# -*- test-case-name: yadis.test.test_xri -*-
"""Utility functions for handling XRIs.

@see: XRI Syntax v2.0 at the U{OASIS XRI Technical Committee<http://www.oasis-open.org/committees/tc_home.php?wg_abbrev=xri>}
"""

import re

XRI_AUTHORITIES = ['!', '=', '@', '+', '$', '(']

try:
    unichr(0x10000)
except ValueError:
    # narrow python build
    UCSCHAR = [
        (0xA0, 0xD7FF),
        (0xF900, 0xFDCF),
        (0xFDF0, 0xFFEF),
        ]

    IPRIVATE = [
        (0xE000, 0xF8FF),
        ]
else:
    UCSCHAR = [
        (0xA0, 0xD7FF),
        (0xF900, 0xFDCF),
        (0xFDF0, 0xFFEF),
        (0x10000, 0x1FFFD),
        (0x20000, 0x2FFFD),
        (0x30000, 0x3FFFD),
        (0x40000, 0x4FFFD),
        (0x50000, 0x5FFFD),
        (0x60000, 0x6FFFD),
        (0x70000, 0x7FFFD),
        (0x80000, 0x8FFFD),
        (0x90000, 0x9FFFD),
        (0xA0000, 0xAFFFD),
        (0xB0000, 0xBFFFD),
        (0xC0000, 0xCFFFD),
        (0xD0000, 0xDFFFD),
        (0xE1000, 0xEFFFD),
        ]

    IPRIVATE = [
        (0xE000, 0xF8FF),
        (0xF0000, 0xFFFFD),
        (0x100000, 0x10FFFD),
        ]


_escapeme_re = re.compile('[%s]' % (''.join(
    map(lambda (m, n): u'%s-%s' % (unichr(m), unichr(n)),
        UCSCHAR + IPRIVATE)),))


def identifierScheme(identifier):
    """Determine if this identifier is an XRI or URI.

    @returns: C{"XRI"} or C{"URI"}
    """
    if identifier.startswith('xri://') or identifier[0] in XRI_AUTHORITIES:
        return "XRI"
    else:
        return "URI"


def toIRINormal(xri):
    """Transform an XRI to IRI-normal form."""
    if not xri.startswith('xri://'):
        xri = 'xri://' + xri
    return escapeForIRI(xri)


_xref_re = re.compile('\((.*?)\)')


def _escape_xref(xref_match):
    """Escape things that need to be escaped if they're in a cross-reference.
    """
    xref = xref_match.group()
    xref = xref.replace('/', '%2F')
    xref = xref.replace('?', '%3F')
    xref = xref.replace('#', '%23')
    return xref


def escapeForIRI(xri):
    """Escape things that need to be escaped when transforming to an IRI."""
    xri = xri.replace('%', '%25')
    xri = _xref_re.sub(_escape_xref, xri)
    return xri


def toURINormal(xri):
    """Transform an XRI to URI normal form."""
    return iriToURI(toIRINormal(xri))


def _percentEscapeUnicode(char_match):
    c = char_match.group()
    return ''.join(['%%%X' % (ord(octet),) for octet in c.encode('utf-8')])


def iriToURI(iri):
    """Transform an IRI to a URI by escaping unicode."""
    # According to RFC 3987, section 3.1, "Mapping of IRIs to URIs"
    return _escapeme_re.sub(_percentEscapeUnicode, iri)


def providerIsAuthoritative(providerID, canonicalID):
    """Is this provider ID authoritative for this XRI?

    @returntype: bool
    """
    # XXX: can't use rsplit until we require python >= 2.4.
    lastbang = canonicalID.rindex('!')
    parent = canonicalID[:lastbang]
    return parent == providerID


def rootAuthority(xri):
    """Return the root authority for an XRI.

    Example::

        rootAuthority("xri://@example") == "xri://@"

    @type xri: unicode
    @returntype: unicode
    """
    if xri.startswith('xri://'):
        xri = xri[6:]
    authority = xri.split('/', 1)[0]
    if authority[0] == '(':
        # Cross-reference.
        # XXX: This is incorrect if someone nests cross-references so there
        #   is another close-paren in there.  Hopefully nobody does that
        #   before we have a real xriparse function.  Hopefully nobody does
        #   that *ever*.
        root = authority[:authority.index(')') + 1]
    elif authority[0] in XRI_AUTHORITIES:
        # Other XRI reference.
        root = authority[0]
    else:
        # IRI reference.  XXX: Can IRI authorities have segments?
        segments = authority.split('!')
        segments = reduce(list.__add__,
            map(lambda s: s.split('*'), segments))
        root = segments[0]

    return XRI(root)


def XRI(xri):
    """An XRI object allowing comparison of XRI.

    Ideally, this would do full normalization and provide comparsion
    operators as per XRI Syntax.  Right now, it just does a bit of
    canonicalization by ensuring the xri scheme is present.

    @param xri: an xri string
    @type xri: unicode
    """
    if not xri.startswith('xri://'):
        xri = 'xri://' + xri
    return xri