Mercurial > libervia-backend
annotate sat/plugins/plugin_misc_text_syntaxes.py @ 3788:12317ba98d99
cli: new `blocking` commands:
commands to list blocked entities, block or unblock one or more entities
rel 367
author | Goffi <goffi@goffi.org> |
---|---|
date | Fri, 27 May 2022 12:15:04 +0200 |
parents | 33d75cd3c371 |
children | 524856bd7b19 |
rev | line source |
---|---|
3028 | 1 #!/usr/bin/env python3 |
3137 | 2 |
665
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 |
3479 | 5 # Copyright (C) 2009-2021 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 |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
20 from functools import partial |
3075
501a1a3c8594
plugin text syntaxes: don't use anymore deprecated cgi.escape
Goffi <goffi@goffi.org>
parents:
3040
diff
changeset
|
21 from html import escape |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
22 import re |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
23 from typing import Set |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
24 |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
25 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
|
26 from twisted.internet.threads import deferToThread |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
27 |
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 |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
29 from sat.core.constants import Const as C |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
30 from sat.core.i18n import D_, _ |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
31 from sat.core.log import getLogger |
2782
b17e6fa1e607
core (XMLUI): new XHTMLBox widget:
Goffi <goffi@goffi.org>
parents:
2781
diff
changeset
|
32 from sat.tools import xml_tools |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
33 |
1542
94901070478e
plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents:
1458
diff
changeset
|
34 try: |
94901070478e
plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents:
1458
diff
changeset
|
35 from lxml import html |
94901070478e
plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents:
1458
diff
changeset
|
36 from lxml.html import clean |
2786
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
37 from lxml import etree |
1542
94901070478e
plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents:
1458
diff
changeset
|
38 except ImportError: |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
39 raise exceptions.MissingModule( |
3028 | 40 "Missing module lxml, please download/install it from http://lxml.de/" |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
41 ) |
674
fb0b1100c908
plugin text_syntaxes: fixed clean_xhml (it now return XHTML instead of HTML)
Goffi <goffi@goffi.org>
parents:
665
diff
changeset
|
42 |
2873
e1207b8ad97c
plugin text syntaxes: disable raw HTML parsing in mardown by default
Goffi <goffi@goffi.org>
parents:
2869
diff
changeset
|
43 log = getLogger(__name__) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
44 |
771 | 45 CATEGORY = D_("Composition") |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
46 NAME = "Syntax" |
2869
148d30147890
plugin text syntaxes: fixed default syntax
Goffi <goffi@goffi.org>
parents:
2786
diff
changeset
|
47 _SYNTAX_XHTML = "xhtml" # must be lower case |
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
|
48 _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
|
49 |
692
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
50 # TODO: check/adapt following list |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
51 # 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
|
52 STYLES_WHITELIST = ( |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
53 "azimuth", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
54 "background-color", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
55 "border-bottom-color", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
56 "border-collapse", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
57 "border-color", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
58 "border-left-color", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
59 "border-right-color", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
60 "border-top-color", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
61 "clear", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
62 "color", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
63 "cursor", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
64 "direction", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
65 "display", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
66 "elevation", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
67 "float", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
68 "font", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
69 "font-family", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
70 "font-size", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
71 "font-style", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
72 "font-variant", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
73 "font-weight", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
74 "height", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
75 "letter-spacing", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
76 "line-height", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
77 "overflow", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
78 "pause", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
79 "pause-after", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
80 "pause-before", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
81 "pitch", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
82 "pitch-range", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
83 "richness", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
84 "speak", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
85 "speak-header", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
86 "speak-numeral", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
87 "speak-punctuation", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
88 "speech-rate", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
89 "stress", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
90 "text-align", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
91 "text-decoration", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
92 "text-indent", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
93 "unicode-bidi", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
94 "vertical-align", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
95 "voice-family", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
96 "volume", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
97 "white-space", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
98 "width", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
99 ) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
100 |
2786
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
101 # cf. https://www.w3.org/TR/html/syntax.html#void-elements |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
102 VOID_ELEMENTS = ( |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
103 "area", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
104 "base", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
105 "br", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
106 "col", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
107 "embed", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
108 "hr", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
109 "img", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
110 "input", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
111 "keygen", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
112 "link", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
113 "menuitem", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
114 "meta", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
115 "param", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
116 "source", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
117 "track", |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
118 "wbr") |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
119 |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
120 SAFE_ATTRS = html.defs.safe_attrs.union({"style", "poster", "controls"}) - {"id"} |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
121 SAFE_CLASSES = { |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
122 # those classes are used for code highlighting |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
123 "bp", "c", "ch", "cm", "cp", "cpf", "cs", "dl", "err", "fm", "gd", "ge", "get", "gh", |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
124 "gi", "go", "gp", "gr", "gs", "gt", "gu", "highlight", "hll", "il", "k", "kc", "kd", |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
125 "kn", "kp", "kr", "kt", "m", "mb", "mf", "mh", "mi", "mo", "na", "nb", "nc", "nd", |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
126 "ne", "nf", "ni", "nl", "nn", "no", "nt", "nv", "o", "ow", "s", "sa", "sb", "sc", |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
127 "sd", "se", "sh", "si", "sr", "ss", "sx", "vc", "vg", "vi", "vm", "w", "write", |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
128 } |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
129 STYLES_VALUES_REGEX = ( |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
130 r"^(" |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
131 + "|".join( |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
132 [ |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
133 "([a-z-]+)", # alphabetical names |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
134 "(#[0-9a-f]+)", # hex value |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
135 "(\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
|
136 "rgb\( *((\d+(.\d+)?), *){2}(\d+(.\d+)?) *\)", # rgb function |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
137 "rgba\( *((\d+(.\d+)?), *){3}(\d+(.\d+)?) *\)", # rgba function |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
138 ] |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
139 ) |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
140 + ") *(!important)?$" |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
141 ) # we accept "!important" at the end |
692
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
142 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
|
143 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
144 PLUGIN_INFO = { |
2145
33c8c4973743
core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents:
2106
diff
changeset
|
145 C.PI_NAME: "Text syntaxes", |
2780
85d3240a400f
plugin text syntaxes: changed import name to TEXT_SYNTAX (better with underscore for autocompletion)
Goffi <goffi@goffi.org>
parents:
2771
diff
changeset
|
146 C.PI_IMPORT_NAME: "TEXT_SYNTAXES", |
2145
33c8c4973743
core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents:
2106
diff
changeset
|
147 C.PI_TYPE: "MISC", |
3726
33d75cd3c371
plugin XEP-0060, XEP-0163, XEP-0277, text syntaxes: make those plugins usable with components
Goffi <goffi@goffi.org>
parents:
3709
diff
changeset
|
148 C.PI_MODES: C.PLUG_MODE_BOTH, |
2145
33c8c4973743
core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents:
2106
diff
changeset
|
149 C.PI_PROTOCOLS: [], |
33c8c4973743
core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents:
2106
diff
changeset
|
150 C.PI_DEPENDENCIES: [], |
33c8c4973743
core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents:
2106
diff
changeset
|
151 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
|
152 C.PI_HANDLER: "no", |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
153 C.PI_DESCRIPTION: _( |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
154 """Management of various text syntaxes (XHTML-IM, Markdown, etc)""" |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
155 ), |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
156 } |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
157 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
158 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
159 class TextSyntaxes(object): |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
160 """ Text conversion class |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
161 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
|
162 """ |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
163 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
164 OPT_DEFAULT = "DEFAULT" |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
165 OPT_HIDDEN = "HIDDEN" |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
166 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
|
167 SYNTAX_XHTML = _SYNTAX_XHTML |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
168 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
|
169 SYNTAX_TEXT = "text" |
2869
148d30147890
plugin text syntaxes: fixed default syntax
Goffi <goffi@goffi.org>
parents:
2786
diff
changeset
|
170 # default_syntax must be lower case |
2324
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
171 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
|
172 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
173 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
174 def __init__(self, host): |
993
301b342c697a
core: use of the new core.log module:
Goffi <goffi@goffi.org>
parents:
968
diff
changeset
|
175 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
|
176 self.host = host |
3620
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
177 self.syntaxes = {} |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
178 |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
179 self.params = """ |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
180 <params> |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
181 <individual> |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
182 <category name="%(category_name)s" label="%(category_label)s"> |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
183 <param name="%(name)s" label="%(label)s" type="list" security="0"> |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
184 %(options)s |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
185 </param> |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
186 </category> |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
187 </individual> |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
188 </params> |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
189 """ |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
190 |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
191 self.params_data = { |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
192 "category_name": CATEGORY, |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
193 "category_label": _(CATEGORY), |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
194 "name": NAME, |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
195 "label": _(NAME), |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
196 "syntaxes": self.syntaxes, |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
197 } |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
198 |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
199 self.addSyntax( |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
200 self.SYNTAX_XHTML, |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
201 lambda xhtml: defer.succeed(xhtml), |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
202 lambda xhtml: defer.succeed(xhtml), |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
203 TextSyntaxes.OPT_NO_THREAD, |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
204 ) |
1826
d80ccf4bf201
plugin blog import dotclear: this plugin import Dotclear 2 backups
Goffi <goffi@goffi.org>
parents:
1811
diff
changeset
|
205 # 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
|
206 # 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
|
207 self.addSyntax( |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
208 self.SYNTAX_TEXT, |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
209 lambda text: escape(text), |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
210 lambda xhtml: self._removeMarkups(xhtml), |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
211 [TextSyntaxes.OPT_HIDDEN], |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
212 ) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
213 try: |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
214 import markdown, html2text |
2873
e1207b8ad97c
plugin text syntaxes: disable raw HTML parsing in mardown by default
Goffi <goffi@goffi.org>
parents:
2869
diff
changeset
|
215 from markdown.extensions import Extension |
e1207b8ad97c
plugin text syntaxes: disable raw HTML parsing in mardown by default
Goffi <goffi@goffi.org>
parents:
2869
diff
changeset
|
216 |
e1207b8ad97c
plugin text syntaxes: disable raw HTML parsing in mardown by default
Goffi <goffi@goffi.org>
parents:
2869
diff
changeset
|
217 # XXX: we disable raw HTML parsing by default, to avoid parsing error |
e1207b8ad97c
plugin text syntaxes: disable raw HTML parsing in mardown by default
Goffi <goffi@goffi.org>
parents:
2869
diff
changeset
|
218 # when the user is not aware of markdown and HTML |
e1207b8ad97c
plugin text syntaxes: disable raw HTML parsing in mardown by default
Goffi <goffi@goffi.org>
parents:
2869
diff
changeset
|
219 class EscapeHTML(Extension): |
e1207b8ad97c
plugin text syntaxes: disable raw HTML parsing in mardown by default
Goffi <goffi@goffi.org>
parents:
2869
diff
changeset
|
220 def extendMarkdown(self, md): |
e1207b8ad97c
plugin text syntaxes: disable raw HTML parsing in mardown by default
Goffi <goffi@goffi.org>
parents:
2869
diff
changeset
|
221 md.preprocessors.deregister('html_block') |
e1207b8ad97c
plugin text syntaxes: disable raw HTML parsing in mardown by default
Goffi <goffi@goffi.org>
parents:
2869
diff
changeset
|
222 md.inlinePatterns.deregister('html') |
841
831f208b4ea3
plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents:
836
diff
changeset
|
223 |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
224 def _html2text(html, baseurl=""): |
841
831f208b4ea3
plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents:
836
diff
changeset
|
225 h = html2text.HTML2Text(baseurl=baseurl) |
831f208b4ea3
plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents:
836
diff
changeset
|
226 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
|
227 return h.handle(html) |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
228 |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
229 self.addSyntax( |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
230 self.SYNTAX_MARKDOWN, |
2878
a02ad4bc0a6d
plugin text syntaxes: activated useful markdown extensions:
Goffi <goffi@goffi.org>
parents:
2873
diff
changeset
|
231 partial(markdown.markdown, |
a02ad4bc0a6d
plugin text syntaxes: activated useful markdown extensions:
Goffi <goffi@goffi.org>
parents:
2873
diff
changeset
|
232 extensions=[ |
a02ad4bc0a6d
plugin text syntaxes: activated useful markdown extensions:
Goffi <goffi@goffi.org>
parents:
2873
diff
changeset
|
233 EscapeHTML(), |
3028 | 234 'nl2br', |
235 'codehilite', | |
236 'fenced_code', | |
237 'sane_lists', | |
238 'tables', | |
2878
a02ad4bc0a6d
plugin text syntaxes: activated useful markdown extensions:
Goffi <goffi@goffi.org>
parents:
2873
diff
changeset
|
239 ], |
a02ad4bc0a6d
plugin text syntaxes: activated useful markdown extensions:
Goffi <goffi@goffi.org>
parents:
2873
diff
changeset
|
240 extension_configs = { |
3028 | 241 "codehilite": { |
242 "css_class": "highlight", | |
2878
a02ad4bc0a6d
plugin text syntaxes: activated useful markdown extensions:
Goffi <goffi@goffi.org>
parents:
2873
diff
changeset
|
243 } |
a02ad4bc0a6d
plugin text syntaxes: activated useful markdown extensions:
Goffi <goffi@goffi.org>
parents:
2873
diff
changeset
|
244 }), |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
245 _html2text, |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
246 [TextSyntaxes.OPT_DEFAULT], |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
247 ) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
248 except ImportError: |
3028 | 249 log.warning("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
|
250 log.info( |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
251 "You can download/install them from https://pythonhosted.org/Markdown/ " |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
252 "and https://github.com/Alir3z4/html2text/" |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
253 ) |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
254 host.bridge.addMethod( |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
255 "syntaxConvert", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
256 ".plugin", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
257 in_sign="sssbs", |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
258 out_sign="s", |
3028 | 259 async_=True, |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
260 method=self.convert, |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
261 ) |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
262 host.bridge.addMethod( |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
263 "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
|
264 ) |
2782
b17e6fa1e607
core (XMLUI): new XHTMLBox widget:
Goffi <goffi@goffi.org>
parents:
2781
diff
changeset
|
265 if xml_tools.cleanXHTML is None: |
3028 | 266 log.debug("Installing cleaning method") |
2782
b17e6fa1e607
core (XMLUI): new XHTMLBox widget:
Goffi <goffi@goffi.org>
parents:
2781
diff
changeset
|
267 xml_tools.cleanXHTML = self.cleanXHTML |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
268 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
269 def _updateParamOptions(self): |
3620
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
270 data_synt = self.syntaxes |
2324
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
271 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
|
272 syntaxes = [] |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
273 |
3028 | 274 for syntax in list(data_synt.keys()): |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
275 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
|
276 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
|
277 syntaxes.append(syntax) |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
278 |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
279 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
|
280 options = [] |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
281 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
282 for syntax in syntaxes: |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
283 selected = 'selected="true"' if syntax == default_synt else "" |
3028 | 284 options.append('<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
|
285 |
3620
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
286 self.params_data["options"] = "\n".join(options) |
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
287 self.host.memory.updateParams(self.params % self.params_data) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
288 |
702
a25db3fe3959
plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents:
699
diff
changeset
|
289 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
|
290 """ 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
|
291 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
292 @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
|
293 @return: profile selected syntax |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
294 """ |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
295 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
|
296 |
3028 | 297 def _logError(self, failure, action="converting syntax"): |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
298 log.error( |
3028 | 299 "Error while {action}: {failure}".format(action=action, failure=failure) |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
300 ) |
2106
5874da3811b7
plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents:
1934
diff
changeset
|
301 return failure |
5874da3811b7
plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents:
1934
diff
changeset
|
302 |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
303 def cleanStyle(self, styles_raw: str) -> str: |
2781
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
304 """"Clean unsafe CSS styles |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
305 |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
306 Remove styles not in the whitelist, or where the value doesn't match the regex |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
307 @param styles_raw: CSS styles |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
308 @return: cleaned styles |
2781
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
309 """ |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
310 styles: List[str] = styles_raw.split(";") |
2781
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
311 cleaned_styles = [] |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
312 for style in styles: |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
313 try: |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
314 key, value = style.split(":") |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
315 except ValueError: |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
316 continue |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
317 key = key.lower().strip() |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
318 if key not in STYLES_WHITELIST: |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
319 continue |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
320 value = value.lower().strip() |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
321 if not STYLES_ACCEPTED_VALUE.match(value): |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
322 continue |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
323 if value == "none": |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
324 continue |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
325 cleaned_styles.append((key, value)) |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
326 return "; ".join( |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
327 ["%s: %s" % (key_, value_) for key_, value_ in cleaned_styles] |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
328 ) |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
329 |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
330 def cleanClasses(self, classes_raw: str) -> str: |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
331 """Remove any non whitelisted class |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
332 |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
333 @param classes_raw: classes set on an element |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
334 @return: remaining classes (can be empty string) |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
335 """ |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
336 return " ".join(SAFE_CLASSES.intersection(classes_raw.split())) |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
337 |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
338 def cleanXHTML(self, xhtml): |
2781
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
339 """Clean XHTML text by removing potentially dangerous/malicious parts |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
340 |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
341 @param xhtml(unicode, lxml.etree._Element): raw HTML/XHTML text to clean |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
342 @return (unicode): cleaned XHTML |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
343 """ |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
344 |
3028 | 345 if isinstance(xhtml, str): |
2786
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
346 try: |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
347 xhtml_elt = html.fromstring(xhtml) |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
348 except etree.ParserError as e: |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
349 if not xhtml.strip(): |
3028 | 350 return "" |
351 log.error("Can't clean XHTML: {xhtml}".format(xhtml=xhtml)) | |
2786
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
352 raise e |
2781
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
353 elif isinstance(xhtml, html.HtmlElement): |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
354 xhtml_elt = xhtml |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
355 else: |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
356 log.error("Only strings and HtmlElements can be cleaned") |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
357 raise exceptions.DataError |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
358 cleaner = clean.Cleaner( |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
359 style=False, add_nofollow=False, safe_attrs=SAFE_ATTRS |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
360 ) |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
361 xhtml_elt = cleaner.clean_html(xhtml_elt) |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
362 for elt in xhtml_elt.xpath("//*[@style]"): |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
363 elt.set("style", self.cleanStyle(elt.get("style"))) |
3693
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
364 for elt in xhtml_elt.xpath("//*[@class]"): |
0bbdc50aa405
plugin text syntaxes: remove `id` attributes and whitelist allowed classes:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
365 elt.set("class", self.cleanClasses(elt.get("class"))) |
2786
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
366 # we remove self-closing elements for non-void elements |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
367 for element in xhtml_elt.iter(tag=etree.Element): |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
368 if not element.text: |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
369 if element.tag in VOID_ELEMENTS: |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
370 element.text = None |
be8405795e09
plugin text syntaxes: handle empty content in cleanXHTML + don't use self-closing tags for non-void elements.
Goffi <goffi@goffi.org>
parents:
2782
diff
changeset
|
371 else: |
3028 | 372 element.text = '' |
373 return html.tostring(xhtml_elt, encoding=str, method="xml") | |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
374 |
3040 | 375 def convert(self, text, syntax_from, syntax_to=_SYNTAX_XHTML, safe=True, |
376 profile=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
|
377 """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
|
378 |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
379 @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
|
380 @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
|
381 @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
|
382 @param safe: clean resulting XHTML to avoid malicious code if True |
2781
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
383 @param profile: needed only when syntax_from or syntax_to is set to |
816be0a23877
plugin text syntaxes: cleanStyle is an independent method, cleanXHTML is now blocking (no need to launch thread for that)
Goffi <goffi@goffi.org>
parents:
2780
diff
changeset
|
384 _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
|
385 @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
|
386 """ |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
387 # 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
|
388 # when dealing with XHTML |
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
389 # 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
|
390 |
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
|
391 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
|
392 syntax_from = self.getCurrentSyntax(profile) |
2324
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
393 else: |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
394 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
|
395 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
|
396 syntax_to = self.getCurrentSyntax(profile) |
2324
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
397 else: |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
398 syntax_to = syntax_to.lower().strip() |
3620
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
399 syntaxes = self.syntaxes |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
400 if syntax_from not in syntaxes: |
2324
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
401 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
|
402 if syntax_to not in syntaxes: |
2324
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
403 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
|
404 d = None |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
405 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
406 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
|
407 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
|
408 else: |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
409 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
|
410 |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
411 # 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
|
412 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
413 if safe: |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
414 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
|
415 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
416 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
|
417 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
|
418 else: |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
419 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
|
420 |
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
|
421 # 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
|
422 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
|
423 return d |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
424 |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
425 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
|
426 """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
|
427 |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
428 @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
|
429 @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
|
430 @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
|
431 @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
|
432 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
|
433 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
|
434 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
|
435 """ |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
436 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
|
437 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
|
438 raise ValueError( |
3028 | 439 "{} and {} are mutually exclusive".format( |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
440 TextSyntaxes.OPT_HIDDEN, TextSyntaxes.OPT_DEFAULT |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
441 ) |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
442 ) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
443 |
3620
f568f304c982
plugin text syntaxes: remove side effect on init:
Goffi <goffi@goffi.org>
parents:
3479
diff
changeset
|
444 syntaxes = self.syntaxes |
2324
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
445 key = name.lower().strip() |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
446 if key in syntaxes: |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
447 raise exceptions.ConflictError( |
3028 | 448 "This syntax key already exists: {}".format(key) |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
449 ) |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
450 syntaxes[key] = { |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
451 "name": name, |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
452 "to": to_xhtml_cb, |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
453 "from": from_xhtml_cb, |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
454 "flags": flags, |
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
455 } |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
456 if TextSyntaxes.OPT_DEFAULT in flags: |
2869
148d30147890
plugin text syntaxes: fixed default syntax
Goffi <goffi@goffi.org>
parents:
2786
diff
changeset
|
457 TextSyntaxes.default_syntax = key |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
458 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
459 self._updateParamOptions() |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
460 |
2324
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
461 def getSyntax(self, name): |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
462 """get syntax key corresponding to a name |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
463 |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
464 @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
|
465 """ |
2324
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
466 key = name.lower().strip() |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
467 if key in self.syntaxes: |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
468 return key |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
469 raise exceptions.NotFound |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
470 |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
471 def _removeMarkups(self, xhtml): |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
472 """Remove XHTML markups from the given string. |
fe922e6fabd4
plugin text syntaxes: various improvments:
Goffi <goffi@goffi.org>
parents:
2145
diff
changeset
|
473 |
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
|
474 @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
|
475 @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
|
476 """ |
2624
56f94936df1e
code style reformatting using black
Goffi <goffi@goffi.org>
parents:
2562
diff
changeset
|
477 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
|
478 cleaned = cleaner.clean_html(html.fromstring(xhtml)) |
3028 | 479 return html.tostring(cleaned, encoding=str, method="text") |