processor 0.5.0

A microframework to build source -> filter -> action workflows.

Badges

Simple rules

Python processor is a tool for creating chained pipelines for dataprocessing. It have very few key concepts:

Data object

Any python dict with two required fields: source and type.

Source

An iterable sequence of data objects or a function which returns data objects. See full list of sources in the docs.

Output

A function which accepts a data object as input and could output another. See full list of outputs in the docs. (or same) data object as result.

Predicate

Pipeline consists from sources outputs, but predicate decides which data object should be processed by which output.

Quick example

Here is example of pipeline which reads IMAP folder and sends all emails to Slack chat:

run_pipeline(
    sources=[sources.imap('imap.gmail.com'
                          'username',
                          'password'
                          'INBOX')],
    rules=[(for_any_message, [email_to_slack, outputs.slack(SLACK_URL)])])

Here you construct a pipeline, which uses sources.imap for reading imap folder “INBOX” of [e-mail]. Function for_any_message is a predicate saying something like that: lambda data_object: True. In more complex case predicates could be used for routing dataobjects to different processors.

Functions email_to_slack and outputs.slack(SLACK_URL) are processors. First one is a simple function which accepts data object, returned by imap source and transforming it to the data object which could be used by slack.output. We need that because slack requires a different set of fields. Call to outputs.slack(SLACK_URL) returns a function which gets an object and send it to the specified Slack’s endpoint.

It is just example, for working snippets, continue reading this documention ;-)

Note

By the way, did you know there is a Lisp dialect which runs on Python virtual machine? It’s name is HyLang, and python processor is written in this language.

Installation

Create a virtual environment with python3::

virtualenv --python=python3 env
source env/bin/activate

If you are on OSX, then install lxml on OSX separately::

STATIC_DEPS=true pip install lxml

Then install the processor::

pip install processor

Usage

Now create an executable python script, where you’ll place your pipline’s configuration. For example, this simple code creates a process line which searches new results in Twitter and outputs them to console. Of cause, you can output them not only to console, but also post by email, to Slack chat or everywhere else if there is an output for it:

#!env/bin/python3
import os
from processor import run_pipeline, sources, outputs
from twiggy_goodies.setup import setup_logging


for_any_message = lambda msg: True

def prepare(tweet):
    return {'text': tweet['text'],
            'from': tweet['user']['screen_name']}

setup_logging('twitter.log')

run_pipeline(
    sources=[sources.twitter.search(
        'My Company',
        consumer_key='***', consumer_secret='***',
        access_token='***', access_secret='***',
        )],
    rules=[(for_any_message, [prepare, outputs.debug()])])

Running this code, will fetch new results for search by query My Company and output them on the screen. Of course, you could use any other output, supported by the processor. Browse online documentation to find out which sources and outputs are supported and for to configure them.

Ideas for Sources and Outputs

• web-hook endpoint (in progress).

• tail source which reads file and outputs lines appeared in a file between invocations or is able to emulate tail -f behaviour. Python module tailer could be used here.

• grep output – a filter to grep some fields using patterns. With tail and grep you could build a pipeline which watch on a log and send errors by email or to the chat.

• xmpp output.

• irc output.

• rss/atom feed reader.

• weather source which tracks tomorrow’s weather forecast and outputs a message if it was changed significantly, for example from “sunny” to “rainy”.

• github some integrations with github API?

• jira or other task tracker of your choice?

• suggest your ideas!

Documentation

https://python-processor.readthedocs.org/

Development

To run the all tests run:

tox

Authors

• Alexander Artemenko - http://dev.svetlyak.ru

Changelog

0.5.0 (2015-04-15)

Good news, everyone! New output was added - email. Now Processor is able to notify you via email about any event.

0.4.0 (2015-04-06)

• Function run_pipline was simplified and now accepts only one source and one ouput. To implement more complex pipelines, use sources.mix and outputs.fanout helpers.

0.3.0 (2015-04-01)

• Added a web.hook source.

• Now source could be not only a iterable object, but any function which returns values.

0.2.1 (2015-03-30)

Fixed error in import-or-error macro, which prevented from using 3-party libraries.

0.2.0 (2015-03-30)

Most 3-party libraries are optional now. If you want to use some extension which requires external library, it will issue an error and call sys.exit(1) until you satisfy this requirement.

This should make life easier for thouse, who does not want to use rss output which requires feedgen which requires lxml which is hard to build because it is C extension.

0.1.0 (2015-03-18)

• First release on PyPI.