Declare charset of XHTML file.

charsets.py

#!/usr/bin/env python
#-*- coding:utf-8 -*-

from sys import argv, stderr
from lxml import etree


xhtml_namespace = "http://www.w3.org/1999/xhtml"
content_type_attrs = {
    "http-equiv": "Content-Type",
    "content": "application/xhtml+xml; charset=utf-8"
}


def xhtml_declare_charset(filename):
    print "* %s" % filename
    parser = etree.XMLParser(remove_blank_text=True)
    html_tag = etree.parse(filename, parser)
    head_tag = html_tag.find("{%s}head" % xhtml_namespace)
    meta_tag = etree.Element("{%s}meta" % xhtml_namespace, content_type_attrs)
    head_tag.append(meta_tag)
    html_tag.write(filename, pretty_print=True, encoding="utf-8")


if __name__ == "__main__":
    if len(argv) < 2:
        print >> stderr, "usage: %s [filename_1] [filename_2]\n" % argv[0]
    for filename in argv[1:]:
        xhtml_declare_charset(filename)

charsets.sh

#!/usr/bin/env sh

find . -type f -name "*.xhtml" -exec python charsets.py {} +