| 1 |
# coding: utf8 |
| 2 |
""" |
| 3 |
tinycss.decoding |
| 4 |
---------------- |
| 5 |
|
| 6 |
Decoding stylesheets from bytes to Unicode. |
| 7 |
http://www.w3.org/TR/CSS21/syndata.html#charset |
| 8 |
|
| 9 |
:copyright: (c) 2012 by Simon Sapin. |
| 10 |
:license: BSD, see LICENSE for more details. |
| 11 |
""" |
| 12 |
|
| 13 |
|
| 14 |
import operator |
| 15 |
import re |
| 16 |
|
| 17 |
from polyglot.binary import from_hex_bytes |
| 18 |
|
| 19 |
__all__ = ['decode'] # Everything else is implementation detail |
| 20 |
|
| 21 |
|
| 22 |
def decode(css_bytes, protocol_encoding=None, |
| 23 |
linking_encoding=None, document_encoding=None): |
| 24 |
""" |
| 25 |
Determine the character encoding from the passed metadata and the |
| 26 |
``@charset`` rule in the stylesheet (if any); and decode accordingly. |
| 27 |
If no encoding information is available or decoding fails, |
| 28 |
decoding defaults to UTF-8 and then fall back on ISO-8859-1. |
| 29 |
|
| 30 |
:param css_bytes: |
| 31 |
a CSS stylesheet as a byte string |
| 32 |
:param protocol_encoding: |
| 33 |
The "charset" parameter of a "Content-Type" HTTP header (if any), |
| 34 |
or similar metadata for other protocols. |
| 35 |
:param linking_encoding: |
| 36 |
``<link charset="">`` or other metadata from the linking mechanism |
| 37 |
(if any) |
| 38 |
:param document_encoding: |
| 39 |
Encoding of the referring style sheet or document (if any) |
| 40 |
:return: |
| 41 |
A tuple of an Unicode string, with any BOM removed, and the |
| 42 |
encoding that was used. |
| 43 |
|
| 44 |
""" |
| 45 |
if protocol_encoding: |
| 46 |
css_unicode = try_encoding(css_bytes, protocol_encoding) |
| 47 |
if css_unicode is not None: |
| 48 |
return css_unicode, protocol_encoding |
| 49 |
for encoding, pattern in ENCODING_MAGIC_NUMBERS: |
| 50 |
match = pattern(css_bytes) |
| 51 |
if match: |
| 52 |
has_at_charset = isinstance(encoding, tuple) |
| 53 |
if has_at_charset: |
| 54 |
extract, endianness = encoding |
| 55 |
encoding = extract(match.group(1)) |
| 56 |
# Get an ASCII-only unicode value. |
| 57 |
# This is the only thing that works on both Python 2 and 3 |
| 58 |
# for bytes.decode() |
| 59 |
# Non-ASCII encoding names are invalid anyway, |
| 60 |
# but make sure they stay invalid. |
| 61 |
encoding = encoding.decode('ascii', 'replace') |
| 62 |
encoding = encoding.replace('\ufffd', '?') |
| 63 |
if encoding.replace('-', '').replace('_', '').lower() in [ |
| 64 |
'utf16', 'utf32']: |
| 65 |
encoding += endianness |
| 66 |
encoding = encoding.encode('ascii', 'replace').decode('ascii') |
| 67 |
css_unicode = try_encoding(css_bytes, encoding) |
| 68 |
if css_unicode and not (has_at_charset and not |
| 69 |
css_unicode.startswith('@charset "')): |
| 70 |
return css_unicode, encoding |
| 71 |
break |
| 72 |
for encoding in [linking_encoding, document_encoding]: |
| 73 |
if encoding: |
| 74 |
css_unicode = try_encoding(css_bytes, encoding) |
| 75 |
if css_unicode is not None: |
| 76 |
return css_unicode, encoding |
| 77 |
css_unicode = try_encoding(css_bytes, 'UTF-8') |
| 78 |
if css_unicode is not None: |
| 79 |
return css_unicode, 'UTF-8' |
| 80 |
return try_encoding(css_bytes, 'ISO-8859-1', fallback=False), 'ISO-8859-1' |
| 81 |
|
| 82 |
|
| 83 |
def try_encoding(css_bytes, encoding, fallback=True): |
| 84 |
if fallback: |
| 85 |
try: |
| 86 |
css_unicode = css_bytes.decode(encoding) |
| 87 |
# LookupError means unknown encoding |
| 88 |
except (UnicodeDecodeError, LookupError): |
| 89 |
return None |
| 90 |
else: |
| 91 |
css_unicode = css_bytes.decode(encoding) |
| 92 |
if css_unicode and css_unicode[0] == '\ufeff': |
| 93 |
# Remove any Byte Order Mark |
| 94 |
css_unicode = css_unicode[1:] |
| 95 |
return css_unicode |
| 96 |
|
| 97 |
|
| 98 |
def hex2re(hex_data): |
| 99 |
return re.escape(from_hex_bytes(hex_data.replace(' ', '').encode('ascii'))) |
| 100 |
|
| 101 |
|
| 102 |
class Slicer: |
| 103 |
"""Slice()[start:stop:end] == slice(start, stop, end)""" |
| 104 |
def __getitem__(self, slice_): |
| 105 |
return operator.itemgetter(slice_) |
| 106 |
|
| 107 |
|
| 108 |
Slice = Slicer() |
| 109 |
|
| 110 |
|
| 111 |
# List of (bom_size, encoding, pattern) |
| 112 |
# bom_size is in bytes and can be zero |
| 113 |
# encoding is a string or (slice_, endianness) for "as specified" |
| 114 |
# slice_ is a slice object.How to extract the specified |
| 115 |
|
| 116 |
ENCODING_MAGIC_NUMBERS = [ |
| 117 |
((Slice[:], ''), re.compile( |
| 118 |
hex2re('EF BB BF 40 63 68 61 72 73 65 74 20 22') |
| 119 |
+ b'([^\x22]*?)' |
| 120 |
+ hex2re('22 3B')).match), |
| 121 |
|
| 122 |
('UTF-8', re.compile( |
| 123 |
hex2re('EF BB BF')).match), |
| 124 |
|
| 125 |
((Slice[:], ''), re.compile( |
| 126 |
hex2re('40 63 68 61 72 73 65 74 20 22') |
| 127 |
+ b'([^\x22]*?)' |
| 128 |
+ hex2re('22 3B')).match), |
| 129 |
|
| 130 |
((Slice[1::2], '-BE'), re.compile( |
| 131 |
hex2re('FE FF 00 40 00 63 00 68 00 61 00 72 00 73 00 65 00' |
| 132 |
'74 00 20 00 22') |
| 133 |
+ b'((\x00[^\x22])*?)' |
| 134 |
+ hex2re('00 22 00 3B')).match), |
| 135 |
|
| 136 |
((Slice[1::2], '-BE'), re.compile( |
| 137 |
hex2re('00 40 00 63 00 68 00 61 00 72 00 73 00 65 00 74 00' |
| 138 |
'20 00 22') |
| 139 |
+ b'((\x00[^\x22])*?)' |
| 140 |
+ hex2re('00 22 00 3B')).match), |
| 141 |
|
| 142 |
((Slice[::2], '-LE'), re.compile( |
| 143 |
hex2re('FF FE 40 00 63 00 68 00 61 00 72 00 73 00 65 00 74' |
| 144 |
'00 20 00 22 00') |
| 145 |
+ b'(([^\x22]\x00)*?)' |
| 146 |
+ hex2re('22 00 3B 00')).match), |
| 147 |
|
| 148 |
((Slice[::2], '-LE'), re.compile( |
| 149 |
hex2re('40 00 63 00 68 00 61 00 72 00 73 00 65 00 74 00 20' |
| 150 |
'00 22 00') |
| 151 |
+ b'(([^\x22]\x00)*?)' |
| 152 |
+ hex2re('22 00 3B 00')).match), |
| 153 |
|
| 154 |
((Slice[3::4], '-BE'), re.compile( |
| 155 |
hex2re('00 00 FE FF 00 00 00 40 00 00 00 63 00 00 00 68 00' |
| 156 |
'00 00 61 00 00 00 72 00 00 00 73 00 00 00 65 00 00' |
| 157 |
'00 74 00 00 00 20 00 00 00 22') |
| 158 |
+ b'((\x00\x00\x00[^\x22])*?)' |
| 159 |
+ hex2re('00 00 00 22 00 00 00 3B')).match), |
| 160 |
|
| 161 |
((Slice[3::4], '-BE'), re.compile( |
| 162 |
hex2re('00 00 00 40 00 00 00 63 00 00 00 68 00 00 00 61 00' |
| 163 |
'00 00 72 00 00 00 73 00 00 00 65 00 00 00 74 00 00' |
| 164 |
'00 20 00 00 00 22') |
| 165 |
+ b'((\x00\x00\x00[^\x22])*?)' |
| 166 |
+ hex2re('00 00 00 22 00 00 00 3B')).match), |
| 167 |
|
| 168 |
|
| 169 |
# Python does not support 2143 or 3412 endianness, AFAIK. |
| 170 |
# I guess we could fix it up ourselves but meh. Patches welcome. |
| 171 |
|
| 172 |
# ((Slice[2::4], '-2143'), re.compile( |
| 173 |
# hex2re('00 00 FF FE 00 00 40 00 00 00 63 00 00 00 68 00 00' |
| 174 |
# '00 61 00 00 00 72 00 00 00 73 00 00 00 65 00 00 00' |
| 175 |
# '74 00 00 00 20 00 00 00 22 00') |
| 176 |
# + b'((\x00\x00[^\x22]\x00)*?)' |
| 177 |
# + hex2re('00 00 22 00 00 00 3B 00')).match), |
| 178 |
|
| 179 |
# ((Slice[2::4], '-2143'), re.compile( |
| 180 |
# hex2re('00 00 40 00 00 00 63 00 00 00 68 00 00 00 61 00 00' |
| 181 |
# '00 72 00 00 00 73 00 00 00 65 00 00 00 74 00 00 00' |
| 182 |
# '20 00 00 00 22 00') |
| 183 |
# + b'((\x00\x00[^\x22]\x00)*?)' |
| 184 |
# + hex2re('00 00 22 00 00 00 3B 00')).match), |
| 185 |
|
| 186 |
# ((Slice[1::4], '-3412'), re.compile( |
| 187 |
# hex2re('FE FF 00 00 00 40 00 00 00 63 00 00 00 68 00 00 00' |
| 188 |
# '61 00 00 00 72 00 00 00 73 00 00 00 65 00 00 00 74' |
| 189 |
# '00 00 00 20 00 00 00 22 00 00') |
| 190 |
# + b'((\x00[^\x22]\x00\x00)*?)' |
| 191 |
# + hex2re('00 22 00 00 00 3B 00 00')).match), |
| 192 |
|
| 193 |
# ((Slice[1::4], '-3412'), re.compile( |
| 194 |
# hex2re('00 40 00 00 00 63 00 00 00 68 00 00 00 61 00 00 00' |
| 195 |
# '72 00 00 00 73 00 00 00 65 00 00 00 74 00 00 00 20' |
| 196 |
# '00 00 00 22 00 00') |
| 197 |
# + b'((\x00[^\x22]\x00\x00)*?)' |
| 198 |
# + hex2re('00 22 00 00 00 3B 00 00')).match), |
| 199 |
|
| 200 |
((Slice[::4], '-LE'), re.compile( |
| 201 |
hex2re('FF FE 00 00 40 00 00 00 63 00 00 00 68 00 00 00 61' |
| 202 |
'00 00 00 72 00 00 00 73 00 00 00 65 00 00 00 74 00' |
| 203 |
'00 00 20 00 00 00 22 00 00 00') |
| 204 |
+ b'(([^\x22]\x00\x00\x00)*?)' |
| 205 |
+ hex2re('22 00 00 00 3B 00 00 00')).match), |
| 206 |
|
| 207 |
((Slice[::4], '-LE'), re.compile( |
| 208 |
hex2re('40 00 00 00 63 00 00 00 68 00 00 00 61 00 00 00 72' |
| 209 |
'00 00 00 73 00 00 00 65 00 00 00 74 00 00 00 20 00' |
| 210 |
'00 00 22 00 00 00') |
| 211 |
+ b'(([^\x22]\x00\x00\x00)*?)' |
| 212 |
+ hex2re('22 00 00 00 3B 00 00 00')).match), |
| 213 |
|
| 214 |
('UTF-32-BE', re.compile( |
| 215 |
hex2re('00 00 FE FF')).match), |
| 216 |
|
| 217 |
('UTF-32-LE', re.compile( |
| 218 |
hex2re('FF FE 00 00')).match), |
| 219 |
|
| 220 |
# ('UTF-32-2143', re.compile( |
| 221 |
# hex2re('00 00 FF FE')).match), |
| 222 |
|
| 223 |
# ('UTF-32-3412', re.compile( |
| 224 |
# hex2re('FE FF 00 00')).match), |
| 225 |
|
| 226 |
('UTF-16-BE', re.compile( |
| 227 |
hex2re('FE FF')).match), |
| 228 |
|
| 229 |
('UTF-16-LE', re.compile( |
| 230 |
hex2re('FF FE')).match), |
| 231 |
|
| 232 |
|
| 233 |
# Some of there are supported by Python, but I didn’t bother. |
| 234 |
# You know the story with patches ... |
| 235 |
|
| 236 |
# # as specified, transcoded from EBCDIC to ASCII |
| 237 |
# ('as_specified-EBCDIC', re.compile( |
| 238 |
# hex2re('7C 83 88 81 99 A2 85 A3 40 7F') |
| 239 |
# + b'([^\x7F]*?)' |
| 240 |
# + hex2re('7F 5E')).match), |
| 241 |
|
| 242 |
# # as specified, transcoded from IBM1026 to ASCII |
| 243 |
# ('as_specified-IBM1026', re.compile( |
| 244 |
# hex2re('AE 83 88 81 99 A2 85 A3 40 FC') |
| 245 |
# + b'([^\xFC]*?)' |
| 246 |
# + hex2re('FC 5E')).match), |
| 247 |
|
| 248 |
# # as specified, transcoded from GSM 03.38 to ASCII |
| 249 |
# ('as_specified-GSM_03.38', re.compile( |
| 250 |
# hex2re('00 63 68 61 72 73 65 74 20 22') |
| 251 |
# + b'([^\x22]*?)' |
| 252 |
# + hex2re('22 3B')).match), |
| 253 |
] |