Mercurial > repos > shellac > sam_consensus_v3

annotate env/lib/python3.9/site-packages/lxml/html/clean.py @ 0:4f3585e2f14b draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
author shellac
date Mon, 22 Mar 2021 18:12:50 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
1 # cython: language_level=3str
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
2
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
3 """A cleanup tool for HTML.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
4
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
5 Removes unwanted tags and content. See the `Cleaner` class for
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
6 details.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
7 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
8
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
9 from __future__ import absolute_import
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
10
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
11 import copy
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
12 import re
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
13 import sys
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
14 try:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
15 from urlparse import urlsplit
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
16 from urllib import unquote_plus
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
17 except ImportError:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
18 # Python 3
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
19 from urllib.parse import urlsplit, unquote_plus
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
20 from lxml import etree
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
21 from lxml.html import defs
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
22 from lxml.html import fromstring, XHTML_NAMESPACE
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
23 from lxml.html import xhtml_to_html, _transform_result
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
24
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
25 try:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
26 unichr
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
27 except NameError:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
28 # Python 3
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
29 unichr = chr
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
30 try:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
31 unicode
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
32 except NameError:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
33 # Python 3
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
34 unicode = str
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
35 try:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
36 basestring
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
37 except NameError:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
38 basestring = (str, bytes)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
39
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
40
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
41 __all__ = ['clean_html', 'clean', 'Cleaner', 'autolink', 'autolink_html',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
42 'word_break', 'word_break_html']
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
43
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
44 # Look at http://code.sixapart.com/trac/livejournal/browser/trunk/cgi-bin/cleanhtml.pl
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
45 # Particularly the CSS cleaning; most of the tag cleaning is integrated now
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
46 # I have multiple kinds of schemes searched; but should schemes be
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
47 # whitelisted instead?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
48 # max height?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
49 # remove images? Also in CSS? background attribute?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
50 # Some way to whitelist object, iframe, etc (e.g., if you want to
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
51 # allow *just* embedded YouTube movies)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
52 # Log what was deleted and why?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
53 # style="behavior: ..." might be bad in IE?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
54 # Should we have something for just <meta http-equiv>? That's the worst of the
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
55 # metas.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
56 # UTF-7 detections? Example:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
57 # <HEAD><META HTTP-EQUIV="CONTENT-TYPE" CONTENT="text/html; charset=UTF-7"> </HEAD>+ADw-SCRIPT+AD4-alert('XSS');+ADw-/SCRIPT+AD4-
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
58 # you don't always have to have the charset set, if the page has no charset
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
59 # and there's UTF7-like code in it.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
60 # Look at these tests: http://htmlpurifier.org/live/smoketests/xssAttacks.php
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
61
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
62
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
63 # This is an IE-specific construct you can have in a stylesheet to
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
64 # run some Javascript:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
65 _replace_css_javascript = re.compile(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
66 r'expression\s*\(.*?\)', re.S|re.I).sub
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
67
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
68 # Do I have to worry about @\nimport?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
69 _replace_css_import = re.compile(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
70 r'@\s*import', re.I).sub
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
71
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
72 _looks_like_tag_content = re.compile(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
73 r'</?[a-zA-Z]+|\son[a-zA-Z]+\s*=',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
74 *((re.ASCII,) if sys.version_info[0] >= 3 else ())).search
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
75
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
76 # All kinds of schemes besides just javascript: that can cause
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
77 # execution:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
78 _is_image_dataurl = re.compile(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
79 r'^data:image/.+;base64', re.I).search
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
80 _is_possibly_malicious_scheme = re.compile(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
81 r'(?:javascript|jscript|livescript|vbscript|data|about|mocha):',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
82 re.I).search
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
83 def _is_javascript_scheme(s):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
84 if _is_image_dataurl(s):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
85 return None
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
86 return _is_possibly_malicious_scheme(s)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
87
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
88 _substitute_whitespace = re.compile(r'[\s\x00-\x08\x0B\x0C\x0E-\x19]+').sub
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
89 # FIXME: should data: be blocked?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
90
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
91 # FIXME: check against: http://msdn2.microsoft.com/en-us/library/ms537512.aspx
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
92 _conditional_comment_re = re.compile(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
93 r'\[if[\s\n\r]+.*?][\s\n\r]*>', re.I|re.S)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
94
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
95 _find_styled_elements = etree.XPath(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
96 "descendant-or-self::*[@style]")
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
97
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
98 _find_external_links = etree.XPath(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
99 ("descendant-or-self::a [normalize-space(@href) and substring(normalize-space(@href),1,1) != '#'] |"
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
100 "descendant-or-self::x:a[normalize-space(@href) and substring(normalize-space(@href),1,1) != '#']"),
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
101 namespaces={'x':XHTML_NAMESPACE})
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
102
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
103
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
104 class Cleaner(object):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
105 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
106 Instances cleans the document of each of the possible offending
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
107 elements. The cleaning is controlled by attributes; you can
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
108 override attributes in a subclass, or set them in the constructor.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
109
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
110 ``scripts``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
111 Removes any ``<script>`` tags.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
112
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
113 ``javascript``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
114 Removes any Javascript, like an ``onclick`` attribute. Also removes stylesheets
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
115 as they could contain Javascript.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
116
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
117 ``comments``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
118 Removes any comments.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
119
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
120 ``style``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
121 Removes any style tags.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
122
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
123 ``inline_style``
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
124 Removes any style attributes. Defaults to the value of the ``style`` option.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
125
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
126 ``links``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
127 Removes any ``<link>`` tags
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
128
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
129 ``meta``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
130 Removes any ``<meta>`` tags
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
131
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
132 ``page_structure``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
133 Structural parts of a page: ``<head>``, ``<html>``, ``<title>``.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
134
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
135 ``processing_instructions``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
136 Removes any processing instructions.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
137
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
138 ``embedded``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
139 Removes any embedded objects (flash, iframes)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
140
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
141 ``frames``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
142 Removes any frame-related tags
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
143
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
144 ``forms``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
145 Removes any form tags
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
146
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
147 ``annoying_tags``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
148 Tags that aren't *wrong*, but are annoying. ``<blink>`` and ``<marquee>``
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
149
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
150 ``remove_tags``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
151 A list of tags to remove. Only the tags will be removed,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
152 their content will get pulled up into the parent tag.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
153
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
154 ``kill_tags``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
155 A list of tags to kill. Killing also removes the tag's content,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
156 i.e. the whole subtree, not just the tag itself.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
157
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
158 ``allow_tags``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
159 A list of tags to include (default include all).
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
160
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
161 ``remove_unknown_tags``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
162 Remove any tags that aren't standard parts of HTML.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
163
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
164 ``safe_attrs_only``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
165 If true, only include 'safe' attributes (specifically the list
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
166 from the feedparser HTML sanitisation web site).
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
167
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
168 ``safe_attrs``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
169 A set of attribute names to override the default list of attributes
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
170 considered 'safe' (when safe_attrs_only=True).
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
171
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
172 ``add_nofollow``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
173 If true, then any <a> tags will have ``rel="nofollow"`` added to them.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
174
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
175 ``host_whitelist``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
176 A list or set of hosts that you can use for embedded content
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
177 (for content like ``<object>``, ``<link rel="stylesheet">``, etc).
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
178 You can also implement/override the method
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
179 ``allow_embedded_url(el, url)`` or ``allow_element(el)`` to
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
180 implement more complex rules for what can be embedded.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
181 Anything that passes this test will be shown, regardless of
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
182 the value of (for instance) ``embedded``.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
183
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
184 Note that this parameter might not work as intended if you do not
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
185 make the links absolute before doing the cleaning.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
186
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
187 Note that you may also need to set ``whitelist_tags``.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
188
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
189 ``whitelist_tags``:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
190 A set of tags that can be included with ``host_whitelist``.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
191 The default is ``iframe`` and ``embed``; you may wish to
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
192 include other tags like ``script``, or you may want to
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
193 implement ``allow_embedded_url`` for more control. Set to None to
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
194 include all tags.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
195
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
196 This modifies the document *in place*.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
197 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
198
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
199 scripts = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
200 javascript = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
201 comments = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
202 style = False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
203 inline_style = None
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
204 links = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
205 meta = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
206 page_structure = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
207 processing_instructions = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
208 embedded = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
209 frames = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
210 forms = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
211 annoying_tags = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
212 remove_tags = None
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
213 allow_tags = None
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
214 kill_tags = None
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
215 remove_unknown_tags = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
216 safe_attrs_only = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
217 safe_attrs = defs.safe_attrs
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
218 add_nofollow = False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
219 host_whitelist = ()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
220 whitelist_tags = {'iframe', 'embed'}
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
221
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
222 def __init__(self, **kw):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
223 not_an_attribute = object()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
224 for name, value in kw.items():
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
225 default = getattr(self, name, not_an_attribute)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
226 if (default is not None and default is not True and default is not False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
227 and not isinstance(default, (frozenset, set, tuple, list))):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
228 raise TypeError(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
229 "Unknown parameter: %s=%r" % (name, value))
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
230 setattr(self, name, value)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
231 if self.inline_style is None and 'inline_style' not in kw:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
232 self.inline_style = self.style
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
233
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
234 if kw.get("allow_tags"):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
235 if kw.get("remove_unknown_tags"):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
236 raise ValueError("It does not make sense to pass in both "
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
237 "allow_tags and remove_unknown_tags")
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
238 self.remove_unknown_tags = False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
239
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
240 # Used to lookup the primary URL for a given tag that is up for
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
241 # removal:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
242 _tag_link_attrs = dict(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
243 script='src',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
244 link='href',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
245 # From: http://java.sun.com/j2se/1.4.2/docs/guide/misc/applet.html
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
246 # From what I can tell, both attributes can contain a link:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
247 applet=['code', 'object'],
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
248 iframe='src',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
249 embed='src',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
250 layer='src',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
251 # FIXME: there doesn't really seem like a general way to figure out what
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
252 # links an <object> tag uses; links often go in <param> tags with values
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
253 # that we don't really know. You'd have to have knowledge about specific
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
254 # kinds of plugins (probably keyed off classid), and match against those.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
255 ##object=?,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
256 # FIXME: not looking at the action currently, because it is more complex
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
257 # than than -- if you keep the form, you should keep the form controls.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
258 ##form='action',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
259 a='href',
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
260 )
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
261
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
262 def __call__(self, doc):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
263 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
264 Cleans the document.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
265 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
266 try:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
267 getroot = doc.getroot
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
268 except AttributeError:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
269 pass # Element instance
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
270 else:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
271 doc = getroot() # ElementTree instance, instead of an element
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
272 # convert XHTML to HTML
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
273 xhtml_to_html(doc)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
274 # Normalize a case that IE treats <image> like <img>, and that
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
275 # can confuse either this step or later steps.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
276 for el in doc.iter('image'):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
277 el.tag = 'img'
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
278 if not self.comments:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
279 # Of course, if we were going to kill comments anyway, we don't
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
280 # need to worry about this
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
281 self.kill_conditional_comments(doc)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
282
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
283 kill_tags = set(self.kill_tags or ())
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
284 remove_tags = set(self.remove_tags or ())
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
285 allow_tags = set(self.allow_tags or ())
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
286
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
287 if self.scripts:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
288 kill_tags.add('script')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
289 if self.safe_attrs_only:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
290 safe_attrs = set(self.safe_attrs)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
291 for el in doc.iter(etree.Element):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
292 attrib = el.attrib
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
293 for aname in attrib.keys():
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
294 if aname not in safe_attrs:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
295 del attrib[aname]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
296 if self.javascript:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
297 if not (self.safe_attrs_only and
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
298 self.safe_attrs == defs.safe_attrs):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
299 # safe_attrs handles events attributes itself
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
300 for el in doc.iter(etree.Element):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
301 attrib = el.attrib
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
302 for aname in attrib.keys():
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
303 if aname.startswith('on'):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
304 del attrib[aname]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
305 doc.rewrite_links(self._remove_javascript_link,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
306 resolve_base_href=False)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
307 # If we're deleting style then we don't have to remove JS links
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
308 # from styles, otherwise...
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
309 if not self.inline_style:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
310 for el in _find_styled_elements(doc):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
311 old = el.get('style')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
312 new = _replace_css_javascript('', old)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
313 new = _replace_css_import('', new)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
314 if self._has_sneaky_javascript(new):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
315 # Something tricky is going on...
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
316 del el.attrib['style']
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
317 elif new != old:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
318 el.set('style', new)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
319 if not self.style:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
320 for el in list(doc.iter('style')):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
321 if el.get('type', '').lower().strip() == 'text/javascript':
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
322 el.drop_tree()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
323 continue
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
324 old = el.text or ''
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
325 new = _replace_css_javascript('', old)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
326 # The imported CSS can do anything; we just can't allow:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
327 new = _replace_css_import('', new)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
328 if self._has_sneaky_javascript(new):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
329 # Something tricky is going on...
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
330 el.text = '/* deleted */'
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
331 elif new != old:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
332 el.text = new
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
333 if self.comments:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
334 kill_tags.add(etree.Comment)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
335 if self.processing_instructions:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
336 kill_tags.add(etree.ProcessingInstruction)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
337 if self.style:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
338 kill_tags.add('style')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
339 if self.inline_style:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
340 etree.strip_attributes(doc, 'style')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
341 if self.links:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
342 kill_tags.add('link')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
343 elif self.style or self.javascript:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
344 # We must get rid of included stylesheets if Javascript is not
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
345 # allowed, as you can put Javascript in them
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
346 for el in list(doc.iter('link')):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
347 if 'stylesheet' in el.get('rel', '').lower():
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
348 # Note this kills alternate stylesheets as well
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
349 if not self.allow_element(el):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
350 el.drop_tree()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
351 if self.meta:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
352 kill_tags.add('meta')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
353 if self.page_structure:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
354 remove_tags.update(('head', 'html', 'title'))
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
355 if self.embedded:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
356 # FIXME: is <layer> really embedded?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
357 # We should get rid of any <param> tags not inside <applet>;
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
358 # These are not really valid anyway.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
359 for el in list(doc.iter('param')):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
360 parent = el.getparent()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
361 while parent is not None and parent.tag not in ('applet', 'object'):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
362 parent = parent.getparent()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
363 if parent is None:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
364 el.drop_tree()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
365 kill_tags.update(('applet',))
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
366 # The alternate contents that are in an iframe are a good fallback:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
367 remove_tags.update(('iframe', 'embed', 'layer', 'object', 'param'))
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
368 if self.frames:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
369 # FIXME: ideally we should look at the frame links, but
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
370 # generally frames don't mix properly with an HTML
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
371 # fragment anyway.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
372 kill_tags.update(defs.frame_tags)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
373 if self.forms:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
374 remove_tags.add('form')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
375 kill_tags.update(('button', 'input', 'select', 'textarea'))
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
376 if self.annoying_tags:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
377 remove_tags.update(('blink', 'marquee'))
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
378
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
379 _remove = []
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
380 _kill = []
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
381 for el in doc.iter():
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
382 if el.tag in kill_tags:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
383 if self.allow_element(el):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
384 continue
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
385 _kill.append(el)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
386 elif el.tag in remove_tags:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
387 if self.allow_element(el):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
388 continue
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
389 _remove.append(el)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
390
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
391 if _remove and _remove[0] == doc:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
392 # We have to drop the parent-most tag, which we can't
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
393 # do. Instead we'll rewrite it:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
394 el = _remove.pop(0)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
395 el.tag = 'div'
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
396 el.attrib.clear()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
397 elif _kill and _kill[0] == doc:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
398 # We have to drop the parent-most element, which we can't
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
399 # do. Instead we'll clear it:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
400 el = _kill.pop(0)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
401 if el.tag != 'html':
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
402 el.tag = 'div'
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
403 el.clear()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
404
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
405 _kill.reverse() # start with innermost tags
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
406 for el in _kill:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
407 el.drop_tree()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
408 for el in _remove:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
409 el.drop_tag()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
410
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
411 if self.remove_unknown_tags:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
412 if allow_tags:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
413 raise ValueError(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
414 "It does not make sense to pass in both allow_tags and remove_unknown_tags")
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
415 allow_tags = set(defs.tags)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
416 if allow_tags:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
417 # make sure we do not remove comments/PIs if users want them (which is rare enough)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
418 if not self.comments:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
419 allow_tags.add(etree.Comment)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
420 if not self.processing_instructions:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
421 allow_tags.add(etree.ProcessingInstruction)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
422
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
423 bad = []
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
424 for el in doc.iter():
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
425 if el.tag not in allow_tags:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
426 bad.append(el)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
427 if bad:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
428 if bad[0] is doc:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
429 el = bad.pop(0)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
430 el.tag = 'div'
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
431 el.attrib.clear()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
432 for el in bad:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
433 el.drop_tag()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
434 if self.add_nofollow:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
435 for el in _find_external_links(doc):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
436 if not self.allow_follow(el):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
437 rel = el.get('rel')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
438 if rel:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
439 if ('nofollow' in rel
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
440 and ' nofollow ' in (' %s ' % rel)):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
441 continue
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
442 rel = '%s nofollow' % rel
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
443 else:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
444 rel = 'nofollow'
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
445 el.set('rel', rel)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
446
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
447 def allow_follow(self, anchor):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
448 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
449 Override to suppress rel="nofollow" on some anchors.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
450 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
451 return False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
452
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
453 def allow_element(self, el):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
454 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
455 Decide whether an element is configured to be accepted or rejected.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
456
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
457 :param el: an element.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
458 :return: true to accept the element or false to reject/discard it.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
459 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
460 if el.tag not in self._tag_link_attrs:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
461 return False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
462 attr = self._tag_link_attrs[el.tag]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
463 if isinstance(attr, (list, tuple)):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
464 for one_attr in attr:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
465 url = el.get(one_attr)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
466 if not url:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
467 return False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
468 if not self.allow_embedded_url(el, url):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
469 return False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
470 return True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
471 else:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
472 url = el.get(attr)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
473 if not url:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
474 return False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
475 return self.allow_embedded_url(el, url)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
476
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
477 def allow_embedded_url(self, el, url):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
478 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
479 Decide whether a URL that was found in an element's attributes or text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
480 if configured to be accepted or rejected.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
481
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
482 :param el: an element.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
483 :param url: a URL found on the element.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
484 :return: true to accept the URL and false to reject it.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
485 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
486 if self.whitelist_tags is not None and el.tag not in self.whitelist_tags:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
487 return False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
488 scheme, netloc, path, query, fragment = urlsplit(url)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
489 netloc = netloc.lower().split(':', 1)[0]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
490 if scheme not in ('http', 'https'):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
491 return False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
492 if netloc in self.host_whitelist:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
493 return True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
494 return False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
495
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
496 def kill_conditional_comments(self, doc):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
497 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
498 IE conditional comments basically embed HTML that the parser
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
499 doesn't normally see. We can't allow anything like that, so
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
500 we'll kill any comments that could be conditional.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
501 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
502 has_conditional_comment = _conditional_comment_re.search
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
503 self._kill_elements(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
504 doc, lambda el: has_conditional_comment(el.text),
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
505 etree.Comment)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
506
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
507 def _kill_elements(self, doc, condition, iterate=None):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
508 bad = []
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
509 for el in doc.iter(iterate):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
510 if condition(el):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
511 bad.append(el)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
512 for el in bad:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
513 el.drop_tree()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
514
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
515 def _remove_javascript_link(self, link):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
516 # links like "j a v a s c r i p t:" might be interpreted in IE
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
517 new = _substitute_whitespace('', unquote_plus(link))
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
518 if _is_javascript_scheme(new):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
519 # FIXME: should this be None to delete?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
520 return ''
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
521 return link
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
522
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
523 _substitute_comments = re.compile(r'/\*.*?\*/', re.S).sub
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
524
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
525 def _has_sneaky_javascript(self, style):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
526 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
527 Depending on the browser, stuff like ``e x p r e s s i o n(...)``
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
528 can get interpreted, or ``expre/* stuff */ssion(...)``. This
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
529 checks for attempt to do stuff like this.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
530
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
531 Typically the response will be to kill the entire style; if you
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
532 have just a bit of Javascript in the style another rule will catch
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
533 that and remove only the Javascript from the style; this catches
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
534 more sneaky attempts.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
535 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
536 style = self._substitute_comments('', style)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
537 style = style.replace('\\', '')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
538 style = _substitute_whitespace('', style)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
539 style = style.lower()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
540 if 'javascript:' in style:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
541 return True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
542 if 'expression(' in style:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
543 return True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
544 if '</noscript' in style:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
545 # e.g. '<noscript><style><a title="</noscript><img src=x onerror=alert(1)>">'
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
546 return True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
547 if _looks_like_tag_content(style):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
548 # e.g. '<math><style><img src=x onerror=alert(1)></style></math>'
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
549 return True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
550 return False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
551
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
552 def clean_html(self, html):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
553 result_type = type(html)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
554 if isinstance(html, basestring):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
555 doc = fromstring(html)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
556 else:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
557 doc = copy.deepcopy(html)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
558 self(doc)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
559 return _transform_result(result_type, doc)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
560
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
561 clean = Cleaner()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
562 clean_html = clean.clean_html
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
563
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
564 ############################################################
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
565 ## Autolinking
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
566 ############################################################
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
567
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
568 _link_regexes = [
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
569 re.compile(r'(?P<body>https?://(?P<host>[a-z0-9._-]+)(?:/[/\-_.,a-z0-9%&?;=~]*)?(?:\([/\-_.,a-z0-9%&?;=~]*\))?)', re.I),
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
570 # This is conservative, but autolinking can be a bit conservative:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
571 re.compile(r'mailto:(?P<body>[a-z0-9._-]+@(?P<host>[a-z0-9_.-]+[a-z]))', re.I),
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
572 ]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
573
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
574 _avoid_elements = ['textarea', 'pre', 'code', 'head', 'select', 'a']
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
575
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
576 _avoid_hosts = [
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
577 re.compile(r'^localhost', re.I),
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
578 re.compile(r'\bexample\.(?:com|org|net)$', re.I),
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
579 re.compile(r'^127\.0\.0\.1$'),
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
580 ]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
581
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
582 _avoid_classes = ['nolink']
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
583
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
584 def autolink(el, link_regexes=_link_regexes,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
585 avoid_elements=_avoid_elements,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
586 avoid_hosts=_avoid_hosts,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
587 avoid_classes=_avoid_classes):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
588 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
589 Turn any URLs into links.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
590
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
591 It will search for links identified by the given regular
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
592 expressions (by default mailto and http(s) links).
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
593
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
594 It won't link text in an element in avoid_elements, or an element
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
595 with a class in avoid_classes. It won't link to anything with a
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
596 host that matches one of the regular expressions in avoid_hosts
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
597 (default localhost and 127.0.0.1).
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
598
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
599 If you pass in an element, the element's tail will not be
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
600 substituted, only the contents of the element.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
601 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
602 if el.tag in avoid_elements:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
603 return
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
604 class_name = el.get('class')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
605 if class_name:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
606 class_name = class_name.split()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
607 for match_class in avoid_classes:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
608 if match_class in class_name:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
609 return
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
610 for child in list(el):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
611 autolink(child, link_regexes=link_regexes,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
612 avoid_elements=avoid_elements,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
613 avoid_hosts=avoid_hosts,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
614 avoid_classes=avoid_classes)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
615 if child.tail:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
616 text, tail_children = _link_text(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
617 child.tail, link_regexes, avoid_hosts, factory=el.makeelement)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
618 if tail_children:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
619 child.tail = text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
620 index = el.index(child)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
621 el[index+1:index+1] = tail_children
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
622 if el.text:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
623 text, pre_children = _link_text(
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
624 el.text, link_regexes, avoid_hosts, factory=el.makeelement)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
625 if pre_children:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
626 el.text = text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
627 el[:0] = pre_children
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
628
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
629 def _link_text(text, link_regexes, avoid_hosts, factory):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
630 leading_text = ''
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
631 links = []
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
632 last_pos = 0
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
633 while 1:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
634 best_match, best_pos = None, None
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
635 for regex in link_regexes:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
636 regex_pos = last_pos
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
637 while 1:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
638 match = regex.search(text, pos=regex_pos)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
639 if match is None:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
640 break
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
641 host = match.group('host')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
642 for host_regex in avoid_hosts:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
643 if host_regex.search(host):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
644 regex_pos = match.end()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
645 break
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
646 else:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
647 break
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
648 if match is None:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
649 continue
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
650 if best_pos is None or match.start() < best_pos:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
651 best_match = match
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
652 best_pos = match.start()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
653 if best_match is None:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
654 # No more matches
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
655 if links:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
656 assert not links[-1].tail
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
657 links[-1].tail = text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
658 else:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
659 assert not leading_text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
660 leading_text = text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
661 break
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
662 link = best_match.group(0)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
663 end = best_match.end()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
664 if link.endswith('.') or link.endswith(','):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
665 # These punctuation marks shouldn't end a link
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
666 end -= 1
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
667 link = link[:-1]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
668 prev_text = text[:best_match.start()]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
669 if links:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
670 assert not links[-1].tail
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
671 links[-1].tail = prev_text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
672 else:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
673 assert not leading_text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
674 leading_text = prev_text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
675 anchor = factory('a')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
676 anchor.set('href', link)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
677 body = best_match.group('body')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
678 if not body:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
679 body = link
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
680 if body.endswith('.') or body.endswith(','):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
681 body = body[:-1]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
682 anchor.text = body
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
683 links.append(anchor)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
684 text = text[end:]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
685 return leading_text, links
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
686
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
687 def autolink_html(html, *args, **kw):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
688 result_type = type(html)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
689 if isinstance(html, basestring):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
690 doc = fromstring(html)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
691 else:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
692 doc = copy.deepcopy(html)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
693 autolink(doc, *args, **kw)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
694 return _transform_result(result_type, doc)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
695
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
696 autolink_html.__doc__ = autolink.__doc__
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
697
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
698 ############################################################
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
699 ## Word wrapping
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
700 ############################################################
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
701
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
702 _avoid_word_break_elements = ['pre', 'textarea', 'code']
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
703 _avoid_word_break_classes = ['nobreak']
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
704
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
705 def word_break(el, max_width=40,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
706 avoid_elements=_avoid_word_break_elements,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
707 avoid_classes=_avoid_word_break_classes,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
708 break_character=unichr(0x200b)):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
709 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
710 Breaks any long words found in the body of the text (not attributes).
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
711
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
712 Doesn't effect any of the tags in avoid_elements, by default
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
713 ``<textarea>`` and ``<pre>``
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
714
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
715 Breaks words by inserting &#8203;, which is a unicode character
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
716 for Zero Width Space character. This generally takes up no space
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
717 in rendering, but does copy as a space, and in monospace contexts
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
718 usually takes up space.
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
719
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
720 See http://www.cs.tut.fi/~jkorpela/html/nobr.html for a discussion
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
721 """
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
722 # Character suggestion of &#8203 comes from:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
723 # http://www.cs.tut.fi/~jkorpela/html/nobr.html
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
724 if el.tag in _avoid_word_break_elements:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
725 return
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
726 class_name = el.get('class')
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
727 if class_name:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
728 dont_break = False
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
729 class_name = class_name.split()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
730 for avoid in avoid_classes:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
731 if avoid in class_name:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
732 dont_break = True
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
733 break
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
734 if dont_break:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
735 return
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
736 if el.text:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
737 el.text = _break_text(el.text, max_width, break_character)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
738 for child in el:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
739 word_break(child, max_width=max_width,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
740 avoid_elements=avoid_elements,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
741 avoid_classes=avoid_classes,
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
742 break_character=break_character)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
743 if child.tail:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
744 child.tail = _break_text(child.tail, max_width, break_character)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
745
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
746 def word_break_html(html, *args, **kw):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
747 result_type = type(html)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
748 doc = fromstring(html)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
749 word_break(doc, *args, **kw)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
750 return _transform_result(result_type, doc)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
751
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
752 def _break_text(text, max_width, break_character):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
753 words = text.split()
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
754 for word in words:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
755 if len(word) > max_width:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
756 replacement = _insert_break(word, max_width, break_character)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
757 text = text.replace(word, replacement)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
758 return text
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
759
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
760 _break_prefer_re = re.compile(r'[^a-z]', re.I)
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
761
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
762 def _insert_break(word, width, break_character):
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
763 orig_word = word
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
764 result = ''
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
765 while len(word) > width:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
766 start = word[:width]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
767 breaks = list(_break_prefer_re.finditer(start))
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
768 if breaks:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
769 last_break = breaks[-1]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
770 # Only walk back up to 10 characters to find a nice break:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
771 if last_break.end() > width-10:
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
772 # FIXME: should the break character be at the end of the
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
773 # chunk, or the beginning of the next chunk?
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
774 start = word[:last_break.end()]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
775 result += start + break_character
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
776 word = word[len(start):]
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
777 result += word
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
778 return result
4f3585e2f14b "planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
shellac
parents:
diff changeset
779