🌐 AI搜索 & 代理 主页
Skip to content

Commit 5d493a5

Browse files
rubysrubys
committed
Update sanitizer to html5; accommodate usage as a filter
1 parent a71989b commit 5d493a5

File tree

1 file changed

+87
-59
lines changed

1 file changed

+87
-59
lines changed

‎src/html5lib/sanitizer.py‎

Lines changed: 87 additions & 59 deletions
Original file line numberDiff line numberDiff line change
@@ -7,15 +7,19 @@
77
class HTMLSanitizerMixin(object):
88
""" sanitization of XHTML+MathML+SVG and of inline style attributes."""
99

10-
acceptable_elements = ['a', 'abbr', 'acronym', 'address', 'area', 'b',
11-
'big', 'blockquote', 'br', 'button', 'caption', 'center', 'cite',
12-
'code', 'col', 'colgroup', 'dd', 'del', 'dfn', 'dir', 'div', 'dl', 'dt',
13-
'em', 'fieldset', 'font', 'form', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
14-
'hr', 'i', 'img', 'input', 'ins', 'kbd', 'label', 'legend', 'li', 'map',
15-
'menu', 'ol', 'optgroup', 'option', 'p', 'pre', 'q', 's', 'samp',
16-
'select', 'small', 'span', 'strike', 'strong', 'sub', 'sup', 'table',
17-
'tbody', 'td', 'textarea', 'tfoot', 'th', 'thead', 'tr', 'tt', 'u',
18-
'ul', 'var']
10+
acceptable_elements = ['a', 'abbr', 'acronym', 'address', 'area',
11+
'article', 'aside', 'audio', 'b', 'big', 'blockquote', 'br', 'button',
12+
'canvas', 'caption', 'center', 'cite', 'code', 'col', 'colgroup',
13+
'command', 'datagrid', 'datalist', 'dd', 'del', 'details', 'dfn',
14+
'dialog', 'dir', 'div', 'dl', 'dt', 'em', 'event-source', 'fieldset',
15+
'figure', 'footer', 'font', 'form', 'header', 'h1', 'h2', 'h3', 'h4',
16+
'h5', 'h6', 'hr', 'i', 'img', 'input', 'ins', 'keygen', 'kbd',
17+
'label', 'legend', 'li', 'm', 'map', 'menu', 'meter', 'multicol',
18+
'nav', 'nextid', 'ol', 'output', 'optgroup', 'option', 'p', 'pre',
19+
'progress', 'q', 's', 'samp', 'section', 'select', 'small', 'sound',
20+
'source', 'spacer', 'span', 'strike', 'strong', 'sub', 'sup', 'table',
21+
'tbody', 'td', 'textarea', 'time', 'tfoot', 'th', 'thead', 'tr', 'tt',
22+
'u', 'ul', 'var', 'video']
1923

2024
mathml_elements = ['maction', 'math', 'merror', 'mfrac', 'mi',
2125
'mmultiscripts', 'mn', 'mo', 'mover', 'mpadded', 'mphantom',
@@ -24,24 +28,35 @@ class HTMLSanitizerMixin(object):
2428
'munderover', 'none']
2529

2630
svg_elements = ['a', 'animate', 'animateColor', 'animateMotion',
27-
'animateTransform', 'circle', 'defs', 'desc', 'ellipse', 'font-face',
28-
'font-face-name', 'font-face-src', 'g', 'glyph', 'hkern',
31+
'animateTransform', 'clipPath', 'circle', 'defs', 'desc', 'ellipse',
32+
'font-face', 'font-face-name', 'font-face-src', 'g', 'glyph', 'hkern',
2933
'linearGradient', 'line', 'marker', 'metadata', 'missing-glyph',
3034
'mpath', 'path', 'polygon', 'polyline', 'radialGradient', 'rect',
3135
'set', 'stop', 'svg', 'switch', 'text', 'title', 'tspan', 'use']
3236

3337
acceptable_attributes = ['abbr', 'accept', 'accept-charset', 'accesskey',
34-
'action', 'align', 'alt', 'axis', 'border', 'cellpadding',
35-
'cellspacing', 'char', 'charoff', 'charset', 'checked', 'cite', 'class',
36-
'clear', 'cols', 'colspan', 'color', 'compact', 'coords', 'datetime',
37-
'dir', 'disabled', 'enctype', 'for', 'frame', 'headers', 'height',
38-
'href', 'hreflang', 'hspace', 'id', 'ismap', 'label', 'lang',
39-
'longdesc', 'maxlength', 'media', 'method', 'multiple', 'name',
40-
'nohref', 'noshade', 'nowrap', 'prompt', 'readonly', 'rel', 'rev',
41-
'rows', 'rowspan', 'rules', 'scope', 'selected', 'shape', 'size',
42-
'span', 'src', 'start', 'style', 'summary', 'tabindex', 'target',
43-
'title', 'type', 'usemap', 'valign', 'value', 'vspace', 'width',
44-
'xml:lang']
38+
'action', 'align', 'alt', 'autocomplete', 'autofocus', 'axis',
39+
'background', 'balance', 'bgcolor', 'bgproperties', 'border',
40+
'bordercolor', 'bordercolordark', 'bordercolorlight', 'bottompadding',
41+
'cellpadding', 'cellspacing', 'ch', 'challenge', 'char', 'charoff',
42+
'choff', 'charset', 'checked', 'cite', 'class', 'clear', 'color',
43+
'cols', 'colspan', 'compact', 'contenteditable', 'controls', 'coords',
44+
'data', 'datafld', 'datapagesize', 'datasrc', 'datetime', 'default',
45+
'delay', 'dir', 'disabled', 'draggable', 'dynsrc', 'enctype', 'end',
46+
'face', 'for', 'form', 'frame', 'galleryimg', 'gutter', 'headers',
47+
'height', 'hidefocus', 'hidden', 'high', 'href', 'hreflang', 'hspace',
48+
'icon', 'id', 'inputmode', 'ismap', 'keytype', 'label', 'leftspacing',
49+
'lang', 'list', 'longdesc', 'loop', 'loopcount', 'loopend',
50+
'loopstart', 'low', 'lowsrc', 'max', 'maxlength', 'media', 'method',
51+
'min', 'multiple', 'name', 'nohref', 'noshade', 'nowrap', 'open',
52+
'optimum', 'pattern', 'ping', 'point-size', 'prompt', 'pqg',
53+
'radiogroup', 'readonly', 'rel', 'repeat-max', 'repeat-min',
54+
'replace', 'required', 'rev', 'rightspacing', 'rows', 'rowspan',
55+
'rules', 'scope', 'selected', 'shape', 'size', 'span', 'src', 'start',
56+
'step', 'style', 'summary', 'suppress', 'tabindex', 'target',
57+
'template', 'title', 'toppadding', 'type', 'unselectable', 'usemap',
58+
'urn', 'valign', 'value', 'variable', 'volume', 'vspace', 'vrml',
59+
'width', 'wrap', 'xml:lang']
4560

4661
mathml_attributes = ['actiontype', 'align', 'columnalign', 'columnalign',
4762
'columnalign', 'columnlines', 'columnspacing', 'columnspan', 'depth',
@@ -54,43 +69,45 @@ class HTMLSanitizerMixin(object):
5469
'xlink:type', 'xmlns', 'xmlns:xlink']
5570

5671
svg_attributes = ['accent-height', 'accumulate', 'additive', 'alphabetic',
57-
'arabic-form', 'ascent', 'attributeName', 'attributeType',
58-
'baseProfile', 'bbox', 'begin', 'by', 'calcMode', 'cap-height',
59-
'class', 'color', 'color-rendering', 'content', 'cx', 'cy', 'd', 'dx',
60-
'dy', 'descent', 'display', 'dur', 'end', 'fill', 'fill-opacity',
61-
'fill-rule', 'font-family', 'font-size', 'font-stretch', 'font-style',
62-
'font-variant', 'font-weight', 'from', 'fx', 'fy', 'g1', 'g2',
63-
'glyph-name', 'gradientUnits', 'hanging', 'height', 'horiz-adv-x',
64-
'horiz-origin-x', 'id', 'ideographic', 'k', 'keyPoints',
65-
'keySplines', 'keyTimes', 'lang', 'marker-end', 'marker-mid',
66-
'marker-start', 'markerHeight', 'markerUnits', 'markerWidth',
67-
'mathematical', 'max', 'min', 'name', 'offset', 'opacity', 'orient',
68-
'origin', 'overline-position', 'overline-thickness', 'panose-1',
69-
'path', 'pathLength', 'points', 'preserveAspectRatio', 'r', 'refX',
70-
'refY', 'repeatCount', 'repeatDur', 'requiredExtensions',
71-
'requiredFeatures', 'restart', 'rotate', 'rx', 'ry', 'slope',
72-
'stemh', 'stemv', 'stop-color', 'stop-opacity',
73-
'strikethrough-position', 'strikethrough-thickness', 'stroke',
74-
'stroke-dasharray', 'stroke-dashoffset', 'stroke-linecap',
75-
'stroke-linejoin', 'stroke-miterlimit', 'stroke-opacity',
76-
'stroke-width', 'systemLanguage', 'target', 'text-anchor', 'to',
77-
'transform', 'type', 'u1', 'u2', 'underline-position',
78-
'underline-thickness', 'unicode', 'unicode-range', 'units-per-em',
79-
'values', 'version', 'viewBox', 'visibility', 'width', 'widths', 'x',
80-
'x-height', 'x1', 'x2', 'xlink:actuate', 'xlink:arcrole',
81-
'xlink:href', 'xlink:role', 'xlink:show', 'xlink:title',
82-
'xlink:type', 'xml:base', 'xml:lang', 'xml:space', 'xmlns',
83-
'xmlns:xlink', 'y', 'y1', 'y2', 'zoomAndPan']
72+
'arabic-form', 'ascent', 'attributeName', 'attributeType',
73+
'baseProfile', 'bbox', 'begin', 'by', 'calcMode', 'cap-height',
74+
'class', 'clip-path', 'color', 'color-rendering', 'content', 'cx',
75+
'cy', 'd', 'dx', 'dy', 'descent', 'display', 'dur', 'end', 'fill',
76+
'fill-opacity', 'fill-rule', 'font-family', 'font-size',
77+
'font-stretch', 'font-style', 'font-variant', 'font-weight', 'from',
78+
'fx', 'fy', 'g1', 'g2', 'glyph-name', 'gradientUnits', 'hanging',
79+
'height', 'horiz-adv-x', 'horiz-origin-x', 'id', 'ideographic', 'k',
80+
'keyPoints', 'keySplines', 'keyTimes', 'lang', 'marker-end',
81+
'marker-mid', 'marker-start', 'markerHeight', 'markerUnits',
82+
'markerWidth', 'mathematical', 'max', 'min', 'name', 'offset',
83+
'opacity', 'orient', 'origin', 'overline-position',
84+
'overline-thickness', 'panose-1', 'path', 'pathLength', 'points',
85+
'preserveAspectRatio', 'r', 'refX', 'refY', 'repeatCount',
86+
'repeatDur', 'requiredExtensions', 'requiredFeatures', 'restart',
87+
'rotate', 'rx', 'ry', 'slope', 'stemh', 'stemv', 'stop-color',
88+
'stop-opacity', 'strikethrough-position', 'strikethrough-thickness',
89+
'stroke', 'stroke-dasharray', 'stroke-dashoffset', 'stroke-linecap',
90+
'stroke-linejoin', 'stroke-miterlimit', 'stroke-opacity',
91+
'stroke-width', 'systemLanguage', 'target', 'text-anchor', 'to',
92+
'transform', 'type', 'u1', 'u2', 'underline-position',
93+
'underline-thickness', 'unicode', 'unicode-range', 'units-per-em',
94+
'values', 'version', 'viewBox', 'visibility', 'width', 'widths', 'x',
95+
'x-height', 'x1', 'x2', 'xlink:actuate', 'xlink:arcrole',
96+
'xlink:href', 'xlink:role', 'xlink:show', 'xlink:title', 'xlink:type',
97+
'xml:base', 'xml:lang', 'xml:space', 'xmlns', 'xmlns:xlink', 'y',
98+
'y1', 'y2', 'zoomAndPan']
8499

85100
attr_val_is_uri = ['href', 'src', 'cite', 'action', 'longdesc',
86-
'xlink:href', 'xml:base']
101+
'xlink:href', 'xml:base']
87102

88103
svg_attr_val_allows_ref = ['clip-path', 'color-profile', 'cursor', 'fill',
89-
'filter', 'marker', 'marker-start', 'marker-mid', 'marker-end', 'mask', 'stroke']
104+
'filter', 'marker', 'marker-start', 'marker-mid', 'marker-end',
105+
'mask', 'stroke']
90106

91-
svg_allow_local_href = ['altGlyph', 'animate', 'animateColor', 'animateMotion',
92-
'animateTransform', 'cursor', 'feImage', 'filter', 'linearGradient', 'pattern',
93-
'radialGradient', 'textpath', 'tref', 'set', 'use']
107+
svg_allow_local_href = ['altGlyph', 'animate', 'animateColor',
108+
'animateMotion', 'animateTransform', 'cursor', 'feImage', 'filter',
109+
'linearGradient', 'pattern', 'radialGradient', 'textpath', 'tref',
110+
'set', 'use']
94111

95112
acceptable_css_properties = ['azimuth', 'background-color',
96113
'border-bottom-color', 'border-collapse', 'border-color',
@@ -140,7 +157,13 @@ class HTMLSanitizerMixin(object):
140157
# sanitize_html('<a href="javascript: sucker();">Click here for $100</a>')
141158
# => <a>Click here for $100</a>
142159
def sanitize_token(self, token):
143-
if token["type"] in (tokenTypes["StartTag"], tokenTypes["EndTag"],
160+
161+
# accommodate filters which use token_type differently
162+
token_type = token["type"]
163+
if token_type in tokenTypes.keys():
164+
token_type = tokenTypes[token_type]
165+
166+
if token_type in (tokenTypes["StartTag"], tokenTypes["EndTag"],
144167
tokenTypes["EmptyTag"]):
145168
if token["name"] in self.allowed_elements:
146169
if token.has_key("data"):
@@ -172,19 +195,24 @@ def sanitize_token(self, token):
172195
token["data"] = [[name,val] for name,val in attrs.items()]
173196
return token
174197
else:
175-
if token["type"] == tokenTypes["EndTag"]:
198+
if token_type == tokenTypes["EndTag"]:
176199
token["data"] = "</%s>" % token["name"]
177200
elif token["data"]:
178201
attrs = ''.join([' %s="%s"' % (k,escape(v)) for k,v in token["data"]])
179202
token["data"] = "<%s%s>" % (token["name"],attrs)
180203
else:
181204
token["data"] = "<%s>" % token["name"]
182-
if token["selfClosing"]:
205+
if token.get("selfClosing"):
183206
token["data"]=token["data"][:-1] + "/>"
184-
token["type"] = tokenTypes["Characters"]
207+
208+
if token["type"] in tokenTypes.keys():
209+
token["type"] = "Characters"
210+
else:
211+
token["type"] = tokenTypes["Characters"]
212+
185213
del token["name"]
186214
return token
187-
elif token["type"] == tokenTypes["Comment"]:
215+
elif token_type == tokenTypes["Comment"]:
188216
pass
189217
else:
190218
return token

0 commit comments

Comments
 (0)