forum/utils/diff.py

   1 #!/usr/bin/python2.2
   2 """HTML Diff: http://www.aaronsw.com/2002/diff
   3 Rough code, badly documented. Send me comments and patches."""
   4
   5 __author__ = 'Aaron Swartz <me@aaronsw.com>'
   6 __copyright__ = '(C) 2003 Aaron Swartz. GNU GPL 2.'
   7 __version__ = '0.22'
   8
   9 import difflib, string
  10
  11 def isTag(x): return x[0] == "<" and x[-1] == ">"
  12
  13 def textDiff(a, b):
  14         """Takes in strings a and b and returns a human-readable HTML diff."""
  15
  16         out = []
  17         a, b = html2list(a), html2list(b)
  18         s = difflib.SequenceMatcher(None, a, b)
  19         for e in s.get_opcodes():
  20                 if e[0] == "replace":
  21                         # @@ need to do something more complicated here
  22                         # call textDiff but not for html, but for some html... ugh
  23                         # gonna cop-out for now
  24                         out.append('<del>'+''.join(a[e[1]:e[2]]) + '</del><ins>'+''.join(b[e[3]:e[4]])+"</ins>")
  25                 elif e[0] == "delete":
  26                         out.append('<del >'+ ''.join(a[e[1]:e[2]]) + "</del>")
  27                 elif e[0] == "insert":
  28                         out.append('<ins >'+''.join(b[e[3]:e[4]]) + "</ins>")
  29                 elif e[0] == "equal":
  30                         out.append(''.join(b[e[3]:e[4]]))
  31                 else:
  32                         raise "Um, something's broken. I didn't expect a '" + `e[0]` + "'."
  33         return ''.join(out)
  34
  35 def html2list(x, b=0):
  36         mode = 'char'
  37         cur = ''
  38         out = []
  39         for c in x:
  40                 if mode == 'tag':
  41                         if c == '>':
  42                                 if b: cur += ']'
  43                                 else: cur += c
  44                                 out.append(cur); cur = ''; mode = 'char'
  45                         else: cur += c
  46                 elif mode == 'char':
  47                         if c == '<':
  48                                 out.append(cur)
  49                                 if b: cur = '['
  50                                 else: cur = c
  51                                 mode = 'tag'
  52                         elif c in string.whitespace: out.append(cur+c); cur = ''
  53                         else: cur += c
  54         out.append(cur)
  55         return filter(lambda x: x is not '', out)
  56
  57 if __name__ == '__main__':
  58         import sys
  59         try:
  60                 a, b = sys.argv[1:3]
  61         except ValueError:
  62                 print "htmldiff: highlight the differences between two html files"
  63                 print "usage: " + sys.argv[0] + " a b"
  64                 sys.exit(1)
  65         print textDiff(open(a).read(), open(b).read())
  66