unicode/php-serialized-recode.py

   1 #!/usr/bin/python
   2 # Outil de recodage de données PHP sérialisées (latin1→utf8)
   3 # (utilisé pour recoder les fichiers .dat de gallery)
   4 # Statut : experimental
   5 import sys, re
   6
   7 PATTERN = r's:([0-9]+):"([^"]+)"'
   8
   9 def serialized_recode(from_charset, to_charset):
  10     def serialized_sub(m):
  11         n, s = m.groups()
  12         if int(n) != len(s):
  13             print "Error: int(%s) != len(%s)" % (n, s)
  14         else:
  15             s = s.decode(from_charset).encode(to_charset)
  16             n = len(s)
  17         return 's:%s:"%s"' % (n, s)
  18     return serialized_sub
  19
  20 standard_sub = serialized_recode('latin1', 'utf8')
  21
  22 if len(sys.argv) < 2:
  23     s = sys.stdin.read()
  24     s = re.sub(PATTERN, standard_sub, s)
  25     sys.stdout.write(s)
  26 else:
  27     for a in sys.argv[1:]:
  28         f = file(a, 'r+')
  29         s = f.read()
  30         s = re.sub(PATTERN, standard_sub, s)
  31         f.seek(0)
  32         f.write(s)
  33         f.truncate()
  34         f.close()