Mercurial > libervia-backend

annotate src/plugins/plugin_blog_import.py @ 2432:c6d859df4b27

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
plugin XEP-0060: use new findServiceEntity to get default pubsub service
author Goffi <goffi@goffi.org>
date Wed, 15 Nov 2017 08:55:47 +0100
parents 8b37a62336c3
children 91bbad17fd53
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1934
2daf7b4c6756 use of /usr/bin/env instead of /usr/bin/python in shebang
Goffi <goffi@goffi.org>
parents: 1852
diff changeset
1 #!/usr/bin/env python2
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
2 # -*- coding: utf-8 -*-
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
3
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
4 # SàT plugin for import external blogs
2414
8b37a62336c3 misc: date update (yes it's a bit late :p )
Goffi <goffi@goffi.org>
parents: 2396
diff changeset
5 # Copyright (C) 2009-2017 Jérôme Poisson (goffi@goffi.org)
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
6
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
7 # This program is free software: you can redistribute it and/or modify
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
8 # it under the terms of the GNU Affero General Public License as published by
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
9 # the Free Software Foundation, either version 3 of the License, or
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
10 # (at your option) any later version.
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
11
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
12 # This program is distributed in the hope that it will be useful,
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
13 # but WITHOUT ANY WARRANTY; without even the implied warranty of
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
14 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
15 # GNU Affero General Public License for more details.
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
16
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
17 # You should have received a copy of the GNU Affero General Public License
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
18 # along with this program. If not, see <http://www.gnu.org/licenses/>.
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
19
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
20
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
21 from sat.core.i18n import _
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
22 from sat.core.constants import Const as C
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
23 from sat.core.log import getLogger
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
24 log = getLogger(__name__)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
25 from twisted.internet import defer
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
26 from twisted.web import client as web_client
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
27 from twisted.words.xish import domish
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
28 from sat.core import exceptions
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
29 from sat.tools import xml_tools
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
30 import os
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
31 import os.path
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
32 import tempfile
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
33 import urlparse
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
34 import shortuuid
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
35
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
36
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
37 PLUGIN_INFO = {
2145
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2108
diff changeset
38 C.PI_NAME: "blog import",
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2108
diff changeset
39 C.PI_IMPORT_NAME: "BLOG_IMPORT",
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
40 C.PI_TYPE: (C.PLUG_TYPE_BLOG, C.PLUG_TYPE_IMPORT),
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
41 C.PI_DEPENDENCIES: ["IMPORT", "XEP-0060", "XEP-0277", "TEXT-SYNTAXES", "UPLOAD"],
2145
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2108
diff changeset
42 C.PI_MAIN: "BlogImportPlugin",
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2108
diff changeset
43 C.PI_HANDLER: "no",
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2108
diff changeset
44 C.PI_DESCRIPTION: _(u"""Blog import management:
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
45 This plugin manage the different blog importers which can register to it, and handle generic importing tasks.""")
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
46 }
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
47
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
48 OPT_HOST = 'host'
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
49 OPT_UPLOAD_IMAGES = 'upload_images'
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
50 OPT_UPLOAD_IGNORE_HOST = 'upload_ignore_host'
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
51 OPT_IGNORE_TLS = 'ignore_tls_errors'
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
52 URL_REDIRECT_PREFIX = 'url_redirect_'
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
53
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
54
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
55 class BlogImportPlugin(object):
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
56 BOOL_OPTIONS = (OPT_UPLOAD_IMAGES, OPT_IGNORE_TLS)
2396
66baa687c682 plugins tickets import, jp (ticket/import): implemented mapping:
Goffi <goffi@goffi.org>
parents: 2370
diff changeset
57 JSON_OPTIONS = ()
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
58 OPT_DEFAULTS = {OPT_UPLOAD_IMAGES: True,
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
59 OPT_IGNORE_TLS: False}
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
60
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
61 def __init__(self, host):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
62 log.info(_("plugin Blog Import initialization"))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
63 self.host = host
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
64 self._u = host.plugins['UPLOAD']
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
65 self._p = host.plugins['XEP-0060']
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
66 self._m = host.plugins['XEP-0277']
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
67 self._s = self.host.plugins['TEXT-SYNTAXES']
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
68 host.plugins['IMPORT'].initialize(self, u'blog')
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
69
2370
2c2b826b0bb3 plugin import: node can now be specified + added a "session" dict to keep import session data:
Goffi <goffi@goffi.org>
parents: 2369
diff changeset
70 def importItem(self, client, item_import_data, session, options, return_data, service, node):
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
71 """importItem specialized for blog import
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
72
2370
2c2b826b0bb3 plugin import: node can now be specified + added a "session" dict to keep import session data:
Goffi <goffi@goffi.org>
parents: 2369
diff changeset
73 @param item_import_data(dict):
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
74 * mandatory keys:
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
75 'blog' (dict): microblog data of the blog post (cf. http://wiki.goffi.org/wiki/Bridge_API_-_Microblogging/en)
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
76 the importer MUST NOT create node or call XEP-0277 plugin itself
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
77 'comments*' key MUST NOT be used in this microblog_data, see bellow for comments
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
78 It is recommanded to use a unique id in the "id" key which is constant per blog item,
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
79 so if the import fail, a new import will overwrite the failed items and avoid duplicates.
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
80
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
81 'comments' (list[list[dict]],None): Dictionaries must have the same keys as main item (i.e. 'blog' and 'comments')
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
82 a list of list is used because XEP-0277 can handler several comments nodes,
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
83 but in most cases, there will we only one item it the first list (something like [[{comment1_data},{comment2_data}, ...]])
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
84 blog['allow_comments'] must be True if there is any comment, and False (or not present) if comments are not allowed.
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
85 If allow_comments is False and some comments are present, an exceptions.DataError will be raised
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
86 * optional keys:
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
87 'url' (unicode): former url of the post (only the path, without host part)
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
88 if present the association to the new path will be displayed to user, so it can make redirections if necessary
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
89 @param options(dict, None): Below are the generic options,
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
90 blog importer can have specific ones. All options have unicode values
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
91 generic options:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
92 - OPT_HOST (unicode): original host
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
93 - OPT_UPLOAD_IMAGES (bool): upload images to XMPP server if True
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
94 see OPT_UPLOAD_IGNORE_HOST.
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
95 Default: True
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
96 - OPT_UPLOAD_IGNORE_HOST (unicode): don't upload images from this host
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
97 - OPT_IGNORE_TLS (bool): ignore TLS error for image upload.
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
98 Default: False
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
99 @param return_data(dict): will contain link between former posts and new items
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
100
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
101 """
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
102 mb_data = item_import_data['blog']
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
103 try:
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
104 item_id = mb_data['id']
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
105 except KeyError:
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
106 item_id = mb_data['id'] = unicode(shortuuid.uuid())
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
107
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
108 try:
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
109 # we keep the link between old url and new blog item
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
110 # so the user can redirect its former blog urls
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
111 old_uri = item_import_data['url']
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
112 except KeyError:
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
113 pass
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
114 else:
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
115 new_uri = return_data[URL_REDIRECT_PREFIX + old_uri] = self._p.getNodeURI(
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
116 service if service is not None else client.jid.userhostJID(),
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
117 node or self._m.namespace,
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
118 item_id)
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
119 log.info(u"url link from {old} to {new}".format(
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
120 old=old_uri, new=new_uri))
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
121
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
122 return mb_data
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
123
2370
2c2b826b0bb3 plugin import: node can now be specified + added a "session" dict to keep import session data:
Goffi <goffi@goffi.org>
parents: 2369
diff changeset
124 def importSubItems(self, client, item_import_data, mb_data, session, options):
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
125 # comments data
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
126 if len(item_import_data['comments']) != 1:
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
127 raise NotImplementedError(u"can't manage multiple comment links")
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
128 allow_comments = C.bool(mb_data.get('allow_comments', C.BOOL_FALSE))
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
129 if allow_comments:
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
130 comments_service, comments_node = self._m.getCommentsService(client), self._m.getCommentsNode(mb_data['id'])
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
131 mb_data['comments_service'] = comments_service.full()
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
132 mb_data['comments_node'] = comments_node
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
133 recurse_kwargs = {
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
134 'items_import_data':item_import_data['comments'][0],
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
135 'service':comments_service,
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
136 'node':comments_node}
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
137 return recurse_kwargs
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
138 else:
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
139 if item_import_data['comments'][0]:
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
140 raise exceptions.DataError(u"allow_comments set to False, but comments are there")
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
141 return None
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
142
2370
2c2b826b0bb3 plugin import: node can now be specified + added a "session" dict to keep import session data:
Goffi <goffi@goffi.org>
parents: 2369
diff changeset
143 def publishItem(self, client, mb_data, service, node, session):
2369
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
144 log.debug(u"uploading item [{id}]: {title}".format(id=mb_data['id'], title=mb_data.get('title','')))
cdaa58e14553 plugin import: generic data import plugin:
Goffi <goffi@goffi.org>
parents: 2145
diff changeset
145 return self._m.send(client, mb_data, service, node)
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
146
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
147 @defer.inlineCallbacks
2370
2c2b826b0bb3 plugin import: node can now be specified + added a "session" dict to keep import session data:
Goffi <goffi@goffi.org>
parents: 2369
diff changeset
148 def itemFilters(self, client, mb_data, session, options):
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
149 """Apply filters according to options
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
150
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
151 modify mb_data in place
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
152 @param posts_data(list[dict]): data as returned by importer callback
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
153 @param options(dict): dict as given in [blogImport]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
154 """
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
155 # FIXME: blog filters don't work on text content
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
156 # TODO: text => XHTML conversion should handler links with <a/>
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
157 # filters can then be used by converting text to XHTML
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
158 if not options:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
159 return
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
160
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
161 # we want only XHTML content
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
162 for prefix in ('content',): # a tuple is use, if title need to be added in the future
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
163 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
164 rich = mb_data['{}_rich'.format(prefix)]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
165 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
166 pass
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
167 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
168 if '{}_xhtml'.format(prefix) in mb_data:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
169 raise exceptions.DataError(u"importer gave {prefix}_rich and {prefix}_xhtml at the same time, this is not allowed".format(prefix=prefix))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
170 # we convert rich syntax to XHTML here, so we can handle filters easily
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
171 converted = yield self._s.convert(rich, self._s.getCurrentSyntax(client.profile), safe=False)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
172 mb_data['{}_xhtml'.format(prefix)] = converted
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
173 del mb_data['{}_rich'.format(prefix)]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
174
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
175 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
176 mb_data['txt']
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
177 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
178 pass
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
179 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
180 if '{}_xhtml'.format(prefix) in mb_data:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
181 log.warning(u"{prefix}_text will be replaced by converted {prefix}_xhtml, so filters can be handled".format(prefix=prefix))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
182 del mb_data['{}_text'.format(prefix)]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
183 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
184 log.warning(u"importer gave a text {prefix}, blog filters don't work on text {prefix}".format(prefix=prefix))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
185 return
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
186
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
187 # at this point, we have only XHTML version of content
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
188 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
189 top_elt = xml_tools.ElementParser()(mb_data['content_xhtml'], namespace=C.NS_XHTML)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
190 except domish.ParserError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
191 # we clean the xml and try again our luck
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
192 cleaned = yield self._s.cleanXHTML(mb_data['content_xhtml'])
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
193 top_elt = xml_tools.ElementParser()(cleaned, namespace=C.NS_XHTML)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
194 opt_host = options.get(OPT_HOST)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
195 if opt_host:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
196 # we normalise the domain
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
197 parsed_host = urlparse.urlsplit(opt_host)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
198 opt_host = urlparse.urlunsplit((parsed_host.scheme or 'http', parsed_host.netloc or parsed_host.path, '', '', ''))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
199
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
200 tmp_dir = tempfile.mkdtemp()
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
201 try:
1844
489b968b3723 plugin blog_import_dokuwiki: also uses the generic image uploader from blog_import (when media_repo is empty and OPT_UPLOAD_IMAGES is True)
souliane <souliane@mailoo.org>
parents: 1839
diff changeset
202 # TODO: would be nice to also update the hyperlinks to these images, e.g. when you have <a href="{url}"><img src="{url}"></a>
2108
70f23bc7859b core (xml_tools): fixed findAll
Goffi <goffi@goffi.org>
parents: 1934
diff changeset
203 for img_elt in xml_tools.findAll(top_elt, names=[u'img']):
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
204 yield self.imgFilters(client, img_elt, options, opt_host, tmp_dir)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
205 finally:
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
206 os.rmdir(tmp_dir) # XXX: tmp_dir should be empty, or something went wrong
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
207
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
208 # we now replace the content with filtered one
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
209 mb_data['content_xhtml'] = top_elt.toXml()
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
210
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
211 @defer.inlineCallbacks
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
212 def imgFilters(self, client, img_elt, options, opt_host, tmp_dir):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
213 """Filters handling images
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
214
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
215 url without host are fixed (if possible)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
216 according to options, images are uploaded to XMPP server
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
217 @param img_elt(domish.Element): <img/> element to handle
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
218 @param options(dict): filters options
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
219 @param opt_host(unicode): normalised host given in options
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
220 @param tmp_dir(str): path to temp directory
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
221 """
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
222 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
223 url = img_elt['src']
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
224 if url[0] == u'/':
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
225 if not opt_host:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
226 log.warning(u"host was not specified, we can't deal with src without host ({url}) and have to ignore the following <img/>:\n{xml}"
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
227 .format(url=url, xml=img_elt.toXml()))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
228 return
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
229 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
230 url = urlparse.urljoin(opt_host, url)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
231 filename = url.rsplit('/',1)[-1].strip()
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
232 if not filename:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
233 raise KeyError
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
234 except (KeyError, IndexError):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
235 log.warning(u"ignoring invalid img element: {}".format(img_elt.toXml()))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
236 return
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
237
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
238 # we change the url for the normalized one
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
239 img_elt['src'] = url
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
240
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
241 if options.get(OPT_UPLOAD_IMAGES, False):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
242 # upload is requested
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
243 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
244 ignore_host = options[OPT_UPLOAD_IGNORE_HOST]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
245 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
246 pass
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
247 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
248 # host is the ignored one, we skip
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
249 parsed_url = urlparse.urlsplit(url)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
250 if ignore_host in parsed_url.hostname:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
251 log.info(u"Don't upload image at {url} because of {opt} option".format(
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
252 url=url, opt=OPT_UPLOAD_IGNORE_HOST))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
253 return
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
254
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
255 # we download images and re-upload them via XMPP
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
256 tmp_file = os.path.join(tmp_dir, filename).encode('utf-8')
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
257 upload_options = {'ignore_tls_errors': options.get(OPT_IGNORE_TLS, False)}
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
258
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
259 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
260 yield web_client.downloadPage(url.encode('utf-8'), tmp_file)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
261 filename = filename.replace(u'%', u'_') # FIXME: tmp workaround for a bug in prosody http upload
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
262 dummy, download_d = yield self._u.upload(client, tmp_file, filename, options=upload_options)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
263 download_url = yield download_d
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
264 except Exception as e:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
265 log.warning(u"can't download image at {url}: {reason}".format(url=url, reason=e))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
266 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
267 img_elt['src'] = download_url
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
268
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
269 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
270 os.unlink(tmp_file)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
271 except OSError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
272 pass