https://github.com/1stvamp/trequests

A Tornado async HTTP/HTTPS client adaptor for python-requests
https://github.com/1stvamp/trequests

Last synced: about 1 month ago
JSON representation

A Tornado async HTTP/HTTPS client adaptor for python-requests

• Host: GitHub
• URL: https://github.com/1stvamp/trequests
• Owner: 1stvamp
• License: bsd-3-clause
• Created: 2013-09-24T19:40:25.000Z (almost 11 years ago)
• Default Branch: master
• Last Pushed: 2019-12-16T13:57:09.000Z (almost 5 years ago)
• Last Synced: 2024-04-24T14:08:57.199Z (5 months ago)
• Language: Python
• Homepage:
• Size: 20.5 KB
• Stars: 129
• Watchers: 10
• Forks: 12
• Open Issues: 3
• Metadata Files:
  • Readme: README.rst
  • Changelog: CHANGELOG
  • License: LICENSE

Awesome Lists containing this project

• starred-awesome - trequests - A Tornado async HTTP/HTTPS client adaptor for python-requests (Python)

README

        
trequests

=========

.. image:: https://travis-ci.org/1stvamp/trequests.png?branch=master

A Tornado async HTTP/HTTPS client adapter for python-requests.

The problem

-----------

You enjoy using `Tornado `_ to build fast non-blocking web applications, and you want to use a library from PyPI that makes a few HTTP requests, but pretty much every dev and their dog uses `Requests `_ to make HTTP requests (rightly so, because it's *awesome*), but requests has no knowledge of the event loop nor can it yield when a socket blocks, which means any time you try to use a library like that it begins to block your request handling and grud-knows what other worlds of pain.

The solution

------------

Luckily there are solutions, one such is to use the `greenlet `_ module to wrap blocking operations and swap Tornado coroutines at the right time, there is even the handy `tornalet `_ module which handles this for you.

To make life even easier, you lucky lucky people, I've created ``trequests``, an async Requests adapter which uses greenlets (via tornalet) and the inbuilt non-blocking HTTP client methos in Tornado, to make any call to a library (utilizing Requests) non-blocking.

Installation

------------

.. code-block:: bash

    

    $ pip install trequests

  

Usage

-----

  

.. code-block:: python

    

    # Assume bobs_big_data uses python-requests for HTTP requests

    import bobs_big_data

    

    from tornado.web import RequestHandler

    from trequests import setup_session

    from tornalet import tornalet

    

    # Tell requests to use our AsyncHTTPadapter for the default

    # session instance, you can also pass you own through

    setup_session()

    

    class WebHandler(RequestHandler):

        @tornalet

        def get(self):

            data = {'foo': 'bar'}

            # This will now unblock the current coroutine, like magic

            response = bobs_big_data.BigData(data).post()

            return self.write(response)

Tests

-----

To run the basic testsuite hit up `python setup.py test`.

Caveats

-------

``trequests`` has been used in production in a large scale metrics application, and is a very small and quite simple module.

**However** I've released it as ``0.9.x`` mainly because it's missing 100% compatibility with the Requests adapter API, most noticeably *cookie jar* and *session* support, which I will improve (or please send me a pull request if you fancy adding support), and release as a ``1.x`` branch when I have the time.

Also at the moment the ``setup_session`` utility actually monkey patches the ``session`` utility functions in Requests, as this was the only way I could see to override the mounts on "default" session instances (e.g. those created for every call when a session isn't provided). I'm hoping to change this in the future.