Output a track's Last-Modified stamp in ISO 8601 format, as MPD does.
Output nothing if track has no last-modified stamp.
The test has to use datetime to work out what the output will look like,
because it is local-time zone dependant.
Output the last_modified timestamp from mopidy's track model to mpd clients
in the same format as mpd uses - yyyy-mm-ddTHH:MM:SS
Outputs nothing for Last-Modified if last_modified is None
Address issues raised in review:
Fix formatting by shortening function name to concat_multi_values
Change comments and variable names to reflect generic nature of function
Fix typos in tests
Default to single quotes for strings
- Split into smaller test cases more or less per command
- Created a BasePopulatedTracklistTestCase with a sensible setUp
- Modified test cases to work with the common tracklist state
- Replaced all calls to tracklist.add(tracks=...) with uris=...
- Test tracklist ordering in more compact way that also gives better error
messages
Seeks will now fail when the duration is None, this is an approximation to if
the track is seekable or not. This check is need as otherwise seeking a radio
stream will trigger the next track.
If the track truly isn't seekable despite having a duration we should still
fail as GStreamer will reject the seek.
...and not False, because the mute state is unknown (None) and not
unmuted (False) when there is no mixer.
Note that this change does not affect the MPD responses.
Default blacklist set to listall and listallinfo.
This change has been done to avoid clients being able to call "bad" MPD
commands which are often misused to try and keep a client db. Note that
this change will break some MPD clients, but the blacklist can be controlled
via config to allow opting out for now.
Newer versions of the protocol have removed this tag, so we should as well.
This also works around the issue of #881 which was breaking things with
newlines in comment fields.
The readcomments command seems to replace this, but it seems to only care about
specific extra tagtypes, not the general comment tag we normally collect when
scanning things.
This was disabled together with a bunch of other changes without any
explanation in commit f24ca36e5a. I'm
guessing that this wasn't intentional, and no test covered the case.
Previously, test_list_album_with_artist_name would only test that the
command didn't fail. Now it also checks that the response is correct.
That is, that the response contains albums.
This makes the test detect the error reported in #817.
Replace the playlist uri/name mapping with a mapping for all the items
in browse. Check the mapping when browsing a path.
This has the added benefit of serving as a cache for browse, so we don't
need to traverse the path parts and lookup each of them for each call to
browse.
This provides a class which works as a command registry. Functions are added to
the registry through a method on the class `add`. Add acts as a decorator
taking in the name of the command and optionally type converters for arguments.
When handling requests, we can now tokenize the line, and then just pass the
arguments to our commands helper. It will lookup the correct handler and apply
any validation before calling the original function.
For the sake of testing the original function is not wrapped. This the
functions, and anything testing them directly can simply assume pre-converted
data as long as type annotations are in place.
A sample usage would be:
@protocol.commands.add('addid', position=protocol.integer)
def addid(context, uri, position=None)
pass
def handle_request(line):
tokens = protocol.tokenizer(line)
context = ...
result = protocol.commands.call(tokens, context=context)
- Adds tests for correctness of tokenizer (which also would have shown shlex
wouldn't have worked).
- Should conform with the original's behavior, though we won't be able to match
the error messages without a lot of extra work as a non-regexp version is
likely a no go on python due to speed.
