guppy_basecaller: env/lib/python3.7/site-packages/bleach/sanitizer.py annotate

annotate env/lib/python3.7/site-packages/bleach/sanitizer.py @ 3:758bc20232e8 draft

"planemo upload commit 2a0fe2cc28b09e101d37293e53e82f61762262ec"

author	shellac
date	Thu, 14 May 2020 16:20:52 -0400
parents	26e78fe6e8c4
children

rev	line source
0 26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	1 from __future__ import unicode_literals
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	2
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	3 from itertools import chain
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	4 import re
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	5
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	6 import six
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	7 from six.moves.urllib.parse import urlparse
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	8 from xml.sax.saxutils import unescape
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	9
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	10 from bleach import html5lib_shim
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	11 from bleach.utils import alphabetize_attributes, force_unicode
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	12
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	13
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	14 #: List of allowed tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	15 ALLOWED_TAGS = [
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	16 'a',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	17 'abbr',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	18 'acronym',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	19 'b',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	20 'blockquote',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	21 'code',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	22 'em',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	23 'i',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	24 'li',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	25 'ol',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	26 'strong',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	27 'ul',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	28 ]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	29
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	30
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	31 #: Map of allowed attributes by tag
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	32 ALLOWED_ATTRIBUTES = {
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	33 'a': ['href', 'title'],
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	34 'abbr': ['title'],
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	35 'acronym': ['title'],
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	36 }
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	37
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	38 #: List of allowed styles
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	39 ALLOWED_STYLES = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	40
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	41 #: List of allowed protocols
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	42 ALLOWED_PROTOCOLS = ['http', 'https', 'mailto']
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	43
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	44 #: Invisible characters--0 to and including 31 except 9 (tab), 10 (lf), and 13 (cr)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	45 INVISIBLE_CHARACTERS = ''.join([chr(c) for c in chain(range(0, 9), range(11, 13), range(14, 32))])
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	46
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	47 #: Regexp for characters that are invisible
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	48 INVISIBLE_CHARACTERS_RE = re.compile(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	49 '[' + INVISIBLE_CHARACTERS + ']',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	50 re.UNICODE
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	51 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	52
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	53 #: String to replace invisible characters with. This can be a character, a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	54 #: string, or even a function that takes a Python re matchobj
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	55 INVISIBLE_REPLACEMENT_CHAR = '?'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	56
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	57
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	58 class Cleaner(object):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	59 """Cleaner for cleaning HTML fragments of malicious content
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	60
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	61 This cleaner is a security-focused function whose sole purpose is to remove
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	62 malicious content from a string such that it can be displayed as content in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	63 a web page.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	64
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	65 To use::
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	66
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	67 from bleach.sanitizer import Cleaner
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	68
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	69 cleaner = Cleaner()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	70
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	71 for text in all_the_yucky_things:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	72 sanitized = cleaner.clean(text)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	73
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	74 .. Note::
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	75
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	76 This cleaner is not designed to use to transform content to be used in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	77 non-web-page contexts.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	78
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	79 .. Warning::
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	80
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	81 This cleaner is not thread-safe--the html parser has internal state.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	82 Create a separate cleaner per thread!
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	83
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	84
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	85 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	86
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	87 def __init__(self, tags=ALLOWED_TAGS, attributes=ALLOWED_ATTRIBUTES,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	88 styles=ALLOWED_STYLES, protocols=ALLOWED_PROTOCOLS, strip=False,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	89 strip_comments=True, filters=None):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	90 """Initializes a Cleaner
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	91
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	92 :arg list tags: allowed list of tags; defaults to
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	93 ``bleach.sanitizer.ALLOWED_TAGS``
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	94
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	95 :arg dict attributes: allowed attributes; can be a callable, list or dict;
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	96 defaults to ``bleach.sanitizer.ALLOWED_ATTRIBUTES``
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	97
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	98 :arg list styles: allowed list of css styles; defaults to
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	99 ``bleach.sanitizer.ALLOWED_STYLES``
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	100
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	101 :arg list protocols: allowed list of protocols for links; defaults
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	102 to ``bleach.sanitizer.ALLOWED_PROTOCOLS``
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	103
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	104 :arg bool strip: whether or not to strip disallowed elements
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	105
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	106 :arg bool strip_comments: whether or not to strip HTML comments
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	107
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	108 :arg list filters: list of html5lib Filter classes to pass streamed content through
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	109
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	110 .. seealso:: http://html5lib.readthedocs.io/en/latest/movingparts.html#filters
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	111
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	112 .. Warning::
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	113
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	114 Using filters changes the output of ``bleach.Cleaner.clean``.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	115 Make sure the way the filters change the output are secure.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	116
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	117 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	118 self.tags = tags
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	119 self.attributes = attributes
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	120 self.styles = styles
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	121 self.protocols = protocols
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	122 self.strip = strip
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	123 self.strip_comments = strip_comments
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	124 self.filters = filters or []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	125
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	126 self.parser = html5lib_shim.BleachHTMLParser(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	127 tags=self.tags,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	128 strip=self.strip,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	129 consume_entities=False,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	130 namespaceHTMLElements=False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	131 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	132 self.walker = html5lib_shim.getTreeWalker('etree')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	133 self.serializer = html5lib_shim.BleachHTMLSerializer(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	134 quote_attr_values='always',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	135 omit_optional_tags=False,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	136 escape_lt_in_attrs=True,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	137
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	138 # We want to leave entities as they are without escaping or
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	139 # resolving or expanding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	140 resolve_entities=False,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	141
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	142 # Bleach has its own sanitizer, so don't use the html5lib one
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	143 sanitize=False,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	144
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	145 # Bleach sanitizer alphabetizes already, so don't use the html5lib one
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	146 alphabetical_attributes=False,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	147 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	148
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	149 def clean(self, text):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	150 """Cleans text and returns sanitized result as unicode
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	151
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	152 :arg str text: text to be cleaned
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	153
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	154 :returns: sanitized text as unicode
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	155
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	156 :raises TypeError: if ``text`` is not a text type
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	157
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	158 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	159 if not isinstance(text, six.string_types):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	160 message = "argument cannot be of '{name}' type, must be of text type".format(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	161 name=text.__class__.__name__)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	162 raise TypeError(message)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	163
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	164 if not text:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	165 return ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	166
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	167 text = force_unicode(text)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	168
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	169 dom = self.parser.parseFragment(text)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	170 filtered = BleachSanitizerFilter(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	171 source=self.walker(dom),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	172
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	173 # Bleach-sanitizer-specific things
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	174 attributes=self.attributes,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	175 strip_disallowed_elements=self.strip,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	176 strip_html_comments=self.strip_comments,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	177
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	178 # html5lib-sanitizer things
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	179 allowed_elements=self.tags,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	180 allowed_css_properties=self.styles,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	181 allowed_protocols=self.protocols,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	182 allowed_svg_properties=[],
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	183 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	184
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	185 # Apply any filters after the BleachSanitizerFilter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	186 for filter_class in self.filters:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	187 filtered = filter_class(source=filtered)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	188
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	189 return self.serializer.render(filtered)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	190
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	191
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	192 def attribute_filter_factory(attributes):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	193 """Generates attribute filter function for the given attributes value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	194
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	195 The attributes value can take one of several shapes. This returns a filter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	196 function appropriate to the attributes value. One nice thing about this is
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	197 that there's less if/then shenanigans in the ``allow_token`` method.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	198
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	199 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	200 if callable(attributes):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	201 return attributes
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	202
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	203 if isinstance(attributes, dict):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	204 def _attr_filter(tag, attr, value):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	205 if tag in attributes:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	206 attr_val = attributes[tag]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	207 if callable(attr_val):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	208 return attr_val(tag, attr, value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	209
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	210 if attr in attr_val:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	211 return True
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	212
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	213 if '*' in attributes:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	214 attr_val = attributes['*']
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	215 if callable(attr_val):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	216 return attr_val(tag, attr, value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	217
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	218 return attr in attr_val
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	219
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	220 return False
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	221
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	222 return _attr_filter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	223
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	224 if isinstance(attributes, list):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	225 def _attr_filter(tag, attr, value):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	226 return attr in attributes
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	227
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	228 return _attr_filter
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	229
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	230 raise ValueError('attributes needs to be a callable, a list or a dict')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	231
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	232
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	233 class BleachSanitizerFilter(html5lib_shim.SanitizerFilter):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	234 """html5lib Filter that sanitizes text
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	235
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	236 This filter can be used anywhere html5lib filters can be used.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	237
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	238 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	239 def __init__(self, source, attributes=ALLOWED_ATTRIBUTES,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	240 strip_disallowed_elements=False, strip_html_comments=True,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	241 **kwargs):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	242 """Creates a BleachSanitizerFilter instance
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	243
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	244 :arg Treewalker source: stream
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	245
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	246 :arg list tags: allowed list of tags; defaults to
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	247 ``bleach.sanitizer.ALLOWED_TAGS``
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	248
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	249 :arg dict attributes: allowed attributes; can be a callable, list or dict;
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	250 defaults to ``bleach.sanitizer.ALLOWED_ATTRIBUTES``
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	251
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	252 :arg list styles: allowed list of css styles; defaults to
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	253 ``bleach.sanitizer.ALLOWED_STYLES``
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	254
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	255 :arg list protocols: allowed list of protocols for links; defaults
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	256 to ``bleach.sanitizer.ALLOWED_PROTOCOLS``
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	257
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	258 :arg bool strip_disallowed_elements: whether or not to strip disallowed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	259 elements
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	260
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	261 :arg bool strip_html_comments: whether or not to strip HTML comments
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	262
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	263 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	264 self.attr_filter = attribute_filter_factory(attributes)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	265 self.strip_disallowed_elements = strip_disallowed_elements
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	266 self.strip_html_comments = strip_html_comments
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	267
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	268 return super(BleachSanitizerFilter, self).__init__(source, **kwargs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	269
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	270 def sanitize_stream(self, token_iterator):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	271 for token in token_iterator:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	272 ret = self.sanitize_token(token)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	273
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	274 if not ret:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	275 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	276
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	277 if isinstance(ret, list):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	278 for subtoken in ret:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	279 yield subtoken
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	280 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	281 yield ret
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	282
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	283 def merge_characters(self, token_iterator):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	284 """Merge consecutive Characters tokens in a stream"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	285 characters_buffer = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	286
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	287 for token in token_iterator:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	288 if characters_buffer:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	289 if token['type'] == 'Characters':
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	290 characters_buffer.append(token)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	291 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	292 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	293 # Merge all the characters tokens together into one and then
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	294 # operate on it.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	295 new_token = {
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	296 'data': ''.join([char_token['data'] for char_token in characters_buffer]),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	297 'type': 'Characters'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	298 }
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	299 characters_buffer = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	300 yield new_token
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	301
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	302 elif token['type'] == 'Characters':
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	303 characters_buffer.append(token)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	304 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	305
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	306 yield token
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	307
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	308 new_token = {
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	309 'data': ''.join([char_token['data'] for char_token in characters_buffer]),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	310 'type': 'Characters'
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	311 }
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	312 yield new_token
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	313
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	314 def __iter__(self):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	315 return self.merge_characters(self.sanitize_stream(html5lib_shim.Filter.__iter__(self)))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	316
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	317 def sanitize_token(self, token):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	318 """Sanitize a token either by HTML-encoding or dropping.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	319
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	320 Unlike sanitizer.Filter, allowed_attributes can be a dict of {'tag':
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	321 ['attribute', 'pairs'], 'tag': callable}.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	322
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	323 Here callable is a function with two arguments of attribute name and
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	324 value. It should return true of false.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	325
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	326 Also gives the option to strip tags instead of encoding.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	327
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	328 :arg dict token: token to sanitize
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	329
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	330 :returns: token or list of tokens
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	331
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	332 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	333 token_type = token['type']
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	334 if token_type in ['StartTag', 'EndTag', 'EmptyTag']:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	335 if token['name'] in self.allowed_elements:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	336 return self.allow_token(token)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	337
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	338 elif self.strip_disallowed_elements:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	339 return None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	340
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	341 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	342 if 'data' in token:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	343 # Alphabetize the attributes before calling .disallowed_token()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	344 # so that the resulting string is stable
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	345 token['data'] = alphabetize_attributes(token['data'])
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	346 return self.disallowed_token(token)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	347
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	348 elif token_type == 'Comment':
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	349 if not self.strip_html_comments:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	350 return token
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	351 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	352 return None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	353
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	354 elif token_type == 'Characters':
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	355 return self.sanitize_characters(token)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	356
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	357 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	358 return token
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	359
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	360 def sanitize_characters(self, token):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	361 """Handles Characters tokens
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	362
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	363 Our overridden tokenizer doesn't do anything with entities. However,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	364 that means that the serializer will convert all ``&`` in Characters
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	365 tokens to ``&``.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	366
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	367 Since we don't want that, we extract entities here and convert them to
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	368 Entity tokens so the serializer will let them be.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	369
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	370 :arg token: the Characters token to work on
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	371
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	372 :returns: a list of tokens
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	373
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	374 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	375 data = token.get('data', '')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	376
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	377 if not data:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	378 return token
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	379
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	380 data = INVISIBLE_CHARACTERS_RE.sub(INVISIBLE_REPLACEMENT_CHAR, data)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	381 token['data'] = data
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	382
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	383 # If there isn't a & in the data, we can return now
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	384 if '&' not in data:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	385 return token
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	386
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	387 new_tokens = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	388
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	389 # For each possible entity that starts with a "&", we try to extract an
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	390 # actual entity and re-tokenize accordingly
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	391 for part in html5lib_shim.next_possible_entity(data):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	392 if not part:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	393 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	394
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	395 if part.startswith('&'):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	396 entity = html5lib_shim.match_entity(part)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	397 if entity is not None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	398 if entity == 'amp':
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	399 # LinkifyFilter can't match urls across token boundaries
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	400 # which is problematic with & since that shows up in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	401 # querystrings all the time. This special-cases &
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	402 # and converts it to a & and sticks it in as a
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	403 # Characters token. It'll get merged with surrounding
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	404 # tokens in the BleachSanitizerfilter.__iter__ and
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	405 # escaped in the serializer.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	406 new_tokens.append({'type': 'Characters', 'data': '&'})
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	407 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	408 new_tokens.append({'type': 'Entity', 'name': entity})
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	409
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	410 # Length of the entity plus 2--one for & at the beginning
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	411 # and one for ; at the end
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	412 remainder = part[len(entity) + 2:]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	413 if remainder:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	414 new_tokens.append({'type': 'Characters', 'data': remainder})
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	415 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	416
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	417 new_tokens.append({'type': 'Characters', 'data': part})
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	418
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	419 return new_tokens
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	420
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	421 def sanitize_uri_value(self, value, allowed_protocols):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	422 """Checks a uri value to see if it's allowed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	423
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	424 :arg value: the uri value to sanitize
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	425 :arg allowed_protocols: list of allowed protocols
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	426
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	427 :returns: allowed value or None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	428
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	429 """
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	430 # NOTE(willkg): This transforms the value into one that's easier to
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	431 # match and verify, but shouldn't get returned since it's vastly
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	432 # different than the original value.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	433
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	434 # Convert all character entities in the value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	435 new_value = html5lib_shim.convert_entities(value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	436
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	437 # Nix backtick, space characters, and control characters
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	438 new_value = re.sub(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	439 r"[`\000-\040\177-\240\s]+",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	440 '',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	441 new_value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	442 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	443
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	444 # Remove REPLACEMENT characters
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	445 new_value = new_value.replace('\ufffd', '')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	446
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	447 # Lowercase it--this breaks the value, but makes it easier to match
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	448 # against
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	449 new_value = new_value.lower()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	450
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	451 try:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	452 # Drop attributes with uri values that have protocols that aren't
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	453 # allowed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	454 parsed = urlparse(new_value)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	455 except ValueError:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	456 # URI is impossible to parse, therefore it's not allowed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	457 return None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	458
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	459 if parsed.scheme:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	460 # If urlparse found a scheme, check that
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	461 if parsed.scheme in allowed_protocols:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	462 return value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	463
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	464 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	465 # Allow uris that are just an anchor
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	466 if new_value.startswith('#'):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	467 return value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	468
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	469 # Handle protocols that urlparse doesn't recognize like "myprotocol"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	470 if ':' in new_value and new_value.split(':')[0] in allowed_protocols:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	471 return value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	472
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	473 # If there's no protocol/scheme specified, then assume it's "http"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	474 # and see if that's allowed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	475 if 'http' in allowed_protocols:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	476 return value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	477
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	478 return None
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	479
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	480 def allow_token(self, token):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	481 """Handles the case where we're allowing the tag"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	482 if 'data' in token:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	483 # Loop through all the attributes and drop the ones that are not
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	484 # allowed, are unsafe or break other rules. Additionally, fix
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	485 # attribute values that need fixing.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	486 #
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	487 # At the end of this loop, we have the final set of attributes
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	488 # we're keeping.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	489 attrs = {}
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	490 for namespaced_name, val in token['data'].items():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	491 namespace, name = namespaced_name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	492
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	493 # Drop attributes that are not explicitly allowed
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	494 #
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	495 # NOTE(willkg): We pass in the attribute name--not a namespaced
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	496 # name.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	497 if not self.attr_filter(token['name'], name, val):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	498 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	499
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	500 # Drop attributes with uri values that use a disallowed protocol
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	501 # Sanitize attributes with uri values
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	502 if namespaced_name in self.attr_val_is_uri:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	503 new_value = self.sanitize_uri_value(val, self.allowed_protocols)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	504 if new_value is None:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	505 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	506 val = new_value
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	507
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	508 # Drop values in svg attrs with non-local IRIs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	509 if namespaced_name in self.svg_attr_val_allows_ref:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	510 new_val = re.sub(r'url\s$\s[^#\s][^)]+?$',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	511 ' ',
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	512 unescape(val))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	513 new_val = new_val.strip()
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	514 if not new_val:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	515 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	516
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	517 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	518 # Replace the val with the unescaped version because
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	519 # it's a iri
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	520 val = new_val
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	521
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	522 # Drop href and xlink:href attr for svg elements with non-local IRIs
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	523 if (None, token['name']) in self.svg_allow_local_href:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	524 if namespaced_name in [
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	525 (None, 'href'), (html5lib_shim.namespaces['xlink'], 'href')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	526 ]:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	527 if re.search(r'^\s*[^#\s]', val):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	528 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	529
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	530 # If it's a style attribute, sanitize it
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	531 if namespaced_name == (None, 'style'):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	532 val = self.sanitize_css(val)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	533
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	534 # At this point, we want to keep the attribute, so add it in
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	535 attrs[namespaced_name] = val
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	536
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	537 token['data'] = alphabetize_attributes(attrs)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	538
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	539 return token
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	540
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	541 def disallowed_token(self, token):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	542 token_type = token["type"]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	543 if token_type == "EndTag":
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	544 token["data"] = "</%s>" % token["name"]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	545
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	546 elif token["data"]:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	547 assert token_type in ("StartTag", "EmptyTag")
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	548 attrs = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	549 for (ns, name), v in token["data"].items():
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	550 # If we end up with a namespace, but no name, switch them so we
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	551 # have a valid name to use.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	552 if ns and not name:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	553 ns, name = name, ns
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	554
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	555 # Figure out namespaced name if the namespace is appropriate
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	556 # and exists; if the ns isn't in prefixes, then drop it.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	557 if ns is None or ns not in html5lib_shim.prefixes:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	558 namespaced_name = name
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	559 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	560 namespaced_name = '%s:%s' % (html5lib_shim.prefixes[ns], name)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	561
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	562 attrs.append(' %s="%s"' % (
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	563 namespaced_name,
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	564 # NOTE(willkg): HTMLSerializer escapes attribute values
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	565 # already, so if we do it here (like HTMLSerializer does),
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	566 # then we end up double-escaping.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	567 v)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	568 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	569 token["data"] = "<%s%s>" % (token["name"], ''.join(attrs))
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	570
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	571 else:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	572 token["data"] = "<%s>" % token["name"]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	573
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	574 if token.get("selfClosing"):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	575 token["data"] = token["data"][:-1] + "/>"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	576
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	577 token["type"] = "Characters"
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	578
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	579 del token["name"]
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	580 return token
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	581
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	582 def sanitize_css(self, style):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	583 """Sanitizes css in style tags"""
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	584 # Convert entities in the style so that it can be parsed as CSS
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	585 style = html5lib_shim.convert_entities(style)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	586
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	587 # Drop any url values before we do anything else
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	588 style = re.compile(r'url\s$\s[^\s)]+?\s$\s').sub(' ', style)
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	589
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	590 # The gauntlet of sanitization
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	591
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	592 # Validate the css in the style tag and if it's not valid, then drop
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	593 # the whole thing.
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	594 parts = style.split(';')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	595 gauntlet = re.compile(
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	596 r"""^( # consider a style attribute value as composed of:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	597 [/:,#%!.\s\w] # a non-newline character
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	598 \|\w-\w # 3 characters in the form \w-\w
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	599 \|'[\s\w]+'\s* # a single quoted string of [\s\w]+ with trailing space
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	600 \|"[\s\w]+" # a double quoted string of [\s\w]+
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	601 \|$[\d,%\.\s]+$ # a parenthesized string of one or more digits, commas, periods, percent signs, or whitespace e.g. from 'color: hsl(30,100%,50%)''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	602 )*$""",
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	603 flags=re.U \| re.VERBOSE
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	604 )
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	605
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	606 for part in parts:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	607 if not gauntlet.match(part):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	608 return ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	609
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	610 if not re.match(r"^\s([-\w]+\s:[^:;](;\s\|$))*$", style):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	611 return ''
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	612
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	613 clean = []
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	614 for prop, value in re.findall(r'([-\w]+)\s:\s([^:;]*)', style):
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	615 if not value:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	616 continue
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	617
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	618 if prop.lower() in self.allowed_css_properties:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	619 clean.append(prop + ': ' + value + ';')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	620
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	621 elif prop.lower() in self.allowed_svg_properties:
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	622 clean.append(prop + ': ' + value + ';')
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	623
26e78fe6e8c4 "planemo upload commit c699937486c35866861690329de38ec1a5d9f783" shellac parents: diff changeset	624 return ' '.join(clean)

Mercurial > repos > shellac > guppy_basecaller

annotate env/lib/python3.7/site-packages/bleach/sanitizer.py @ 3:758bc20232e8 draft