check_whence.py

   1 #!/usr/bin/python
   2
   3 import os, re, sys
   4 from io import open
   5
   6 def list_whence():
   7     with open('WHENCE', encoding='utf-8') as whence:
   8         for line in whence:
   9             match = re.match(r'(?:File|Source):\s*"(.*)"', line)
  10             if match:
  11                 yield match.group(1)
  12                 continue
  13             match = re.match(r'(?:File|Source):\s*(\S*)', line)
  14             if match:
  15                 yield match.group(1)
  16                 continue
  17             match = re.match(r'Licen[cs]e: (?:.*\bSee (.*) for details\.?|(\S*))\n',
  18                              line)
  19             if match:
  20                 if match.group(1):
  21                     for name in re.split(r', | and ', match.group(1)):
  22                         yield name
  23                     continue
  24                 if match.group(2):
  25                     # Just one word - may or may not be a filename
  26                     if not re.search(r'unknown|distributable', match.group(2),
  27                                      re.IGNORECASE):
  28                         yield match.group(2)
  29                         continue
  30
  31 def list_git():
  32     with os.popen('git ls-files') as git_files:
  33         for line in git_files:
  34             yield line.rstrip('\n')
  35
  36 def main():
  37     ret = 0
  38     whence_list = list(list_whence())
  39     known_files = set(name for name in whence_list if not name.endswith('/')) | \
  40                   set(['check_whence.py', 'configure', 'Makefile',
  41                        'README', 'copy-firmware.sh', 'WHENCE'])
  42     known_prefixes = set(name for name in whence_list if name.endswith('/'))
  43     git_files = set(list_git())
  44
  45     for name in sorted(list(known_files - git_files)):
  46         sys.stderr.write('E: %s listed in WHENCE does not exist\n' % name)
  47         ret = 1
  48
  49     for name in sorted(list(git_files - known_files)):
  50         # Ignore subdirectory changelogs and GPG detached signatures
  51         if (name.endswith('/ChangeLog') or
  52             (name.endswith('.asc') and name[:-4] in known_files)):
  53             continue
  54
  55         # Ignore unknown files in known directories
  56         for prefix in known_prefixes:
  57             if name.startswith(prefix):
  58                 break
  59         else:
  60             sys.stderr.write('E: %s not listed in WHENCE\n' % name)
  61             ret = 1
  62     return ret
  63
  64 if __name__ == '__main__':
  65     sys.exit(main())