Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
pythonic filesystem library
branch: master

This branch is even with stedolan:master

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.

pathfinder - pythonic filesystem library

Python's standard library includes many, many modules for dealing with files. Ironically, none of these are particularly "pythonic".

There's os, which among other things can create, delete and move files. To copy them, you'll need shutil's copy or copy2 (slightly different, of course). To list files, you might use the os or the glob modules. You can get file metadata with os.stat, but you'll have to use an entirely different module to interpret its results.

pathfinder uses these standard modules to make a library that can be used without constant reference to the documentation.

It's currently alpha software. I'm using the "one giant file" method of distribution, so you can install it by putting somewhere handy. There'll be a proper pip package soon!

It's all slightly unix-flavoured at the moment. It sits on top of the cross-platform primitives provided by Python, so it should more-or-less work on Windows, but this hasn't been tested.

pathfinder is licensed under the MIT license.


A ''path'' represents a file, which may or may not exist.

>>> p = path('/home/stephen/coding/pathfinder/')

You can do lots of things with path objects

>>> if p.exists:
...     print p.owner, p.last_modify_time
stephen 2012-02-12 15:40:20.328579

If you have a path object for a directory, you can use the / operator to look at things inside the directory.

>>> p = path('/home/stephen/coding')
>>> print p / 'pathfinder'
<Directory "/home/stephen/coding/pathfinder">
>>> print p / 'pathfinder' / ''
<File "/home/stephen/coding/pathfinder/">

'/' interprets the second argument relative to the first, so

>>> path('/home/stephen') / 'coding/pathfinder'
<Directory "/home/stephen/coding/pathfinder">
>>> path('/home/stephen') / '/etc/passwd'
<File "/etc/passwd">

'%' is its inverse, and gives you the first path as a relative path from the second.

>>> path('/home/stephen/coding/pathfinder') % '/home/stephen'
<Directory 'coding/pathfinder'>
>>> path('/home/stephen/coding/pathfinder') % '/etc'
<Directory '../home/stephen/coding/pathfinder'>

You can list things in a directory by just iterating over the object.

>>> for f in p:
...     print f.basename

or by using the more advanced ''.find'' method (see below)

>>> for f in p.find("Pathfinder*", 
...                 exclude=["*~", "*.pyc", ".git"], 
...                 ignore_case=True):

Manipulating the filesystem

You can inspect a path with path.exists, path.is_directory, path.is_file, path.is_symlink, path.size and various others.

path.last_modify_time returns an actual Python datetime object. For symlinks, path.link_target shows where they point, and path.final_link_target follows as many steps are necessary to get to the end of a chain of symlinks.

>>> if path('README.markdown').newer_than(path('')):
...     print "whatever the hell you just did, document it!"

You can modify the filesystem with path.mkdir(), path.symlink(), path.chown() and friends. path.chown can be used to change the user or group or both, and takes either a numeric ID or a name.

>>> p.chown(user = "stephen", group = "stephen")

Unix permissions

''.owner'' and ''.group'' give the owner and group of a file as strings (use ''.owner_uid'') and (''.group_uid'') for numeric IDs.

>>> path('/etc/passwd').group

''.perms'' will give the unix permissions of the file at a particular path.

>>> path('/etc/passwd').perms
<0644 -rw-r--r-->

perms.user will restrict show only the user permissions, perms.write only the write permissions, and so on. So, we can find all world-executable setuid root programs in '/bin' with:

>>> [f for f in path('/bin') 
     if and f.perms.setuid and f.owner == 'root']
[<File "/bin/ping6">,
 <File "/bin/fusermount">,
 <File "/bin/su">,
 <File "/bin/mount">,
 <File "/bin/umount">,
 <File "/bin/ping">]

Reading and writing

pathfinder provides the absolute easiest method ever of reading and writing files.

>>> p = path('myfile')
>>> p.contents = "Look at me! I'm a file!"
>>> print p.contents
Look at me! I'm a file!

You can also go the traditional route and open the file:

>>> f = path('myfile').open("r")

What you get back is a ''pathfinder.filehandle'' object, which is just a little bit more awesome than a standard Python file. It implements the standard Python file methods (so you can pass it to things that expect a "file-like object").

You can seek around and read and write parts of a filehandle:

>>> f = path('myfile').open("w+")
>>> f.write("Hello world")
>>> print f[0:5]
>>> f[6:11] = "again"
>>> print f[:]
"Hello again"

Atomic update

There's a nice technique for modifying a file, where instead of directly writing the file you create a new file, write that one, then change its name to the target filename and replace the original. On Unix-y systems, this is guaranteed atomic: any other program will see either the entire old file or the entire new file, never a half-written file or a missing file.

On Windows, it's not atomic but still useful: no other process will see a half-written file, although they may observe a missing file in the instant during the file move.

Using pathfinder and Python's with statement, this is easily done:

>>> with mypath.atomic_update() as f:
...     f.write('lots of interesting data')
...     f.write('and some more')

"f" will point to a temporary file in the same directory as 'mypath', and at the end of the 'with' statement the new file will replace the old. If the code in the with statement throws an exception, or the Python process mysteriously dies, then the old file will remain intact.

Finding stuff

pathfinder offers a friendlier alternative to the slightly awkward os.walk and the deeply unpleasant os.path.walk.

>>> p = path('/home/stephen/coding')
>>> p.find(include = ["*.py", "docs/**/*.html"],
           exclude = ".git",
           visit_dirs = "before",
           ignore_case = True,
           max_depth = 3)
  • include specifies what to search for, as a list of string patterns. The patterns can contain wildcards: "?" matches any single character, "*" any sequence of characters and "**" any sequence of directories. So, the pattern "coding/*/*.html" matches "coding/project1/info.html" but not "coding/project1/docs/help.html", while "coding/**/*.html" matches both. Patterns may also be specified as callables, which will be invoked to determine whether a file is worth returning. If there's only one pattern, it need not be passed in a list.

  • exclude is much the same, but specifies paths to avoid. Excluded paths won't even be recursed into.

  • visit_dirs can be 'before', 'after' or False, and specifies whether directories are returned before or after their contents, or not at all.

  • ignore_case and max_depth should hopefully be reasonably clear. :)

find returns an iterator yielding path objects.

Something went wrong with that request. Please try again.