This is part two of our look at using Python with the API of the Twitter social networking website. (Part one here)

I am assuming you are using a GNOME-based Unix/Linux system. If you use something else then sorry, you might want to read one of my other posts.

So in Twitter, you can have updates of what your friends are doing appear on the Twitter website (like this), however, it is more fun to make these updates come to you.

So I wrote a little program in Python that makes your Twitter updates pop-up on the desktop, like this:

This uses a feature of the Gnome desktop called 'notify', as explained by Andy W, and then followed up by me two posts ago.

Without further ado, lets get started.

1. Get python-notify

There are three very small but important dependencies. Firstly, you need the Python bindings for libnotify. These are available from your Linux distribution (you might even have them installed them already):

You will need setuptools as well. I am sure you already have that installed, but it cannot hurt to double-check.

The two computers I have here are Ubuntu and Gentoo:

On Ubuntu or Debian:

sudo apt-get install python-notify python-setuptools

On Gentoo:

sudo emerge notify-python setuptools

On your Linux/Unix, you can alter the commands slightly so it matches your package management system.

2. Get json and Twitter bindings

Now we need two more, simplejson and python-twitter

From any Linux Distribution, run:

sudo easy_install simplejson

sudo easy_install python-twitter

3. Twitter account

Next you will need to sign up to a Twitter account and add at least one friend (e.g. zeth0), in the spooky terminology of twitter, to add a friend you "follow them".

4. Get my script

Next you need to download a copy of my script and make the file extension .py. Get it from here, or type:

wget http://zeth.me.uk/python/twitnotify.txt -O twitnotify.py

chmod +x twitnotify.py

Now you need to edit the top of the script so that your username and password are correct. For historical reasons, I'm an Emacs man, but if you don't have a favourite editor, then gedit is pretty easy to use:

gedit twitnotify.py

5. Engage!

Now we are finally ready to run the script:

./twitnotify.py

If you have a lot of friends then there will be lots of updates the first time. The updates and pictures are all cached, so you don't have to download those again.

Six degrees of separation

After I made that script, I tried to be more adventurous and wrote a new script that found other people that shared your interests.

It would spider recursively through your friends, followers and their friends and followers, and their friends and followers, and their friends and followers (and so on), and returned a set of updates based on whatever keywords you specified.

It worked pretty well and the results were really fun. However, what I didn't know was that Twitter has quite sharp rate limits, so when this script stopped working I was combing through the code to find out what had gone wrong. It turned out that my code was fine but that my username had just been suspended from Twitter.

Twitter rate usage

Your username is allowed 70 requests per 60 sixty minute time period, starting from the first request.

Using my original twitnotify.py script posed no problem, the script usually uses three requests each time (one for normal updates, one for replies and one for personal "direct" messages).

So if you update every 10 minutes, then that will be about 18 requests an hour, easily under your 70 maximum.

Getting the photos of your friends is included in the above, but the first time you get a private message from a new person who is not a friend, it will take one request to download the photo.

My second script, the one to find new friends according to your interests, easily ate up 70 requests in one pass. So I had to give that up.

Fortunately, it seems that you are automatically un-suspended from Twitter after a couple of hours, and having your username temporally suspended from the API does not affect logging into the Twitter website.

Discuss this post - leave a comment