Use bleach to sanitize HTML; allows use with newer versions of html5lib.

This fixes #296
This commit is contained in:
Jelmer Vernooij 2017-03-02 01:22:58 +00:00 committed by gloomy-ghost
parent 8e37a88d6d
commit f1a4478272
4 changed files with 16 additions and 48 deletions

View File

@ -59,7 +59,6 @@ class TestHTML(unittest.TestCase):
print("Hello, World")
</code></pre>""")
@unittest.skipIf(html.HTML5LIB_VERSION <= html.HTML5LIB_SIMPLETREE, "backport")
def test_sanitizer(self):
sanitizer = html.Sanitizer(elements=[], attributes=[])
examples = [
@ -73,11 +72,10 @@ class TestHTML(unittest.TestCase):
for (input, expected) in examples:
if isinstance(expected, list):
self.assertIn(html.sanitize(sanitizer, input), expected)
self.assertIn(sanitizer.sanitize(input), expected)
else:
self.assertEqual(html.sanitize(sanitizer, input), expected)
self.assertEqual(sanitizer.sanitize(input), expected)
@unittest.skipIf(html.HTML5LIB_VERSION <= html.HTML5LIB_SIMPLETREE, "backport")
def test_sanitizer_extensions(self):
sanitizer = html.Sanitizer(elements=["img"], attributes=["src"])
examples = [
@ -85,7 +83,7 @@ class TestHTML(unittest.TestCase):
('<script src="doge.js"></script>', '')]
for (input, expected) in examples:
self.assertEqual(html.sanitize(sanitizer, input), expected)
self.assertEqual(sanitizer.sanitize(input), expected)
def test_render(self):
conf = config.new({

View File

@ -6,61 +6,30 @@ import pkg_resources
from distutils.version import LooseVersion as Version
HTML5LIB_VERSION = Version(pkg_resources.get_distribution("html5lib").version)
HTML5LIB_SIMPLETREE = Version("0.95")
import html5lib
from html5lib.sanitizer import HTMLSanitizer
from html5lib.serializer import HTMLSerializer
import bleach
import misaka
def Sanitizer(elements, attributes):
class Inner(HTMLSanitizer):
class Sanitizer(object):
def __init__(self, elements, attributes):
# attributes found in Sundown's HTML serializer [1]
# except for <img> tag,
# because images are not generated anyways.
#
# [1] https://github.com/vmg/sundown/blob/master/html/html.c
allowed_elements = ["a", "p", "hr", "br", "ol", "ul", "li",
self.elements = ["a", "p", "hr", "br", "ol", "ul", "li",
"pre", "code", "blockquote",
"del", "ins", "strong", "em",
"h1", "h2", "h3", "h4", "h5", "h6",
"table", "thead", "tbody", "th", "td"] + elements
# href for <a> and align for <table>
allowed_attributes = ["align", "href"] + attributes
self.attributes = ["align", "href"] + attributes
# remove disallowed tokens from the output
def disallowed_token(self, token, token_type):
return None
return Inner
def sanitize(tokenizer, document):
parser = html5lib.HTMLParser(tokenizer=tokenizer)
domtree = parser.parseFragment(document)
if HTML5LIB_VERSION > HTML5LIB_SIMPLETREE:
builder = "etree"
for link in domtree.findall(".//{http://www.w3.org/1999/xhtml}a"):
if link.get('href', None):
link.set("rel", "nofollow noopener")
else:
builder = "simpletree"
stream = html5lib.treewalkers.getTreeWalker(builder)(domtree)
serializer = HTMLSerializer(
quote_attr_values=True, omit_optional_tags=False)
return serializer.render(stream)
def sanitize(self, text):
return bleach.clean(text, tags=self.elements,
attributes=self.attributes, strip=True)
def Markdown(extensions=("strikethrough", "superscript", "autolink",
@ -100,7 +69,7 @@ class Markup(object):
conf.getlist("allowed-elements"),
conf.getlist("allowed-attributes"))
self._render = lambda text: sanitize(sanitizer, parser(text))
self._render = lambda text: sanitizer.sanitize(parser(text))
def render(self, text):
return self._render(text)

View File

@ -5,8 +5,8 @@ import sys
from setuptools import setup, find_packages
requires = ['itsdangerous', 'Jinja2', 'misaka>=2.0,<3.0', 'html5lib<0.9999999',
'werkzeug>=0.9']
requires = ['itsdangerous', 'Jinja2', 'misaka>=2.0,<3.0', 'html5lib',
'werkzeug>=0.9', 'bleach']
if sys.version_info < (2, 7):
raise SystemExit("Python 2 versions < 2.7 are not supported.")

View File

@ -15,7 +15,8 @@ deps =
[testenv:debian]
deps=
html5lib==0.95
bleach
html5lib
ipaddr==2.1.10
itsdangerous==0.22
misaka==1.0.2