check_whence.py

   1 #!/usr/bin/python
   2
   3 import os, re, sys
   4
   5 def list_whence():
   6     with open('WHENCE') as whence:
   7         for line in whence:
   8             match = re.match(r'(?:File|Link|Source):\s*(\S*)', line)
   9             if match:
  10                 yield match.group(1)
  11                 continue
  12             match = re.match(r'Licen[cs]e: (?:.*\bSee (.*) for details\.?|(\S*))\n',
  13                              line)
  14             if match:
  15                 if match.group(1):
  16                     for name in re.split(r', | and ', match.group(1)):
  17                         yield name
  18                     continue
  19                 if match.group(2):
  20                     # Just one word - may or may not be a filename
  21                     if not re.search(r'unknown|distributable', match.group(2),
  22                                      re.IGNORECASE):
  23                         yield match.group(2)
  24                         continue
  25
  26 def list_git():
  27     with os.popen('git ls-files') as git_files:
  28         for line in git_files:
  29             yield line.rstrip('\n')
  30
  31 def main():
  32     whence_list = list(list_whence())
  33     known_files = set(name for name in whence_list if not name.endswith('/')) | \
  34                   set(['check_whence.py', 'configure', 'Makefile',
  35                        'README', 'WHENCE'])
  36     known_prefixes = set(name for name in whence_list if name.endswith('/'))
  37     git_files = set(list_git())
  38
  39     for name in sorted(list(known_files - git_files)):
  40         sys.stderr.write('E: %s listed in WHENCE does not exist\n' % name)
  41
  42     for name in sorted(list(git_files - known_files)):
  43         # Ignore subdirectory changelogs and GPG detached signatures
  44         if (name.endswith('/ChangeLog') or
  45             (name.endswith('.asc') and name[:-4] in known_files)):
  46             continue
  47
  48         # Ignore unknown files in known directories
  49         for prefix in known_prefixes:
  50             if name.startswith(prefix):
  51                 break
  52         else:
  53             sys.stderr.write('E: %s not listed in WHENCE\n' % name)
  54
  55 if __name__ == '__main__':
  56     main()