* use single-threaded image resizing in native code so we have more control over goroutines
* implement parallel-free versions of image transform functions also
* remove debug code
* use a single instance of wazero runtime and compiled modules
* remove test output 🤦
* undo process-{media,emoji} changes
* update test runner to include wazero compilation cache
* sign drone.yml
---------
Co-authored-by: tobi <tobi.smethurst@protonmail.com>
* on account update, ensure that public key has not changed
* change expected error message
* also support the case of changing account keys when expired (not waiting for handshake)
* tweak account update hardening logic, add tests for updating account with pubkey expired
* add check for whether incoming data was via federator, accepting keys if so
* use freshest window for federated account updates + comment about it
* include pinned status when incrementing / decrementing status counts
* remove the pinned increment on status creation
* code comments
* microoptimize decr
* wrap thumbnailing code to handle generation natively where possible
* more code comments!
* add even more code comments!
* add code comments about blurhash generation
* maintain image rotation if contained in exif data
* move rotation before resizing
* ensure pix_fmt actually selected by ffprobe, check for alpha layer with gifs
* use linear instead of nearest-neighbour for resizing
* work with image "orientation" instead of "rotation". use default 75% quality for both webp and jpeg generation
* add header to new file
* use thumb extension when getting thumb mime type
* update test models and tests with new media processing
* add suggested code comments
* add note about thumbnail filter count reducing memory usage
* [bugfix] send back Sec-Websocket-Protocol header for streaming WebSocket
Chrome expects the selected Sec-Websocket-Protocol to be sent back
on the WebSocket upgrade request (RFC6455 1.9).
* fiddle a bit to avoid getting headers multiple times
* add some explanatory notes
---------
Co-authored-by: tobi <tobi.smethurst@protonmail.com>
* add test for emoji update image
* update emoji recache to set the instance account id
* don't refresh emoji if only not cached. in that case literally just recache
* code comment
* rename + move a few things
* add some more code comments, and rename some functions to make logic a bit clearer
* remove unnecessary nil check (the value can be nil)
* comment wording
* remove test data output
* handle the case of caching an emoji which has been refreshed then uncached
* allow overwriting on testrig storage as we do now on regular storage
* fix emoji category ID not getting updated
---------
Co-authored-by: tobi <tobi.smethurst@protonmail.com>
* [chore] Synchronise our robots.txt with upstream
* [feature] Add headers to escape AI crawlers
This adds 2 headers that a number of AI crawlers respect to signal that
content should not be included in their datasets.
* add back exif-terminator and use only for jpeg,png,webp
* fix arguments passed to terminateExif()
* pull in latest exif-terminator
* fix test
* update processed img
---------
Co-authored-by: tobi <tobi.smethurst@protonmail.com>
* Implement Mastodon-compatible roles
- `Account.role` should only be available through verify_credentials for checking current user's permissions
- `Account.role` now carries a Mastodon-compatible permissions bitmap and a marker for whether it should be shown to the public
- `Account.roles` added for *public* display roles (undocumented but stable since Mastodon 4.1)
- Web template now uses only public display roles (no user-visible change here, we already special-cased the `user` role)
* Handle verify_credentials case for default role
* Update JSON exact-match tests
* Address review comments
* Add blocks bit to admin permissions bitmap
* persist queued worker tasks to database on shutdown, fill worker queues from database on startup
* ensure the tasks are sorted by creation time before pushing them
* add migration to insert WorkerTask{} into database, add test for worker task persistence
* add test for recovering worker queues from database
* quick tweak
* whoops we ended up with double cleaner job scheduling
* insert each task separately, because bun is throwing some reflection error??
* add specific checking of cancelled worker contexts
* add http request signing to deliveries recovered from database
* add test for outgoing public key ID being correctly set on delivery
* replace select with Queue.PopCtx()
* get rid of loop now we don't use it
* remove field now we don't use it
* ensure that signing func is set
* header values weren't being copied over 🤦
* use ptr for httpclient.Request in delivery
* move worker queue filling to later in server init process
* fix rebase issues
* make logging less shouty
* use slices.Delete() instead of copying / reslicing
* have database return tasks in ascending order instead of sorting them
* add a 1 minute timeout to persisting worker queues
* Implement followed tags API
* Insert statuses with followed tags into home timelines
* Test following and unfollowing tags
* Correct Swagger path params
* Trim conversation caches
* Migration for followed_tags table
* Followed tag caches and DB implementation
* Lint and tests
* Add missing tag info endpoint, reorganize tag API
* Unwrap boosts when timelining based on tags
* Apply visibility filters to tag followers
* Address review comments
* take into account rotation when generating thumbnail, simplify ffprobe output to show only fields we need
* only show rotation side data
* remove unnecessary comment
* fix code comments
* remove debug logging
