Emanuele Rocca

Useful tools for Python developers

Python is a great language with an impressive number of tools designed to make developers' life easier. Sometimes, however, the problem is getting to know that these tools exist in the first place. By contributing to projects like OpenStack's Nova client and Falcon, I have recently come across some useful tools that can seriously improve the quality of your code.

The first one is called pyflakes, a passive checker of Python programs developed by Phil Frost. What it does is parsing your source files and checking for possible errors such as undefined names and unused imports. Let's consider the following example:

1
2
3
4

import urllib

print "pyflakes example"
urlib.urlopen('http://www.linux.it')

The code above contains a typo, we have misspelled urllib. Here is what pyflakes thinks about our program:

$ pyflakes example.py
example.py:1: 'urllib' imported but unused
example.py:4: undefined name 'urlib'

On line 4 we try to use urlib which is not defined. Also, we import urllib on line 1 and we do nothing with it. Our typo has been spotted! Notice that, even though our program contains a print statement, 'pyflakes example' has not been printed. That is because pyflakes parses the source files it checks, without importing them, making it safe to use on modules with side effects.

pyflakes can be installed with pip or apt-get.

The second tool I want to talk about is Ned Batchelder's coverage.py.

No doubt you write unit tests for your programs. Right? Good. coverage.py is out there to help you checking how much of your program is actually covered.

Let's use as an example codicefiscale, a Python project of mine.

First we install coverage:

pip install coverage

Then we run our unit tests:

$ coverage run --source=codicefiscale tests.py
.......
----------------------------------------------------------------------
Ran 7 tests in 0.003s

We pass the module we want to test with --source=codicefiscale so that coverage will only report information about that specific module.

Now that our tests have been performed successfully it is time to check how much of our code is covered by unit tests:

$ coverage report -m
Name            Stmts   Miss  Cover   Missing
---------------------------------------------
codicefiscale      73      4    95%   61, 67, 95, 100

Not bad, 95% of our module is covered! Still, coverage let us know that 4 lines have not been touched by the unit tests. With this information, we can go write some meaningful test cases that will also cover the missing lines.