robots.org.uk: attachment:pycruft.py of DebianCruft

   1 #!/usr/bin/python
   2 
   3 import glob
   4 import os
   5 import sys
   6 
   7 # dict that maps path -> list of packages that contain the path
   8 files = {}
   9 
  10 for lf in glob.glob ('/var/lib/dpkg/info/*.list'):
  11     package = lf[19:-5]
  12     for file in [line.strip () for line in open (lf)]:
  13         if not file in files:
  14             files[file] = []
  15         files[file].append (package)
  16 
  17 dirs = sys.path[1:] # sys.path[0] is the dir that contains this script
  18 
  19 def raise_error (e):
  20     raise e
  21 
  22 for dir in dirs:
  23     if not os.path.isdir (dir):
  24         continue
  25 
  26     for dirpath, dirnames, filenames in os.walk (dir, onerror = raise_error):
  27         for filename in filenames:
  28             path = os.path.join (dirpath, filename)
  29 
  30             # symlinks are OK as long as they point to an existing file
  31             if os.path.islink (path):
  32                 if not os.path.exists (path):
  33                     print 'broken link: %s' % (path)
  34                 continue
  35 
  36             # .pyc files should be accompanied by a corresponding .py file
  37             if path.endswith ('.pyc'):
  38                 if not os.path.exists (path[:-1]):
  39                     print 'orphaned .pyc: %s' % (path)
  40                 continue
  41 
  42             # other files should be in dpkg's datbase
  43             if path not in files:
  44                 if filename == '__init__.py' and os.path.getsize (path) == 0:
  45                     # possibly a python-central-created __init__.py
  46                     continue
  47 
  48                 print 'untracked: %s' % (path)