ddeutil-workflow 0.0.18__tar.gz → 0.0.20__tar.gz

{ddeutil_workflow-0.0.18/src/ddeutil_workflow.egg-info → ddeutil_workflow-0.0.20}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ddeutil-workflow
-Version: 0.0.18
+Version: 0.0.20
 Summary: Lightweight workflow orchestration with less dependencies
 Author-email: ddeutils <korawich.anu@gmail.com>
 License: MIT
@@ -24,9 +24,9 @@ Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: ddeutil>=0.4.3
 Requires-Dist: ddeutil-io[toml,yaml]>=0.2.3
-Requires-Dist: pydantic==2.9.2
+Requires-Dist: pydantic==2.10.2
 Requires-Dist: python-dotenv==1.0.1
-Requires-Dist: typer<1.0.0,==0.12.5
+Requires-Dist: typer==0.15.1
 Requires-Dist: schedule<2.0.0,==1.2.2
 Provides-Extra: api
 Requires-Dist: fastapi<1.0.0,>=0.115.0; extra == "api"
@@ -46,7 +46,7 @@ for easy to make a simple metadata driven for data workflow orchestration.
 It can to use for data operator by a `.yaml` template.
 
 > [!WARNING]
-> This package provide only orchestration workload task. That mean you should not
+> This package provide only orchestration workload. That mean you should not
 > use the workflow stage to process any large volume data which use lot of compute
 > resource. :cold_sweat:
 
@@ -58,10 +58,10 @@ configuration. It called **Metadata Driven Data Workflow**.
 
 **:pushpin: <u>Rules of This Workflow engine</u>**:
 
-1. Minimum frequency unit of scheduling is **1 minute** :warning:
+1. The Minimum frequency unit of scheduling is **1 minute** :warning:
 2. Can not re-run only failed stage and its pending downstream :rotating_light:
 3. All parallel tasks inside workflow engine use Multi-Threading
-   (Because Python 3.13 unlock GIL :unlock:)
+   (Python 3.13 unlock GIL :unlock:)
 
 > [!NOTE]
 > _Disclaimer_: I inspire the dynamic statement from the [**GitHub Action**](https://github.com/features/actions)
@@ -183,7 +183,7 @@ application. If any configuration values do not set yet, it will use default val
 and do not raise any error to you.
 
 | Environment                             | Component | Default                          | Description                                                                                                        | Remark |
-|:----------------------------------------|-----------|----------------------------------|--------------------------------------------------------------------------------------------------------------------|--------|
+|:----------------------------------------|:----------|:---------------------------------|--------------------------------------------------------------------------------------------------------------------|--------|
 | `WORKFLOW_ROOT_PATH`                    | Core      | .                                | The root path of the workflow application.                                                                         |        |
 | `WORKFLOW_CORE_REGISTRY`                | Core      | src.ddeutil.workflow,tests.utils | List of importable string for the hook stage.                                                                      |        |
 | `WORKFLOW_CORE_REGISTRY_FILTER`         | Core      | ddeutil.workflow.utils           | List of importable string for the filter template.                                                                 |        |

{ddeutil_workflow-0.0.18 → ddeutil_workflow-0.0.20}/README.md

@@ -13,7 +13,7 @@ for easy to make a simple metadata driven for data workflow orchestration.
 It can to use for data operator by a `.yaml` template.
 
 > [!WARNING]
-> This package provide only orchestration workload task. That mean you should not
+> This package provide only orchestration workload. That mean you should not
 > use the workflow stage to process any large volume data which use lot of compute
 > resource. :cold_sweat:
 
@@ -25,10 +25,10 @@ configuration. It called **Metadata Driven Data Workflow**.
 
 **:pushpin: <u>Rules of This Workflow engine</u>**:
 
-1. Minimum frequency unit of scheduling is **1 minute** :warning:
+1. The Minimum frequency unit of scheduling is **1 minute** :warning:
 2. Can not re-run only failed stage and its pending downstream :rotating_light:
 3. All parallel tasks inside workflow engine use Multi-Threading
-   (Because Python 3.13 unlock GIL :unlock:)
+   (Python 3.13 unlock GIL :unlock:)
 
 > [!NOTE]
 > _Disclaimer_: I inspire the dynamic statement from the [**GitHub Action**](https://github.com/features/actions)
@@ -150,7 +150,7 @@ application. If any configuration values do not set yet, it will use default val
 and do not raise any error to you.
 
 | Environment                             | Component | Default                          | Description                                                                                                        | Remark |
-|:----------------------------------------|-----------|----------------------------------|--------------------------------------------------------------------------------------------------------------------|--------|
+|:----------------------------------------|:----------|:---------------------------------|--------------------------------------------------------------------------------------------------------------------|--------|
 | `WORKFLOW_ROOT_PATH`                    | Core      | .                                | The root path of the workflow application.                                                                         |        |
 | `WORKFLOW_CORE_REGISTRY`                | Core      | src.ddeutil.workflow,tests.utils | List of importable string for the hook stage.                                                                      |        |
 | `WORKFLOW_CORE_REGISTRY_FILTER`         | Core      | ddeutil.workflow.utils           | List of importable string for the filter template.                                                                 |        |

{ddeutil_workflow-0.0.18 → ddeutil_workflow-0.0.20}/pyproject.toml

@@ -28,9 +28,9 @@ requires-python = ">=3.9.13"
 dependencies = [
     "ddeutil>=0.4.3",
     "ddeutil-io[yaml,toml]>=0.2.3",
-    "pydantic==2.9.2",
+    "pydantic==2.10.2",
     "python-dotenv==1.0.1",
-    "typer==0.12.5,<1.0.0",
+    "typer==0.15.1",
     "schedule==1.2.2,<2.0.0",
 ]
 dynamic = ["version"]

ddeutil_workflow-0.0.20/src/ddeutil/workflow/__about__.py

@@ -0,0 +1 @@
+__version__: str = "0.0.20"

{ddeutil_workflow-0.0.18 → ddeutil_workflow-0.0.20}/src/ddeutil/workflow/__cron.py

@@ -646,12 +646,27 @@ class CronJob:
 
         :param date: An initial date that want to mark as the start point.
         :param tz: A string timezone that want to change on runner.
+
         :rtype: CronRunner
         """
         return CronRunner(self, date, tz=tz)
 
 
 class CronJobYear(CronJob):
+    """The Cron Job Converter with Year extension object that generate datetime
+    dimension of cron job schedule format,
+
+        *  *  *  *  *  *  <command to execute>
+
+        (i)     minute (0 - 59)
+        (ii)    hour (0 - 23)
+        (iii)   day of the month (1 - 31)
+        (iv)    month (1 - 12)
+        (v)     day of the week (0 - 6) (Sunday to Saturday; 7 is also Sunday
+                on some systems)
+        (vi)    year (1990 - 2100)
+    """
+
     cron_length = 6
     cron_units = CRON_UNITS_YEAR
 
@@ -704,9 +719,17 @@ class CronRunner:
         else:
             self.date: datetime = datetime.now(tz=self.tz)
 
+        # NOTE: Add one second if the microsecond value more than 0.
+        if self.date.microsecond > 0:
+            self.date: datetime = self.date.replace(microsecond=0) + timedelta(
+                seconds=1
+            )
+
         # NOTE: Add one minute if the second value more than 0.
         if self.date.second > 0:
-            self.date: datetime = self.date + timedelta(minutes=1)
+            self.date: datetime = self.date.replace(second=0) + timedelta(
+                minutes=1
+            )
 
         self.__start_date: datetime = self.date
         self.cron: CronJob | CronJobYear = cron
@@ -752,7 +775,7 @@ class CronRunner:
                 not self.__shift_date(mode, reverse)
                 for mode in ("year", "month", "day", "hour", "minute")
             ):
-                return copy.deepcopy(self.date.replace(second=0, microsecond=0))
+                return copy.deepcopy(self.date)
 
         raise RecursionError("Unable to find execution time for schedule")
 
@@ -801,6 +824,10 @@ class CronRunner:
             # NOTE: Replace date that less than it mode to zero.
             self.date: datetime = replace_date(self.date, mode, reverse=reverse)
 
+            # NOTE: Replace second and microsecond values that change from
+            #   the replace_date func with reverse flag.
+            self.date: datetime = self.date.replace(second=0, microsecond=0)
+
             if current_value != getattr(self.date, switch[mode]):
                 return mode != "month"
 

{ddeutil_workflow-0.0.18 → ddeutil_workflow-0.0.20}/src/ddeutil/workflow/__init__.py

@@ -15,7 +15,10 @@ from .exceptions import (
     UtilException,
     WorkflowException,
 )
-from .job import Job, Strategy
+from .job import (
+    Job,
+    Strategy,
+)
 from .on import (
     On,
     YearOn,
@@ -24,8 +27,6 @@ from .on import (
 from .scheduler import (
     Schedule,
     ScheduleWorkflow,
-    Workflow,
-    WorkflowTaskData,
 )
 from .stage import (
     BashStage,
@@ -34,7 +35,7 @@ from .stage import (
     PyStage,
     Stage,
     TriggerStage,
-    handler_result,
+    extract_hook,
 )
 from .utils import (
     FILTERS,
@@ -70,3 +71,7 @@ from .utils import (
     str2template,
     tag,
 )
+from .workflow import (
+    Workflow,
+    WorkflowTaskData,
+)

{ddeutil_workflow-0.0.18 → ddeutil_workflow-0.0.20}/src/ddeutil/workflow/conf.py

@@ -33,6 +33,29 @@ load_dotenv()
 env = os.getenv
 
 
+@lru_cache
+def get_logger(name: str):
+    """Return logger object with an input module name.
+
+    :param name: A module name that want to log.
+    """
+    lg = logging.getLogger(name)
+    formatter = logging.Formatter(
+        fmt=(
+            "%(asctime)s.%(msecs)03d (%(name)-10s, %(process)-5d, "
+            "%(thread)-5d) [%(levelname)-7s] %(message)-120s "
+            "(%(filename)s:%(lineno)s)"
+        ),
+        datefmt="%Y-%m-%d %H:%M:%S",
+    )
+    stream = logging.StreamHandler()
+    stream.setFormatter(formatter)
+    lg.addHandler(stream)
+
+    lg.setLevel(logging.DEBUG if config.debug else logging.INFO)
+    return lg
+
+
 class Config:
     """Config object for keeping application configuration on current session
     without changing when if the application still running.
@@ -98,12 +121,14 @@ class Config:
         os.getenv("WORKFLOW_API_ENABLE_ROUTE_SCHEDULE", "true")
     )
 
-    def __init__(self):
+    def __init__(self) -> None:
+        # VALIDATE: the MAX_JOB_PARALLEL value should not less than 0.
         if self.max_job_parallel < 0:
             raise ValueError(
                 f"``MAX_JOB_PARALLEL`` should more than 0 but got "
                 f"{self.max_job_parallel}."
             )
+
         try:
             self.stop_boundary_delta: timedelta = timedelta(
                 **json.loads(self.stop_boundary_delta_str)
@@ -195,6 +220,7 @@ class SimLoad:
         :param conf: A config object.
         :param include:
         :param exclude:
+
         :rtype: Iterator[tuple[str, DictData]]
         """
         exclude: list[str] = exclude or []
@@ -247,12 +273,14 @@ class Loader(SimLoad):
         include: list[str] | None = None,
         exclude: list[str] | None = None,
         **kwargs,
-    ) -> DictData:
+    ) -> Iterator[tuple[str, DictData]]:
         """Override the find class method from the Simple Loader object.
 
         :param obj: A object that want to validate matching before return.
         :param include:
         :param exclude:
+
+        :rtype: Iterator[tuple[str, DictData]]
         """
         return super().finds(
             obj=obj, conf=Config(), include=include, exclude=exclude
@@ -268,6 +296,7 @@ def get_type(t: str, params: Config) -> AnyModelType:
     :param t: A importable type string.
     :param params: A config parameters that use registry to search this
         type.
+
     :rtype: AnyModelType
     """
     try:
@@ -283,29 +312,7 @@ def get_type(t: str, params: Config) -> AnyModelType:
 
 
 config = Config()
-
-
-@lru_cache
-def get_logger(name: str):
-    """Return logger object with an input module name.
-
-    :param name: A module name that want to log.
-    """
-    logger = logging.getLogger(name)
-    formatter = logging.Formatter(
-        fmt=(
-            "%(asctime)s.%(msecs)03d (%(name)-10s, %(process)-5d, "
-            "%(thread)-5d) [%(levelname)-7s] %(message)-120s "
-            "(%(filename)s:%(lineno)s)"
-        ),
-        datefmt="%Y-%m-%d %H:%M:%S",
-    )
-    stream = logging.StreamHandler()
-    stream.setFormatter(formatter)
-    logger.addHandler(stream)
-
-    logger.setLevel(logging.DEBUG if config.debug else logging.INFO)
-    return logger
+logger = get_logger("ddeutil.workflow")
 
 
 class BaseLog(BaseModel, ABC):
@@ -315,8 +322,8 @@ class BaseLog(BaseModel, ABC):
     """
 
     name: str = Field(description="A workflow name.")
-    on: str = Field(description="A cronjob string of this piepline schedule.")
     release: datetime = Field(description="A release datetime.")
+    type: str = Field(description="A running type before logging.")
     context: DictData = Field(
         default_factory=dict,
         description=(
@@ -366,6 +373,8 @@ class FileLog(BaseLog):
         workflow name.
 
         :param name: A workflow name that want to search release logging data.
+
+        :rtype: Iterator[Self]
         """
         pointer: Path = config.root_path / f"./logs/workflow={name}"
         if not pointer.exists():
@@ -387,6 +396,9 @@ class FileLog(BaseLog):
         workflow name and release values. If a release does not pass to an input
         argument, it will return the latest release from the current log path.
 
+        :param name:
+        :param release:
+
         :raise FileNotFoundError:
         :raise NotImplementedError:
 
@@ -411,21 +423,17 @@ class FileLog(BaseLog):
             return cls.model_validate(obj=json.load(f))
 
     @classmethod
-    def is_pointed(
-        cls,
-        name: str,
-        release: datetime,
-        *,
-        queue: list[datetime] | None = None,
-    ) -> bool:
-        """Check this log already point in the destination.
+    def is_pointed(cls, name: str, release: datetime) -> bool:
+        """Check the release log already pointed or created at the destination
+        log path.
 
         :param name: A workflow name.
         :param release: A release datetime.
-        :param queue: A list of queue of datetime that already run in the
-            future.
+
+        :rtype: bool
+        :return: Return False if the release log was not pointed or created.
         """
-        # NOTE: Check environ variable was set for real writing.
+        # NOTE: Return False if enable writing log flag does not set.
         if not config.enable_write_log:
             return False
 
@@ -434,9 +442,7 @@ class FileLog(BaseLog):
             name=name, release=release
         )
 
-        if not queue:
-            return pointer.exists()
-        return pointer.exists() or (release in queue)
+        return pointer.exists()
 
     def pointer(self) -> Path:
         """Return release directory path that was generated from model data.
@@ -459,6 +465,9 @@ class FileLog(BaseLog):
         if not config.enable_write_log:
             return self
 
+        logger.debug(
+            f"({self.run_id}) [LOG]: Start writing log: {self.name!r}."
+        )
         log_file: Path = self.pointer() / f"{self.run_id}.log"
         log_file.write_text(
             json.dumps(

{ddeutil_workflow-0.0.18 → ddeutil_workflow-0.0.20}/src/ddeutil/workflow/exceptions.py

@@ -3,6 +3,10 @@
 # Licensed under the MIT License. See LICENSE in the project root for
 # license information.
 # ------------------------------------------------------------------------------
+"""Exception objects for this package do not do anything because I want to
+create the lightweight workflow package. So, this module do just a exception
+annotate for handle error only.
+"""
 from __future__ import annotations