check_whence.py

   1 #!/usr/bin/python3
   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_whence_files():
  32     with open('WHENCE', encoding='utf-8') as whence:
  33         for line in whence:
  34             match = re.match(r'File:\s*(.*)', line)
  35             if match:
  36                 yield match.group(1).replace("\ ", " ").replace("\"", "")
  37                 continue
  38
  39 def list_links_list():
  40     with open('WHENCE', encoding='utf-8') as whence:
  41         for line in whence:
  42             match = re.match(r'Link:\s*(.*)', line)
  43             if match:
  44                 linkname, target = match.group(1).split("->")
  45
  46                 linkname = linkname.strip().replace("\ ", " ").replace("\"", "")
  47                 target = target.strip().replace("\ ", " ").replace("\"", "")
  48
  49                 # Link target is relative to the link
  50                 target = os.path.join(os.path.dirname(linkname), target)
  51                 target = os.path.normpath(target)
  52
  53                 yield (linkname, target)
  54                 continue
  55
  56 def list_git():
  57     with os.popen('git ls-files') as git_files:
  58         for line in git_files:
  59             yield line.rstrip('\n')
  60
  61 def main():
  62     ret = 0
  63     whence_list = list(list_whence())
  64     whence_files = list(list_whence_files())
  65     links_list = list(list_links_list())
  66     known_files = set(name for name in whence_list if not name.endswith('/')) | \
  67                   set(['check_whence.py', 'configure', 'Makefile',
  68                        'README', 'copy-firmware.sh', 'WHENCE'])
  69     known_prefixes = set(name for name in whence_list if name.endswith('/'))
  70     git_files = set(list_git())
  71
  72     for name in set(name for name in whence_files if name.endswith('/')):
  73         sys.stderr.write('E: %s listed in WHENCE as File, but is directory\n' %
  74                          name)
  75         ret = 1
  76
  77     for name in set(fw for fw in whence_files if whence_files.count(fw) > 1):
  78         sys.stderr.write('E: %s listed in WHENCE twice\n' % name)
  79         ret = 1
  80
  81     for name in set(link for link in whence_files if os.path.islink(link)):
  82         sys.stderr.write('E: %s listed in WHENCE as File, but is a symlink\n' %
  83                          name)
  84         ret = 1
  85
  86     for name in set(link[0] for link in links_list if os.path.islink(link[0])):
  87         sys.stderr.write('E: %s listed in WHENCE as Link, is in tree\n' % name)
  88         ret = 1
  89
  90     for name in sorted(list(known_files - git_files)):
  91         sys.stderr.write('E: %s listed in WHENCE does not exist\n' % name)
  92         ret = 1
  93
  94     # A link can point to another link, or to a file...
  95     valid_targets = set(link[0] for link in links_list) | git_files
  96
  97     # ... or to a directory
  98     for target in set(valid_targets):
  99         dirname = target
 100         while True:
 101             dirname = os.path.dirname(dirname)
 102             if dirname == '':
 103                 break
 104             valid_targets.add(dirname)
 105
 106     for name, target in sorted(links_list):
 107         if target not in valid_targets:
 108             sys.stderr.write('E: target %s of link %s in WHENCE'
 109                              ' does not exist\n' % (target, name))
 110             ret = 1
 111
 112     for name in sorted(list(git_files - known_files)):
 113         # Ignore subdirectory changelogs and GPG detached signatures
 114         if (name.endswith('/ChangeLog') or
 115             (name.endswith('.asc') and name[:-4] in known_files)):
 116             continue
 117
 118         # Ignore unknown files in known directories
 119         for prefix in known_prefixes:
 120             if name.startswith(prefix):
 121                 break
 122         else:
 123             sys.stderr.write('E: %s not listed in WHENCE\n' % name)
 124             ret = 1
 125     return ret
 126
 127 if __name__ == '__main__':
 128     sys.exit(main())