コード例 #1
0
ファイル: looker.py プロジェクト: vipadm/airflow
    def execute(self, context: "Context") -> str:

        self.hook = LookerHook(looker_conn_id=self.looker_conn_id)

        resp = self.hook.start_pdt_build(
            model=self.model,
            view=self.view,
            query_params=self.query_params,
        )

        self.materialization_id = resp.materialization_id

        if not self.materialization_id:
            raise AirflowException(
                f'No `materialization_id` was returned for model: {self.model}, view: {self.view}.'
            )

        self.log.info(
            "PDT materialization job submitted successfully. Job id: %s.",
            self.materialization_id)

        if not self.asynchronous:
            self.hook.wait_for_job(
                materialization_id=self.materialization_id,
                wait_time=self.wait_time,
                timeout=self.wait_timeout,
            )

        return self.materialization_id
コード例 #2
0
ファイル: looker.py プロジェクト: vipadm/airflow
    def poke(self, context: "Context") -> bool:

        self.hook = LookerHook(looker_conn_id=self.looker_conn_id)

        if not self.materialization_id:
            raise AirflowException('Invalid `materialization_id`.')

        # materialization_id is templated var pulling output from start task
        status_dict = self.hook.pdt_build_status(materialization_id=self.materialization_id)
        status = status_dict['status']

        if status == JobStatus.ERROR.value:
            msg = status_dict["message"]
            raise AirflowException(
                f'PDT materialization job failed. Job id: {self.materialization_id}. Message:\n"{msg}"'
            )
        elif status == JobStatus.CANCELLED.value:
            raise AirflowException(
                f'PDT materialization job was cancelled. Job id: {self.materialization_id}.'
            )
        elif status == JobStatus.UNKNOWN.value:
            raise AirflowException(
                f'PDT materialization job has unknown status. Job id: {self.materialization_id}.'
            )
        elif status == JobStatus.DONE.value:
            self.log.debug(
                "PDT materialization job completed successfully. Job id: %s.", self.materialization_id
            )
            return True

        self.log.info("Waiting for PDT materialization job to complete. Job id: %s.", self.materialization_id)
        return False
コード例 #3
0
ファイル: looker.py プロジェクト: wkodate/airflow
class LookerCheckPdtBuildSensor(BaseSensorOperator):
    """
    Check for the state of a previously submitted PDT materialization job.

    :param materialization_id: Required. The materialization job ID to poll. (templated)
    :param looker_conn_id: Required. The connection ID to use connecting to Looker.
    :param cancel_on_kill: Optional. Flag which indicates whether cancel the hook's job or not,
        when on_kill is called.
    """

    template_fields = ["materialization_id"]

    def __init__(self,
                 materialization_id: str,
                 looker_conn_id: str,
                 cancel_on_kill: bool = True,
                 **kwargs) -> None:
        super().__init__(**kwargs)
        self.materialization_id = materialization_id
        self.looker_conn_id = looker_conn_id
        self.cancel_on_kill = cancel_on_kill
        self.hook: Optional[LookerHook] = None

    def poke(self, context: "Context") -> bool:

        self.hook = LookerHook(looker_conn_id=self.looker_conn_id)

        # materialization_id is templated var pulling output from start task
        status_dict = self.hook.pdt_build_status(
            materialization_id=self.materialization_id)
        status = status_dict['status']

        if status == JobStatus.ERROR.value:
            msg = status_dict["message"]
            raise AirflowException(
                f'PDT materialization job failed. Job id: {self.materialization_id}. Message:\n"{msg}"'
            )
        elif status == JobStatus.CANCELLED.value:
            raise AirflowException(
                f'PDT materialization job was cancelled. Job id: {self.materialization_id}.'
            )
        elif status == JobStatus.UNKNOWN.value:
            raise AirflowException(
                f'PDT materialization job has unknown status. Job id: {self.materialization_id}.'
            )
        elif status == JobStatus.DONE.value:
            self.log.debug(
                "PDT materialization job completed successfully. Job id: %s.",
                self.materialization_id)
            return True

        self.log.info(
            "Waiting for PDT materialization job to complete. Job id: %s.",
            self.materialization_id)
        return False

    def on_kill(self):
        if self.materialization_id and self.cancel_on_kill:
            self.hook.stop_pdt_build(
                materialization_id=self.materialization_id)
コード例 #4
0
ファイル: looker.py プロジェクト: vipadm/airflow
class LookerStartPdtBuildOperator(BaseOperator):
    """
    Submits a PDT materialization job to Looker.

    :param looker_conn_id: Required. The connection ID to use connecting to Looker.
    :param model: Required. The model of the PDT to start building.
    :param view: Required. The view of the PDT to start building.
    :param query_params: Optional. Additional materialization parameters.
    :param asynchronous: Optional. Flag indicating whether to wait for the job
        to finish or return immediately.
        This is useful for submitting long running jobs and
        waiting on them asynchronously using the LookerCheckPdtBuildSensor
    :param cancel_on_kill: Optional. Flag which indicates whether cancel the
        hook's job or not, when on_kill is called.
    :param wait_time: Optional. Number of seconds between checks for job to be
        ready. Used only if ``asynchronous`` is False.
    :param wait_timeout: Optional. How many seconds wait for job to be ready.
        Used only if ``asynchronous`` is False.
    """
    def __init__(
        self,
        looker_conn_id: str,
        model: str,
        view: str,
        query_params: Optional[Dict] = None,
        asynchronous: bool = False,
        cancel_on_kill: bool = True,
        wait_time: int = 10,
        wait_timeout: Optional[int] = None,
        **kwargs,
    ) -> None:
        super().__init__(**kwargs)
        self.model = model
        self.view = view
        self.query_params = query_params
        self.looker_conn_id = looker_conn_id
        self.asynchronous = asynchronous
        self.cancel_on_kill = cancel_on_kill
        self.wait_time = wait_time
        self.wait_timeout = wait_timeout
        self.hook: Optional[LookerHook] = None
        self.materialization_id: Optional[str] = None

    def execute(self, context: "Context") -> str:

        self.hook = LookerHook(looker_conn_id=self.looker_conn_id)

        resp = self.hook.start_pdt_build(
            model=self.model,
            view=self.view,
            query_params=self.query_params,
        )

        self.materialization_id = resp.materialization_id

        if not self.materialization_id:
            raise AirflowException(
                f'No `materialization_id` was returned for model: {self.model}, view: {self.view}.'
            )

        self.log.info(
            "PDT materialization job submitted successfully. Job id: %s.",
            self.materialization_id)

        if not self.asynchronous:
            self.hook.wait_for_job(
                materialization_id=self.materialization_id,
                wait_time=self.wait_time,
                timeout=self.wait_timeout,
            )

        return self.materialization_id

    def on_kill(self):
        if self.materialization_id and self.cancel_on_kill:
            self.hook.stop_pdt_build(
                materialization_id=self.materialization_id)