Coverage for airflow.operators.python_operator : 50%

from builtins import str 

from datetime import datetime 

import logging 

from airflow.models import BaseOperator, TaskInstance 

from airflow.utils import apply_defaults, State 

from airflow import settings 

class PythonOperator(BaseOperator): 

    """ 

    Executes a Python callable 

    :param python_callable: A reference to an object that is callable 

    :type python_callable: python callable 

    :param op_kwargs: a dictionary of keyword arguments that will get unpacked 

        in your function 

    :type op_kwargs: dict 

    :param op_args: a list of positional arguments that will get unpacked when 

        calling your callable 

    :type op_args: list 

    :param provide_context: if set to true, Airflow will pass a set of 

        keyword arguments that can be used in your function. This set of 

        kwargs correspond exactly to what you can use in your jinja 

        templates. For this to work, you need to define `**kwargs` in your 

        function header. 

    :type provide_context: bool 

    :param templates_dict: a dictionary where the values are templates that 

        will get templated by the Airflow engine sometime between 

        ``__init__`` and ``execute`` takes place and are made available 

        in your callable's context after the template has been applied 

    :type templates_dict: dict of str 

    :param templates_exts: a list of file extensions to resolve while 

        processing templated fields, for examples ``['.sql', '.hql']`` 

    """ 

    template_fields = ('templates_dict',) 

    template_ext = tuple() 

    ui_color = '#ffefeb' 

    @apply_defaults 

    def __init__( 

            self, 

            python_callable, 

            op_args=None, 

            op_kwargs=None, 

            provide_context=False, 

            templates_dict=None, 

            templates_exts=None, 

            *args, **kwargs): 

        super(PythonOperator, self).__init__(*args, **kwargs) 

        self.python_callable = python_callable 

        self.op_args = op_args or [] 

        self.op_kwargs = op_kwargs or {} 

        self.provide_context = provide_context 

        self.templates_dict = templates_dict 

        if templates_exts: 

            self.template_ext = templates_exts 

    def execute(self, context): 

        if self.provide_context: 

            context.update(self.op_kwargs) 

            context['templates_dict'] = self.templates_dict 

            self.op_kwargs = context 

        return_value = self.python_callable(*self.op_args, **self.op_kwargs) 

        logging.info("Done. Returned value was: " + str(return_value)) 

        return return_value 

class BranchPythonOperator(PythonOperator): 

    """ 

    Allows a workflow to "branch" or follow a single path following the 

    execution of this task. 

    It derives the PythonOperator and expects a Python function that returns 

    the task_id to follow. The task_id returned should point to a task 

    directely downstream from {self}. All other "branches" or 

    directly downstream tasks are marked wit a state of "skipped" so that 

    these paths can't move forward. 

    """ 

    def execute(self, context): 

        branch = super(BranchPythonOperator, self).execute(context) 

        logging.info("Following branch " + branch) 

        logging.info("Marking other directly downstream tasks as failed") 

        session = settings.Session() 

        for task in context['task'].downstream_list: 

            if task.task_id != branch: 

                ti = TaskInstance( 

                    task, execution_date=context['ti'].execution_date) 

                ti.state = State.SKIPPED 

                ti.start_date = datetime.now() 

                ti.end_date = datetime.now() 

                session.merge(ti) 

        session.commit() 

        session.close() 

        logging.info("Done.")