src/cm/utils/string.py
author raph
Mon, 25 Jan 2010 11:34:22 +0100
changeset 119 5e8dda1b7631
permissions -rw-r--r--
recover when tidy trashes: try markdown anyway

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