Tomorrow

Tomorrow

Magic decorator syntax for asynchronous code in Python 2.7.

Please don't actually use this in production. It's more of a thought experiment than anything else, and relies heavily on behavior specific to Python's old style classes. Pull requests, issues, comments and suggestions welcome.

Installation

Tomorrow is conveniently available via pip:

pip install tomorrow

or installable via git clone and setup.py

git clone [email protected]:madisonmay/Tomorrow.git
sudo python setup.py install

To ensure Tomorrow is properly installed, you can run the unittest suite from the project root:

nosetests -v 

Usage

The tomorrow library enables you to utilize the benefits of multi-threading with minimal concern about the implementation details.

Behind the scenes, the library is a thin wrapper around the Future object in concurrent.futures that resolves the Future whenever you try to access any of its attributes.

Enough of the implementation details, let's take a look at how simple it is to speed up an inefficient chunk of blocking code with minimal effort.

Naive Web Scraper

You've collected a list of urls and are looking to download the HTML of the lot. The following is a perfectly reasonable first stab at solving the task.

For the following examples, we'll be using the top sites from the Alexa rankings.

urls = [
    'http://google.com',
    'http://facebook.com',
    'http://youtube.com',
    'http://baidu.com',
    'http://yahoo.com',
]

Right then, let's get on to the code.

import time
import requests

def download(url):
    return requests.get(url)

if __name__ == "__main__":

    start = time.time()
    responses = [download(url) for url in urls]
    html = [response.text for response in responses]
    end = time.time()
    print "Time: %f seconds" % (end - start)

More Efficient Web Scraper

Using tomorrow's decorator syntax, we can define a function that executes in multiple threads. Individual calls to download are non-blocking, but we can largely ignore this fact and write code identically to how we would in a synchronous paradigm.

import time
import requests

from tomorrow import threads

@threads(5)
def download(url):
    return requests.get(url)

if __name__ == "__main__":
    start = time.time()
    responses = [download(url) for url in urls]
    html = [response.text for response in responses]
    end = time.time()
    print "Time: %f seconds" % (end - start)

Awesome! With a single line of additional code (and no explicit threading logic) we can now download websites ~10x as efficiently.

You can also optionally pass in a timeout argument, to prevent hanging on a task that is not guaranteed to return.

import time

from tomorrow import threads

@threads(1, timeout=0.1)
def raises_timeout_error():
    time.sleep(1)

if __name__ == "__main__":
    print raises_timeout_error()

How Does it Work?

Feel free to read the source for a peek behind the scenes -- it's less than 50 lines of code.