Mac/scripts/findgremlins.py

   1 """findgremlins - Search through a folder and subfolders for
   2 text files that have characters with bit 8 set, and print
   3 the filename and a bit of context.
   4
   5 By Just, with a little glue by Jack"""
   6
   7 import macfs
   8 import re
   9 import os
  10 import string
  11 import sys
  12
  13 xpat = re.compile(r"[\200-\377]")
  14
  15 def walk(top, recurse=1):
  16         if os.path.isdir(top):
  17                 if recurse:
  18                         for name in os.listdir(top):
  19                                 path = os.path.join(top, name)
  20                                 walk(path)
  21         else:
  22                 cr, tp = macfs.FSSpec(top).GetCreatorType()
  23                 if tp == 'TEXT' and top[-4:] <> ".hqx":
  24                         data = open(top).read()
  25                         badcount = 0
  26                         for ch in data[:256]:
  27                                 if ord(ch) == 0 or ord(ch) >= 0200:
  28                                         badcount = badcount + 1
  29                         if badcount > 16:
  30                                 print `top`, 'appears to be a binary file'
  31                                 return
  32                         pos = 0
  33                         gotone = 0
  34                         while 1:
  35                                 m = xpat.search(data, pos)
  36                                 if m is None:
  37                                         break
  38                                 if not gotone:
  39                                         print `top`
  40                                         gotone = 1
  41                                 [(i, j)] = m.regs
  42                                 print "     ", string.replace(data[i-15:j+15], '\n', ' ')
  43                                 pos = j
  44
  45 def main():
  46         fss, ok = macfs.GetDirectory()
  47         if ok:
  48                 walk(fss.as_pathname())
  49
  50 if __name__ == '__main__':
  51         main()
  52         sys.exit(1) # So we see the output
  53