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*"(.*)"', line)
   9             if match:
  10                 yield match.group(1)
  11                 continue
  12             match = re.match(r'(?:File|Link|Source):\s*(\S*)', line)
  13             if match:
  14                 yield match.group(1)
  15                 continue
  16             match = re.match(r'Licen[cs]e: (?:.*\bSee (.*) for details\.?|(\S*))\n',
  17                              line)
  18             if match:
  19                 if match.group(1):
  20                     for name in re.split(r', | and ', match.group(1)):
  21                         yield name
  22                     continue
  23                 if match.group(2):
  24                     # Just one word - may or may not be a filename
  25                     if not re.search(r'unknown|distributable', match.group(2),
  26                                      re.IGNORECASE):
  27                         yield match.group(2)
  28                         continue
  29
  30 def list_git():
  31     with os.popen('git ls-files') as git_files:
  32         for line in git_files:
  33             yield line.rstrip('\n')
  34
  35 def main():
  36     whence_list = list(list_whence())
  37     known_files = set(name for name in whence_list if not name.endswith('/')) | \
  38                   set(['check_whence.py', 'configure', 'Makefile',
  39                        'README', 'WHENCE'])
  40     known_prefixes = set(name for name in whence_list if name.endswith('/'))
  41     git_files = set(list_git())
  42
  43     for name in sorted(list(known_files - git_files)):
  44         sys.stderr.write('E: %s listed in WHENCE does not exist\n' % name)
  45
  46     for name in sorted(list(git_files - known_files)):
  47         # Ignore subdirectory changelogs and GPG detached signatures
  48         if (name.endswith('/ChangeLog') or
  49             (name.endswith('.asc') and name[:-4] in known_files)):
  50             continue
  51
  52         # Ignore unknown files in known directories
  53         for prefix in known_prefixes:
  54             if name.startswith(prefix):
  55                 break
  56         else:
  57             sys.stderr.write('E: %s not listed in WHENCE\n' % name)
  58
  59 if __name__ == '__main__':
  60     main()