Awesome

Crontabs

NOTE: I've recently discovered the Rocketry and Huey projects, which you should probably use instead of crontabs. They are just better than crontabs.

Think of crontabs as a quick-and-dirty solution you can throw into one-off python scripts to execute tasks on a cron-like schedule.

Crontabs is a small pure-python library that was inspired by the excellent schedule library for python.

In addition to having a slightly different API, crontabs differs from the schedule module in the following ways.

• You do not need to provide your own event loop.
• Job timing is guaranteed not to drift over time. For example, if you specify to run a job every five minutes, you can rest assured that it will always run at 5, 10, 15, etc. passed the hour with no drift.
• The python functions are all run in child processes. A memory-friendly flag is available to run each iteration of your task in its own process thereby mitigating memory problems due to Python's high watermark issue

Why Crontabs

Python has no shortage of cron-like job scheduling libraries, so why create yet another. The honest answer is that I couldn't find one that met a simple list of criteria.

• Simple installation with no configuration. An extremely robust and scalable solution to this problem already exists. Celery. But for quick and dirty work, I didn't want the hastle of setting up and configuring a broker, which celery requires to do its magic. For simple jobs, I just wanted to pip install and go.
• Human readable interface. I loved the interface provided by the schedule library and wanted something similarly intuitive to use.
• Memory safe for long running jobs. Celery workers can suffer from severe memory bloat due to the way Python manages memory. As of 2017, the recommended solution for this was to periodically restart the workers. Crontabs runs each job in a subprocess. It can optionally also run each iteration of a task in it's own process thereby mitigating the memory bloat issue.
• Simple solution for cron-style workflow and nothing more. I was only interested in supporting cron-like functionality, and wasn't interested in all the other capabilities and guarantees offered by a real task-queue solution like celery.
• Suggestions for improvement welcome. If you encounter a bug or have an improvement that remains within the scope listed above, please feel free to open an issue (or even better... a PR).

Installation

pip install crontabs

Usage

Schedule a single job

from crontabs import Cron, Tab
from datetime import datetime


def my_job(*args, **kwargs):
    print('args={} kwargs={} running at {}'.format(args, kwargs, datetime.now()))


# Will run with a 5 second interval synced to the top of the minute
Cron().schedule(
    Tab(name='run_my_job').every(seconds=5).run(my_job, 'my_arg', my_kwarg='hello')
).go()

Schedule multiple jobs

from crontabs import Cron, Tab
from datetime import datetime


def my_job(*args, **kwargs):
    print('args={} kwargs={} running at {}'.format(args, kwargs, datetime.now()))


# All logging messages are sent to sdtout
Cron().schedule(
    # Turn off logging for job that runs every five seconds
    Tab(name='my_fast_job', verbose=False).every(seconds=5).run(my_job, 'fast', seconds=5),

    # Go ahead and let this job emit logging messages
    Tab(name='my_slow_job').every(seconds=20).run(my_job, 'slow', seconds=20),
).go()

Schedule future job to run repeatedly for a fixed amount of time

from crontabs import Cron, Tab
from datetime import datetime


def my_job(*args, **kwargs):
    print('args={} kwargs={} running at {}'.format(args, kwargs, datetime.now()))


Cron().schedule(
    Tab(
        name='future_job'
    ).every(
        seconds=5
    ).starting(
        '12/27/2017 16:45'  # This argument can either be parsable text or datetime object.
    ).run(
        my_job, 'fast', seconds=5
    )
# max_seconds starts from the moment go is called.  Pad for future run times accordingly.
).go(max_seconds=60)

Cron API

The Cron class has a very small api

Tab API with examples

The api for the Tab class is designed to be composable and readable in plain English. It supports the following "verbs" by invoking methods.

Run a job indefinitely

from crontabs import Cron, Tab
from datetime import datetime


def my_job(name):
    print('Running function with name={}'.format(name))


Cron().schedule(
    Tab(name='forever').every(seconds=5).run(my_job, 'my_func'),
).go()

Run one job indefinitely, another for thirty seconds, and another until 1/1/2030

from crontabs import Cron, Tab
from datetime import datetime


def my_job(name):
    print('Running function with name={}'.format(name))


Cron().schedule(
    Tab(name='forever').run(my_job, 'forever_job').every(seconds=5),
    Tab(name='for_thirty').run(my_job, 'mortal_job').every(seconds=5).lasting(seconds=30),
    Tab(name='real_long').run(my_job, 'long_job').every(seconds=5).until('1/1/2030'),
).go()

Run job every half hour from 9AM to 5PM excluding weekends

from crontabs import Cron, Tab
from datetime import datetime

def my_job(name):
    # Grab an instance of the crontab logger and write to it.
    logger = Cron.get_logger()
    logger.info('Running function with name={}'.format(name))


def business_hours(timestamp):
    return 9 <= timestamp.hour < 17

def weekends(timestamp):
    return timestamp.weekday() > 4


# Run a job every 30 minutes during weekdays.  Stop crontabs after it has been running for a year.
# This will indiscriminately kill every Tab it owns at that time.
Cron().schedule(
    Tab(
        name='my_job'
    ).run(
        my_job, 'my_job'
    ).every(
        minutes=30
    ).during(
        business_hours
    ).excluding(
        weekends
    )
).go(max_seconds=3600 * 24 * 365)

Run test suite with

git clone git@github.com:robdmc/crontabs.git
cd crontabs
pip install -e .[dev]
py.test -s -n 8   # Might need to change the -n amount to pass

Projects by robdmc.

• Pandashells Pandas at the bash command line
• Consecution Pipeline abstraction for Python
• Behold Helping debug large Python projects
• Crontabs Simple scheduling library for Python scripts
• Switchenv Manager for bash environments
• Gistfinder Fuzzy-search your gists