devtools/fix_po_encoding
author Aurelien Campeas <aurelien.campeas@logilab.fr>
Mon, 06 Jan 2014 12:01:35 +0100
changeset 9383 7b3fa5a7a5a3
parent 0 b97547f5f1fa
permissions -rwxr-xr-x
[views/dot] use an inlined svg for schema and workflow graphs These are currently served as "temporary" pngs which are actually consummed immediately. Which means they cannot be requested twice and any further attempt will yield an error in the logs and some end-user surprise. There is no known acceptable workaround for IE-8 (and previous versions). SVGWEB could be workable but it's not trivial to integrate it properly. Closes #3400448.

#!/usr/bin/python

"""usage: fix-po-encodings [filename...]
change the encoding of the po files passed as arguments to utf-8
"""
import sys
import re
import codecs

def change_encoding(filename, target='UTF-8'):
    fdesc = open(filename)
    data = fdesc.read()
    fdesc.close()
    encoding = find_encoding(data)
    if encoding == target:
        return
    data = fix_encoding(data, target)
    data = unicode(data, encoding)
    fdesc = codecs.open(filename, 'wb', encoding=target)
    fdesc.write(data)
    fdesc.close()

def find_encoding(data):
    regexp = re.compile(r'"Content-Type:.* charset=([a-zA-Z0-9-]+)\\n"', re.M)
    mo = regexp.search(data)
    if mo is None:
        raise ValueError('No encoding declaration')
    return mo.group(1)

def fix_encoding(data, target_encoding):
    regexp = re.compile(r'("Content-Type:.* charset=)(.*)(\\n")', re.M)
    return regexp.sub(r'\1%s\3' % target_encoding, data)
    


for filename in sys.argv[1:]:
    print filename
    change_encoding(filename)