Script ConvertIndex_descLatin1ToUTF8_py
[hide private]
[frames] | no frames]

Source Code for Script script-ConvertIndex_descLatin1ToUTF8_py

 1  #!/usr/bin/python 
 2  #Small utility to find all index.desc files and convert them from latin1 to UTF8 
 3  #written by Jerome 23/03/2009 
 4  # $Id$  
 5  import  os, sys 
 6   
7 -def convertToUTF8(unknownStr):
8 enc = None 9 unicStr = u"" 10 for line in unknownStr.split("\n"): 11 if line.find("coding:") == 0: 12 enc = line.split(":", 1)[1].strip() 13 if enc == None: 14 enc = "Latin-1" 15 unicStr = u"coding: UTF-8\n\n" 16 for line in unknownStr.split("\n"): 17 if line.find("coding:") == 0: 18 unicStr += u"coding: UTF-8\n" 19 else: 20 unicStr += line.strip().decode(enc) + u"\n" 21 return unicStr.encode("UTF-8")
22 23 for root, dirs, files in os.walk(os.path.abspath(sys.argv[1])): 24 if "index.desc" in files >= 0: 25 filein = os.path.join(os.path.abspath(sys.argv[1]), root, "index.desc") 26 f = open(filein).read() 27 g = convertToUTF8(f) 28 if f != g[:-1]: 29 open(filein, "w").write(g) 30