Mercurial > libervia-backend

annotate sat/plugins/plugin_misc_text_syntaxes.py @ 2771:003b8b4b56a7

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
date update
author Goffi <goffi@goffi.org>
date Sat, 12 Jan 2019 12:25:31 +0100
parents 56f94936df1e
children 85d3240a400f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1934
2daf7b4c6756 use of /usr/bin/env instead of /usr/bin/python in shebang
Goffi <goffi@goffi.org>
parents: 1867
diff changeset
1 #!/usr/bin/env python2
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
2 # -*- coding: utf-8 -*-
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
3
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
4 # SAT plugin for managing various text syntaxes
2771
003b8b4b56a7 date update
Goffi <goffi@goffi.org>
parents: 2624
diff changeset
5 # Copyright (C) 2009-2019 Jérôme Poisson (goffi@goffi.org)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
6
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
7 # This program is free software: you can redistribute it and/or modify
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
8 # it under the terms of the GNU Affero General Public License as published by
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
9 # the Free Software Foundation, either version 3 of the License, or
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
10 # (at your option) any later version.
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
11
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
12 # This program is distributed in the hope that it will be useful,
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
13 # but WITHOUT ANY WARRANTY; without even the implied warranty of
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
14 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
15 # GNU Affero General Public License for more details.
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
16
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
17 # You should have received a copy of the GNU Affero General Public License
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
18 # along with this program. If not, see <http://www.gnu.org/licenses/>.
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
19
771
bfabeedbf32e core: i18n refactoring:
Goffi <goffi@goffi.org>
parents: 744
diff changeset
20 from sat.core.i18n import _, D_
2145
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2106
diff changeset
21 from sat.core.constants import Const as C
993
301b342c697a core: use of the new core.log module:
Goffi <goffi@goffi.org>
parents: 968
diff changeset
22 from sat.core.log import getLogger
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
23
993
301b342c697a core: use of the new core.log module:
Goffi <goffi@goffi.org>
parents: 968
diff changeset
24 log = getLogger(__name__)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
25
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
26 from twisted.internet import defer
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
27 from twisted.internet.threads import deferToThread
705
6c8a119dcc94 plugin text syntaxes: clean_xhtml now accept lxml's HtmlElement to avoid parsing two times the same xml
Goffi <goffi@goffi.org>
parents: 702
diff changeset
28 from sat.core import exceptions
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
29
1542
94901070478e plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents: 1458
diff changeset
30 try:
94901070478e plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents: 1458
diff changeset
31 from lxml import html
94901070478e plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents: 1458
diff changeset
32 from lxml.html import clean
94901070478e plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents: 1458
diff changeset
33 except ImportError:
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
34 raise exceptions.MissingModule(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
35 u"Missing module lxml, please download/install it from http://lxml.de/"
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
36 )
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
37 from cgi import escape
692
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
38 import re
674
fb0b1100c908 plugin text_syntaxes: fixed clean_xhml (it now return XHTML instead of HTML)
Goffi <goffi@goffi.org>
parents: 665
diff changeset
39
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
40
771
bfabeedbf32e core: i18n refactoring:
Goffi <goffi@goffi.org>
parents: 744
diff changeset
41 CATEGORY = D_("Composition")
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
42 NAME = "Syntax"
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
43 _SYNTAX_XHTML = "XHTML"
744
312a2842b2b8 plugins text-syntaxes: added a default value to use the current user syntax in convert
souliane <souliane@mailoo.org>
parents: 705
diff changeset
44 _SYNTAX_CURRENT = "@CURRENT@"
312a2842b2b8 plugins text-syntaxes: added a default value to use the current user syntax in convert
souliane <souliane@mailoo.org>
parents: 705
diff changeset
45
692
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
46 # TODO: check/adapt following list
1805
3c40fa0dcd7a pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents: 1803
diff changeset
47 # list initialy based on feedparser list (http://pythonhosted.org/feedparser/html-sanitization.html)
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
48 STYLES_WHITELIST = (
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
49 "azimuth",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
50 "background-color",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
51 "border-bottom-color",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
52 "border-collapse",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
53 "border-color",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
54 "border-left-color",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
55 "border-right-color",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
56 "border-top-color",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
57 "clear",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
58 "color",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
59 "cursor",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
60 "direction",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
61 "display",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
62 "elevation",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
63 "float",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
64 "font",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
65 "font-family",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
66 "font-size",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
67 "font-style",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
68 "font-variant",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
69 "font-weight",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
70 "height",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
71 "letter-spacing",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
72 "line-height",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
73 "overflow",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
74 "pause",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
75 "pause-after",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
76 "pause-before",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
77 "pitch",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
78 "pitch-range",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
79 "richness",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
80 "speak",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
81 "speak-header",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
82 "speak-numeral",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
83 "speak-punctuation",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
84 "speech-rate",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
85 "stress",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
86 "text-align",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
87 "text-decoration",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
88 "text-indent",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
89 "unicode-bidi",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
90 "vertical-align",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
91 "voice-family",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
92 "volume",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
93 "white-space",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
94 "width",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
95 )
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
96
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
97 SAFE_ATTRS = html.defs.safe_attrs.union(("style", "poster", "controls"))
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
98 STYLES_VALUES_REGEX = (
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
99 r"^("
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
100 + "|".join(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
101 [
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
102 "([a-z-]+)", # alphabetical names
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
103 "(#[0-9a-f]+)", # hex value
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
104 "(\d+(.\d+)? *(|%|em|ex|px|in|cm|mm|pt|pc))", # values with units (or not)
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
105 "rgb\( *((\d+(.\d+)?), *){2}(\d+(.\d+)?) *\)", # rgb function
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
106 "rgba\( *((\d+(.\d+)?), *){3}(\d+(.\d+)?) *\)", # rgba function
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
107 ]
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
108 )
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
109 + ") *(!important)?$"
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
110 ) # we accept "!important" at the end
692
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
111 STYLES_ACCEPTED_VALUE = re.compile(STYLES_VALUES_REGEX)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
112
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
113 PLUGIN_INFO = {
2145
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2106
diff changeset
114 C.PI_NAME: "Text syntaxes",
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2106
diff changeset
115 C.PI_IMPORT_NAME: "TEXT-SYNTAXES",
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2106
diff changeset
116 C.PI_TYPE: "MISC",
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2106
diff changeset
117 C.PI_PROTOCOLS: [],
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2106
diff changeset
118 C.PI_DEPENDENCIES: [],
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2106
diff changeset
119 C.PI_MAIN: "TextSyntaxes",
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2106
diff changeset
120 C.PI_HANDLER: "no",
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
121 C.PI_DESCRIPTION: _(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
122 """Management of various text syntaxes (XHTML-IM, Markdown, etc)"""
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
123 ),
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
124 }
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
125
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
126
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
127 class TextSyntaxes(object):
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
128 """ Text conversion class
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
129 XHTML utf-8 is used as intermediate language for conversions
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
130 """
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
131
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
132 OPT_DEFAULT = "DEFAULT"
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
133 OPT_HIDDEN = "HIDDEN"
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
134 OPT_NO_THREAD = "NO_THREAD"
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
135 SYNTAX_XHTML = _SYNTAX_XHTML
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
136 SYNTAX_MARKDOWN = "markdown"
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
137 SYNTAX_TEXT = "text"
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
138 syntaxes = {}
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
139 default_syntax = SYNTAX_XHTML
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
140
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
141 params = """
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
142 <params>
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
143 <individual>
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
144 <category name="%(category_name)s" label="%(category_label)s">
968
75f3b3b430ff tools, frontends, memory: param definition and XMLUI handle multi-selection for list widgets:
souliane <souliane@mailoo.org>
parents: 852
diff changeset
145 <param name="%(name)s" label="%(label)s" type="list" security="0">
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
146 %(options)s
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
147 </param>
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
148 </category>
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
149 </individual>
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
150 </params>
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
151 """
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
152
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
153 params_data = {
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
154 "category_name": CATEGORY,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
155 "category_label": _(CATEGORY),
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
156 "name": NAME,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
157 "label": _(NAME),
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
158 "syntaxes": syntaxes,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
159 }
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
160
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
161 def __init__(self, host):
993
301b342c697a core: use of the new core.log module:
Goffi <goffi@goffi.org>
parents: 968
diff changeset
162 log.info(_("Text syntaxes plugin initialization"))
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
163 self.host = host
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
164 self.addSyntax(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
165 self.SYNTAX_XHTML,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
166 lambda xhtml: defer.succeed(xhtml),
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
167 lambda xhtml: defer.succeed(xhtml),
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
168 TextSyntaxes.OPT_NO_THREAD,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
169 )
1826
d80ccf4bf201 plugin blog import dotclear: this plugin import Dotclear 2 backups
Goffi <goffi@goffi.org>
parents: 1811
diff changeset
170 # TODO: text => XHTML should add <a/> to url like in frontends
d80ccf4bf201 plugin blog import dotclear: this plugin import Dotclear 2 backups
Goffi <goffi@goffi.org>
parents: 1811
diff changeset
171 # it's probably best to move sat_frontends.tools.strings to sat.tools.common or similar
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
172 self.addSyntax(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
173 self.SYNTAX_TEXT,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
174 lambda text: escape(text),
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
175 lambda xhtml: self._removeMarkups(xhtml),
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
176 [TextSyntaxes.OPT_HIDDEN],
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
177 )
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
178 try:
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
179 import markdown, html2text
841
831f208b4ea3 plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents: 836
diff changeset
180
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
181 def _html2text(html, baseurl=""):
841
831f208b4ea3 plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents: 836
diff changeset
182 h = html2text.HTML2Text(baseurl=baseurl)
831f208b4ea3 plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents: 836
diff changeset
183 h.body_width = 0 # do not truncate the lines, it breaks the long URLs
831f208b4ea3 plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents: 836
diff changeset
184 return h.handle(html)
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
185
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
186 self.addSyntax(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
187 self.SYNTAX_MARKDOWN,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
188 markdown.markdown,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
189 _html2text,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
190 [TextSyntaxes.OPT_DEFAULT],
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
191 )
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
192 except ImportError:
1805
3c40fa0dcd7a pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents: 1803
diff changeset
193 log.warning(u"markdown or html2text not found, can't use Markdown syntax")
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
194 log.info(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
195 u"You can download/install them from https://pythonhosted.org/Markdown/ and https://github.com/Alir3z4/html2text/"
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
196 )
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
197 host.bridge.addMethod(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
198 "syntaxConvert",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
199 ".plugin",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
200 in_sign="sssbs",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
201 out_sign="s",
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
202 async=True,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
203 method=self.convert,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
204 )
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
205 host.bridge.addMethod(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
206 "syntaxGet", ".plugin", in_sign="s", out_sign="s", method=self.getSyntax
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
207 )
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
208
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
209 def _updateParamOptions(self):
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
210 data_synt = TextSyntaxes.syntaxes
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
211 default_synt = TextSyntaxes.default_syntax
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
212 syntaxes = []
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
213
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
214 for syntax in data_synt.keys():
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
215 flags = data_synt[syntax]["flags"]
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
216 if TextSyntaxes.OPT_HIDDEN not in flags:
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
217 syntaxes.append(syntax)
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
218
1805
3c40fa0dcd7a pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents: 1803
diff changeset
219 syntaxes.sort(key=lambda synt: synt.lower())
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
220 options = []
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
221
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
222 for syntax in syntaxes:
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
223 selected = 'selected="true"' if syntax == default_synt else ""
968
75f3b3b430ff tools, frontends, memory: param definition and XMLUI handle multi-selection for list widgets:
souliane <souliane@mailoo.org>
parents: 852
diff changeset
224 options.append(u'<option value="%s" %s/>' % (syntax, selected))
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
225
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
226 TextSyntaxes.params_data["options"] = u"\n".join(options)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
227 self.host.memory.updateParams(TextSyntaxes.params % TextSyntaxes.params_data)
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
228
702
a25db3fe3959 plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents: 699
diff changeset
229 def getCurrentSyntax(self, profile):
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
230 """ Return the selected syntax for the given profile
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
231
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
232 @param profile: %(doc_profile)s
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
233 @return: profile selected syntax
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
234 """
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
235 return self.host.memory.getParamA(NAME, CATEGORY, profile_key=profile)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
236
2106
5874da3811b7 plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents: 1934
diff changeset
237 def _logError(self, failure, action=u"converting syntax"):
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
238 log.error(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
239 u"Error while {action}: {failure}".format(action=action, failure=failure)
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
240 )
2106
5874da3811b7 plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents: 1934
diff changeset
241 return failure
5874da3811b7 plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents: 1934
diff changeset
242
1805
3c40fa0dcd7a pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents: 1803
diff changeset
243 def cleanXHTML(self, xhtml):
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
244 """ Clean XHTML text by removing potentially dangerous/malicious parts
705
6c8a119dcc94 plugin text syntaxes: clean_xhtml now accept lxml's HtmlElement to avoid parsing two times the same xml
Goffi <goffi@goffi.org>
parents: 702
diff changeset
245 @param xhtml: raw xhtml text to clean (or lxml's HtmlElement)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
246 """
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
247
674
fb0b1100c908 plugin text_syntaxes: fixed clean_xhml (it now return XHTML instead of HTML)
Goffi <goffi@goffi.org>
parents: 665
diff changeset
248 def blocking_cleaning(xhtml):
692
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
249 """ Clean XHTML and style attributes """
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
250
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
251 def clean_style(styles_raw):
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
252 """" Remove styles not in the whitelist,
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
253 or where the value doesn't match the regex """
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
254 styles = styles_raw.split(";")
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
255 cleaned_styles = []
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
256 for style in styles:
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
257 try:
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
258 key, value = style.split(":")
692
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
259 except ValueError:
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
260 continue
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
261 key = key.lower().strip()
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
262 if key not in STYLES_WHITELIST:
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
263 continue
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
264 value = value.lower().strip()
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
265 if not STYLES_ACCEPTED_VALUE.match(value):
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
266 continue
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
267 if value == "none":
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
268 continue
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
269 cleaned_styles.append((key, value))
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
270 return "; ".join(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
271 ["%s: %s" % (key_, value_) for key_, value_ in cleaned_styles]
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
272 )
692
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
273
705
6c8a119dcc94 plugin text syntaxes: clean_xhtml now accept lxml's HtmlElement to avoid parsing two times the same xml
Goffi <goffi@goffi.org>
parents: 702
diff changeset
274 if isinstance(xhtml, basestring):
6c8a119dcc94 plugin text syntaxes: clean_xhtml now accept lxml's HtmlElement to avoid parsing two times the same xml
Goffi <goffi@goffi.org>
parents: 702
diff changeset
275 xhtml_elt = html.fromstring(xhtml)
6c8a119dcc94 plugin text syntaxes: clean_xhtml now accept lxml's HtmlElement to avoid parsing two times the same xml
Goffi <goffi@goffi.org>
parents: 702
diff changeset
276 elif isinstance(xhtml, html.HtmlElement):
6c8a119dcc94 plugin text syntaxes: clean_xhtml now accept lxml's HtmlElement to avoid parsing two times the same xml
Goffi <goffi@goffi.org>
parents: 702
diff changeset
277 xhtml_elt = xhtml
6c8a119dcc94 plugin text syntaxes: clean_xhtml now accept lxml's HtmlElement to avoid parsing two times the same xml
Goffi <goffi@goffi.org>
parents: 702
diff changeset
278 else:
993
301b342c697a core: use of the new core.log module:
Goffi <goffi@goffi.org>
parents: 968
diff changeset
279 log.error("Only strings and HtmlElements can be cleaned")
705
6c8a119dcc94 plugin text syntaxes: clean_xhtml now accept lxml's HtmlElement to avoid parsing two times the same xml
Goffi <goffi@goffi.org>
parents: 702
diff changeset
280 raise exceptions.DataError
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
281 cleaner = clean.Cleaner(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
282 style=False, add_nofollow=False, safe_attrs=SAFE_ATTRS
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
283 )
692
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
284 xhtml_elt = cleaner.clean_html(xhtml_elt)
e98db42cd78c plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents: 674
diff changeset
285 for elt in xhtml_elt.xpath("//*[@style]"):
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
286 elt.set("style", clean_style(elt.get("style")))
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
287 return html.tostring(xhtml_elt, encoding=unicode, method="xml")
674
fb0b1100c908 plugin text_syntaxes: fixed clean_xhml (it now return XHTML instead of HTML)
Goffi <goffi@goffi.org>
parents: 665
diff changeset
288
2106
5874da3811b7 plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents: 1934
diff changeset
289 d = deferToThread(blocking_cleaning, xhtml)
5874da3811b7 plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents: 1934
diff changeset
290 d.addErrback(self._logError, action=u"cleaning syntax")
5874da3811b7 plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents: 1934
diff changeset
291 return d
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
292
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
293 def convert(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
294 self, text, syntax_from, syntax_to=_SYNTAX_XHTML, safe=True, profile=None
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
295 ):
1803
14a97a5fe1c0 plugin text syntaxes: a non blocking syntax callback can now return a unicode directly instead of a Deferred
Goffi <goffi@goffi.org>
parents: 1766
diff changeset
296 """Convert a text between two syntaxes
14a97a5fe1c0 plugin text syntaxes: a non blocking syntax callback can now return a unicode directly instead of a Deferred
Goffi <goffi@goffi.org>
parents: 1766
diff changeset
297
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
298 @param text: text to convert
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
299 @param syntax_from: source syntax (e.g. "markdown")
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
300 @param syntax_to: dest syntax (e.g.: "XHTML")
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
301 @param safe: clean resulting XHTML to avoid malicious code if True
744
312a2842b2b8 plugins text-syntaxes: added a default value to use the current user syntax in convert
souliane <souliane@mailoo.org>
parents: 705
diff changeset
302 @param profile: needed only when syntax_from or syntax_to is set to _SYNTAX_CURRENT
1803
14a97a5fe1c0 plugin text syntaxes: a non blocking syntax callback can now return a unicode directly instead of a Deferred
Goffi <goffi@goffi.org>
parents: 1766
diff changeset
303 @return(unicode): converted text
14a97a5fe1c0 plugin text syntaxes: a non blocking syntax callback can now return a unicode directly instead of a Deferred
Goffi <goffi@goffi.org>
parents: 1766
diff changeset
304 """
1805
3c40fa0dcd7a pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents: 1803
diff changeset
305 # FIXME: convert should be abled to handle domish.Element directly
3c40fa0dcd7a pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents: 1803
diff changeset
306 # when dealing with XHTML
3c40fa0dcd7a pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents: 1803
diff changeset
307 # TODO: a way for parser to return parsing errors/warnings
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
308
744
312a2842b2b8 plugins text-syntaxes: added a default value to use the current user syntax in convert
souliane <souliane@mailoo.org>
parents: 705
diff changeset
309 if syntax_from == _SYNTAX_CURRENT:
312a2842b2b8 plugins text-syntaxes: added a default value to use the current user syntax in convert
souliane <souliane@mailoo.org>
parents: 705
diff changeset
310 syntax_from = self.getCurrentSyntax(profile)
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
311 else:
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
312 syntax_from = syntax_from.lower().strip()
744
312a2842b2b8 plugins text-syntaxes: added a default value to use the current user syntax in convert
souliane <souliane@mailoo.org>
parents: 705
diff changeset
313 if syntax_to == _SYNTAX_CURRENT:
312a2842b2b8 plugins text-syntaxes: added a default value to use the current user syntax in convert
souliane <souliane@mailoo.org>
parents: 705
diff changeset
314 syntax_to = self.getCurrentSyntax(profile)
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
315 else:
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
316 syntax_to = syntax_to.lower().strip()
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
317 syntaxes = TextSyntaxes.syntaxes
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
318 if syntax_from not in syntaxes:
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
319 raise exceptions.NotFound(syntax_from)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
320 if syntax_to not in syntaxes:
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
321 raise exceptions.NotFound(syntax_to)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
322 d = None
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
323
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
324 if TextSyntaxes.OPT_NO_THREAD in syntaxes[syntax_from]["flags"]:
1803
14a97a5fe1c0 plugin text syntaxes: a non blocking syntax callback can now return a unicode directly instead of a Deferred
Goffi <goffi@goffi.org>
parents: 1766
diff changeset
325 d = defer.maybeDeferred(syntaxes[syntax_from]["to"], text)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
326 else:
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
327 d = deferToThread(syntaxes[syntax_from]["to"], text)
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
328
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
329 # TODO: keep only body element and change it to a div here ?
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
330
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
331 if safe:
1805
3c40fa0dcd7a pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents: 1803
diff changeset
332 d.addCallback(self.cleanXHTML)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
333
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
334 if TextSyntaxes.OPT_NO_THREAD in syntaxes[syntax_to]["flags"]:
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
335 d.addCallback(syntaxes[syntax_to]["from"])
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
336 else:
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
337 d.addCallback(lambda xhtml: deferToThread(syntaxes[syntax_to]["from"], xhtml))
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
338
836
2cc0201b4613 plugin text_syntaxes: rstrip the conversion result to avoid new lines systematically added by converters (e.g. html2text do this)
souliane <souliane@mailoo.org>
parents: 832
diff changeset
339 # converters can add new lines that disturb the microblog change detection
2cc0201b4613 plugin text_syntaxes: rstrip the conversion result to avoid new lines systematically added by converters (e.g. html2text do this)
souliane <souliane@mailoo.org>
parents: 832
diff changeset
340 d.addCallback(lambda text: text.rstrip())
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
341 return d
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
342
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
343 def addSyntax(self, name, to_xhtml_cb, from_xhtml_cb, flags=None):
1803
14a97a5fe1c0 plugin text syntaxes: a non blocking syntax callback can now return a unicode directly instead of a Deferred
Goffi <goffi@goffi.org>
parents: 1766
diff changeset
344 """Add a new syntax to the manager
14a97a5fe1c0 plugin text syntaxes: a non blocking syntax callback can now return a unicode directly instead of a Deferred
Goffi <goffi@goffi.org>
parents: 1766
diff changeset
345
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
346 @param name: unique name of the syntax
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
347 @param to_xhtml_cb: callback to convert from syntax to XHTML
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
348 @param from_xhtml_cb: callback to convert from XHTML to syntax
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
349 @param flags: set of optional flags, can be:
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
350 TextSyntaxes.OPT_DEFAULT: use as the default syntax (replace former one)
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
351 TextSyntaxes.OPT_HIDDEN: do not show in parameters
1803
14a97a5fe1c0 plugin text syntaxes: a non blocking syntax callback can now return a unicode directly instead of a Deferred
Goffi <goffi@goffi.org>
parents: 1766
diff changeset
352 TextSyntaxes.OPT_NO_THREAD: do not defer to thread when converting (the callback may then return a deferred)
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
353 """
1805
3c40fa0dcd7a pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents: 1803
diff changeset
354 flags = flags if flags is not None else []
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
355 if TextSyntaxes.OPT_HIDDEN in flags and TextSyntaxes.OPT_DEFAULT in flags:
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
356 raise ValueError(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
357 u"{} and {} are mutually exclusive".format(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
358 TextSyntaxes.OPT_HIDDEN, TextSyntaxes.OPT_DEFAULT
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
359 )
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
360 )
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
361
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
362 syntaxes = TextSyntaxes.syntaxes
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
363 key = name.lower().strip()
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
364 if key in syntaxes:
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
365 raise exceptions.ConflictError(
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
366 u"This syntax key already exists: {}".format(key)
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
367 )
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
368 syntaxes[key] = {
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
369 "name": name,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
370 "to": to_xhtml_cb,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
371 "from": from_xhtml_cb,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
372 "flags": flags,
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
373 }
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
374 if TextSyntaxes.OPT_DEFAULT in flags:
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
375 TextSyntaxes.default_syntaxe = key
665
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
376
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
377 self._updateParamOptions()
6a64e0a759e6 plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff changeset
378
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
379 def getSyntax(self, name):
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
380 """get syntax key corresponding to a name
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
381
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
382 @raise exceptions.NotFound: syntax doesn't exist
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
383 """
2324
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
384 key = name.lower().strip()
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
385 if key in self.syntaxes:
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
386 return key
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
387 raise exceptions.NotFound
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
388
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
389 def _removeMarkups(self, xhtml):
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
390 """Remove XHTML markups from the given string.
fe922e6fabd4 plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
391
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
392 @param xhtml: the XHTML string to be cleaned
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
393 @return: the cleaned string
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
394 """
2624
56f94936df1e code style reformatting using black
Goffi <goffi@goffi.org>
parents: 2562
diff changeset
395 cleaner = clean.Cleaner(kill_tags=["style"])
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
396 cleaned = cleaner.clean_html(html.fromstring(xhtml))
852
4cc55e05266d plugin text syntaxes: fixed cleaners encoding
Goffi <goffi@goffi.org>
parents: 841
diff changeset
397 return html.tostring(cleaned, encoding=unicode, method="text")