Dealing with Paths
++++++++++++++++++

:Posted: 2009-04-25 14:07
:Tags: Python

One issue I've come across is how to deal with paths. My conclusion is that all
input from the command line which takes the form of a filename or a directory
should immediately be converted into a normalised, absolute path like this:

::

    import os.path
    os.path.normpath(os.path.abspath(path))

This makes paths easier to work with in the application because the same
physical path will always be represented by the same string. I've written a
function called ``uniform_path()`` which does this. Variables are named
according to this convention:

* Normalised paths are named ``path``, (or ``dirpath`` or ``filepath`` if you want to be more specific)
* Directories and filenames without the path component are named ``filename`` or ``dirname``
* Relative paths are named ``rpath``

Relative Paths
--------------

Python 2.6 has an implementation for calculating relative paths called
``os.path.relpath()``. I use the following code to make a ``relpath()``
function available to the module globally, regardless of whether the user is
using Python 2.6 or an earlier version. This version only works on Posix platforms, because it uses ``posixpath``.

::

    import posixpath
    # relpath import (available in Python 2.6 and above)
    try:
        relpath = posixpath.relpath
    except AttributeError:
    
        from posixpath import curdir, sep, pardir, join
    
        def relpath(path, start=curdir):
            """Return a relative version of a path"""
    
            if not path:
                raise ValueError("no path specified")
    
            start_list = posixpath.abspath(start).split(sep)
            path_list = posixpath.abspath(path).split(sep)
    
            # Work out how much of the filepath is shared by start and path.
            i = len(posixpath.commonprefix([start_list, path_list]))
    
            rel_list = [pardir] * (len(start_list)-i) + path_list[i:]
            if not rel_list:
                return curdir
            return join(*rel_list)

Incidentally, ``os.path`` is just an alias which defaults to the correct module
for your platform. On Linux it uses the ``posixpath`` module from the standard
library.