validators.py 文件源码

def unicode_to_ascii_authority(authority):
    """
    Follows the steps in RFC 3490, Section 4 to convert a unicode authority
    string into its ASCII equivalent.
    For example, u'www.Alliancefran\xe7aise.nu' will be converted into
    'www.xn--alliancefranaise-npb.nu'

    Args:
        authority: unicode string, the URL authority component to convert,
            e.g. u'www.Alliancefran\xe7aise.nu'

    Returns:
        string: the US-ASCII character equivalent to the inputed authority,
             e.g. 'www.xn--alliancefranaise-npb.nu'

    Raises:
        Exception: if the function is not able to convert the inputed
            authority

    @author: Jonathan Benn
    """
    # RFC 3490, Section 4, Step 1
    # The encodings.idna Python module assumes that AllowUnassigned == True

    # RFC 3490, Section 4, Step 2
    labels = label_split_regex.split(authority)

    # RFC 3490, Section 4, Step 3
    # The encodings.idna Python module assumes that UseSTD3ASCIIRules == False

    # RFC 3490, Section 4, Step 4
    # We use the ToASCII operation because we are about to put the authority
    # into an IDN-unaware slot
    asciiLabels = []
    import encodings.idna
    for label in labels:
        if label:
            asciiLabels.append(to_native(encodings.idna.ToASCII(label)))
        else:
            # encodings.idna.ToASCII does not accept an empty string, but
            # it is necessary for us to allow for empty labels so that we
            # don't modify the URL
            asciiLabels.append('')
    # RFC 3490, Section 4, Step 5
    return str(reduce(lambda x, y: x + unichr(0x002E) + y, asciiLabels))