mirror of
https://github.com/Kareadita/Kavita.git
synced 2025-05-24 00:52:23 -04:00
5 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
Joseph Milazzo
|
56cf7be799
|
Scan Chunking (#604)
* Some performance refactoring around getting Library and avoid a byte[] copy for getting cover images for epubs. * Initial commit. Rewrote the main series scan loop to use chunks of data at a time. Not fully shaken out. * Hooked in the ability for the UI to react to series being added or removed from the DB. * Cleaned up the messaging in the scan loop to be more clear. * Metadata scan and scan work as expected and populate data to the UI. There is a slow down in speed for overall operation. Scan series and refresh series metadata does not work fully. * Fixed a bug where MangaFiles were not having LastModified Updated correctly, meaning they were opening archives every scan. * Modified the code to be more realistic to the underlying file * Updated ScanService to properly handle deleted files and not result in a higher-level scan. * Shuffled around volume related repo apis to the volume repo rather than being in series. * Rewrote scan series to be much cleaner and more concise on the flow. Fixed an issue in UpdateVolumes such that the debug code to log out removed volumes could throw an exception and actually break updating volumes. * Refactored the code to set MangaFile last modified timestamp into the MangaFile entity. * Added Series Name to ScanSeries event * Added additional checks in ScanSeries to ensure we never go outside the library folder. Added extra debug messages for when a metadata refresh doesn't actually make changes and for when we regen cover images. * More logging statements saying where they originate from. Fixed a critical bug which caused only 1 chunk to ever be processed. * Fixed a concurrency issue with natural sorter which could cause issues in ArchiveService.cs. * Log cleanups * Fixed an issue with logging out total time of a scan. * Only show added toastrs for admins. When kicking off a refresh metadata for series, make sure we regenerate all cover images. * Code smells on benchmark despite it being ignored |
||
|
Joseph Milazzo
|
82b5b599e0
|
Foundational Cover Image Rework (#584)
* Updating wording on card item when total pages is 0, to be just "Cannot Read" since it could be a non-archive file * Refactored cover images to be stored on disk. This first commit has the extraction to disk and the metadata service to handle updating when applicable. * Refactored code to have the actual save to cover image directory done by ImageService. * Implemented the ability to override cover images. * Some cleanup on Image service * Implemented the ability to cleanup old covers nightly * Added a migration to migrate existing covers to new cover image format (files). * Refactored all CoverImages to just be the filename, leaving the Join with Cover directory to higher level code. * Ensure when getting user progress, we pick the first. * Added cleanup cover images for deleted tags. Don't pull any cover images that are blank. * After series update, clear out cover image. No change on UI, but just keeps things clear before metadata refresh hits * Refactored image formats for covers to ImageService. * Fixed an issue where after refactoring how images were stored, the cleanup service was deleting them after each scan. * Changed how ShouldUpdateCoverImage works to check if file exists or not even if cover image is locked. * Fixed unit tests * Added caching back to cover images. * Caching on series as well * Code Cleanup items * Ensure when checking if a file exists in MetadataService, that we join for cover image directory. After we scan library, do one last filter to delete any series that have 0 pages total. * Catch exceptions so we don't run cover migration if this is first time run. * After a scan, only clear out the cache directory and not do a deep clean. * Implemented the ability to backup custom locked covers only. * Fixed unit tests * Trying to figure out why GA crashes when running MetadataServiceTests.cs * Some debugging on GA tests not running * Commented out tests that were causing issues in GA. * Fixed an issue where series cover images wouldn't migrate * Fixed the updating of links to actually do all series and not just locked |
||
|
Joseph Milazzo
|
cf4fd2cb9c
|
Bugs, Enhancements, and Performance (#580)
* Added parser case for "The Duke of Death and His Black Maid - Ch. 177 - The Ball (3).cbz" * Removed a file that is created and modified every test run. * Fixed a bad parser case for "Batman Beyond 02 (of 6) (1999)" which was consuming too many characters * Removed a lot of "Volume" parsing for Comics that don't make sense. This is prep work for the upcoming Comic Rework release. * Reworked a lot of parsing cases for comics based on naming conventions observed from releases found online. * Added a way for external scripts to use a user api key to authenticate * Fixed an issue if the manga only had one page, the bottom menu would be missing page and chapter controls. * Fixed a bug where on small phones, nav bar could overflow due to scroll to top * Tweaked a lot of regex for manga parsing to handle some cases where poorly named files, like "Vol. 03 Ch. 21" would end up parsing as Series "Vol. 03". * Even more handling of parser cases. Manga parser should be as it was but more robust to handle bad naming. * Fixed: Don't force metadata refresh on Scan Series, only on refresh metadata * Implemented the ability to automatically refresh after a series scan based on when server finishes. Remove a duplicate API call from series detail. * Removed another API call for series metadata that isn't needed. * Refactored Message creation to a factory, hardcoded strings are centralized, and RefreshSeriesMetadata sends an event and is refactored to be async. * Fixed a bug when really poorly named files are within a folder that contains the series name, fallback couldn't occur due to it being taken as root folder. Now we detect said condition and will go one level higher, resulting in potentially more I/O, but the series will not be deleted. * Added the Read in Incognito context item for Chapter cards * Skip an additional check for series summary for series that aren't EPUB or Archive formats. * Fixed an issue where cover image generation could occur due to a bad check on LastWriteTime on the underlying file. * Added some extra comic parser tests * Added a ScanLibrary event (not hooked up in UI) * Performance improvement on metadata service. Now when we scan for cover image changes, we emit when a change occurs and only then do we update parent entities (array copy). * Removed an hr from series detail and ensure we update the cover image for series when scan series finishes. * Updated the infinite scroller to use a Flags pattern for the debug mode. Updated a few logical conditions for mobile. * Removed the concurrency check on row progress as if too many calls hit the DB, it will throw, but it doesn't matter. Fixed a bad logic code which could cause scrolling after hitting the bottom of the chapter. * Ensure prefetching uses totalPages + 1 since we pass in totalPages as - 1 from manga reader * Fixed issue where last page of webtoon wouldn't be prefetched due to a < instead of <= on prefetching code * Implemented ability to send images from archives to the UI without incurring any extra memory pressure. * Dropdown menus now have a darker background * Webtoon reader now works on mobile. * Fixed how keyboard presses for up/down/left/right work with MANGA_UD reading mode. See issue #579 * Fixed cont reader for webtoons on mobile * Fixed a small issue where top spacer would too quickly switch to prev chapter * Updated user preferences to use same slider style. Removed some css that is not used. * Added comic parser case for "Saga 001 (2012) (Digital) (Empire-Zone)" * Added accessibility toggle to reading list order and aligned sliders to all use the same style. * Removed a todo for checking on new image serving code. It works great. * Fixed a missing await * Auth guard will now check if an existing toast is present giving same message before poping the toast. * Fixed alignment on phones for reading lists * Moved sorters so they aren't resused between multiple threads. Slightly higher memory footprint. * Fixed a broken unit test * Code smells * More unit test fixing |
||
|
Joseph Milazzo
|
83f8e25478
|
Continuous Reading for Webtoons & I Just Couldn't Stop Coding (#574)
* Fixed an issue from perf tuning where I forgot to send Pages to frontend, breaking reader. * Built out continuous reading for webtoon reader. Still has some issues with triggering. * Refactored GetUserByUsernameAsync to have a new flavor and allow the caller to pass in bitwise flags for what to include. This has a get by username or id variant. Code is much cleaner and snappier as we avoid many extra joins when not needed. * Cleanup old code from UserRepository.cs * Refactored OPDS to use faster API lookups for User * Refactored more code to be cleaner and faster. * Refactored GetNext/Prev ChapterIds to ReaderService. * Refactored Repository methods to their correct entity repos. * Refactored DTOs and overall cleanup of the code. * Added ability to press 'b' to bookmark a page * On hitting last page, save progress forcing last page to be read. Adjusted logic for the top and bottom spacers for triggering next/prev chapter load * When at top or moving between chapters, scrolling down then up will now trigger page load. Show a toastr to inform the user of a change in chapter (it can be really fast to switch) * Cleaned up scroll code * Fixed an issue where loading a chapter with last page bookmarked, we'd load lastpage - 1 * Fixed last page of webtoon reader not being resumed on loading said chapter due to a difference in how max page is handled between infinite scroller and manga reader. * Removed some comments * Book reader shouldn't look at left/right tap to paginate elems for position bookmarking. Missed a few areas for saving while in incognito mode * Added a benchmark to test out a sort code * Updated the read status on reading list to use same style as other places * Refactored GetNextChapterId to bring the average response time from 1.2 seconds to 400ms. * Added a filter to add to list when there are more than 5 reading lists * Added download reading list (will be removed, just saving for later). Fixes around styling on reading lists * Removed ability to download reading lists * Tweaked the logic for infinite scroller to be much smoother loading next/prev chapter. Added a bug marker for a concurrency bug. * Updated the top spacer so that when you hit the top, you stay at the page height and can now just scroll up. * Got the logic for scrolling up. Now just need the CSS then cont infinite scroller will be working * More polishing on infinite scroller * Removed IsSpecial on volumeDto, which is not used anywhere. * Cont Reading inf scroller edition is done. * Code smells and fixed package.json explore script |
||
|
Joseph Milazzo
|
2fd02f0d2b
|
Custom Cover Images (#499)
* Added some documentation. Removed Require Admin Role from Search Tags. Added Summary to be updated on UpdateTag. * Added Swagger xml doc generation to beef up the documentation. Started adding xml comments to the APIs. This is a needed, slow task for upcoming Plugins system. * Implemented the ability to upload a custom series image to override the existing cover image. Refactored some code out to use ImageService and added more documentation * When a page cache fails, delete cache directory so user can try to reload. * Implemented the ability to lock a series cover image such that after user uploads something, it wont get refreshed by Kavita. * Implemented the ability to reset cover image for series by unlocking * Kick off a series refresh after a cover is unlocked. * Ability to press enter to load a url * Ability to reset selection * Cleaned up cover chooser such that reset is nicer, errors inform user to use file upload, series edit modal now doesn't use scrollable body. Mobile tweaks. CoverImageLocked is now sent to the UI. * More css changes to look better * When no bookmarks, don't show both markups * Fixed issues where images wouldn't refresh after cover image was changed. * Implemented the ability to change the cover images for collection tags. * Added property and API for chapter cover image update * Added UI code to prepare for updating cover image for chapters. need to rearrange components * Moved a ton of code around to separate card related screens into their own module. * Implemented the ability to update a chapter/volume cover image * Refactored action for volume to say edit to reflect modal action * Fixed issue where after editing chapter cover image, the underlying card wouldn't update * Fixed an issue where we were passing volumeId to the reset chapter lock. Changed some logic in volume cover image generation. * Automatically apply when you hit reset cover image |