_sdk.setup()
-    # GCR please address https://github.com/ITISFoundation/osparc-simcore/issues/8159
-
-    return CeleryTaskManager(
-        app,
-        settings,
-        RedisTaskInfoStore(redis_client_sdk),
-    )

packages/celery-library/src/celery_library/rpc/_async_jobs.py

@@ -134,9 +134,8 @@ async def result(
 
 @router.expose(reraise_if_error_type=(JobSchedulerError,))
 async def list_jobs(
-    task_manager: TaskManager, filter_: str, job_filter: AsyncJobFilter
+    task_manager: TaskManager, job_filter: AsyncJobFilter
 ) -> list[AsyncJobGet]:
-    _ = filter_
     assert task_manager  # nosec
     task_filter = TaskFilter.model_validate(job_filter.model_dump())
     try:

packages/celery-library/src/celery_library/signals.py

@@ -6,18 +6,15 @@
 from celery.worker.worker import WorkController  # type: ignore[import-untyped]
 from servicelib.celery.app_server import BaseAppServer
 from servicelib.logging_utils import log_context
-from settings_library.celery import CelerySettings
 
-from .common import create_task_manager
 from .utils import get_app_server, set_app_server
 
 _logger = logging.getLogger(__name__)
 
 
 def on_worker_init(
-    app_server: BaseAppServer,
-    celery_settings: CelerySettings,
     sender: WorkController,
+    app_server: BaseAppServer,
     **_kwargs,
 ) -> None:
     startup_complete_event = threading.Event()
@@ -26,21 +23,14 @@ def _init(startup_complete_event: threading.Event) -> None:
         loop = asyncio.new_event_loop()
         asyncio.set_event_loop(loop)
 
-        async def _setup_task_manager():
-            assert sender.app  # nosec
-            assert isinstance(sender.app, Celery)  # nosec
-
-            app_server.task_manager = await create_task_manager(
-                sender.app,
-                celery_settings,
-            )
+        assert sender.app  # nosec
+        assert isinstance(sender.app, Celery)  # nosec
 
-            set_app_server(sender.app, app_server)
+        set_app_server(sender.app, app_server)
 
         app_server.event_loop = loop
 
-        loop.run_until_complete(_setup_task_manager())
-        loop.run_until_complete(app_server.lifespan(startup_complete_event))
+        loop.run_until_complete(app_server.run_until_shutdown(startup_complete_event))
 
     thread = threading.Thread(
         group=None,

packages/celery-library/src/celery_library/task_manager.py

@@ -22,7 +22,6 @@
 from settings_library.celery import CelerySettings
 
 from .errors import TaskNotFoundError
-from .utils import build_task_id
 
 _logger = logging.getLogger(__name__)
 
@@ -50,7 +49,7 @@ async def submit_task(
             msg=f"Submit {task_metadata.name=}: {task_filter=} {task_params=}",
         ):
             task_uuid = uuid4()
-            task_id = build_task_id(task_filter, task_uuid)
+            task_id = task_filter.create_task_id(task_uuid=task_uuid)
             self._celery_app.send_task(
                 task_metadata.name,
                 task_id=task_id,
@@ -74,7 +73,7 @@ async def cancel_task(self, task_filter: TaskFilter, task_uuid: TaskUUID) -> Non
             logging.DEBUG,
             msg=f"task cancellation: {task_filter=} {task_uuid=}",
         ):
-            task_id = build_task_id(task_filter, task_uuid)
+            task_id = task_filter.create_task_id(task_uuid=task_uuid)
             if not await self.task_exists(task_id):
                 raise TaskNotFoundError(task_id=task_id)
 
@@ -96,7 +95,7 @@ async def get_task_result(
             logging.DEBUG,
             msg=f"Get task result: {task_filter=} {task_uuid=}",
         ):
-            task_id = build_task_id(task_filter, task_uuid)
+            task_id = task_filter.create_task_id(task_uuid=task_uuid)
             if not await self.task_exists(task_id):
                 raise TaskNotFoundError(task_id=task_id)
 
@@ -139,7 +138,7 @@ async def get_task_status(
             logging.DEBUG,
             msg=f"Getting task status: {task_filter=} {task_uuid=}",
         ):
-            task_id = build_task_id(task_filter, task_uuid)
+            task_id = task_filter.create_task_id(task_uuid=task_uuid)
             if not await self.task_exists(task_id):
                 raise TaskNotFoundError(task_id=task_id)
 

packages/celery-library/src/celery_library/utils.py

@@ -1,26 +1,8 @@
-from typing import Final
-
 from celery import Celery  # type: ignore[import-untyped]
 from servicelib.celery.app_server import BaseAppServer
-from servicelib.celery.models import TaskFilter, TaskID, TaskUUID
 
 _APP_SERVER_KEY = "app_server"
 
-_TASK_ID_KEY_DELIMITATOR: Final[str] = ":"
-
-
-def build_task_id_prefix(task_filter: TaskFilter) -> str:
-    filter_dict = task_filter.model_dump()
-    return _TASK_ID_KEY_DELIMITATOR.join(
-        [f"{filter_dict[key]}" for key in sorted(filter_dict)]
-    )
-
-
-def build_task_id(task_filter: TaskFilter, task_uuid: TaskUUID) -> TaskID:
-    return _TASK_ID_KEY_DELIMITATOR.join(
-        [build_task_id_prefix(task_filter), f"{task_uuid}"]
-    )
-
 
 def get_app_server(app: Celery) -> BaseAppServer:
     app_server = app.conf[_APP_SERVER_KEY]