| 1 |
# coding: utf8 |
| 2 |
""" |
| 3 |
tinycss.tokenizer |
| 4 |
----------------- |
| 5 |
|
| 6 |
Tokenizer for the CSS core syntax: |
| 7 |
http://www.w3.org/TR/CSS21/syndata.html#tokenization |
| 8 |
|
| 9 |
This is the pure-python implementation. See also speedups.pyx |
| 10 |
|
| 11 |
:copyright: (c) 2012 by Simon Sapin. |
| 12 |
:license: BSD, see LICENSE for more details. |
| 13 |
""" |
| 14 |
|
| 15 |
|
| 16 |
from tinycss import token_data |
| 17 |
|
| 18 |
|
| 19 |
def tokenize_flat(css_source, ignore_comments=True, |
| 20 |
# Make these local variable to avoid global lookups in the loop |
| 21 |
tokens_dispatch=token_data.TOKEN_DISPATCH, |
| 22 |
unicode_unescape=token_data.UNICODE_UNESCAPE, |
| 23 |
newline_unescape=token_data.NEWLINE_UNESCAPE, |
| 24 |
simple_unescape=token_data.SIMPLE_UNESCAPE, |
| 25 |
find_newlines=token_data.FIND_NEWLINES, |
| 26 |
Token=token_data.Token, |
| 27 |
len=len, |
| 28 |
int=int, |
| 29 |
float=float, |
| 30 |
list=list, |
| 31 |
_None=None, |
| 32 |
): |
| 33 |
""" |
| 34 |
:param css_source: |
| 35 |
CSS as an unicode string |
| 36 |
:param ignore_comments: |
| 37 |
if true (the default) comments will not be included in the |
| 38 |
return value |
| 39 |
:return: |
| 40 |
An iterator of :class:`Token` |
| 41 |
|
| 42 |
""" |
| 43 |
|
| 44 |
pos = 0 |
| 45 |
line = 1 |
| 46 |
column = 1 |
| 47 |
source_len = len(css_source) |
| 48 |
tokens = [] |
| 49 |
while pos < source_len: |
| 50 |
char = css_source[pos] |
| 51 |
if char in ':;{}()[]': |
| 52 |
type_ = char |
| 53 |
css_value = char |
| 54 |
else: |
| 55 |
codepoint = min(ord(char), 160) |
| 56 |
for _index, type_, regexp in tokens_dispatch[codepoint]: |
| 57 |
match = regexp(css_source, pos) |
| 58 |
if match is not None: |
| 59 |
# First match is the longest. See comments on TOKENS above. |
| 60 |
css_value = match.group() |
| 61 |
break |
| 62 |
else: |
| 63 |
# No match. |
| 64 |
# "Any other character not matched by the above rules, |
| 65 |
# and neither a single nor a double quote." |
| 66 |
# ... but quotes at the start of a token are always matched |
| 67 |
# by STRING or BAD_STRING. So DELIM is any single character. |
| 68 |
type_ = 'DELIM' |
| 69 |
css_value = char |
| 70 |
length = len(css_value) |
| 71 |
next_pos = pos + length |
| 72 |
|
| 73 |
# A BAD_COMMENT is a comment at EOF. Ignore it too. |
| 74 |
if not (ignore_comments and type_ in ('COMMENT', 'BAD_COMMENT')): |
| 75 |
# Parse numbers, extract strings and URIs, unescape |
| 76 |
unit = _None |
| 77 |
if type_ == 'DIMENSION': |
| 78 |
value = match.group(1) |
| 79 |
value = float(value) if '.' in value else int(value) |
| 80 |
unit = match.group(2) |
| 81 |
unit = simple_unescape(unit) |
| 82 |
unit = unicode_unescape(unit) |
| 83 |
unit = unit.lower() # normalize |
| 84 |
elif type_ == 'PERCENTAGE': |
| 85 |
value = css_value[:-1] |
| 86 |
value = float(value) if '.' in value else int(value) |
| 87 |
unit = '%' |
| 88 |
elif type_ == 'NUMBER': |
| 89 |
value = css_value |
| 90 |
if '.' in value: |
| 91 |
value = float(value) |
| 92 |
else: |
| 93 |
value = int(value) |
| 94 |
type_ = 'INTEGER' |
| 95 |
elif type_ in ('IDENT', 'ATKEYWORD', 'HASH', 'FUNCTION'): |
| 96 |
value = simple_unescape(css_value) |
| 97 |
value = unicode_unescape(value) |
| 98 |
elif type_ == 'URI': |
| 99 |
value = match.group(1) |
| 100 |
if value and value[0] in '"\'': |
| 101 |
value = value[1:-1] # Remove quotes |
| 102 |
value = newline_unescape(value) |
| 103 |
value = simple_unescape(value) |
| 104 |
value = unicode_unescape(value) |
| 105 |
elif type_ == 'STRING': |
| 106 |
value = css_value[1:-1] # Remove quotes |
| 107 |
value = newline_unescape(value) |
| 108 |
value = simple_unescape(value) |
| 109 |
value = unicode_unescape(value) |
| 110 |
# BAD_STRING can only be one of: |
| 111 |
# * Unclosed string at the end of the stylesheet: |
| 112 |
# Close the string, but this is not an error. |
| 113 |
# Make it a "good" STRING token. |
| 114 |
# * Unclosed string at the (unescaped) end of the line: |
| 115 |
# Close the string, but this is an error. |
| 116 |
# Leave it as a BAD_STRING, don’t bother parsing it. |
| 117 |
# See http://www.w3.org/TR/CSS21/syndata.html#parsing-errors |
| 118 |
elif type_ == 'BAD_STRING' and next_pos == source_len: |
| 119 |
type_ = 'STRING' |
| 120 |
value = css_value[1:] # Remove quote |
| 121 |
value = newline_unescape(value) |
| 122 |
value = simple_unescape(value) |
| 123 |
value = unicode_unescape(value) |
| 124 |
else: |
| 125 |
value = css_value |
| 126 |
tokens.append(Token(type_, css_value, value, unit, line, column)) |
| 127 |
|
| 128 |
pos = next_pos |
| 129 |
newlines = find_newlines(css_value) |
| 130 |
if newlines: |
| 131 |
line += len(newlines) |
| 132 |
# Add 1 to have lines start at column 1, not 0 |
| 133 |
column = length - newlines[-1].end() + 1 |
| 134 |
else: |
| 135 |
column += length |
| 136 |
return tokens |
| 137 |
|
| 138 |
|
| 139 |
def regroup(tokens): |
| 140 |
""" |
| 141 |
Match pairs of tokens: () [] {} function() |
| 142 |
(Strings in "" or '' are taken care of by the tokenizer.) |
| 143 |
|
| 144 |
Opening tokens are replaced by a :class:`ContainerToken`. |
| 145 |
Closing tokens are removed. Unmatched closing tokens are invalid |
| 146 |
but left as-is. All nested structures that are still open at |
| 147 |
the end of the stylesheet are implicitly closed. |
| 148 |
|
| 149 |
:param tokens: |
| 150 |
a *flat* iterable of tokens, as returned by :func:`tokenize_flat`. |
| 151 |
:return: |
| 152 |
A tree of tokens. |
| 153 |
|
| 154 |
""" |
| 155 |
# "global" objects for the inner recursion |
| 156 |
pairs = {'FUNCTION': ')', '(': ')', '[': ']', '{': '}'} |
| 157 |
tokens = iter(tokens) |
| 158 |
eof = [False] |
| 159 |
|
| 160 |
def _regroup_inner(stop_at=None, |
| 161 |
tokens=tokens, pairs=pairs, eof=eof, |
| 162 |
ContainerToken=token_data.ContainerToken, |
| 163 |
FunctionToken=token_data.FunctionToken): |
| 164 |
for token in tokens: |
| 165 |
type_ = token.type |
| 166 |
if type_ == stop_at: |
| 167 |
return |
| 168 |
|
| 169 |
end = pairs.get(type_) |
| 170 |
if end is None: |
| 171 |
yield token # Not a grouping token |
| 172 |
else: |
| 173 |
assert not isinstance(token, ContainerToken), ( |
| 174 |
'Token looks already grouped: {0}'.format(token)) |
| 175 |
content = list(_regroup_inner(end)) |
| 176 |
if eof[0]: |
| 177 |
end = '' # Implicit end of structure at EOF. |
| 178 |
if type_ == 'FUNCTION': |
| 179 |
yield FunctionToken(token.type, token.as_css(), end, |
| 180 |
token.value, content, |
| 181 |
token.line, token.column) |
| 182 |
else: |
| 183 |
yield ContainerToken(token.type, token.as_css(), end, |
| 184 |
content, |
| 185 |
token.line, token.column) |
| 186 |
else: |
| 187 |
eof[0] = True # end of file/stylesheet |
| 188 |
return _regroup_inner() |
| 189 |
|
| 190 |
|
| 191 |
def tokenize_grouped(css_source, ignore_comments=True): |
| 192 |
""" |
| 193 |
:param css_source: |
| 194 |
CSS as an unicode string |
| 195 |
:param ignore_comments: |
| 196 |
if true (the default) comments will not be included in the |
| 197 |
return value |
| 198 |
:return: |
| 199 |
An iterator of :class:`Token` |
| 200 |
|
| 201 |
""" |
| 202 |
return regroup(tokenize_flat(css_source, ignore_comments)) |
| 203 |
|
| 204 |
|
| 205 |
# Optional Cython version of tokenize_flat |
| 206 |
# Make both versions available with explicit names for tests. |
| 207 |
python_tokenize_flat = tokenize_flat |
| 208 |
|
| 209 |
try: |
| 210 |
tok = token_data.load_c_tokenizer() |
| 211 |
except (ImportError, RuntimeError): |
| 212 |
c_tokenize_flat = None |
| 213 |
else: |
| 214 |
# Use the c tokenizer by default |
| 215 |
c_tokenize_flat = tokenize_flat = lambda s, ignore_comments=False:tok.tokenize_flat(s, ignore_comments) # type: ignore |