#!/usr/bin/env python
"""Python codec for CSS."""
__docformat__ = 'restructuredtext'
__author__ = 'Walter Doerwald'
__version__ = '$Id: util.py 1114 2008-03-05 13:22:59Z cthedot $'

import sys
import codecs
import marshal

# We're using bits to store all possible candidate encodings (or variants, i.e.
# we have two bits for the variants of UTF-16 and two for the
# variants of UTF-32).
#
# Prefixes for various CSS encodings
# UTF-8-SIG   xEF  xBB  xBF
# UTF-16 (LE) xFF  xFE ~x00|~x00
# UTF-16 (BE) xFE  xFF
# UTF-16-LE    @   x00   @   x00
# UTF-16-BE   x00   @
# UTF-32 (LE) xFF  xFE  x00  x00
# UTF-32 (BE) x00  x00  xFE  xFF
# UTF-32-LE    @   x00  x00  x00
# UTF-32-BE   x00  x00  x00   @
# CHARSET      @    c    h    a  ...


def chars(bytestring):
    return ''.join(chr(byte) for byte in bytestring)


def detectencoding_str(input, final=False):
    """
    Detect the encoding of the byte string ``input``, which contains the
    beginning of a CSS file. This function returns the detected encoding (or
    ``None`` if it hasn't got enough data), and a flag that indicates whether
    that encoding has been detected explicitely or implicitely. To detect the
    encoding the first few bytes are used (or if ``input`` is ASCII compatible
    and starts with a charset rule the encoding name from the rule). "Explicit"
    detection means that the bytes start with a BOM or a charset rule.

    If the encoding can't be detected yet, ``None`` is returned as the encoding.
    ``final`` specifies whether more data will be available in later calls or
    not. If ``final`` is true, ``detectencoding_str()`` will never return
    ``None`` as the encoding.
    """

    # A bit for every candidate
    CANDIDATE_UTF_8_SIG    =   1
    CANDIDATE_UTF_16_AS_LE =   2
    CANDIDATE_UTF_16_AS_BE =   4
    CANDIDATE_UTF_16_LE    =   8
    CANDIDATE_UTF_16_BE    =  16
    CANDIDATE_UTF_32_AS_LE =  32
    CANDIDATE_UTF_32_AS_BE =  64
    CANDIDATE_UTF_32_LE    = 128
    CANDIDATE_UTF_32_BE    = 256
    CANDIDATE_CHARSET      = 512

    candidates = 1023 # all candidates

    #input = chars(input)
    li = len(input)
    if li>=1:
        # Check first byte
        c = input[0]
        if c != b"\xef"[0]:
            candidates &= ~CANDIDATE_UTF_8_SIG
        if c != b"\xff"[0]:
            candidates &= ~(CANDIDATE_UTF_32_AS_LE|CANDIDATE_UTF_16_AS_LE)
        if c != b"\xfe"[0]:
            candidates &= ~CANDIDATE_UTF_16_AS_BE
        if c != b"@"[0]:
            candidates &= ~(CANDIDATE_UTF_32_LE|CANDIDATE_UTF_16_LE|CANDIDATE_CHARSET)
        if c != b"\x00"[0]:
            candidates &= ~(CANDIDATE_UTF_32_AS_BE|CANDIDATE_UTF_32_BE|CANDIDATE_UTF_16_BE)
        if li>=2:
            # Check second byte
            c = input[1]
            if c != b"\xbb"[0]:
                candidates &= ~CANDIDATE_UTF_8_SIG
            if c != b"\xfe"[0]:
                candidates &= ~(CANDIDATE_UTF_16_AS_LE|CANDIDATE_UTF_32_AS_LE)
            if c != b"\xff"[0]:
                candidates &= ~CANDIDATE_UTF_16_AS_BE
            if c != b"\x00"[0]:
                candidates &= ~(CANDIDATE_UTF_16_LE|CANDIDATE_UTF_32_AS_BE|CANDIDATE_UTF_32_LE|CANDIDATE_UTF_32_BE)
            if c != b"@"[0]:
                candidates &= ~CANDIDATE_UTF_16_BE
            if c != b"c"[0]:
                candidates &= ~CANDIDATE_CHARSET
            if li>=3:
                # Check third byte
                c = input[2]
                if c != b"\xbf"[0]:
                    candidates &= ~CANDIDATE_UTF_8_SIG
                if c != b"c"[0]:
                    candidates &= ~CANDIDATE_UTF_16_LE
                if c != b"\x00"[0]:
                    candidates &= ~(CANDIDATE_UTF_32_AS_LE|CANDIDATE_UTF_32_LE|CANDIDATE_UTF_32_BE)
                if c != b"\xfe"[0]:
                    candidates &= ~CANDIDATE_UTF_32_AS_BE
                if c != b"h"[0]:
                    candidates &= ~CANDIDATE_CHARSET
                if li>=4:
                    # Check fourth byte
                    c = input[3]
                    if input[2:4] == b"\x00\x00"[0:2]:
                        candidates &= ~CANDIDATE_UTF_16_AS_LE
                    if c != b"\x00"[0]:
                        candidates &= ~(CANDIDATE_UTF_16_LE|CANDIDATE_UTF_32_AS_LE|CANDIDATE_UTF_32_LE)
                    if c != b"\xff"[0]:
                        candidates &= ~CANDIDATE_UTF_32_AS_BE
                    if c != b"@"[0]:
                        candidates &= ~CANDIDATE_UTF_32_BE
                    if c != b"a"[0]:
                        candidates &= ~CANDIDATE_CHARSET
    if candidates == 0:
        return ("utf-8", False)
    if not (candidates & (candidates-1)): # only one candidate remaining
        if candidates == CANDIDATE_UTF_8_SIG and li >= 3:
            return ("utf-8-sig", True)
        elif candidates == CANDIDATE_UTF_16_AS_LE and li >= 2:
            return ("utf-16", True)
        elif candidates == CANDIDATE_UTF_16_AS_BE and li >= 2:
            return ("utf-16", True)
        elif candidates == CANDIDATE_UTF_16_LE and li >= 4:
            return ("utf-16-le", False)
        elif candidates == CANDIDATE_UTF_16_BE and li >= 2:
            return ("utf-16-be", False)
        elif candidates == CANDIDATE_UTF_32_AS_LE and li >= 4:
            return ("utf-32", True)
        elif candidates == CANDIDATE_UTF_32_AS_BE and li >= 4:
            return ("utf-32", True)
        elif candidates == CANDIDATE_UTF_32_LE and li >= 4:
            return ("utf-32-le", False)
        elif candidates == CANDIDATE_UTF_32_BE and li >= 4:
            return ("utf-32-be", False)
        elif candidates == CANDIDATE_CHARSET and li >= 4:
            prefix = '@charset "'
            charsinput = chars(input) 
            if charsinput[:len(prefix)] == prefix:
                pos = charsinput.find('"', len(prefix))
                if pos >= 0:
                    # TODO: return str and not bytes!
                    return (charsinput[len(prefix):pos], True)
    # if this is the last call, and we haven't determined an encoding yet,
    # we default to UTF-8
    if final:
        return ("utf-8", False)
    return (None, False) # dont' know yet


def detectencoding_unicode(input, final=False):
    """
    Detect the encoding of the unicode string ``input``, which contains the
    beginning of a CSS file. The encoding is detected from the charset rule
    at the beginning of ``input``. If there is no charset rule, ``"utf-8"``
    will be returned.

    If the encoding can't be detected yet, ``None`` is returned. ``final``
    specifies whether more data will be available in later calls or not. If
    ``final`` is true, ``detectencoding_unicode()`` will never return ``None``.
    """
    prefix = '@charset "'
    if input.startswith(prefix):
        pos = input.find('"', len(prefix))
        if pos >= 0:
            return (input[len(prefix):pos], True)
    elif final or not prefix.startswith(input):
        # if this is the last call, and we haven't determined an encoding yet,
        # (or the string definitely doesn't start with prefix) we default to UTF-8
        return ("utf-8", False)
    return (None, False) # don't know yet


def _fixencoding(input, encoding, final=False):
    """
    Replace the name of the encoding in the charset rule at the beginning of
    ``input`` with ``encoding``. If ``input`` doesn't starts with a charset
    rule, ``input`` will be returned unmodified.

    If the encoding can't be found yet, ``None`` is returned. ``final``
    specifies whether more data will be available in later calls or not.
    If ``final`` is true, ``_fixencoding()`` will never return ``None``.
    """
    prefix = '@charset "'
    if len(input) > len(prefix):
        if input.startswith(prefix):
            pos = input.find('"', len(prefix))
            if pos >= 0:
                if encoding.replace("_", "-").lower() == "utf-8-sig":
                    encoding = "utf-8"
                return prefix + encoding + input[pos:]
            # we haven't seen the end of the encoding name yet => fall through
        else:
            return input # doesn't start with prefix, so nothing to fix
    elif not prefix.startswith(input) or final:
        # can't turn out to be a @charset rule later (or there is no "later")
        return input
    if final:
        return input
    return None # don't know yet


def decode(input, errors="strict", encoding=None, force=True):
    try:
        # py 3 only, memory?! object to bytes
        input = input.tobytes()
    except AttributeError as e:
        pass
    
    if encoding is None or not force:
        (_encoding, explicit) = detectencoding_str(input, True)
        if _encoding == "css":
            raise ValueError("css not allowed as encoding name")
        if (explicit and not force) or encoding is None: # Take the encoding from the input
            encoding = _encoding
            
    # NEEDS: change in parse.py (str to bytes!)
    (input, consumed) = codecs.getdecoder(encoding)(input, errors)
    return (_fixencoding(input, str(encoding), True), consumed)


def encode(input, errors="strict", encoding=None):
    consumed = len(input)
    if encoding is None:
        encoding = detectencoding_unicode(input, True)[0]
        if encoding.replace("_", "-").lower() == "utf-8-sig":
            input = _fixencoding(input, "utf-8", True)
    else:
        input = _fixencoding(input, str(encoding), True)
    if encoding == "css":
        raise ValueError("css not allowed as encoding name")
    encoder = codecs.getencoder(encoding)
    return (encoder(input, errors)[0], consumed)


def _bytes2int(bytes):
    # Helper: convert an 8 bit string into an ``int``.
    i = 0
    for byte in bytes:
        i = (i<<8) + ord(byte)
    return i


def _int2bytes(i):
    # Helper: convert an ``int`` into an 8-bit string.
    v = []
    while i:
        v.insert(0, chr(i&0xff))
        i >>= 8
    return "".join(v)


if hasattr(codecs, "IncrementalDecoder"):
    class IncrementalDecoder(codecs.IncrementalDecoder):
        def __init__(self, errors="strict", encoding=None, force=True):
            self.decoder = None
            self.encoding = encoding
            self.force = force
            codecs.IncrementalDecoder.__init__(self, errors)
            # Store ``errors`` somewhere else,
            # because we have to hide it in a property
            self._errors = errors
            self.buffer = b""
            self.headerfixed = False

        def iterdecode(self, input):
            for part in input:
                result = self.decode(part, False)
                if result:
                    yield result
            result = self.decode("", True)
            if result:
                yield result

        def decode(self, input, final=False):
            # We're doing basically the same as a ``BufferedIncrementalDecoder``,
            # but since the buffer is only relevant until the encoding has been
            # detected (in which case the buffer of the underlying codec might
            # kick in), we're implementing buffering ourselves to avoid some
            # overhead.
            if self.decoder is None:
                input = self.buffer + input
                # Do we have to detect the encoding from the input?
                if self.encoding is None or not self.force:
                    (encoding, explicit) = detectencoding_str(input, final)
                    if encoding is None: # no encoding determined yet
                        self.buffer = input # retry the complete input on the next call
                        return "" # no encoding determined yet, so no output
                    elif encoding == "css":
                        raise ValueError("css not allowed as encoding name")
                    if (explicit and not self.force) or self.encoding is None: # Take the encoding from the input
                        self.encoding = encoding
                self.buffer = "" # drop buffer, as the decoder might keep its own
                decoder = codecs.getincrementaldecoder(self.encoding)
                self.decoder = decoder(self._errors)
            if self.headerfixed:
                return self.decoder.decode(input, final)
            # If we haven't fixed the header yet,
            # the content of ``self.buffer`` is a ``unicode`` object
            output = self.buffer + self.decoder.decode(input, final)
            encoding = self.encoding
            if encoding.replace("_", "-").lower() == "utf-8-sig":
                encoding = "utf-8"
            newoutput = _fixencoding(output, str(encoding), final)
            if newoutput is None:
                # retry fixing the @charset rule (but keep the decoded stuff)
                self.buffer = output
                return ""
            self.headerfixed = True
            return newoutput

        def reset(self):
            codecs.IncrementalDecoder.reset(self)
            self.decoder = None
            self.buffer = b""
            self.headerfixed = False

        def _geterrors(self):
            return self._errors

        def _seterrors(self, errors):
            # Setting ``errors`` must be done on the real decoder too
            if self.decoder is not None:
                self.decoder.errors = errors
            self._errors = errors
        errors = property(_geterrors, _seterrors)

        def getstate(self):
            if self.decoder is not None:
                state = (self.encoding, self.buffer, self.headerfixed, True, self.decoder.getstate())
            else:
                state = (self.encoding, self.buffer, self.headerfixed, False, None)
            return ("", _bytes2int(marshal.dumps(state)))

        def setstate(self, state):
            state = _int2bytes(marshal.loads(state[1])) # ignore buffered input
            self.encoding = state[0]
            self.buffer = state[1]
            self.headerfixed = state[2]
            if state[3] is not None:
                self.decoder = codecs.getincrementaldecoder(self.encoding)(self._errors)
                self.decoder.setstate(state[4])
            else:
                self.decoder = None


if hasattr(codecs, "IncrementalEncoder"):
    class IncrementalEncoder(codecs.IncrementalEncoder):
        def __init__(self, errors="strict", encoding=None):
            self.encoder = None
            self.encoding = encoding
            codecs.IncrementalEncoder.__init__(self, errors)
            # Store ``errors`` somewhere else,
            # because we have to hide it in a property
            self._errors = errors
            self.buffer = ""

        def iterencode(self, input):
            for part in input:
                result = self.encode(part, False)
                if result:
                    yield result
            result = self.encode("", True)
            if result:
                yield result

        def encode(self, input, final=False):
            if self.encoder is None:
                input = self.buffer + input
                if self.encoding is not None:
                    # Replace encoding in the @charset rule with the specified one
                    encoding = self.encoding
                    if encoding.replace("_", "-").lower() == "utf-8-sig":
                        encoding = "utf-8"
                    newinput = _fixencoding(input, str(encoding), final)
                    if newinput is None: # @charset rule incomplete => Retry next time
                        self.buffer = input
                        return ""
                    input = newinput
                else:
                    # Use encoding from the @charset declaration
                    self.encoding = detectencoding_unicode(input, final)[0]
                if self.encoding is not None:
                    if self.encoding == "css":
                        raise ValueError("css not allowed as encoding name")
                    info = codecs.lookup(self.encoding)
                    encoding = self.encoding
                    if self.encoding.replace("_", "-").lower() == "utf-8-sig":
                        input = _fixencoding(input, "utf-8", True)
                    self.encoder = info.incrementalencoder(self._errors)
                    self.buffer = ""
                else:
                    self.buffer = input
                    return ""
            return self.encoder.encode(input, final)

        def reset(self):
            codecs.IncrementalEncoder.reset(self)
            self.encoder = None
            self.buffer = ""

        def _geterrors(self):
            return self._errors

        def _seterrors(self, errors):
            # Setting ``errors ``must be done on the real encoder too
            if self.encoder is not None:
                self.encoder.errors = errors
            self._errors = errors
        errors = property(_geterrors, _seterrors)

        def getstate(self):
            if self.encoder is not None:
                state = (self.encoding, self.buffer, True, self.encoder.getstate())
            else:
                state = (self.encoding, self.buffer, False, None)
            return _bytes2int(marshal.dumps(state))

        def setstate(self, state):
            state = _int2bytes(marshal.loads(state))
            self.encoding = state[0]
            self.buffer = state[1]
            if state[2] is not None:
                self.encoder = codecs.getincrementalencoder(self.encoding)(self._errors)
                self.encoder.setstate(state[4])
            else:
                self.encoder = None


class StreamWriter(codecs.StreamWriter):
    def __init__(self, stream, errors="strict", encoding=None, header=False):
        codecs.StreamWriter.__init__(self, stream, errors)
        self.streamwriter = None
        self.encoding = encoding
        self._errors = errors
        self.buffer = ""

    def encode(self, input, errors='strict'):
        li = len(input)
        if self.streamwriter is None:
            input = self.buffer + input
            li = len(input)
            if self.encoding is not None:
                # Replace encoding in the @charset rule with the specified one
                encoding = self.encoding
                if encoding.replace("_", "-").lower() == "utf-8-sig":
                    encoding = "utf-8"
                newinput = _fixencoding(input, str(encoding), False)
                if newinput is None: # @charset rule incomplete => Retry next time
                    self.buffer = input
                    return ("", 0)
                input = newinput
            else:
                # Use encoding from the @charset declaration
                self.encoding = detectencoding_unicode(input, False)[0]
            if self.encoding is not None:
                if self.encoding == "css":
                    raise ValueError("css not allowed as encoding name")
                self.streamwriter = codecs.getwriter(self.encoding)(self.stream, self._errors)
                encoding = self.encoding
                if self.encoding.replace("_", "-").lower() == "utf-8-sig":
                    input = _fixencoding(input, "utf-8", True)
                self.buffer = ""
            else:
                self.buffer = input
                return ("", 0)
        return (self.streamwriter.encode(input, errors)[0], li)

    def _geterrors(self):
        return self._errors

    def _seterrors(self, errors):
        # Setting ``errors`` must be done on the streamwriter too
        try:
            if self.streamwriter is not None:
                self.streamwriter.errors = errors
        except AttributeError as e:
            # TODO: py3 only exception?
            pass
        
        self._errors = errors
    errors = property(_geterrors, _seterrors)


class StreamReader(codecs.StreamReader):
    def __init__(self, stream, errors="strict", encoding=None, force=True):
        codecs.StreamReader.__init__(self, stream, errors)
        self.streamreader = None
        self.encoding = encoding
        self.force = force
        self._errors = errors

    def decode(self, input, errors='strict'):
        if self.streamreader is None:
            if self.encoding is None or not self.force:
                (encoding, explicit) = detectencoding_str(input, False)
                if encoding is None: # no encoding determined yet
                    return ("", 0) # no encoding determined yet, so no output
                elif encoding == "css":
                    raise ValueError("css not allowed as encoding name")
                if (explicit and not self.force) or self.encoding is None: # Take the encoding from the input
                    self.encoding = encoding
            streamreader = codecs.getreader(self.encoding)
            streamreader = streamreader(self.stream, self._errors)
            (output, consumed) = streamreader.decode(input, errors)
            encoding = self.encoding
            if encoding.replace("_", "-").lower() == "utf-8-sig":
                encoding = "utf-8"
            newoutput = _fixencoding(output, str(encoding), False)
            if newoutput is not None:
                self.streamreader = streamreader
                return (newoutput, consumed)
            return ("", 0) # we will create a new streamreader on the next call
        return self.streamreader.decode(input, errors)

    def _geterrors(self):
        return self._errors

    def _seterrors(self, errors):
        # Setting ``errors`` must be done on the streamreader too
        try:
            if self.streamreader is not None:
                self.streamreader.errors = errors
        except AttributeError as e:
            # TODO: py3 only exception?
            pass
        
        self._errors = errors
    errors = property(_geterrors, _seterrors)


if hasattr(codecs, "CodecInfo"):
    # We're running on Python 2.5 or better
    def search_function(name):
        if name == "css":
            return codecs.CodecInfo(
                name="css",
                encode=encode,
                decode=decode,
                incrementalencoder=IncrementalEncoder,
                incrementaldecoder=IncrementalDecoder,
                streamwriter=StreamWriter,
                streamreader=StreamReader,
            )
else:
    # If we're running on Python 2.4, define the utf-8-sig codec here
    def utf8sig_encode(input, errors='strict'):
        return (codecs.BOM_UTF8 + codecs.utf_8_encode(input, errors)[0], len(input))

    def utf8sig_decode(input, errors='strict'):
        prefix = 0
        if input[:3] == codecs.BOM_UTF8:
            input = input[3:]
            prefix = 3
        (output, consumed) = codecs.utf_8_decode(input, errors, True)
        return (output, consumed+prefix)

    class UTF8SigStreamWriter(codecs.StreamWriter):
        def reset(self):
            codecs.StreamWriter.reset(self)
            try:
                del self.encode
            except AttributeError:
                pass

        def encode(self, input, errors='strict'):
            self.encode = codecs.utf_8_encode
            return utf8sig_encode(input, errors)

    class UTF8SigStreamReader(codecs.StreamReader):
        def reset(self):
            codecs.StreamReader.reset(self)
            try:
                del self.decode
            except AttributeError:
                pass

        def decode(self, input, errors='strict'):
            if len(input) < 3 and codecs.BOM_UTF8.startswith(input):
                # not enough data to decide if this is a BOM
                # => try again on the next call
                return ("", 0)
            self.decode = codecs.utf_8_decode
            return utf8sig_decode(input, errors)

    def search_function(name):
        import encodings
        name = encodings.normalize_encoding(name)
        if name == "css":
            return (encode, decode, StreamReader, StreamWriter)
        elif name == "utf_8_sig":
            return (utf8sig_encode, utf8sig_decode, UTF8SigStreamReader, UTF8SigStreamWriter)


codecs.register(search_function)


# Error handler for CSS escaping

def cssescape(exc):
    if not isinstance(exc, UnicodeEncodeError):
        raise TypeError("don't know how to handle %r" % exc)
    return ("".join("\\%06x" % ord(c) for c in exc.object[exc.start:exc.end]), exc.end)

codecs.register_error("cssescape", cssescape)