Simon Richter <[EMAIL PROTECTED]> writes:
> I'd like to write an automatic dependency generator for Python scripts (so
> that you can write "Depends: {pythonlibs:Depends}" into your control file
> and have it replaced by the packages you need to satisfy all imports by
> the Python scripts in debian/<package>. For this, you'd need a database of
> which Module is in which package.
[...]
It seems to me that what you're proposing is much more complicated
than is actually needed.
It's fairly straightforward to determine all the modules directly
imported by providing a custom __import__ in the __builtins__ passed
to execfile(). To get the packages containing these modules, you can
just replace any trailing .pyc by .py on each module's __file__
attribute, and call 'dpkg -S' with the resulting list.
For example, if a script imports just sys and Tkinter, only Tkinter
has a filename for the module, /usr/lib/python2.0/lib-tk/Tkinter.pyc.
Passing this to dpkg gives:
% dpkg -S /usr/lib/python2.0/lib-tk/Tkinter.py
python2-tk
This does require that all the packages required by the script be
installed, which I think is a reasonable assumption. This doesn't
help with build dependencies, though.
--
Carey Evans http://home.clear.net.nz/pages/c.evans/
"Quiet, you'll miss the humorous conclusion."
