Some f-strings where used in translatable text, this has been fixed by using explicit
`format()` call (using a script based on `tokenize`).
As tokenize messes with spaces, a reformating tool (`black`) has been applied to some
files afterwards.
This file is used to set `USE_X_FORWARDED_HOST`, which is necessary to make Weblate work
correctly being an Apache (and potentially other HTTP servers) reverse proxy (otherwise we
have CSRF errors, cf. https://github.com/WeblateOrg/weblate/issues/1912).
a `files` field can now be used to create files on the fly. Its value must be a dict where
key are file names, and values are an other dict with file data.
For now file data can only use the `content` key. Value of content will be written to the file.
If a file already exists, it is not overwritten.
`failure.Failure` is subclass of `BaseException` but not of `Exception`, as a result
`except Exception` doesn't catch it correctly. This is fixed by not wrapping `DataError`
in `failure.Failure` anymore.
quotas can now be specified using the `quotas_json` option of `component file_sharing`
section in settings. This must be a dict where:
- `users` key contains default quotas for all users
- `admins` key contains quotas for administrators (not implemented yet)
- `jids` contain bare JID to quota mapping, to have user-specific quota
The value can be either a int for quota in bytes, or a case insensitive string with an
optional multiplier symbol (e.g. "500 Mio"). `None` can be used for explicit unlimited
quota (which is the default is `users` is not set).
When a file size is too big for quota, upload is refused with an error message indicating
allowed quota, used space, and the size of the file that user wants to upload.
- new `text` argument to specify human readable reason of termination
- handling of `FirstError` (used when a error happens in a DeferredList)
- handling of `StanzaError`
- (XEP-0234): show human readable text of the reason if present
- `parseSize` lets parse a size which can be either a int, or a string with a symbol like
`Mio`
- `getSizeMultiplier` returns a tuple of size with it's multiplier symbol
- `getHumanSize` returns a string with size and adapted multiplier.
Jingle Message Initiation is currently not useful for components, and the message workflow
is different in Libervia between clients and components (as a result, `messageReceived`
trigger is never called, and the session proposal is never answered).
This method does more or less the same thing as sending, but without actually sending the
item, and parse the generated element just after. This way, the triggers are run, and a
preview of the item can be made for the resulting microblog data.
when publishing a blog post without `id` specified, instead of using uuid, a user friendly
ID based on title or content is used. This allow to have better URLs in web frontend.
There is a (very low) risk of name collision if several articles with the same title or
content are published on the same pubsub node, user friendly ID can be deactivated by
using `user_friendly_id=False` in blog data to avoid that.
- if `author` element can't be found, `publisher` attribute, then `IQ`'s `from` attributes
are used as fallback to find author jid
- fix categories (tags) parsing if `author` element is not found
- remove items which have failed parsing from `mbGet` results (instead of using `None`).
