This is (or will be) a consolidated repository of all the stupid Python tricks I have written (and can still find).
A stupid python trick is usually an experiment with an advanced language feature; ostensibly to learn how it works, but usually in order to abuse that feature to write something truly horrible.
I used to be a Perl programmer.
I am also no longer the only person responsible for all of this.
This one got out of hand. It is an attempt to write Fizz Buzz, using as many language features as possible. It uses a bunch of functional techniques, a strategy pattern implementation, and dynamically named closures (I was especially proud of that).
It is PEP-8 compliant.
A friend of mine used to complain that he couldn't run a regex as part of an
if statement's expression if he wanted to use the returned groups (you can
in Perl using implicit variables).
I worked out how to modify the caller's scope to dynamically add a variable containing the result of the pattern match. Mission accomplished.
Ish is a stupid library that allows you to test if a variable is tru-ish or
false-ish.
It exports a single variable called ish, which can be subtracted from True
or False to create a TrueIsh or FalseIsh object. This can be compared with
the value to see if it either:
- Evaluates to
True - Is a string containing the English, French, German, Danish, Dutch,
Afrikaans, Swedish, Norwegian, Portuguese, Irish, Esperanto or Arabic for
either
YesorNo(or a handful of slang words with the same meaning).
If it is a string containing none of these known words, it raises
a ValueError
if 'Yup' == True-ish:
print 'True-ish!'I wanted to see if I could create an emoticon that was valid Python syntax.
It turns out I could! OOO[:-P]
Subclass str, add a bunch of interesting properties and methods, and voila,
easy path management, if a little bit too much voodoo.
Note: I'm not 100% sure I wrote this. Some of it doesn't seem like my style. It's possible I came up with inspiration from looking at someone else's code.
Contains an 'everlasting cache' (which is not very useful), and a (crazy) decorator-based dependency calling mechanism. Warning, may cause seizures!
Only one trick! A metaclass that automatically replaces get_ and set_
methods with properties that call the methods. Too much magic, but a neat trick,
I think.
A deep neural network based facial expression classifier is available. Using it requires Theano (Theano on Github), Lasagne, and OpenCV 3 with Python bindings.
To run a live demo, use IPython Notebook to run
Live Deep Neural Network Demo.ipynb. If you want to train a new model, run
Train deep neural network.ipynb.
To install OpenCV 3 on Windows, download the installer, find the .pyd extension
and put it into your Python interpreter's Lib/site-packages path.
To install it on MacOS X, use brew;
installation instructions for OpenCV 2
(also worked for me for OpenCV v3). You must make OpenCV available to Python in the
same way as windows (put library in lib/site-packages/