Repository navigation
Conversation
…enaming When a scene's file changes at the same path, the scan handler called MigrateHash, which renames the old hash's sprite/preview/transcode files onto the new hash's path instead of regenerating them. Every generator's "does this already exist" check then found a file already sitting there and skipped, leaving stale pre-edit content wearing the new hash's name - and the cover (a DB blob, never touched by MigrateHash) stayed stale too. Add InvalidateGeneratedFiles, used only for the same-path content-change case: deletes the old hash's generated files instead of renaming them, and clears the scene's cover. MigrateHash itself is unchanged and still renames for its real use case (the hash-naming-algorithm migration task, where content is unchanged). With the stale files/cover actually gone, every generator's existing exists-check works correctly on its own for any future regeneration path - scan, manual Generate, or scheduled. Also fixes ScenePlayer never refreshing a scene's video source/duration after such an edit: its player-(re)init effect only fired on scene ID change, so an in-place file edit under the same scene ID left the player showing the pre-edit duration/sources indefinitely. Fixes stashapp#7155 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
lurkfueh
marked this pull request as ready for review
August 14, 2026 07:33
Gykes
requested changes
Aug 30, 2026
Gykes
left a comment
Gykes
left a comment
Collaborator
There was a problem hiding this comment.
Just a quick pass through. No actual testing was done.
| // so stale content doesn't pass as valid for the new content | ||
| InvalidateGeneratedFiles(h.Paths, oldHash) | ||
|
|
||
| for _, s := range existing { |
Collaborator
There was a problem hiding this comment.
What about secondary files? I think this would clear for every scene even if it's the secondary file that gets modified, is that expected? I think this should be guarded to the primary file only, no?
Author
There was a problem hiding this comment.
Could I get more context on what secondary files are or a pointer to what file to start at?
| // where the old hash's generated files are no longer valid and should be | ||
| // deleted rather than renamed onto the new hash. | ||
| func InvalidateGeneratedFiles(p *paths.Paths, hash string) { | ||
| removeSceneFolder(filepath.Join(p.Generated.Markers, hash)) |
Collaborator
There was a problem hiding this comment.
Is this not essentially the same as line 67?
| scenePaths := p.Scene | ||
| removeSceneFile(scenePaths.GetVideoPreviewPath(hash)) | ||
| removeSceneFile(scenePaths.GetWebpPreviewPath(hash)) | ||
| removeSceneFile(scenePaths.GetTranscodePath(hash)) |
Collaborator
There was a problem hiding this comment.
This removes transcodes but your PR description said you're not touching them yet.
Author
There was a problem hiding this comment.
my bad, but do we want to remove them? I think they should be removed since they would now differ from the source.
…rprint comparison Responds to Gykes' review on stashapp#7158 plus follow-up findings from re-reviewing the same changes: - Guard both InvalidateGeneratedFiles and the cover-clear on whether the file that changed is actually a scene's primary file. Sprite/preview/ transcode/cover are generated from a scene's primary file, keyed on its hash - editing a secondary file (or a scene's primary file that happens to share a hash with another scene's untouched primary file, e.g. two files that started out as byte-identical duplicates) must not delete content that's still valid for the unchanged primary file. - Defer InvalidateGeneratedFiles to a post-commit hook instead of running it synchronously inside the DB transaction. Deleting from disk isn't part of the SQL transaction and can't be rolled back with it - if a later step in the same Handle() call fails, the DB now correctly keeps believing the old hash's files are current, because they were never touched. Registered before the existing ScanGenerator hook so deletion still runs before regeneration (post-commit hooks run in registration order). - migrate_hash.go: drop the redundant removeSceneFolder call (same path as the one three lines later), and extract the shared "check exists, tolerating IsNotExist" logic (existsForAction) used by all five exists-then-act call sites instead of duplicating it. - ScenePlayer.tsx: replace the fingerprint-join comparison with fingerprintsMatch, comparing only fingerprint types present in both the previous and current file snapshots. Fixes the false positive Gykes flagged (a phash finishing generation in the background looking like a content edit and resetting playback position) without reintroducing the false positive of a size/mod_time-based key (touch, a non- timestamp-preserving copy, or a backup restore bumping those without changing content) - fingerprints are actual content hashes, immune to both. Also short-circuits to "unchanged" when both snapshots have no fingerprints at all, so a scene whose file hasn't finished fingerprinting yet doesn't get its player reset on every unrelated re-render (e.g. periodic resume_time saves during normal playback). - New tests: TestHandle_ContentChangedAtSamePath_SecondaryFile, _SharedHashSibling, and _RollbackDoesNotDeleteFiles (forces a later failure and asserts the generated files survive). Existing tests switched from mocks.Database.WithTxnCtx (which always rolls back by design) to a new withCommittedTxn helper so the post-commit deletion path is actually exercised. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Condensed fingerprintsMatch's docblock and the effect's guard comment - same information, no restatement/rationale-per-sentence. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Same two spots as the ScenePlayer.tsx trim - the isPrimary guard and the post-commit hook explanation were both over-explained. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
migrate_hash.go: merge removeSceneFile/removeSceneFolder into one removePath helper, and use fsutil.RemoveDir instead of raw os.RemoveAll. scan_test.go: extract newTestScanHandler and seedStaleFile, shared by all four TestHandle_ContentChangedAtSamePath* tests; hoist the oldHash/newHash literals (identical across three of them) to package consts. ScenePlayer.tsx: merge the sceneId/fileFingerprints refs - they were always read, written, and reset together - into one loadedSource ref. Rewrite fingerprintsMatch as a single loop instead of a Map plus an intermediate filtered array. All within code this PR added; nothing pre-existing restructured beyond what the review-feedback commit already touched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Deliberate pass over every comment this PR added - condensing the rest that were still over-explained (doc comments restating the same point across multiple sentences, test comments repeating what the adjacent one already said). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Deletes generated content when a file is modified (i.e. trimmed) but the file path remains the same. Also modifies the UI scene player to track media fingerprint instead of just scene id when deciding to update the player.
Related Issue
Fixes #7155
Testing
pkg/scene/scan_test.gocovering my changesScreenshots
Checklist
AI Usage Disclosure
used claude to make the code changes, tested and verified that the changes work
Additional Context
Note, this does not touch transcodes (I think they should be removed also but want to hear from maintainers)