Tools/scripts/findnocoding.py

   1 #!/usr/bin/env python
   2
   3 """List all those Python files that require a coding directive
   4
   5 Usage: nocoding.py dir1 [dir2...]
   6 """
   7
   8 __author__ = "Oleg Broytmann, Georg Brandl"
   9
  10 import sys, os, re, getopt
  11
  12 # our pysource module finds Python source files
  13 try:
  14     import pysource
  15 except ImportError:
  16     # emulate the module with a simple os.walk
  17     class pysource:
  18         has_python_ext = looks_like_python = can_be_compiled = None
  19         def walk_python_files(self, paths, *args, **kwargs):
  20             for path in paths:
  21                 if os.path.isfile(path):
  22                     yield path.endswith(".py")
  23                 elif os.path.isdir(path):
  24                     for root, dirs, files in os.walk(path):
  25                         for filename in files:
  26                             if filename.endswith(".py"):
  27                                 yield os.path.join(root, filename)
  28     pysource = pysource()
  29
  30
  31     print >>sys.stderr, ("The pysource module is not available; "
  32                          "no sophisticated Python source file search will be done.")
  33
  34
  35 decl_re = re.compile(r"coding[=:]\s*([-\w.]+)")
  36
  37 def get_declaration(line):
  38     match = decl_re.search(line)
  39     if match:
  40         return match.group(1)
  41     return ''
  42
  43 def has_correct_encoding(text, codec):
  44     try:
  45         unicode(text, codec)
  46     except UnicodeDecodeError:
  47         return False
  48     else:
  49         return True
  50
  51 def needs_declaration(fullpath):
  52     try:
  53         infile = open(fullpath, 'rU')
  54     except IOError: # Oops, the file was removed - ignore it
  55         return None
  56
  57     line1 = infile.readline()
  58     line2 = infile.readline()
  59
  60     if get_declaration(line1) or get_declaration(line2):
  61         # the file does have an encoding declaration, so trust it
  62         infile.close()
  63         return False
  64
  65     # check the whole file for non-ASCII characters
  66     rest = infile.read()
  67     infile.close()
  68
  69     if has_correct_encoding(line1+line2+rest, "ascii"):
  70         return False
  71
  72     return True
  73
  74
  75 usage = """Usage: %s [-cd] paths...
  76     -c: recognize Python source files trying to compile them
  77     -d: debug output""" % sys.argv[0]
  78
  79 try:
  80     opts, args = getopt.getopt(sys.argv[1:], 'cd')
  81 except getopt.error, msg:
  82     print >>sys.stderr, msg
  83     print >>sys.stderr, usage
  84     sys.exit(1)
  85
  86 is_python = pysource.looks_like_python
  87 debug = False
  88
  89 for o, a in opts:
  90     if o == '-c':
  91         is_python = pysource.can_be_compiled
  92     elif o == '-d':
  93         debug = True
  94
  95 if not args:
  96     print >>sys.stderr, usage
  97     sys.exit(1)
  98
  99 for fullpath in pysource.walk_python_files(args, is_python):
 100     if debug:
 101         print "Testing for coding: %s" % fullpath
 102     result = needs_declaration(fullpath)
 103     if result:
 104         print fullpath