When pipe playback is started, but no data is written to the pipe, the input
loop would bring the cpu to 100%. This fix limits the loop like it was before
player refactor.
This adds a new settings component for user configurable options that
can be changed through the JSON API.
The settings are stored in the admin db table and not in the conf-file.
Since it is unknown how to do real sync on Chromecast, this commit instead adds
a primitive delay to the stream, so that it is at least somewhat closer to
Airplay/local audio.
Also some cleanup of unused stuff.
* [db,jsonapi] case insensitive directory/file listing
* [jsonapi] file listing of playlist uses same VPATH ordering as per directory and files
* [db,jsonapi] sorting via existing S_VPATH
* [db] replace LOWER with COLLATE NOCASE
pb_suspend() + pb_resume() during track changes made the playback status
incorrect, i.e. pb_session.source_list/playing_now would not match what the
input was actually writing. This attempts to solve it by resetting the
session when pb_suspend() is called, so that the input, input_buffer and
source_list come into sync.
If playback was paused during the very last part of the track, the rest of the
track would be read into the input buffer and the input would be closed. With
this commit the input will not be reopened.
Also allow input_flush to be called with null argument.
Fixes bugs which were due to incorrect handling of unsigned integer wrap-around:
1. Calling packet_resend() with seqnum + len greater than UINT16_MAX => infinite loop
2. Calling rtp_packet_get() with session->seqnum - seqnum greater than pktbuf_next => wrong packet
This fixes a bug from commit 37ce8dd6 where seek_http (which is called when
pausing playback) for non-seekable streams would return -1, thus signalling
an error, even though it is not. The player would think that the stream
could not be played and then skip to the next item.
The fix in commit 3928ab6 broke resuming from an underrun, since it meant that
pb_resume() would flush the input buffer. With this fix it is possible to call
input_resume(), which will not flush the buffer if the source is already open.
Also renamed some functions in player.c for consistency.
For dates that require context (ie today, yesterday, N days ago etc) we want the
underlying SQL to respect the current time when running query; a query that
requests items for 'today' should only find matches for the time it was run.
Current implementation would generated a fixed date (at the time the SMARTPL is
inserted into db) in the playlist table where as this commit understands the
context of the date.
* Fix "clicks" during playback, especially on low buffer size devices
Bug had two causes: Trying to write to the prebuf ringbuffer when it was full
and writing new audio to the device without first having drained the prebuf,
thus writing out of order.
* Use snd_pcm_drain() so alsa doesn't report underrun on playback session end
Removes SNDRV_PCM_IOCTL_SYNC_PTR errors
* Fix missing error check of the return value from snd_pcm_avail (now use snd_pcm_avail_delay)
The previous solution would use subqueries to count the number of items and
streams in each playlist, which means that response time gets pretty slow if
there are many playlists.
This commit also includes a number of lesser db code changes.
Commit b3bfb0a and e1993bc change the triggers and calculation of id's in a way
that is not backwards compatible, so we need to make major schema upgrade.
The purpose of this is to support library backends making their own
calculation of these id's, which is relevant if they have more information
available than just album_artist and album.
This also removes a bunch of sqlite extension code plus some triggers, which
in itself is probably an improvement.
Replace reading_next and reading_prev with a list of sources, so that we can
deal with short tracks, i.e. tracks where reading ends before playback starts.
With short tracks reading ends before playback starts, so event_read_eof comes
before event_play_start, which causes playing_now to point to a null
reading_now.
With this change it will point to a non-null reading_prev, but note that in the
hopefully rare case of multiple short tracks, the playing_now pointer will
still be incorrect.
ffmpeg changed the behaviour of avcodec_default_get_format() so that it picks
AV_PIX_FMT_MONOBLACK instead of AV_PIX_FMT_RGB24 for the png encoder. That
makes the function of no use to us, so now the pix formats are just hardcoded
in the settings instead.
This change is preparation to use ffmpeg's resampling capabilities to keep local
audio in sync (by up/downsampling slightly). This requires that sample rates are
not fixed for a transcode profile.
Added benefit of this is that we don't need quite as many xcode profiles.
Previously input_metadata_get() would retrieve artwork from the source being
read currently, which might not be the one that triggered the FLAG_METADATA
event. So to fix this the metadata is now read by the input module itself when
the METADATA event happens, and the result is stored with the marker.
The commit also includes a timer so that the input thread does loop forever
if the player never starts reading.
Also some refactoring of metadata + abolish input_metadata_get and
input_quality_get. The latter in an attempt to treat the two in the same way.
In the output implementations playback_stop() was somewhat redundant,
since device_stop() does the same.
The timer should make sure that we always close outputs (previously
they were in some cases kept open).
The commit also includes some renaming.
After an underrun the player doesn't read, so that meant input_wait would
wait a second before allowing the input to write, even though the input_buffer
was not full
+ remember to flush in source_start(), since the input won't do it if
input_now_reading has already been closed (e.g. if starting a new track
while playback is at the end of another track)
Player now stops 10 secs after stop command and 10 mins after pause. At
that time the outputs have probably cut the connection themselves, but
that might be ok (needs testing).
* Also call full_cb() from input_wait if buffer is full
* Make read_deficit count missing bytes instead of clock ticks
* Make read_deficit a part of the playback session
The time stamp was getting set too late, because if pos was zero the first
reads then it would be overwritten, but it shouldn't because the loop will
catch up even if the initial reads have zero samples.
* Drop output_sessions, was just a pointer to the actual session anyway
* Drop the old write, flush and stop prototypes
* Some minor changes/renaming
Purpose of this is also to fix a race condition in player.c where it
could try to start two sessions on the same speaker. This could happen
because outputs_device_start() in line 2093 is conditional on device->session
which however is false while a device is starting up.
outputs_playback_start() had the problem that was not consistently invoked: If
for instance local audio playback was running and a Airplay device was then
activated, the raop's playback_start would never be invoked (and vice versa,
of course).
Instead, the player now writes the presentation timestamp every time to the
output, so it doesn't need to keep track of it from the start.
* Untie Airtunes stuff further from player and non-Airplay outputs
* Change raop.c to use rtp_common.c (step 1)
* Change heartbeat of player to 100 ticks/sec, since we have untied from
Airtunes 352 samples per packet (which equals 126 ticks/sec at 44100)
Still a lot to be done in the player, since the rtptime's in it don't
are probably broken.
Output module can now take input data in multiple quality levels, and
can resample to those output modules that would require a certain quality
level, like raop.c would
Extends the http_client_ctx to hold the response code for a request.
Also adds the content type header, if it was a https request (using
libcurl instead of libevent)
This adds a new timestamp value "db_modified" into the admin db table.
In addition to the existing "db_update" admin value, this value is also
updated if rating, play-/skip-count or seek changes for a
media_info_file (files db table).
This should improve the caching behavior in clients of the JSON API
(especially the player web interface) in refreshing its data if some of
this values changes.
New endpoint is PUT api/library/tracks/[id] and supported query
parameters are:
- rating: with values between 0 and 100
- play_count: with values "reset" (resets play_count and skip_count) or
"increment" (increments play_count)
"Play Date" tag was seconds since 1904 (an Apple Mac HFS+ timestamp), not a
Unix timestamp as we assumed. Seems Apple themselves realised that wasn't a
great idea (+ not a proper plist date type), and therefore provide "Play Date
UTC" as an alternative.
If a file gets updated/rescanned we generally don't want to reset the above
values. This commit adds DB_FLAG_NO_ZERO, which marks a field so that
db_file_update() will only update it if the new value is non-zero (i.e. the
caller probably has a "better" value).
This change means that we will use album_artist even if compilation_artist
is configured, thus compilation_artist will only be used to override artist.
The constraints on songalbumid and songartistid where changed with v20, so we
need to make sure they take effect when upgrading.
This commit tries to do the table recreation like sqlite recommends and without
manually crafted copy queries that are probably prone to errors.
Since we are recreating anyway, this commit also reorders the columns slightly.
It also includes auto-drop/recreation of triggers (should really have been its
own commit) during upgrade, like is already done with indices.
The idea here is to make sure the fixing up of tags is done in a consistent
manner. For strings, this means stuff like trimming and empty strings -> null
are applied the same unless there are special exception rules set. It also
means that defaults are applied the same across structs, e.g. "Unknown artist"
for both mfi->artist and queue_item->artist.
The change is also necessary because we want to remove trimming from the sql
query and instead implement it ourselves.
is not the current playing item.
This happens if the input already switched to the next item in the queue
starting to stream it to the outputs (2 second buffer) while the outputs
are still playing the last seconds of the old item.
Add check against the special file_id DB_MEDIA_FILE_NON_PERSISTENT_ID to
identify if a queue item is not in the library. And always prefer the
artwork url in the queue item before the artwork for the library file.
Some remotes don't respond as expected to the test. Retune will give connection
refused, because the test is made too quickly, before the service is running.
Even if we delay the test it won't work because Retune crashes.
Since the false mdns advertisements are only seen on Airplay, we only do the
test there.
The unconfigurable resync period of 10 seconds was not frequent
enough to keep my own ALSA device in sync with the AirPlay stream.
Now the period is configurable. The default is still at 10
seconds, to prevent any change in behavior unless opted in by
the user.
Currently the adjustment causes a tiny "click" distortion in the
ALSA output, so it is better to make the check as infrequent as
possible, while still being frequent enough to stay in sync
over lengthy sessions of playback.
Added source_sample_rate, target_sample_rate to alsa_session.
This is a first step toward rendering ALSA at a different
sampling rate than the AirPlay stream, so that (a) we will
be able to dynamically adjust the ALSA sampling rate for an
improved sync algorithm, and (b) later, a more generalized
resampling algorithm can accommodate very different hardware
sampling rates like 22050 Hz or 48000 Hz.
Reworked alsa_session_free() so that it can be used to
tear down a partially initialized alsa_session if an
error occurs in the middle of alsa_session_make().
This simplifies the error handling logic in alsa_session_make().
This refactoring will be helpful later when resampling is added,
because more data structures will be dynamically allocated
during initialization.
Signed-off-by: Don Cross <cosinekitty@gmail.com>
The current logic in httpd_dacp.c cannot handle non persistent items
correctly. The items are always shown with the dummy_mfi with "unkown
artist" etc.
Ommits useless update query for playcount for items that are not in the
library. Also avoids trying to scrobble these items (fixes error log
message "lastfm: Scrobble failed, track id 9999999 is unknown")
The connection test would not catch "No route to host", as this is returned
through the value-result buffer.
This fix might partially solve issue #498.
AirPlay 2 devices like Sonos One and AirPort Express with firmware 7.8
require auth-setup before ANNOUNCE, otherwise they will return 403.
Also fixed a problem where metadata did not get sent when toggling
a speaker on/off if we were playing an endless stream.
Avoid calling cache_daap_add() when hreq->user_agent is NULL
(user agent is not provided by the client), because it will
trigger a segfault when strdup() is called with that NULL pointer.
Fixes#571.
Avahi gives us several ipv6 addresses for an ATV4 that aren't actually
connectable. Here we simply try to make a connection to the address,
and if it is not possible within a timeout we ignore the announcement.
We also now don't start a mdns record browser if the address from the
resolver passes the connection test and isn't link-local.
