lamenor.py - Pasteur

Filename:

Private:

Filetype:

Indentation:

def getEncoding(s):
    try:
        match = re.search(_charset_re, s, re.MULTILINE)
        if match:
            return match.group('charset')[1:-1]
    except:
        match = re.search(_charset_re.encode(), s, re.MULTILINE)
        if match:
            return match.group('charset').decode()[1:-1]

    try:
        import charade.universaldetector
        u = charade.universaldetector.UniversalDetector()
        u.feed(s)
        u.close()
        return u.result['encoding']
    except:
        return None