Add a new view called ImageGroup that will handle all advanced image
hacks from now on.
This includes the indicator (which is now animated), any selection
indicators, and the weirdness of the album song image. All of that
is now handled by ImageGroup. This is the culmination of probably
a day and a half of wrangling with android insanity and having to
remove a lot of what I liked about the indicator in order to make
this work on a basic level.
The only major bug I am currently aware of with this is that the
indicator is bugged out on Lollipop devices due to bad vectors.
Again.
I never want to do this again. I cannot believe that adding a basic
indicator took this long and required so much stupid hacks and
inefficient code. And then google wonders why android apps are so
visually unappealing and janky and laggy. Hm. Must be that devs aren't
using the brand new FooBarBlasterFlow library!
Add a playing indicator to cover art.
This is simply to improve the general aestethics of this view. Of
course, the current way I implement this is incredibly stupid and I
plan to replace it.
Completely rework the way Auxio handles icons.
This is mostly two changes:
1. Removing ImageButton/StyledImageButton for MaterialButton. This is
done by abusing MaterialButton's theming options to make it only show
an icon.
2. Standardizing icon sizes into small, medium, and large categories.
Small is the default, Medium and Large are for edge-cases like the
playback icons which look horrible at 24dp.
3. Abusing the Toolbar to make it follow Material 3 guidelines. This
mostly involved removing the strange icon sizing and correctly padding
the view.
4. Reworking the playback bar to use more, smaller icons, making it
more like a Toolbar in the process (which I like).
Supply a view to WindowInsetsCompat.toWindowInsetsCompat through all
calls in the app.
This does nothing, but it would resolve future bugs if I were to use
the most exotic window inset types.
On the playback par, pick the larger inset out of the system bar
insets or system gesture insets.
This is to fix an issue where android just does not provide correct
gesture insets on Android 12L, and [maybe???] fix what is going on
in #149.
Make the home toolbar collapse on scroll.
I was planning to implement this back in 2.0.0, but visual bugs stopped
me. Now, knowing how much space the the Toolbar + Tab combination takes
up on smaller devices, I think it would be better to have this in
general.
Rework PrimitiveBackingData into SyncBackingData.
Google says that you should not use DiffUtil on the main thread, as it
is far too slow. In practice, the time it takes to diff is negligable,
even more so when costly diffs are filled in with a replace call
instead. Not only that, but switching to a synced list differ fixees
the horrible nightmare that is the queue adapter. Generally a win-win
for most use-cases in Auxio.
Rework the submitList animation to be less resource intensive and nicer
looking (at the cost of scroll positioning)
notifyDatasetChanged is slow and has no animation, but list diffing is
chaotic and basically useless outside of search. However, clearing the
adapter and then populating it with new items seems to work quite well,
albeit with the scroll position being lost sadly. Switch to that.
Extend the range of tags Auxio is capable of recognizing, removing some
edge-cases in the process.
This is mostly three changes:
1. Auxio no longer tolerates tracks/discs/years that are 0. These are
usually never valid.
2. Auxio now has support for original date/year values. This are niche
tags, but I was inspired to implement it by this HN discussion at
https://news.ycombinator.com/item?id=31659799 simply because it
differentiates Auxio.
3. Auxio can handle the more esoteric tag variations in vorbis, such
as "YEAR" or "ALBUM ARTIST".
Add a toggle for edge-to-edge mode.
Normally we would want to enable edge-to-edge by default (in fact, we
still do). However, some phones (once again, samsung) don't provide
Auxio with actual window insets. As a result, we need to add a toggle
so that it can be disabled on busted devices.
Did you know that even when Auxio has it's edge-to-edge functionality
busted, Samsung Music works just fine? Very interesting.
Resolves#149.
Use androidx compat tools when setting up edge-to-edge and applying
window insets.
This is mostly a readability tweak ahead of making edge-to-edge a
setting.
Remove the version-specific display name hacks added previously.
Turns out RELATIVE_PATH only includes the directory structure. Oh.
Since this invalidates the purpose of version-specific display name
hacks, we just remove it.
Do not add a callback to a MetadataRetriever call in ExoPlayerBackend.
This fixes an issue where if an error occurs while a future completes,
the app will crash entirely. This is because Futures run on a different
Executor and thus exceptions will not bubble up to Indexer.
Make the detail UIs respond to automatic rescanning events.
This currently has no effect, but is instead a preparation step for the
seamless addition of automatic rescanning.
Make IndexerService only auto-index music when there is no library AND
no loading has occured.
This resolves an issue where if IndexerService dies after a failed
index, IndexerService would try to index music regardless, as there was
no library present.
Add support for excluding directories on other volumes, at least from
Android Q onwards.
Previously, Auxio only supported excluding the primary volume. This was
mostly out of laziness, as the excluded directory implementation was
shamelessly copied from Phonograph. This commit completely refactors
the excluded directory system, dumpstering the old database (which was
overkill anyway) for a new system based on SharedPreferences that is
actually capable of handling external volumes.
Now, limitations regarding external volumes still apply below Android
Q, as the VOLUME_NAME field does not exist on those versions. However,
this should resolve at least one major complaint regarding the
excluded directory system. Now theres just all of the other complaints.
Resolves#134.
Fix an issue with highlighting in the detail UIs taht stemmed from the
ViewHolder position being cached.
Resorting could have changed the ViewHolder position, so caching it
will result in ViewHolders being incorrectly updated.
Rework genre parsing to properly align with the ID3v2 standard.
The ID3v2 genre specification is weirdly complex. Auxio ignored most
of that and just parsed out the most common cases for genres (digits
and (digits)). For the sake of completeness, rework Auxio's genre
parser to handle the weirder cases like RX/CR, multiple values,
and escaping, with an implementation roughly based on Mutagen's
genre parser.
Improve Indexer's state management by splitting up the current loading
state and the last response.
This is intended to resolve a bug where if the UI task and
IndexerService are both killed, the Indexer state would become
indeterminate and the library would not show. Resolve this by keeping
track of whatever the last completed state was and falling back to it
whenever the loading process is canceled.
Add a dedicated service towards the loading of the music library.
This new service was created for two reasons:
1. Music loading is slow and resource-intensive, so putting it on the
ViewModel layer just didn't seem right and made it vulnerable to the
OS simply stopping the loading process.
2. For automatic rescanning [#72], there must be something watching
the music library and waiting for a change in the background. This
would require a service as that is probably the least insane way to
do that kind of background work.
I have no garuntees how viable the service might be. If anything, it
might be halted by some insane android restriction or issue that
makes it more or less impossible to use for most apps, and I will have
to largely drop truly automatic rescanning.
Rework loading management into a new object, Indexer, which should act
as the base for all future loading changes.
Indexer tries to solve a two issues with music loading:
1. The issue of what occurs when tasks must be restarted or cancelled,
which is common with music loading.
2. Trying to find a good strategy to mediate between service and
ViewModel components in a sensible manner.
Indexer also rolls in a lot of the universal music loading code
alongside this, as much of MusicStore's loading state went unused
by app components and only raised technical challenges when reworking
it.
Rework genre ordering to be alphabetical based on album, but without
all of the mosaic being taken up by one artist.
With the ordering changes that were done previously, genre images
greatly differed compared to previously. May as well use this breakage
as an oppertunity to improve genre ordering. In this case, we want to
order albums alphabetically, as usual, but we also want to prevent
one artist from taking up the whole mosaic. Fix that by deduplicating
artists in the set of grouped albums.
Add an indicator to gague the current music loading progress.
This is actually a lot harder to implement than it might seem, not only
due to UI state issues, but also due to the fact that MusicStore needs
to keep it's state sane across a myriad of possible events that could
occur while loading music. This system seems like a good stopgap until
a full service-backed implementation can be created.
Apply the notifyItemChanged fix everywhere by making it an explicit
part of the RecyclerView framework.
This way, implementing future selection and rewrite behavior will be
much easier, as the payload argument is available in every adapter
implementation.
Fix two major highlighting bugs based around the janky and stupid way
I would handle highlighting previously.
Previously, I would index the views of a RecyclerView in order to
highlight viewholders. In retrospect this was a pretty bad idea,
as viewholders could be in a weird limbo state where they are bound,
but not accessible. I mean, it's in the name. It's a Recycling View.
Fortunately, google actually knew what they were doing and provided
a way to mutate viewholders at runtime using notifyItemChanged.
And the API actually makes sense! Wow! Migrate all detail adapters
to a system that uses notifyItemChanged instead of the terrible
pre-existing system.
Fix a state restore issue that would cause the parent to restore
incorrectly.
At some point, I accidentally used the index for the PlaybackMode field
when restoring the playbackState. This resulted in the playback mode
effectively reverting to ALL_SONGS and causing a number of subtle
issues.
Further refine the Indexer and ExoPlayerBackend implementations.
These fixes were primarily focused on ensuring stable grouping through
stable sorting order, and more graceful handling of edge cases in
ExoPlayerBackend.
Fix a regression where media button controls would no longer work
with GadgetBridge.
In 4a79de4, I assumed that my code was redundant, as there should be
some kind of reciever handling media button events already. I was
completely wrong. Re-add the reciever code (with extra checks) to
make controls work again.
Switch from LiveData to StateFlow.
While LiveData is a pretty good data storage/observer mechanism, it has
a few flaws:
- Values are always nullable in LiveData, even if you make them
non-null.
- LiveData can only be mutated on Dispatchers.Main, which frustrates
possible additions like a more fine-grained music status system.
- LiveData's perks are exclusive to ViewModels, which made coupling
with shared objects somewhat cumbersome.
StateFlow solves all of these by being a native coroutine solution with
proper android bindings. Use it instead.
Move the loading screen into the home view.
Previously, we would use a Snackbar to track the music loading state.
This ended up being a pretty stupid and buggy idea, and would get even
worse with the new music loader changes. Instead, we re-implement the
loading screen into the home view to generally be more sane and
extendable compared to previously.
Make addCallback calls (or similar) initalize the caller by calling the
specific callback methods.
This is primarily intended as a fix to an issue where the music library
would actually reload during a warm start. This is extremely bad, but I
hope it doesn't make the app too unusable until the next version.
Add an ExoPlayer-based metadata backend.
This backend finally allows Auxio to parse metadata internally, which
has been a highly requested feature given how many unfixable MediaStore
issues it would fix.
However, this is not fully ready for production yet. The loading
process becomes much larger with manual parsing, the current
implementation still allocates picture metadata (terrible for
efficiency), and there is no good indicator in the app to keep
track of music loading.
As a result, this backend is disabled until it can be fully integrated
into Auxio's architecture.
Move out the MediaStoreCompat interface into a full interface called
Backend.
In preparation for direct metadata parsing, it would be useful to
create some kind of object system to properly handle the capabilities
of each metadata indexing mode. Backend fulfills that by allowing
each object to implement their own query and then loading routine.
This system is designed somewhat strangely. This is firstly because
the ExoPlayer metadata backend will have to plug in to the original
MediaStore backend, so making methods more granular allows the
ExoPlayer backend to avoid some of the stupid inefficiencies from
the actual MediaStore backend, such as the genre loading process.
We also want to separate the steps of loading music in order to
more adequately show the current loading process to the user in
a future change.
Implement a safe slider wrapper that does not crash with invalid values
as often.
Slider is a terrible component that is not designed with Auxio's
use-case in the slightest. Instead of gracefully degrading with invalid
values, it just crashes the entire app, which is horrible for UX.
Since SeekBar is a useless buggy version-specific sh******ed mess too,
we have no choice but to wrap Slider in a safe view layout that
hopefully hacks with the input enough to not crash the app when doing
simple seeking actions.
I hate android so much.
Resolves#140.
Add disc number sorting to sorts that originally sorted by track.
Forgot to add disc sorting to the other sort modes when adding disc
number support. Fix that.
Create separate routines to parse disc and track numbers in the most
optimal manner.
On API 30 (Android R) and above, ensure that we are using the superior
CD_TRACK_NUMBER and DISC_NUMBER database fields, while below API 30,
stick with the normal TRACK field.
Revert the introduction of the thin/tiny widgets, but keep the new
cover layout I created while working on them.
There is simply no way I can cram controls and metadata within the
size bucket that the thin widget occupies. I have decided to give up
and revert the widget to it's old form.
I understand why the thin widget is not appealing. However, the sizing
at which a widget can properly accomodate a taller widget is just too
precise and not really large enough to justify it's existance.
Also use the gesture insets on playback panel.
Turns out when the panel uses normal insets, the playback panel is
extremely close to the system bars. Fix this by using the bigger
gesture insets.
Resolves#137.
Rework the smallest widgets to have no textual metadata, alongside
adding a new widget size class.
Turns out the last tiny widget redesign made the controls far too
small. Instead, remove the textual metadata and make these widgets
more akin to the small and wide widgets. This also introduces a new
side class alongside that with more controls, again to make it
similar to the other size classes.
Re-add accent customization on Android 12 and above.
Previously, I disabled accent customization since I thought they were
more or less useless with the new Material You dynamic colors system.
Turns out I severely underestimated how horribly OEMs would botch the
dynamic colors system. Guess I was blinded by my adherence to the pixel
line. Re-add the accent customization for those who do not have a good
dynamic color palette at all.
Resolves#131.
Rework the tiny widget to cram even more information into it.
The tiny widget is nominally meant for edge cases like exceptionally
small screens or landscape mode, but apparently it's triggered on some
devices in normal use because of platform fragmentation and OEM insanity.
Update the tiny widget layout with some new buttons in order to make it
more usable in mnormal use. This is still nowehre near ideal. For
example, when triggering the layout on my device, it ends up squishing
the buttons. But it should probably work better outside of those edge
cases.
Hack around more insane lollipop bugs, such as:
- The angular auxio icon crashing the system UI
- Optimized icons being corrupted
- Setting image alpha not working properly
I really wish I could drop support for this horrible version, but I
either have to wait for a major library to drop support or for the
usage numbers to reach 1%.
Replace the hodge-podge framework of state restoration and URI playing
with a single "delayed action" system.
Auxio's initialization routine is a total cluster----. This is mostly
because it involves multiple asynchronous operations such as music
loading, state restore, and service starting which tend to make it
highly prone to race conditions and other insanity.
In particular, the way Auxio would attempt to restore playback and
handle file opening was a spaghetti pile of bad API boundaries and
dubious UI code. This has not changed. I want to move this routine
to the service, but it's lifecycle is also sh------ed to such an
extent where that would be nearly impossible. Instead, this commit
introduces a new "delayed action" system that bites the bullet and
allowes PlaybackViewModel to accept a context and an action in
return for initializing playback...eventually.
I tried my best to eliminate as much memory leaks as I physically
could here, but could only go so far. Still though, even this insane
system is better than the UI-level LiveData shenanigans I did
previously, and actually works compared to the broken android
components that google keeps wanting you to use.
Remove references to android system strings, in favor of in-house
translations.
Previously Auxio would use the `android.R.string.ok` and
`android.R.string.cancel` strings to represent Ok and Cancel
respectively, but these system strings are actually untranslated on
some devices, so it is better for i18n if we use our own strings
for such.
Implement a UI frontend for customizing the ReplayGain pre-amp value.
This finally completes Auxio's ReplayGain implementation. Not only
that, it also shows how Auxio can use positive ReplayGain values,
unlike other apps.
As a side-note, this also fiddles with the dialog style somewhat.
I got carried away.
Resolves#114.
Implement an internal setting for a ReplayGain pre-amp setting.
Pre-amp is a lot like above reference volume regarding Auxio's
ReplayGain implementation, where I want to implement it in order
to allow ReplayGain to graduate from being labeled "experimental".
No UI frontend has been implemented just yet.
Add sorting modes for duration and song count.
This was requested previously in the now-closed UI/UX changes
megathread, however I have only gotten to it now.
Add recovery code to the music indexer in the case that Android doesn't
provide the DISPLAY_NAME field.
Nominally this should never happen, but OEMs will OEM and apparently
this does happen on some devices. Try to recover by grokking DATA for
a file name.
Rework the sort implementation to allow Auxio to leverage it's sorting
capabilities in a more powerful manner.
This is mostly the removal of stupid redunant methods and the change of
Sort overrides to sort in-place. This just gives us the option to avoid
full blown list copies in cases where such is reasonable.
Finalize the disc number implementation within Auxio.
This is probably one of the most widely-requested features outside
of playlisting. This implementation also adds some more fine grained
sorting modes for disc numbers in particular, which actually removes
some of the quirkiness of the Sort class.
Resolves#96.
Change the indexer to parse out the disc number field from
AudioColumns.TRACK.
On newer versions, I hope to further modify the loader to use
the API 30+ CD_TRACK_NUMBER field, as it is more versatile than
TRACK.
Add the UI and data components for a disc number value within the album
detail view.
This is the first-step of a multi-step addition to finally implement
disc numbers.
Make MediaButtonReceiver an empty class without any behavior.
The previous manner in which Auxio would handle media button events
was actually bad, as in certain cases (like a rewind or if there was
no song playing) Auxio would actually crash, as the service would
not start a foreground state in time.
Fix this by removing that functionality. This should hopefully still
retain the current state.
Try to rework the playback service components to reduce race conditions
and improve readability.
This changeset has gone through a number of changes. I originally
wanted to unify all cover loading under a single "Component Manager",
but this turned out to be stupid given that the three service components
are different in nearly every way. Instead I just reworked them all
individually by introducing a new less-data race-prone image loading
framework, and moving around a bunch of code I was planning to move
around.
Do some miscellanious formatting reworks.
1. Remove all instances of m in favor of _. _ is only used when names
collide or if something should be internal.
2. Make fragments apply their own click listeners.
3. Remove instances of inc/dec and replace them with the more
straightfoward + 1 or - 1.
Split off the "songs loaded" about item into it's own card called
"library statistics"
This card includes the song, album, artist, and genre counts,
alongside a total duration of the music library. This is just more
informative and useful to the user.
Resolves#121.
Rework state restore to be more coherent and in line with the new
member layout of the general
This primarily involves making the index the primary database attribute
in favor of song, with the old song id field becoming a sanity check
field.
Fix miscellanious bugs and clean the component code.
Currently the components are in a strange state. They are a big ball of
mud with inconsistent lifecycles and callbacks. I want to find a way to
unify them under a single lifecycle, but the competing nature of them
makes this extremely difficult.
Rename LoopMode to RepeatMode, which is generally much clearer
than the previous name.
This changes all non-breaking instances of "loop" in the app with
"repeat".
Rework the asynchronous aspects of MusicStore to rely on a more
conventional callback system.
In preparation for automatic rescanning, it would be more elegant
for music consumers to use a callback, so that updates can be
registered easier than if a co-routine continued to be used.
Convert MusicStore to this new model and rework the accesses of
MusicStore in-app to line up with this new system.
Create a ViewModel for the more complicated navigation pathways.
Normally, navigation was fragmented along a complicated stretch of
fragment hacks and DetailViewModel's navToItem attitbute, both of which
were not really that ideal. Dumpster them for a single, unified
viewmodel for the more complicated navigation situations. This removes
much of the duplicate navigation logic and is likely much more
maintainable for future situations.
Rework the style of all album covers in the app to be more in line with
the new track number style.
This is mostly comprised of adding a new background to all cover views
and rescaling error icons to be smaller than they would normally be.
This also includes a change in the cover/track background color from
colorSurfaceVariant to colorOnSurfaceInverse, which seems to provide
the best visibility in all cases.
These changes also apply to the track number views.
Update the album song layout to be more alike to other songs.
Recently I migrated the TextView in the album songs to use 48dp sizing,
like the other song views. However, this resulted in a lot of empty
space that felt off. Fix this by adding a light background to the track
number, which fills the room it takes up a bit more. It also hopefully
primes the track number to take an indicator once multi-select is
added.
Remove extraneous encodings from ReplayGainAudioProcessor.
According to the ExoPlayer developers, an AudioProcessor is only
provided 16-bit PCM data. This makes all of the other encodings I
had implemented in ReplayGainAudioProcessor useless, as their
cases would never run. Remove those extraneous encodings and just
just 16-bit PCM.
Implement support for positive ReplayGain values.
Turns out the blocker for this with the new AudioProcessor was that
I did not properly clamp PCM data when I manipulated the data,
resulting in target samples like 75491 being truncated to lower
values like 9955, which resulted in popping. This is a niche addition,
but also puts Auxio in a category that no other (FOSS) android music
player currently occupies. Yay.
Resolves#115.
Create an AudioProcessor implementation for ReplayGain.
Now that ExoPlayer handles AudioFocus, the ReplayGain implementation
would conflict with the changes that ducking would make to the volume.
To fix this, migrate the ReplayGain implementation to a dedicated
audio processor. This not only resolves this system, but also opens the
door for positive ReplayGain values in the future. Currently however,
our method for modifying the bitstream results in popping with values
above the reference volume, so some more work must be done in that
regard.
Rework audio focus to rely on the native ExoPlayer implementation
instead of a custom implementation.
Previously, we avoided ExoPlayer's AudioFocus system as it never
played after a transient lost. A few versions later now through,
now it does, so we may as well switch to it. This does introduce
a bug where ReplayGain functionality will conflict with audio
focus, but I hope to eliminate this with #115 as I switch to
an AudioProcessor instead of a callback.
Audit null safety to remove extraneous and stupid calls while
optimizing certain checks here and there.
This commit is primarily centered around the introduction of a new
utility: unlikelyToBeNull. This call uses requireNotNull on debug
builds and !! on release builds, which allows for bug catching in
an easier manner on normal builds while also allowing for
optimizations on the release builds.
Migrate all esoteric adapters to the new RecyclerView framework.
One of the shortcomings with the previous RecyclerView utilities was
how more esoteric adapters with data that does not implement Item
could not use the utlities. The new system, by comparison, is capable
of taking any type of data, so we can no migrate all of the esoteric
adapters to the new system.
Fix an issue where album covers fail to load when not explicitly
specifying a size.
At some point, Coil changed some internal scaling logic to depend on
this weird pseudo-screen size value that caused the bitmap to fail
to load only on in widget creation process. Who knows why. Fix this
by explicitly specifying the size in all cases.
