Mercurial > libervia-backend
annotate src/plugins/plugin_misc_text_syntaxes.py @ 2182:087eec4c6c07
memory (persistent, sqlite): better private values handling + new LazyPersistentBinaryDict:
- merged private values method handling in sqlite, and added a keys arguments to get only some keys
- LazyPersistentBinaryDict allow to get values in database only when they are needed, saving memory for big data
author | Goffi <goffi@goffi.org> |
---|---|
date | Sun, 12 Mar 2017 19:33:17 +0100 |
parents | 33c8c4973743 |
children | fe922e6fabd4 |
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 |
1766 | 5 # Copyright (C) 2009-2016 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 | 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 |
301b342c697a
core: use of the new core.log module:
Goffi <goffi@goffi.org>
parents:
968
diff
changeset
|
23 log = getLogger(__name__) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
24 |
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 |
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
|
27 from sat.core import exceptions |
1542
94901070478e
plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents:
1458
diff
changeset
|
28 try: |
94901070478e
plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents:
1458
diff
changeset
|
29 from lxml import html |
94901070478e
plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents:
1458
diff
changeset
|
30 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
|
31 except ImportError: |
94901070478e
plugins: added new MissingModule exceptions to plugins using third party modules
Goffi <goffi@goffi.org>
parents:
1458
diff
changeset
|
32 raise exceptions.MissingModule(u"Missing module lxml, please download/install it from http://lxml.de/") |
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
|
33 from cgi import escape |
692
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
34 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
|
35 |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
36 |
771 | 37 CATEGORY = D_("Composition") |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
38 NAME = "Syntax" |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
39 _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
|
40 _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
|
41 |
692
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
42 # TODO: check/adapt following list |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
43 # list initialy based on feedparser list (http://pythonhosted.org/feedparser/html-sanitization.html) |
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
44 STYLES_WHITELIST = ("azimuth", "background-color", "border-bottom-color", "border-collapse", "border-color", "border-left-color", "border-right-color", "border-top-color", "clear", "color", "cursor", "direction", "display", "elevation", "float", "font", "font-family", "font-size", "font-style", "font-variant", "font-weight", "height", "letter-spacing", "line-height", "overflow", "pause", "pause-after", "pause-before", "pitch", "pitch-range", "richness", "speak", "speak-header", "speak-numeral", "speak-punctuation", "speech-rate", "stress", "text-align", "text-decoration", "text-indent", "unicode-bidi", "vertical-align", "voice-family", "volume", "white-space", "width") |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
45 |
1867
47108a4f3a70
plugin text syntaxes: added 'poster' and 'controls' in allowed attributes for cleanXHTML
Goffi <goffi@goffi.org>
parents:
1847
diff
changeset
|
46 SAFE_ATTRS = html.defs.safe_attrs.union(('style', 'poster', 'controls')) |
692
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
47 STYLES_VALUES_REGEX = r'^(' + '|'.join(['([a-z-]+)', # alphabetical names |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
48 '(#[0-9a-f]+)', # hex value |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
49 '(\d+(.\d+)? *(|%|em|ex|px|in|cm|mm|pt|pc))', # values with units (or not) |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
50 'rgb\( *((\d+(.\d+)?), *){2}(\d+(.\d+)?) *\)', # rgb function |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
51 'rgba\( *((\d+(.\d+)?), *){3}(\d+(.\d+)?) *\)', # rgba function |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
52 ]) + ') *(!important)?$' # we accept "!important" at the end |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
53 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
|
54 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
55 PLUGIN_INFO = { |
2145
33c8c4973743
core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents:
2106
diff
changeset
|
56 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
|
57 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
|
58 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
|
59 C.PI_PROTOCOLS: [], |
33c8c4973743
core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents:
2106
diff
changeset
|
60 C.PI_DEPENDENCIES: [], |
33c8c4973743
core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents:
2106
diff
changeset
|
61 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
|
62 C.PI_HANDLER: "no", |
33c8c4973743
core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents:
2106
diff
changeset
|
63 C.PI_DESCRIPTION: _("""Management of various text syntaxes (XHTML-IM, Markdown, etc)""") |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
64 } |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
65 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
66 class UnknownSyntax(Exception): |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
67 pass |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
68 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
69 class TextSyntaxes(object): |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
70 """ Text conversion class |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
71 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
|
72 """ |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
73 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
74 OPT_DEFAULT = "DEFAULT" |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
75 OPT_HIDDEN = "HIDDEN" |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
76 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
|
77 SYNTAX_XHTML = _SYNTAX_XHTML |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
78 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
|
79 SYNTAX_TEXT = "text" |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
80 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
81 params = """ |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
82 <params> |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
83 <individual> |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
84 <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
|
85 <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
|
86 %(options)s |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
87 </param> |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
88 </category> |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
89 </individual> |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
90 </params> |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
91 """ |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
92 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
93 params_data = { |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
94 'category_name': CATEGORY, |
771 | 95 'category_label': _(CATEGORY), |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
96 'name': NAME, |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
97 'label': _(NAME), |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
98 'syntaxes': {}, |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
99 } |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
100 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
101 def __init__(self, host): |
993
301b342c697a
core: use of the new core.log module:
Goffi <goffi@goffi.org>
parents:
968
diff
changeset
|
102 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
|
103 self.host = host |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
104 self.syntaxes = {} |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
105 self.addSyntax(self.SYNTAX_XHTML, lambda xhtml: defer.succeed(xhtml), lambda xhtml: defer.succeed(xhtml), |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
106 TextSyntaxes.OPT_NO_THREAD) |
1826
d80ccf4bf201
plugin blog import dotclear: this plugin import Dotclear 2 backups
Goffi <goffi@goffi.org>
parents:
1811
diff
changeset
|
107 # 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
|
108 # it's probably best to move sat_frontends.tools.strings to sat.tools.common or similar |
832
c4b22aedb7d7
plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents:
811
diff
changeset
|
109 self.addSyntax(self.SYNTAX_TEXT, lambda text: escape(text), lambda xhtml: self._removeMarkups(xhtml), [TextSyntaxes.OPT_HIDDEN]) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
110 try: |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
111 import markdown, html2text |
841
831f208b4ea3
plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents:
836
diff
changeset
|
112 |
831f208b4ea3
plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents:
836
diff
changeset
|
113 def _html2text(html, baseurl=''): |
831f208b4ea3
plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents:
836
diff
changeset
|
114 h = html2text.HTML2Text(baseurl=baseurl) |
831f208b4ea3
plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents:
836
diff
changeset
|
115 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
|
116 return h.handle(html) |
831f208b4ea3
plugin text_syntaxes: html2text was breaking the long URLs
souliane <souliane@mailoo.org>
parents:
836
diff
changeset
|
117 self.addSyntax(self.SYNTAX_MARKDOWN, markdown.markdown, _html2text, [TextSyntaxes.OPT_DEFAULT]) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
118 except ImportError: |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
119 log.warning(u"markdown or html2text not found, can't use Markdown syntax") |
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
120 log.info(u"You can download/install them from https://pythonhosted.org/Markdown/ and https://github.com/Alir3z4/html2text/") |
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
|
121 host.bridge.addMethod("syntaxConvert", ".plugin", in_sign='sssbs', out_sign='s', |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
122 async=True, method=self.convert) |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
123 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
124 def _updateParamOptions(self): |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
125 data_synt = TextSyntaxes.params_data['syntaxes'] |
1847
d8c6c55aafd7
plugin text syntaxes: fixed defaut syntax handling
Goffi <goffi@goffi.org>
parents:
1826
diff
changeset
|
126 default_synt = TextSyntaxes.params_data.get('default', _SYNTAX_XHTML) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
127 syntaxes = [] |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
128 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
129 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
|
130 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
|
131 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
|
132 syntaxes.append(syntax) |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
133 |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
134 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
|
135 options = [] |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
136 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
137 for syntax in syntaxes: |
1847
d8c6c55aafd7
plugin text syntaxes: fixed defaut syntax handling
Goffi <goffi@goffi.org>
parents:
1826
diff
changeset
|
138 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
|
139 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
|
140 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
141 TextSyntaxes.params_data["options"] = u'\n'.join(options) |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
142 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
|
143 |
702
a25db3fe3959
plugin XEP-0071: rich messages management for sendMessage
Goffi <goffi@goffi.org>
parents:
699
diff
changeset
|
144 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
|
145 """ 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
|
146 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
147 @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
|
148 @return: profile selected syntax |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
149 """ |
699
e819630c6737
plugin text syntaxes: fixed bad parameter order for getParamA
Goffi <goffi@goffi.org>
parents:
695
diff
changeset
|
150 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
|
151 |
2106
5874da3811b7
plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents:
1934
diff
changeset
|
152 def _logError(self, failure, action=u"converting syntax"): |
5874da3811b7
plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents:
1934
diff
changeset
|
153 log.error(u"Error while {action}: {failure}".format(action=action, failure=failure)) |
5874da3811b7
plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents:
1934
diff
changeset
|
154 return failure |
5874da3811b7
plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents:
1934
diff
changeset
|
155 |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
156 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
|
157 """ 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
|
158 @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
|
159 """ |
674
fb0b1100c908
plugin text_syntaxes: fixed clean_xhml (it now return XHTML instead of HTML)
Goffi <goffi@goffi.org>
parents:
665
diff
changeset
|
160 def blocking_cleaning(xhtml): |
692
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
161 """ Clean XHTML and style attributes """ |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
162 |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
163 def clean_style(styles_raw): |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
164 """" Remove styles not in the whitelist, |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
165 or where the value doesn't match the regex """ |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
166 styles = styles_raw.split(";") |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
167 cleaned_styles = [] |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
168 for style in styles: |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
169 try: |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
170 key, value = style.split(':') |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
171 except ValueError: |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
172 continue |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
173 key = key.lower().strip() |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
174 if key not in STYLES_WHITELIST: |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
175 continue |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
176 value = value.lower().strip() |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
177 if not STYLES_ACCEPTED_VALUE.match(value): |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
178 continue |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
179 if value == "none": |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
180 continue |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
181 cleaned_styles.append((key, value)) |
1458
832846fefe85
plugin text-syntaxes: minor variable renaming
Goffi <goffi@goffi.org>
parents:
1396
diff
changeset
|
182 return "; ".join(["%s: %s" % (key_, value_) for key_, value_ in cleaned_styles]) |
692
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
183 |
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
|
184 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
|
185 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
|
186 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
|
187 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
|
188 else: |
993
301b342c697a
core: use of the new core.log module:
Goffi <goffi@goffi.org>
parents:
968
diff
changeset
|
189 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
|
190 raise exceptions.DataError |
674
fb0b1100c908
plugin text_syntaxes: fixed clean_xhml (it now return XHTML instead of HTML)
Goffi <goffi@goffi.org>
parents:
665
diff
changeset
|
191 cleaner = clean.Cleaner(style=False, |
fb0b1100c908
plugin text_syntaxes: fixed clean_xhml (it now return XHTML instead of HTML)
Goffi <goffi@goffi.org>
parents:
665
diff
changeset
|
192 add_nofollow=False, |
692
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
193 safe_attrs=SAFE_ATTRS) |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
194 xhtml_elt = cleaner.clean_html(xhtml_elt) |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
195 for elt in xhtml_elt.xpath("//*[@style]"): |
e98db42cd78c
plugin text syntaxes: styles sanitisation
Goffi <goffi@goffi.org>
parents:
674
diff
changeset
|
196 elt.set("style", clean_style(elt.get('style'))) |
852
4cc55e05266d
plugin text syntaxes: fixed cleaners encoding
Goffi <goffi@goffi.org>
parents:
841
diff
changeset
|
197 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
|
198 |
2106
5874da3811b7
plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents:
1934
diff
changeset
|
199 d = deferToThread(blocking_cleaning, xhtml) |
5874da3811b7
plugin text syntaxes: log error on cleanXHTML failure
Goffi <goffi@goffi.org>
parents:
1934
diff
changeset
|
200 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
|
201 return d |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
202 |
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
|
203 def convert(self, text, syntax_from, syntax_to=_SYNTAX_XHTML, safe=True, 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
|
204 """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
|
205 |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
206 @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
|
207 @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
|
208 @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
|
209 @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
|
210 @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
|
211 @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
|
212 """ |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
213 # 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
|
214 # when dealing with XHTML |
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
215 # 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
|
216 |
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
|
217 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
|
218 syntax_from = self.getCurrentSyntax(profile) |
312a2842b2b8
plugins text-syntaxes: added a default value to use the current user syntax in convert
souliane <souliane@mailoo.org>
parents:
705
diff
changeset
|
219 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
|
220 syntax_to = self.getCurrentSyntax(profile) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
221 syntaxes = TextSyntaxes.params_data['syntaxes'] |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
222 if syntax_from not in syntaxes: |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
223 raise UnknownSyntax(syntax_from) |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
224 if syntax_to not in syntaxes: |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
225 raise UnknownSyntax(syntax_to) |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
226 d = None |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
227 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
228 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
|
229 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
|
230 else: |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
231 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
|
232 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
233 #TODO: keep only body element and change it to a div here ? |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
234 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
235 if safe: |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
236 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
|
237 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
238 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
|
239 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
|
240 else: |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
241 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
|
242 |
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
|
243 # 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
|
244 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
|
245 return d |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
246 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
247 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
|
248 """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
|
249 |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
250 @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
|
251 @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
|
252 @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
|
253 @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
|
254 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
|
255 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
|
256 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
|
257 """ |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
258 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
|
259 if TextSyntaxes.OPT_HIDDEN in flags and TextSyntaxes.OPT_DEFAULT in 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
|
260 raise ValueError(u"{} and {} are mutually exclusive".format(TextSyntaxes.OPT_HIDDEN, TextSyntaxes.OPT_DEFAULT)) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
261 |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
262 syntaxes = TextSyntaxes.params_data['syntaxes'] |
1805
3c40fa0dcd7a
pluging text syntaxes: various minor improvments:
Goffi <goffi@goffi.org>
parents:
1803
diff
changeset
|
263 if name in syntaxes: |
1811
90ae9a2462c2
plugin text syntaxes: fixes typo
souliane <souliane@mailoo.org>
parents:
1805
diff
changeset
|
264 raise exceptions.ConflictError(u"This syntax name already exists: {}".format(name)) |
665
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
265 syntaxes[name] = {"to": to_xhtml_cb, "from": from_xhtml_cb, "flags": flags} |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
266 if TextSyntaxes.OPT_DEFAULT in flags: |
1847
d8c6c55aafd7
plugin text syntaxes: fixed defaut syntax handling
Goffi <goffi@goffi.org>
parents:
1826
diff
changeset
|
267 TextSyntaxes.params_data['default'] = name |
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 self._updateParamOptions() |
6a64e0a759e6
plugin text syntaxes: this plugin manage rich text syntaxes conversions and cleaning.
Goffi <goffi@goffi.org>
parents:
diff
changeset
|
270 |
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
|
271 def _removeMarkups(self, xhtml): |
c4b22aedb7d7
plugin groupblog, XEP-0071, XEP-0277, text_syntaxes: manage raw/rich/xhtml data for content/title:
souliane <souliane@mailoo.org>
parents:
811
diff
changeset
|
272 """ |
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
|
273 Remove XHTML markups from the given 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
|
274 @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
|
275 @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
|
276 """ |
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
|
277 cleaner = clean.Cleaner(kill_tags=['style']) |
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
|
278 cleaned = cleaner.clean_html(html.fromstring(xhtml)) |
852
4cc55e05266d
plugin text syntaxes: fixed cleaners encoding
Goffi <goffi@goffi.org>
parents:
841
diff
changeset
|
279 return html.tostring(cleaned, encoding=unicode, method="text") |