Keeps the changelog entries that f45d0ad31d removed for the release under
Unreleased, since main still has the WebGPU reader and the new image
decoder.
(cherry picked from commit 79fc3a0a8a)
* Add Comick Tracker
Comick specifically requires a native reverse-domain scheme (just
`mihon://` is not accepted), so I went with `app.mihon://` which works
fine.
I added a new `intent-filter` for this because adding the scheme to
the existing `intent-filter` would effectively create all permutations
of schemes and hosts (i.e. it'd also accept `app.mihon://anilist-auth`
which we don't use).
A real client ID for Mihon's actual client is still missing.
* Changelog
Conflicts:
CHANGELOG.md
* Fix library adding failing with 400
I made this change after testing because I thought it'd have no
consequences.
Don't do that.
Comick does not allow `progress: {number: null}`, only
`progress: null` or `progress: {number: 123}`.
* Adjust to fit with changes in #4068
* Cut search query to fit Comick's 100 char limit
* Mihon client ID
* Close fire-forget requests according to c46317f0
* Add refresh button to tracker for new profile info
As-is, allows a user to manually refresh their current user data from
the tracker in order to update Mihon's knowledge of remote settings
(e.g. nickname changes, AL/MB/Kitsu score types, etc.).
* Catch errors to prevent app crashing
Surfaces errors as toasts in the same way library title refreshes do.
* spotless
* [skip ci] Remove leftover debugging logs
* Fix Kitsu user attribute access
* Spotless
* Lowercase Kitsu score type on refresh
Maybe prevent needing another #3887 for this.
* Center tracker vertically name if not logged in
Looks much better than the name being stuck at the top with empty
space below it.
* Don't show Refresh button for Enhanced trackers
* [skip ci] Add changelog entry
* Use & write Year, Month, Day info in ComicInfo
Currently, users can only modify the date Mihon displays in the
chapter list by manipulating the _last modified date_ of the file.
This is obviously inconvenient, especially since the ComicInfo spec
includes Year, Month, and Day fields for this exact purpose.
If present, this reads the date at start of day in the current default
system time zone (which matches how we use Chapter.dateUpload with
relativeDateText, for example).
If Year is not specified, the entire date process is skipped.
If Month is not specified, it is set to January.
If Day is not specified, it is set to the first.
The parsed date (if any) is set as the Chapter's `date_upload` which
is used for display in the app.
Additionally, Mihon now includes this info in the written
ComicInfo.xml file for downloaded chapters. This will be convenient
for many more users, not least of which those who move
source-downloaded chapters to Local Source to read from there.
---
I used kotlinx.datetime because it's what we use for date parsing in
non-i18n places already.
* Changelog
A leftover of what I assume was the v1 API behaviour handling manually
multiplied the `expires` values by 1,000, supposedly to make it a
millisecond value. However, AL tokens have been saved as milliseconds
since at least the v2 API implementation in 2018, so this has made
every expiry comparison ludicrously impossible.
Example:
millisecond timestamp:
1788529931115 (2026-09-04T13:52:11.115Z)
*1,000
1788529931115000 (+58646-04-04T21:45:15Z)
So any expiry comparisons would say "yes this token is still totally
valid", causing Mihon to send expired credentials and AL to respond
with a 401.
With these changes (which include some loosely associated clean-up),
we now correctly identify expired credentials before we attempt any AL
requests & actually show a toast with the "Token expired." error
message for the first time in 8 years or so?
I was able to keep the new ALOAuth shape compatible with
already-serialized data, though whenever a user (re-)links AL from now
on, both the `token_type` and `expiresIn` fields won't exist any more.
Since they had no use in the current code anyway, I don't consider
this a problem.
I took the liberty of adding basic instructions to the error message
but these can't be i18n'd properly so I kept it short and sweet.
---
Side effect of expired credentials is Mihon erasing the token in
storage, meaning AL disappears from the tracker sheet until re-linked.
I would say this is acceptable, but we might see panicked users in
support saying "Mihon deleted my AL links" or something.
While there are still some nasty corners (multiple shapes of error
response, unhandled error types on their end, etc.), this is a
functional, feature-parity replacement for the JSON:API API we were
using before.
According to the GraphQL docs, searching by title still goes via
Algolia.
Add support for different Kitsu rating scales
Includes a database migration because we have been scaling Kitsu's
2-20 `ratingTwenty` integer value to a 1-10 (step 0.5) value and
stored that in the database.
I'm also "rounding" the values to the next lowest valid one in the
current rating system to mirror Kitsu's behaviour for this. Both Kitsu
and the app keep the misfit value around until the score is otherwise
edited, at which point the interface constrains the user selection to
only valid values (for the current system).
Maybe the RatingSystem data class is overkill, but a previous version
of mine had three constants for each system that had to be manually
associated at each corner. Encapsulating those in a little map of data
classes seemed the more ergonomic solution.
This was included for all other trackers in #3776 but these two had
prefix search before.
It's a small change, but it can go a long way when using `id:` with
copy-pasted IDs from elsewhere which come with starting or trailing
whitespace.
Also added author info for `id:` prefix search results since those hit
the single-series endpoint instead of `/search`, which does not
include author information (I wish it did).
Maybe unintuitive but the info is nice to have when available.
* Add id: prefix search to AniList
* Add id: prefix search to Bangumi
* Add id: prefix search to Kitsu
* Add id: prefix search to MangaUpdates
* Add id: prefix search to Shikimori
* Add id: prefix search to Hikka
Since Hikka doesn't use integer IDs and instead relies solely on a slug system,
users have to supply the slug to the `id:` search (so not `id:123` but
`id:sousou-no-frieren-97e05f` to get the Frieren manga specifically)
It's asymmetrical to the other trackers, but I thought keeping the prefix
consistent was more important than to be true to the tracker's internal
workings.
* Parse MU string IDs as base36 Longs
Turns out that's what the random strings in the URL are.
Co-authored-by: BrutuZ <7104931+BrutuZ@users.noreply.github.com>
* [skip ci] Add Changelog
---------
Co-authored-by: BrutuZ <7104931+BrutuZ@users.noreply.github.com>
* Fix wrong MangaBaka scores for step sizes > 1
We're actually sending the index of the score type in the list, not
the score itself.
Overriding indexToScore solves that issue.
Also refactored the ranges to be constants for consistency.
* Changelog
* Fix Mangabaka dates drifting in negative timezones
This was only partially handled in #3047, but MB returns full datetime
strings with a truncated time component that's always midnight UTC.
So when Mihon then parses the returned dates, it gets the day wrong in
any negative offset timezones.
With this, Mihon now discards the time component and instead takes the
date component + start-of-day in the device timezone.
* Add Changelog
* Fix relative date display
DatePeriod.days provides the number of days in the DatePeriod specifically, not the total number.
LocalDate.daysUntil provides what we were looking for.
* Add Changelog
This has the side effect of forcing the date to something concrete.
For Year-only: 1st of January that year
Ex: 2025 -> 2025-01-01
For Year+Month: 1st of said month
Ex: 2024-11 -> 2024-11-01
MAL allows setting Day+Month and Day+Year, but both of those are sent
as `null` in the API, which means Mihon will not show a date at all.
This has the side effect of Mihon sending up the date as `null` the
next time info is sent to MAL, which will clear the date fields
completely.
Co-authored-by: AntsyLich <59261191+AntsyLich@users.noreply.github.com>
Init the reader from the view model instead of the activity so it isn't
cancelled by a configuration change mid init, which left the retained
view model half initialized and needsInit() returning false.
Assisted-by: Claude Code:claude-opus-5
- Spoof `Sec-CH-UA` client hints to match the user agent
- Fix `X-Requested-With` spoofing leaking to unrelated callers
- Update default user agent to Chrome 149
* Add category filters for Updates tab
Works just like Library Updates.
Provides a tab in the Updates tab filter dialog to allow for including
& excluding of updates belonging to a title from a given category.
Excludes overpower Includes, as with Library Updates.
A little helpful text informs users about this.
* Changelog
* [skip ci] Remove unused string
Leftover from before I realized the Default category will always be
available for in-/exclusion.
* Make :includeEmpty/:excludeEmpty true Boolean params
* "include" -> "included"
Consistency with #3607.
* Remove DISTINCT from getRecentUpdatesWithFilters
Apparently not needed anymore (though it was required at some point).
* [skip ci] Fix Changelog
* Fix backup restore dropping library entries
History restore resolved chapters by URL alone, so a backup holding the
same entry twice under one source matched multiple chapter rows and threw
`ResultSet returned more than 1 row`. Scope the lookups to the entry.
Entries are restored in chunks of 100 sharing one transaction. SQLDelight
fails the enclosing transaction when a nested one fails, so catching the
per-entry exception did not contain it and the whole chunk rolled back,
losing up to 100 entries per bad one. Retry entry by entry on failure.
Refs #647
* Update changelog
---------
Co-authored-by: Naji Astier <na-ji@users.noreply.github.com>
AntsyLich