Skip to content

Commit 85256cf

Browse files
gsneddersgsnedders
committed
Fix html5lib#72: Move the sanitizer to purely be a filter.
This drops support for the tokenizing side of thing, which is sadly the only side that works in previous releases.
1 parent fdcdea8 commit 85256cf

File tree

3 files changed

+260
-273
lines changed

3 files changed

+260
-273
lines changed

‎CHANGES.rst

+8
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,14 @@
11
Change Log
22
----------
33

4+
1.0
5+
~~~
6+
7+
Released on September XX, 2013
8+
9+
* Sanitizer change XXX
10+
11+
412
1.0b3
513
~~~~~
614

‎html5lib/filters/sanitizer.py

+252-2
Original file line numberDiff line numberDiff line change
@@ -1,12 +1,262 @@
11
from __future__ import absolute_import, division, unicode_literals
22

3+
import re
4+
from xml.sax.saxutils import escape, unescape
5+
36
from . import _base
4-
from ..sanitizer import HTMLSanitizerMixin
7+
from ..constants import tokenTypes
8+
9+
10+
class Filter(_base.Filter):
11+
""" sanitization of XHTML+MathML+SVG and of inline style attributes."""
12+
13+
acceptable_elements = ['a', 'abbr', 'acronym', 'address', 'area',
14+
'article', 'aside', 'audio', 'b', 'big', 'blockquote', 'br', 'button',
15+
'canvas', 'caption', 'center', 'cite', 'code', 'col', 'colgroup',
16+
'command', 'datagrid', 'datalist', 'dd', 'del', 'details', 'dfn',
17+
'dialog', 'dir', 'div', 'dl', 'dt', 'em', 'event-source', 'fieldset',
18+
'figcaption', 'figure', 'footer', 'font', 'form', 'header', 'h1',
19+
'h2', 'h3', 'h4', 'h5', 'h6', 'hr', 'i', 'img', 'input', 'ins',
20+
'keygen', 'kbd', 'label', 'legend', 'li', 'm', 'map', 'menu', 'meter',
21+
'multicol', 'nav', 'nextid', 'ol', 'output', 'optgroup', 'option',
22+
'p', 'pre', 'progress', 'q', 's', 'samp', 'section', 'select',
23+
'small', 'sound', 'source', 'spacer', 'span', 'strike', 'strong',
24+
'sub', 'sup', 'table', 'tbody', 'td', 'textarea', 'time', 'tfoot',
25+
'th', 'thead', 'tr', 'tt', 'u', 'ul', 'var', 'video']
26+
27+
mathml_elements = ['maction', 'math', 'merror', 'mfrac', 'mi',
28+
'mmultiscripts', 'mn', 'mo', 'mover', 'mpadded', 'mphantom',
29+
'mprescripts', 'mroot', 'mrow', 'mspace', 'msqrt', 'mstyle', 'msub',
30+
'msubsup', 'msup', 'mtable', 'mtd', 'mtext', 'mtr', 'munder',
31+
'munderover', 'none']
32+
33+
svg_elements = ['a', 'animate', 'animateColor', 'animateMotion',
34+
'animateTransform', 'clipPath', 'circle', 'defs', 'desc', 'ellipse',
35+
'font-face', 'font-face-name', 'font-face-src', 'g', 'glyph', 'hkern',
36+
'linearGradient', 'line', 'marker', 'metadata', 'missing-glyph',
37+
'mpath', 'path', 'polygon', 'polyline', 'radialGradient', 'rect',
38+
'set', 'stop', 'svg', 'switch', 'text', 'title', 'tspan', 'use']
39+
40+
acceptable_attributes = ['abbr', 'accept', 'accept-charset', 'accesskey',
41+
'action', 'align', 'alt', 'autocomplete', 'autofocus', 'axis',
42+
'background', 'balance', 'bgcolor', 'bgproperties', 'border',
43+
'bordercolor', 'bordercolordark', 'bordercolorlight', 'bottompadding',
44+
'cellpadding', 'cellspacing', 'ch', 'challenge', 'char', 'charoff',
45+
'choff', 'charset', 'checked', 'cite', 'class', 'clear', 'color',
46+
'cols', 'colspan', 'compact', 'contenteditable', 'controls', 'coords',
47+
'data', 'datafld', 'datapagesize', 'datasrc', 'datetime', 'default',
48+
'delay', 'dir', 'disabled', 'draggable', 'dynsrc', 'enctype', 'end',
49+
'face', 'for', 'form', 'frame', 'galleryimg', 'gutter', 'headers',
50+
'height', 'hidefocus', 'hidden', 'high', 'href', 'hreflang', 'hspace',
51+
'icon', 'id', 'inputmode', 'ismap', 'keytype', 'label', 'leftspacing',
52+
'lang', 'list', 'longdesc', 'loop', 'loopcount', 'loopend',
53+
'loopstart', 'low', 'lowsrc', 'max', 'maxlength', 'media', 'method',
54+
'min', 'multiple', 'name', 'nohref', 'noshade', 'nowrap', 'open',
55+
'optimum', 'pattern', 'ping', 'point-size', 'poster', 'pqg', 'preload',
56+
'prompt', 'radiogroup', 'readonly', 'rel', 'repeat-max', 'repeat-min',
57+
'replace', 'required', 'rev', 'rightspacing', 'rows', 'rowspan',
58+
'rules', 'scope', 'selected', 'shape', 'size', 'span', 'src', 'start',
59+
'step', 'style', 'summary', 'suppress', 'tabindex', 'target',
60+
'template', 'title', 'toppadding', 'type', 'unselectable', 'usemap',
61+
'urn', 'valign', 'value', 'variable', 'volume', 'vspace', 'vrml',
62+
'width', 'wrap', 'xml:lang']
63+
64+
mathml_attributes = ['actiontype', 'align', 'columnalign', 'columnalign',
65+
'columnalign', 'columnlines', 'columnspacing', 'columnspan', 'depth',
66+
'display', 'displaystyle', 'equalcolumns', 'equalrows', 'fence',
67+
'fontstyle', 'fontweight', 'frame', 'height', 'linethickness', 'lspace',
68+
'mathbackground', 'mathcolor', 'mathvariant', 'mathvariant', 'maxsize',
69+
'minsize', 'other', 'rowalign', 'rowalign', 'rowalign', 'rowlines',
70+
'rowspacing', 'rowspan', 'rspace', 'scriptlevel', 'selection',
71+
'separator', 'stretchy', 'width', 'width', 'xlink:href', 'xlink:show',
72+
'xlink:type', 'xmlns', 'xmlns:xlink']
73+
74+
svg_attributes = ['accent-height', 'accumulate', 'additive', 'alphabetic',
75+
'arabic-form', 'ascent', 'attributeName', 'attributeType',
76+
'baseProfile', 'bbox', 'begin', 'by', 'calcMode', 'cap-height',
77+
'class', 'clip-path', 'color', 'color-rendering', 'content', 'cx',
78+
'cy', 'd', 'dx', 'dy', 'descent', 'display', 'dur', 'end', 'fill',
79+
'fill-opacity', 'fill-rule', 'font-family', 'font-size',
80+
'font-stretch', 'font-style', 'font-variant', 'font-weight', 'from',
81+
'fx', 'fy', 'g1', 'g2', 'glyph-name', 'gradientUnits', 'hanging',
82+
'height', 'horiz-adv-x', 'horiz-origin-x', 'id', 'ideographic', 'k',
83+
'keyPoints', 'keySplines', 'keyTimes', 'lang', 'marker-end',
84+
'marker-mid', 'marker-start', 'markerHeight', 'markerUnits',
85+
'markerWidth', 'mathematical', 'max', 'min', 'name', 'offset',
86+
'opacity', 'orient', 'origin', 'overline-position',
87+
'overline-thickness', 'panose-1', 'path', 'pathLength', 'points',
88+
'preserveAspectRatio', 'r', 'refX', 'refY', 'repeatCount',
89+
'repeatDur', 'requiredExtensions', 'requiredFeatures', 'restart',
90+
'rotate', 'rx', 'ry', 'slope', 'stemh', 'stemv', 'stop-color',
91+
'stop-opacity', 'strikethrough-position', 'strikethrough-thickness',
92+
'stroke', 'stroke-dasharray', 'stroke-dashoffset', 'stroke-linecap',
93+
'stroke-linejoin', 'stroke-miterlimit', 'stroke-opacity',
94+
'stroke-width', 'systemLanguage', 'target', 'text-anchor', 'to',
95+
'transform', 'type', 'u1', 'u2', 'underline-position',
96+
'underline-thickness', 'unicode', 'unicode-range', 'units-per-em',
97+
'values', 'version', 'viewBox', 'visibility', 'width', 'widths', 'x',
98+
'x-height', 'x1', 'x2', 'xlink:actuate', 'xlink:arcrole',
99+
'xlink:href', 'xlink:role', 'xlink:show', 'xlink:title', 'xlink:type',
100+
'xml:base', 'xml:lang', 'xml:space', 'xmlns', 'xmlns:xlink', 'y',
101+
'y1', 'y2', 'zoomAndPan']
102+
103+
attr_val_is_uri = ['href', 'src', 'cite', 'action', 'longdesc', 'poster',
104+
'xlink:href', 'xml:base']
105+
106+
svg_attr_val_allows_ref = ['clip-path', 'color-profile', 'cursor', 'fill',
107+
'filter', 'marker', 'marker-start', 'marker-mid', 'marker-end',
108+
'mask', 'stroke']
109+
110+
svg_allow_local_href = ['altGlyph', 'animate', 'animateColor',
111+
'animateMotion', 'animateTransform', 'cursor', 'feImage', 'filter',
112+
'linearGradient', 'pattern', 'radialGradient', 'textpath', 'tref',
113+
'set', 'use']
114+
115+
acceptable_css_properties = ['azimuth', 'background-color',
116+
'border-bottom-color', 'border-collapse', 'border-color',
117+
'border-left-color', 'border-right-color', 'border-top-color', 'clear',
118+
'color', 'cursor', 'direction', 'display', 'elevation', 'float', 'font',
119+
'font-family', 'font-size', 'font-style', 'font-variant', 'font-weight',
120+
'height', 'letter-spacing', 'line-height', 'overflow', 'pause',
121+
'pause-after', 'pause-before', 'pitch', 'pitch-range', 'richness',
122+
'speak', 'speak-header', 'speak-numeral', 'speak-punctuation',
123+
'speech-rate', 'stress', 'text-align', 'text-decoration', 'text-indent',
124+
'unicode-bidi', 'vertical-align', 'voice-family', 'volume',
125+
'white-space', 'width']
126+
127+
acceptable_css_keywords = ['auto', 'aqua', 'black', 'block', 'blue',
128+
'bold', 'both', 'bottom', 'brown', 'center', 'collapse', 'dashed',
129+
'dotted', 'fuchsia', 'gray', 'green', '!important', 'italic', 'left',
130+
'lime', 'maroon', 'medium', 'none', 'navy', 'normal', 'nowrap', 'olive',
131+
'pointer', 'purple', 'red', 'right', 'solid', 'silver', 'teal', 'top',
132+
'transparent', 'underline', 'white', 'yellow']
5133

134+
acceptable_svg_properties = ['fill', 'fill-opacity', 'fill-rule',
135+
'stroke', 'stroke-width', 'stroke-linecap', 'stroke-linejoin',
136+
'stroke-opacity']
137+
138+
acceptable_protocols = ['ed2k', 'ftp', 'http', 'https', 'irc',
139+
'mailto', 'news', 'gopher', 'nntp', 'telnet', 'webcal',
140+
'xmpp', 'callto', 'feed', 'urn', 'aim', 'rsync', 'tag',
141+
'ssh', 'sftp', 'rtsp', 'afs']
142+
143+
# subclasses may define their own versions of these constants
144+
allowed_elements = acceptable_elements + mathml_elements + svg_elements
145+
allowed_attributes = acceptable_attributes + mathml_attributes + svg_attributes
146+
allowed_css_properties = acceptable_css_properties
147+
allowed_css_keywords = acceptable_css_keywords
148+
allowed_svg_properties = acceptable_svg_properties
149+
allowed_protocols = acceptable_protocols
6150

7-
class Filter(_base.Filter, HTMLSanitizerMixin):
8151
def __iter__(self):
9152
for token in _base.Filter.__iter__(self):
10153
token = self.sanitize_token(token)
11154
if token:
12155
yield token
156+
157+
# Sanitize the +html+, escaping all elements not in ALLOWED_ELEMENTS, and
158+
# stripping out all # attributes not in ALLOWED_ATTRIBUTES. Style
159+
# attributes are parsed, and a restricted set, # specified by
160+
# ALLOWED_CSS_PROPERTIES and ALLOWED_CSS_KEYWORDS, are allowed through.
161+
# attributes in ATTR_VAL_IS_URI are scanned, and only URI schemes specified
162+
# in ALLOWED_PROTOCOLS are allowed.
163+
#
164+
# sanitize_html('<script> do_nasty_stuff() </script>')
165+
# => &lt;script> do_nasty_stuff() &lt;/script>
166+
# sanitize_html('<a href="javascript: sucker();">Click here for $100</a>')
167+
# => <a>Click here for $100</a>
168+
def sanitize_token(self, token):
169+
170+
# accommodate filters which use token_type differently
171+
token_type = token["type"]
172+
if token_type in list(tokenTypes.keys()):
173+
token_type = tokenTypes[token_type]
174+
175+
if token_type in (tokenTypes["StartTag"], tokenTypes["EndTag"],
176+
tokenTypes["EmptyTag"]):
177+
if token["name"] in self.allowed_elements:
178+
return self.allowed_token(token, token_type)
179+
else:
180+
return self.disallowed_token(token, token_type)
181+
elif token_type == tokenTypes["Comment"]:
182+
pass
183+
else:
184+
return token
185+
186+
def allowed_token(self, token, token_type):
187+
if "data" in token:
188+
attrs = dict([(name, val) for name, val in
189+
token["data"][::-1]
190+
if name in self.allowed_attributes])
191+
for attr in self.attr_val_is_uri:
192+
if attr not in attrs:
193+
continue
194+
val_unescaped = re.sub("[`\000-\040\177-\240\s]+", '',
195+
unescape(attrs[attr])).lower()
196+
# remove replacement characters from unescaped characters
197+
val_unescaped = val_unescaped.replace("\ufffd", "")
198+
if (re.match("^[a-z0-9][-+.a-z0-9]*:", val_unescaped) and
199+
(val_unescaped.split(':')[0] not in
200+
self.allowed_protocols)):
201+
del attrs[attr]
202+
for attr in self.svg_attr_val_allows_ref:
203+
if attr in attrs:
204+
attrs[attr] = re.sub(r'url\s*\(\s*[^#\s][^)]+?\)',
205+
' ',
206+
unescape(attrs[attr]))
207+
if (token["name"] in self.svg_allow_local_href and
208+
'xlink:href' in attrs and re.search('^\s*[^#\s].*',
209+
attrs['xlink:href'])):
210+
del attrs['xlink:href']
211+
if 'style' in attrs:
212+
attrs['style'] = self.sanitize_css(attrs['style'])
213+
token["data"] = [[name, val] for name, val in list(attrs.items())]
214+
return token
215+
216+
def disallowed_token(self, token, token_type):
217+
if token_type == tokenTypes["EndTag"]:
218+
token["data"] = "</%s>" % token["name"]
219+
elif token["data"]:
220+
attrs = ''.join([' %s="%s"' % (k, escape(v)) for k, v in token["data"]])
221+
token["data"] = "<%s%s>" % (token["name"], attrs)
222+
else:
223+
token["data"] = "<%s>" % token["name"]
224+
if token.get("selfClosing"):
225+
token["data"] = token["data"][:-1] + "/>"
226+
227+
if token["type"] in list(tokenTypes.keys()):
228+
token["type"] = "Characters"
229+
else:
230+
token["type"] = tokenTypes["Characters"]
231+
232+
del token["name"]
233+
return token
234+
235+
def sanitize_css(self, style):
236+
# disallow urls
237+
style = re.compile('url\s*\(\s*[^\s)]+?\s*\)\s*').sub(' ', style)
238+
239+
# gauntlet
240+
if not re.match("""^([:,;#%.\sa-zA-Z0-9!]|\w-\w|'[\s\w]+'|"[\s\w]+"|\([\d,\s]+\))*$""", style):
241+
return ''
242+
if not re.match("^\s*([-\w]+\s*:[^:;]*(;\s*|$))*$", style):
243+
return ''
244+
245+
clean = []
246+
for prop, value in re.findall("([-\w]+)\s*:\s*([^:;]*)", style):
247+
if not value:
248+
continue
249+
if prop.lower() in self.allowed_css_properties:
250+
clean.append(prop + ': ' + value + ';')
251+
elif prop.split('-')[0].lower() in ['background', 'border', 'margin',
252+
'padding']:
253+
for keyword in value.split():
254+
if not keyword in self.acceptable_css_keywords and \
255+
not re.match("^(#[0-9a-f]+|rgb\(\d+%?,\d*%?,?\d*%?\)?|\d{0,2}\.?\d{0,2}(cm|em|ex|in|mm|pc|pt|px|%|,|\))?)$", keyword):
256+
break
257+
else:
258+
clean.append(prop + ': ' + value + ';')
259+
elif prop.lower() in self.allowed_svg_properties:
260+
clean.append(prop + ': ' + value + ';')
261+
262+
return ' '.join(clean)

0 commit comments

Comments
 (0)