annotate src/plugins/plugin_xep_0071.py @ 940:92e41e7c7e00

misc: precisions about pyjamas in README4PACKAGERS
author Goffi <goffi@goffi.org>
date Fri, 28 Mar 2014 18:06:51 +0100
parents c897c8d321b3
children 301b342c697a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
1 #!/usr/bin/python
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
2 # -*- coding: utf-8 -*-
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
3
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
4 # SAT plugin for Publish-Subscribe (xep-0071)
811
1fe00f0c9a91 dates update
Goffi <goffi@goffi.org>
parents: 771
diff changeset
5 # Copyright (C) 2009, 2010, 2011, 2012, 2013, 2014 Jérôme Poisson (goffi@goffi.org)
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
6
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
7 # This program is free software: you can redistribute it and/or modify
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
8 # it under the terms of the GNU Affero General Public License as published by
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
9 # the Free Software Foundation, either version 3 of the License, or
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
10 # (at your option) any later version.
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
11
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
12 # This program is distributed in the hope that it will be useful,
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
13 # but WITHOUT ANY WARRANTY; without even the implied warranty of
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
14 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
15 # GNU Affero General Public License for more details.
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
16
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
17 # You should have received a copy of the GNU Affero General Public License
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
18 # along with this program. If not, see <http://www.gnu.org/licenses/>.
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
19
771
bfabeedbf32e core: i18n refactoring:
Goffi <goffi@goffi.org>
parents: 702
diff changeset
20 from sat.core.i18n import _
832
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
21 from sat.core import exceptions
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
22 from logging import debug, info, error
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
23
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
24 from wokkel import disco, pubsub, iwokkel
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
25 from zope.interface import implements
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
26 # from lxml import etree
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
27 from lxml import html
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
28 try:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
29 from twisted.words.protocols.xmlstream import XMPPHandler
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
30 except ImportError:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
31 from wokkel.subprotocols import XMPPHandler
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
32
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
33 NS_XHTML_IM = 'http://jabber.org/protocol/xhtml-im'
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
34 NS_XHTML = 'http://www.w3.org/1999/xhtml'
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
35
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
36 PLUGIN_INFO = {
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
37 "name": "XHTML-IM Plugin",
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
38 "import_name": "XEP-0071",
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
39 "type": "XEP",
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
40 "protocols": ["XEP-0071"],
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
41 "dependencies": ["TEXT-SYNTAXES"],
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
42 "main": "XEP_0071",
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
43 "handler": "yes",
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
44 "description": _("""Implementation of XHTML-IM""")
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
45 }
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
46
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
47 allowed = {
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
48 "a": set(["href", "style", "type"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
49 "blockquote": set(["style"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
50 "body": set(["style"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
51 "br": set([]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
52 "cite": set(["style"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
53 "em": set([]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
54 "img": set(["alt", "height", "src", "style", "width"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
55 "li": set(["style"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
56 "ol": set(["style"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
57 "p": set(["style"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
58 "span": set(["style"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
59 "strong": set([]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
60 "ul": set(["style"]),
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
61 }
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
62
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
63 styles_allowed = ["background-color", "color", "font-family", "font-size", "font-style", "font-weight", "margin-left", "margin-right", "text-align", "text-decoration"]
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
64
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
65 blacklist = ['script'] # tag that we have to kill (we don't keep content)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
66
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
67
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
68 class XEP_0071(object):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
69 SYNTAX_XHTML_IM = "XHTML-IM"
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
70
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
71 def __init__(self, host):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
72 info(_("XHTML-IM plugin initialization"))
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
73 self.host = host
702
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
74 self.synt_plg = self.host.plugins["TEXT-SYNTAXES"]
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
75 self.synt_plg.addSyntax(self.SYNTAX_XHTML_IM, lambda xhtml: xhtml, self.XHTML2XHTML_IM, [self.synt_plg.OPT_HIDDEN])
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
76 host.trigger.add("MessageReceived", self.messageReceivedTrigger)
702
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
77 host.trigger.add("sendMessage", self.sendMessageTrigger)
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
78
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
79 def getHandler(self, profile):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
80 return XEP_0071_handler(self)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
81
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
82 def _messagePostTreat(self, data, body_elt):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
83 """ Callback which manage the post treatment of the message in case of XHTML-IM found
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
84 @param data: data send by MessageReceived trigger through post_treat deferred
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
85 @param xhtml_im: XHTML-IM body element found
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
86 @return: the data with the extra parameter updated
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
87 """
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
88 #TODO: check if text only body is empty, then try to convert XHTML-IM to pure text and show a warning message
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
89 def converted(xhtml):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
90 data['extra']['xhtml'] = xhtml
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
91 return data
702
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
92 d = self.synt_plg.convert(body_elt.toXml(), self.SYNTAX_XHTML_IM, safe=True)
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
93 d.addCallback(converted)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
94 return d
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
95
702
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
96 def _sendMessageAddRich(self, mess_data, profile):
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
97 """ Construct XHTML-IM node and add it XML element
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
98 @param mess_data: message data as sended by sendMessage callback
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
99 """
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
100 def syntax_converted(xhtml_im):
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
101 message_elt = mess_data['xml']
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
102 html_elt = message_elt.addElement('html', NS_XHTML_IM)
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
103 body_elt = html_elt.addElement('body', NS_XHTML)
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
104 body_elt.addRawXml(xhtml_im)
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
105 mess_data['extra']['xhtml'] = xhtml_im
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
106 return mess_data
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
107
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
108 syntax = self.synt_plg.getCurrentSyntax(profile)
832
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
109 rich = mess_data['extra'].get('rich', '')
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
110 xhtml = mess_data['extra'].get('xhtml', '')
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
111 if rich:
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
112 d = self.synt_plg.convert(rich, syntax, self.SYNTAX_XHTML_IM)
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
113 if xhtml:
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
114 raise exceptions.DataError(_("Can't have xhtml and rich content at the same time"))
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
115 if xhtml:
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
116 d = self.synt_plg.clean_xhtml(xhtml)
702
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
117 d.addCallback(syntax_converted)
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
118 return d
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
119
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
120 def messageReceivedTrigger(self, message, post_treat, profile):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
121 """ Check presence of XHTML-IM in message
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
122 """
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
123 try:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
124 html_elt = message.elements(NS_XHTML_IM, 'html').next()
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
125 body_elt = html_elt.elements(NS_XHTML, 'body').next()
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
126 # OK, we have found rich text
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
127 post_treat.addCallback(self._messagePostTreat, body_elt)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
128 except StopIteration:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
129 # No XHTML-IM
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
130 pass
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
131 return True
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
132
922
c897c8d321b3 core: sendMessageTrigger now manage pre and post treatments, which happen before or after XML generation
Goffi <goffi@goffi.org>
parents: 832
diff changeset
133 def sendMessageTrigger(self, mess_data, pre_xml_treatments, post_xml_treatments, profile):
702
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
134 """ Check presence of rich text in extra
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
135 """
832
c4b22aedb7d7 plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents: 811
diff changeset
136 if 'rich' in mess_data['extra'] or 'xhtml' in mess_data['extra']:
922
c897c8d321b3 core: sendMessageTrigger now manage pre and post treatments, which happen before or after XML generation
Goffi <goffi@goffi.org>
parents: 832
diff changeset
137 post_xml_treatments.addCallback(self._sendMessageAddRich, profile)
702
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
138 return True
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 701
diff changeset
139
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
140 def _purgeStyle(self, styles_raw):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
141 """ Remove unauthorised styles according to the XEP-0071
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
142 @param styles_raw: raw styles (value of the style attribute)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
143 """
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
144 purged = []
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
145
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
146 styles = [style.strip().split(':') for style in styles_raw.split(';')]
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
147
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
148 for style_tuple in styles:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
149 if len(style_tuple) != 2:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
150 continue
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
151 name, value = style_tuple
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
152 name = name.strip()
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
153 if name not in styles_allowed:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
154 continue
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
155 purged.append((name, value.strip()))
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
156
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
157 return u'; '.join([u"%s: %s" % data for data in purged])
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
158
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
159 def XHTML2XHTML_IM(self, xhtml):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
160 """ Convert XHTML document to XHTML_IM subset
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
161 @param xhtml: raw xhtml to convert
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
162 """
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
163 # TODO: more clever tag replacement (replace forbidden tags with equivalents when possible)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
164
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
165 parser = html.HTMLParser(remove_comments=True, encoding='utf-8')
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
166 root = html.fromstring(xhtml, parser=parser)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
167 body_elt = root.find('body')
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
168 if body_elt is None:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
169 # we use the whole XML as body if no body element is found
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
170 body_elt = html.Element('body')
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
171 body_elt.append(root)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
172 else:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
173 body_elt.attrib.clear()
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
174
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
175 allowed_tags = allowed.keys()
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
176 to_strip = []
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
177 for elem in body_elt.iter():
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
178 if elem.tag not in allowed_tags:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
179 to_strip.append(elem)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
180 else:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
181 # we remove unallowed attributes
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
182 attrib = elem.attrib
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
183 att_to_remove = set(attrib).difference(allowed[elem.tag])
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
184 for att in att_to_remove:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
185 del(attrib[att])
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
186 if "style" in attrib:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
187 attrib["style"] = self._purgeStyle(attrib["style"])
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
188
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
189 for elem in to_strip:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
190 if elem.tag in blacklist:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
191 #we need to remove the element and all descendants
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
192 debug(u"removing black listed tag: %s" % (elem.tag))
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
193 elem.drop_tree()
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
194 else:
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
195 elem.drop_tag()
701
98b2400e17d6 plugin XEP-0071: XHTML2XHTML_IM don't return the <body> root tag anymore.
Goffi <goffi@goffi.org>
parents: 668
diff changeset
196 if len(body_elt) !=1:
98b2400e17d6 plugin XEP-0071: XHTML2XHTML_IM don't return the <body> root tag anymore.
Goffi <goffi@goffi.org>
parents: 668
diff changeset
197 root_elt = body_elt
98b2400e17d6 plugin XEP-0071: XHTML2XHTML_IM don't return the <body> root tag anymore.
Goffi <goffi@goffi.org>
parents: 668
diff changeset
198 body_elt.tag = "p"
98b2400e17d6 plugin XEP-0071: XHTML2XHTML_IM don't return the <body> root tag anymore.
Goffi <goffi@goffi.org>
parents: 668
diff changeset
199 else:
98b2400e17d6 plugin XEP-0071: XHTML2XHTML_IM don't return the <body> root tag anymore.
Goffi <goffi@goffi.org>
parents: 668
diff changeset
200 root_elt = body_elt[0]
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
201
701
98b2400e17d6 plugin XEP-0071: XHTML2XHTML_IM don't return the <body> root tag anymore.
Goffi <goffi@goffi.org>
parents: 668
diff changeset
202 return html.tostring(root_elt, encoding='unicode', method='xml')
668
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
203
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
204 class XEP_0071_handler(XMPPHandler):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
205 implements(iwokkel.IDisco)
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
206
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
207 def __init__(self, plugin_parent):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
208 self.plugin_parent = plugin_parent
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
209 self.host = plugin_parent.host
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
210
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
211 def getDiscoInfo(self, requestor, target, nodeIdentifier=''):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
212 return [disco.DiscoFeature(NS_XHTML_IM)]
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
213
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
214 def getDiscoItems(self, requestor, target, nodeIdentifier=''):
7bb50096d225 plugin XEP_0071: first draft
Goffi <goffi@goffi.org>
parents:
diff changeset
215 return []