Some cleanup commands:

To add a space after each comma:
sed -e "s/\(\w\),\(\w\)/\1, \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\w\),\(\"\)/\1, \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\"\),\(\w\)/\1, \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\"\),\(\"\)/\1, \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\([)}\]]\),\(\w\)/\1, \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\([)}\]]\),\([\[{(]\)/\1, \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\w\),\([\[{(]\)/\1, \2/g" -i *.py printrun/*.py printrun/*/*.py

To add spaces around each =:
sed -e "s/\(\w\)=\(\w\)/\1 = \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\w\)=\(\"\)/\1 = \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\w\)=\((\)/\1 = \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\w\)=\((\)/\1 = \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\w\)=\([\[{(]\)/\1 = \2/g" -i *.py printrun/*.py printrun/*/*.py

To add spaces around each ==:
sed -e "s/\(\w\)==\(\w\)/\1 == \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\w\)==\(\"\)/\1 == \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\(\w\)==\((\)/\1 == \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\()\)==\(\w\)/\1 == \2/g" -i *.py printrun/*.py printrun/*/*.py
sed -e "s/\()\)==\((\)/\1 == \2/g" -i *.py printrun/*.py printrun/*/*.py

Obviously this is not a perfect solution, it WILL break the code. Juste check the diff and fix what's wrong before committing.

Flake8 checking:
Flake8 can be used to check the coding style of the project.
The current source code (as of July 23rd 2013) has been checked using the following command:
    flake8 . --statistics --count --ignore=E251,E701,E302,E501 --exclude=.svn,CVS,.bzr,.hg,.git,__pycache__,./printrun/cairosvg
This call ignores 4 kind of errors (E501: line being greater than 80 chars,
E701: multiple statements on one line (usually this is if ...: ...), E302:
wrong number of blank lines between functions, E251: unexpected spaces around
keywoard/parameter equals), the rest of the errors and warnings should be
killed as much as possible. Long lines should be avoided too.