- blog importers are plugin which register to this one with the "register" method
- blog import manage uploading the given blog data to a PubSub server
- filters can be used.
- current filters fixe the <img> src without host, and upload images to XMPP server
fileUpload is used by external frontends, while upload can be used by frontends. upload return in addition to the progress_id a download_d Deferred which fire with URI when file is uploaded.
changed option ignore-tls-errors to ignore_tls_errors.
- if comments*, comments*_node or comments*_service are provided, they are used instead of generated values
- moved node and service generation to getCommentNode and getCommentService, so they can be used by an external plugin
- while it is mandatory to clean input data, output data sent by frontends should and need to be clean
- cleaning all output data means that we can't send be sure that all the XHTML we send is not modified
- it's a waste of resources as cleaning today reparse the data with lxml
- behaviour may change in the future (e.g. having an option to use cleaning on output data)
- escape_chars regex has been fixed (an \ was missing for backslash escaping)
- escape_chars is now splited in consts so XHTML => dc_wiki escape less characters, for better readability
- fixed bad length check in DCWikiParser.parser_link
dotclear wiki syntax should now be usable
- use tuple instead of list for STYLES_WHITELIST
- specify where to get python makdown and html2text
- renamed clean_xhtml to cleanXHTML for consistency
- check name for conflict in addSyntax
- removed useless XEP-0059 recommendation
- added XEP-0313 as a recommended plugin
- parseExtra now manage mam_* keys (mam_filter_[name] to use a filter)
- if a mam request is in 'mam' key of extra parameted in getItems, a MAM request is done, but the result is returned in the same way as for normal getItems, making like more easy for other plugins and for frontends
- moved common namespaces to core.constants
- removed useless dependencies and recommendations
- removed all bride methods as they where there only for testing purpose
- don't use a plugin dict to store MAM clients, put them in profiles' clients instead
- removed message trigger in favor of an observer which is triggered only when specific queryid is used
- new getArchives method which grab all result and return them all at once
- don't use getRSMResponse anymore: rsm.PubSubClient use a different signature which return items and RSMResponse, this is better than the previous method as if the requested don't call getRSMResponse, there is no memory leak anymore.
- use RSMResponse directly instead of using ext_data, as other extension (like MAM) will be probably be managed with an other workflow
- don't do a useless deep copy of pubsub.PubSubRequest._parameters on each instance anymore
- RSMResponse.parse do now manage parsing of <set/> element directly
- restored original tmp.wokkel.pubsub.PubSubService._toResponse_items
- changed arguments order in RSMRequest.__init__ to have most common arguments first
- added __str__ methods
- better parsing/toElement
- better handling of optional elements/attributes (count/index)
- _toResponse_items handler RSM set elements without modifying original pubsub _toResponse_items
- renamed parse to fromElement for coherency
with XEP-0260, we need to manage 2 hashes (one with SHA1(SID + Initiator JID + Responder JID) and the other with SHA1(SID + Responder JID + Initiator JID)). This was not handled correclty, resulting in transfer failure in several cases.
- this was here for retro-compatibility with messages previously posted from Libervia
- but this is not appropriate and will add some extra <br /> e.g. in messages from Movim
- \n in XHTML content should not be converted to <br /> by the frontend
- save it directly in the Atom payload for better compatibility with other clients
