src/cm/utils/string_utils.py
author raph
Fri, 05 Feb 2010 16:01:32 +0100
changeset 149 0f2c5744b39b
parent 119 src/cm/utils/string.py@5e8dda1b7631
child 175 4f072edc51a1
permissions -rw-r--r--
cleanup diff files / add experimental diff

import chardet

def to_unicode(input):
    if type(input) == str:
        res = None
        for encoding in [chardet.detect(input)['encoding'], 'utf8', 'latin1']:
            try:
                res = unicode(input, encoding)
                break;
            except UnicodeDecodeError:
                pass
        if not res:
            raise Exception('UnicodeDecodeError: could not decode')
        return res
    return input