Mercurial > libervia-backend

annotate src/plugins/plugin_blog_import.py @ 2240:4b66f070cfbb

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
jp (uri): new command to parse or generate XMPP URIs
author Goffi <goffi@goffi.org>
date Fri, 19 May 2017 12:34:33 +0200
parents 33c8c4973743
children cdaa58e14553
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
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
5 # Copyright (C) 2009-2016 Jérôme Poisson (goffi@goffi.org)
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
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
20 from sat.core.i18n import _
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
21 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
22 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
23 log = getLogger(__name__)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
24 from twisted.internet import defer
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
25 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
26 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
27 from sat.core import exceptions
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
28 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
29 import collections
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
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
34 import uuid
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",
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2108
diff changeset
40 C.PI_TYPE: C.PLUG_TYPE_BLOG,
33c8c4973743 core (plugins): added missing contants + use of new constants in PLUGIN_INFO
Goffi <goffi@goffi.org>
parents: 2108
diff changeset
41 C.PI_DEPENDENCIES: ["XEP-0060", "XEP-0277", "TEXT-SYNTAXES", "UPLOAD"],
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:
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
45 This plugin manage the different blog importers which can register to it, and handler generic importing tasks.""")
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 BOOL_OPTIONS = (OPT_UPLOAD_IMAGES, OPT_IGNORE_TLS)
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
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
56 BlogImporter = collections.namedtuple('BlogImporter', ('callback', 'short_desc', 'long_desc'))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
57
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
58
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
59 class BlogImportPlugin(object):
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._importers = {}
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
65 self._u = host.plugins['UPLOAD']
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
66 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
67 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
68 self._s = self.host.plugins['TEXT-SYNTAXES']
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
69 host.bridge.addMethod("blogImport", ".plugin", in_sign='ssa{ss}ss', out_sign='s', method=self._blogImport, async=True)
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
70 host.bridge.addMethod("blogImportList", ".plugin", in_sign='', out_sign='a(ss)', method=self.listImporters)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
71 host.bridge.addMethod("blogImportDesc", ".plugin", in_sign='s', out_sign='(ss)', method=self.getDescription)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
72
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
73 def getProgress(self, progress_id, profile):
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
74 client = self.host.getClient(profile)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
75 return client._blogImport_progress[progress_id]
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
76
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
77 def listImporters(self):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
78 importers = self._importers.keys()
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
79 importers.sort()
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
80 return [(name, self._importers[name].short_desc) for name in self._importers]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
81
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
82 def getDescription(self, name):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
83 """Return import short and long descriptions
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
84
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
85 @param name(unicode): blog importer name
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
86 @return (tuple[unicode,unicode]): short and long description
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
87 """
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
88 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
89 importer = self._importers[name]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
90 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
91 raise exceptions.NotFound(u"Blog importer not found [{}]".format(name))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
92 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
93 return importer.short_desc, importer.long_desc
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
94
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
95 def _blogImport(self, name, location, options, pubsub_service='', profile=C.PROF_KEY_DEFAULT):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
96 client = self.host.getClient(profile)
1839
cdecf553e051 frontends (jp/blog), plugin blog_import: fixes:
souliane <souliane@mailoo.org>
parents: 1831
diff changeset
97 options = {key: unicode(value) for key, value in options.iteritems()}
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
98 for option in BOOL_OPTIONS:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
99 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
100 options[option] = C.bool(options[option])
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
101 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
102 pass
1839
cdecf553e051 frontends (jp/blog), plugin blog_import: fixes:
souliane <souliane@mailoo.org>
parents: 1831
diff changeset
103 return self.blogImport(client, unicode(name), unicode(location), options)
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
104
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
105 @defer.inlineCallbacks
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
106 def blogImport(self, client, name, location, options=None, pubsub_service=None):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
107 """Import a blog
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
108
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
109 @param name(unicode): name of the blog importer
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
110 @param location(unicode): location of the blog data to import
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
111 can be an url, a file path, or anything which make sense
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
112 check importer description for more details
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
113 @param options(dict, None): extra options. Below are the generic options,
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
114 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
115 generic options:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
116 - OPT_HOST (unicode): original host
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
117 - 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
118 see OPT_UPLOAD_IGNORE_HOST.
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
119 Default: True
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
120 - 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
121 - 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
122 Default: False
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
123 @param pubsub_service(jid.JID, None): jid of the PubSub service where blog must be imported
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
124 None to use profile's server
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
125 @return (unicode): progress id
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
126 """
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
127 if options is None:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
128 options = {}
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
129 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
130 for opt_name, opt_default in ((OPT_UPLOAD_IMAGES, True),
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
131 (OPT_IGNORE_TLS, False)):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
132 # we want an filled options dict, with all empty or False values removed
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
133 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
134 value =options[opt_name]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
135 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
136 if opt_default:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
137 options[opt_name] = opt_default
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
138 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
139 if not value:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
140 del options[opt_name]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
141 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
142 importer = self._importers[name]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
143 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
144 raise exceptions.NotFound(u"Importer [{}] not found".format(name))
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
145 posts_data, posts_count = yield importer.callback(client, location, options)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
146 url_redirect = {}
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
147 progress_id = unicode(uuid.uuid4())
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
148 try:
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
149 progress_data = client._blogImport_progress
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
150 except AttributeError:
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
151 progress_data = client._blogImport_progress = {}
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
152 progress_data[progress_id] = {u'position': '0'}
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
153 if posts_count is not None:
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
154 progress_data[progress_id]['size'] = unicode(posts_count)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
155 metadata = {'name': u'{}: {}'.format(name, location),
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
156 'direction': 'out',
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
157 'type': 'BLOG_IMPORT'
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
158 }
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
159 self.host.registerProgressCb(progress_id, self.getProgress, metadata, profile=client.profile)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
160 self.host.bridge.progressStarted(progress_id, metadata, client.profile)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
161 self._recursiveImport(client, posts_data, progress_id, options, url_redirect)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
162 defer.returnValue(progress_id)
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
163
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
164 @defer.inlineCallbacks
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
165 def _recursiveImport(self, client, posts_data, progress_id, options, url_redirect, service=None, node=None, depth=0):
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
166 """Do the upload recursively
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
167
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
168 @param posts_data(list): list of data as specified in [register]
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
169 @param options(dict): import options
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
170 @param url_redirect(dict): link between former posts and new items
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
171 @param service(jid.JID, None): PubSub service to use
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
172 @param node(unicode, None): PubSub node to use
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
173 @param depth(int): level of recursion
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
174 """
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
175 for idx, data in enumerate(posts_data):
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
176 # data checks/filters
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
177 mb_data = data['blog']
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
178 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
179 item_id = mb_data['id']
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
180 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
181 item_id = mb_data['id'] = unicode(uuid.uuid4())
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
182
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
183 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
184 # we keep the link between old url and new blog item
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
185 # so the user can redirect its former blog urls
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
186 old_uri = data['url']
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
187 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
188 pass
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
189 else:
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
190 new_uri = url_redirect[old_uri] = self._p.getNodeURI(
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
191 service if service is not None else client.jid.userhostJID(),
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
192 node or self._m.namespace,
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
193 item_id)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
194 log.info(u"url link from {old} to {new}".format(
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
195 old=old_uri, new=new_uri))
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
196
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
197 yield self.blogFilters(client, mb_data, options)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
198
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
199 # comments data
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
200 if len(data['comments']) != 1:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
201 raise NotImplementedError(u"can't manage multiple comment links")
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
202 allow_comments = C.bool(mb_data.get('allow_comments', C.BOOL_FALSE))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
203 if allow_comments:
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
204 comments_service, comments_node = self._m.getCommentsService(client), self._m.getCommentsNode(item_id)
1852
569c48e4cf0d plugin blog import: mb_data handle unicode only, so comments_service must be a unicode and not a JID
Goffi <goffi@goffi.org>
parents: 1844
diff changeset
205 mb_data['comments_service'] = comments_service.full()
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
206 mb_data['comments_node'] = comments_node
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
207 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
208 if data['comments'][0]:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
209 raise exceptions.DataError(u"allow_comments set to False, but comments are there")
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 # post upload
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
212 depth or log.debug(u"uploading item [{id}]: {title}".format(id=mb_data['id'], title=mb_data.get('title','')))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
213 yield self._m.send(mb_data, service, node, profile=client.profile)
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 # comments upload
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
216 depth or log.debug(u"uploading comments")
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
217 if allow_comments:
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
218 yield self._recursiveImport(client, data['comments'][0], progress_id, options, url_redirect, service=comments_service, node=comments_node, depth=depth+1)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
219 if depth == 0:
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
220 client._blogImport_progress[progress_id]['position'] = unicode(idx+1)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
221
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
222 if depth == 0:
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
223 self.host.bridge.progressFinished(progress_id,
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
224 {u'{}{}'.format(URL_REDIRECT_PREFIX, old): new for old, new in url_redirect.iteritems()},
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
225 client.profile)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
226 self.host.removeProgressCb(progress_id, client.profile)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
227 del client._blogImport_progress[progress_id]
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
228
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
229 @defer.inlineCallbacks
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
230 def blogFilters(self, client, mb_data, options):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
231 """Apply filters according to options
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
232
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
233 modify mb_data in place
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
234 @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
235 @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
236 """
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
237 # 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
238 # 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
239 # 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
240 if not options:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
241 return
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
242
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
243 # we want only XHTML content
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
244 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
245 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
246 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
247 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
248 pass
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
249 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
250 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
251 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
252 # 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
253 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
254 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
255 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
256
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
257 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
258 mb_data['txt']
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
259 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
260 pass
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
261 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
262 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
263 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
264 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
265 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
266 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
267 return
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 # 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
270 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
271 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
272 except domish.ParserError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
273 # 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
274 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
275 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
276 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
277 if opt_host:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
278 # we normalise the domain
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
279 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
280 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
281
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
282 tmp_dir = tempfile.mkdtemp()
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
283 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
284 # 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
285 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
286 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
287 finally:
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
288 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
289
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
290 # 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
291 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
292
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
293 @defer.inlineCallbacks
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
294 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
295 """Filters handling images
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
296
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
297 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
298 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
299 @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
300 @param options(dict): filters options
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
301 @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
302 @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
303 """
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
304 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
305 url = img_elt['src']
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
306 if url[0] == u'/':
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
307 if not opt_host:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
308 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
309 .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
310 return
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
311 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
312 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
313 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
314 if not filename:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
315 raise KeyError
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
316 except (KeyError, IndexError):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
317 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
318 return
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
319
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
320 # 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
321 img_elt['src'] = url
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
322
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
323 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
324 # upload is requested
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
325 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
326 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
327 except KeyError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
328 pass
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
329 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
330 # 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
331 parsed_url = urlparse.urlsplit(url)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
332 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
333 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
334 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
335 return
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
336
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
337 # 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
338 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
339 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
340
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
341 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
342 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
343 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
344 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
345 download_url = yield download_d
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
346 except Exception as e:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
347 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
348 else:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
349 img_elt['src'] = download_url
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
350
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
351 try:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
352 os.unlink(tmp_file)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
353 except OSError:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
354 pass
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
355
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
356 def register(self, name, callback, short_desc='', long_desc=''):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
357 """Register a blogImport method
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
358
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
359 @param name(unicode): unique importer name, should indicate the blogging software it handler and always lowercase
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
360 @param callback(callable): method to call:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
361 the signature must be (client, location, options) (cf. [blogImport])
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
362 the importer must return a tuple with (posts_data, posts_count)
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
363
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
364 posts_data is an iterable of dict which must have the following keys:
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
365 'blog' (dict): microblog data of the blog post (cf. http://wiki.goffi.org/wiki/Bridge_API_-_Microblogging/en)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
366 the importer MUST NOT create node or call XEP-0277 plugin itself
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
367 'comments*' key MUST NOT be used in this microblog_data, see bellow for comments
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
368 It is recommanded to use a unique id in the "id" key which is constant per blog item,
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
369 so if the import fail, a new import will overwrite the failed items and avoid duplicates.
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
370
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
371 'comments' (list[list[dict]],None): Dictionaries must have the same keys as main item (i.e. 'blog' and 'comments')
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
372 a list of list is used because XEP-0277 can handler several comments nodes,
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
373 but in most cases, there will we only one item it the first list (something like [[{comment1_data},{comment2_data}, ...]])
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
374 blog['allow_comments'] must be True if there is any comment, and False (or not present) if comments are not allowed.
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
375 If allow_comments is False and some comments are present, a exceptions.DataError will be raised
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
376 the import MAY optionally have the following keys:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
377 'url' (unicode): former url of the post (only the path, without host part)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
378 if present the association to the new path will be displayed to user, so it can make redirections if necessary
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
379
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
380 posts_count (int, None) indicate the total number of posts (without comments)
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
381 useful to display a progress indicator when the iterator is a generator
1831
68c0dc13d821 plugin blog import, XEP-0277: progress + redirect:
Goffi <goffi@goffi.org>
parents: 1825
diff changeset
382 use None if you can't guess the total number of blog posts
1825
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
383 @param short_desc(unicode): one line description of the importer
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
384 @param long_desc(unicode): long description of the importer, its options, etc.
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
385 """
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
386 name = name.lower()
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
387 if name in self._importers:
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
388 raise exceptions.ConflictError(u"A blog importer with the name {} already exsit".format(name))
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
389 self._importers[name] = BlogImporter(callback, short_desc, long_desc)
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
390
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
391 def unregister(self, name):
4e51f21c687f plugin blog import: this plugin is the base handling blog importers:
Goffi <goffi@goffi.org>
parents:
diff changeset
392 del self._importers[name]