fix(story): no rechazar el MP3 custom transcodificado ni fallar el cover del catálogo

[cursor]

Resumen ejecutivo

Qué cambia

El import de MP3 custom desde el disco acepta el WAV que deja /upload-audio. La generación de cover también resuelve un tema del workspace/catálogo, no solo uploads/audio/.

Para qué sirve

En #245, Import custom MP3 desde el dispositivo sube el archivo, el servidor transcodifica mp3→wav y commitStoryAudioChoice(..., requireMp3) rechazaba ese artefacto. El cover del picker compartido guardaba solo el basename; si el tema vivía en el workspace, MiniMax respondía 400.

Impacto para el usuario

Volver a poder importar un MP3 desde el disco como candidato de cue. Poder generar un cover usando un audio ya generado en la biblioteca.

Riesgo

• Bajo
• Medio
• Alto

Estado

• Listo para revisión

Summary

isCustomMp3Output now treats /api/v1/uploads/audio/ (and uploads/audio/ paths) as a valid custom-MP3 choice, because /upload-audio transcodes mp3/m4a/aac to wav before onChoose. Catalog non-mp3 rows stay rejected. Cover jobs resolve uploads/audio/{basename} first, then the workspace-confined media path, so a catalog track bound as coverReferenceFilename is found.

Overview

PR #245 routes cover / Lyria / custom MP3 through StoryAudioPicker. Device files upload via AssetInput → uploadAudio. That endpoint always transcodes mp3 to {uuid}.wav and deletes the original. The new requireMp3 guard then saw .wav and returned reject, so every from-device custom import failed after a successful upload.

Cover generation still joined uploads/audio/ + basename only. A library song stored in the workspace (typical MiniMax output) produced a success notice in the UI and then Upload a valid reference song before generating a cover.

Detailed changes

Backend

resolve_story_cover_audio in media_paths.py. Both MiniMax music candidate endpoints use _story_cover_reference_path instead of _safe_join(uploads/audio, basename) only.

UI and Wizard

isUploadedAudioOutput + isCustomMp3Output accept the post-transcode upload artifact. Catalog wav/flac via /api/v1/file/ still reject.

Data, provenance and compatibility

No sidecar change. Custom candidates keep model: custom-audio-upload. Cover still stores basename; the server now finds that name in uploads/audio or the workspace.

Files and ownership

• ui/src/features/stories/storyAudioPick.ts — accept transcoded uploads as custom MP3
• ui/tests/storyAudioPick.test.mjs — device wav upload still applies; catalog wav still rejected
• app/services/media_paths.py — cover resolver
• app/_launch_runtime.py — both music-candidate endpoints
• tests/test_media_path_security.py — upload-then-workspace cover resolution + AST check

Validation

• Date (UTC): 2026-09-07

• Base SHA: 12b07db (grok/agente2-story-audio-06ffec07)

• Head SHA: 2a50c15

• Validation scope: focused

• python scripts/verify_clean_repo.py

• python -m compileall -q app/services/media_paths.py

• Focused Python tests: python3 -m pytest tests/test_media_path_security.py tests/test_story_lab_audio_ui.py -q — 11 + 17 passed

• cd ui && npm run i18n:check

• UI tests: npx tsx --tsconfig tsconfig.app.json --test tests/storyAudioPick.test.mjs — 5/5 pass

• cd ui && npm run lint -- --max-warnings=0

• cd ui && npm run build

• git diff --check

• E2E/smoke checks: N/A (picker commit + path resolver; no GPU)

Code quality

• Score: pending CI
• Complexity trend: pending CI
• Production LOC trend: pending CI
• Regression versus main: pending CI
• Ratchet: pending CI

CI and review

• CI of this HEAD: pending
• Independent agent review of this HEAD: pending
• Human merge click (operational, not code review): pending
• Pending

Coste de la tarea

• Tests simulados: 0 tokens externos
• Tests reales: N/A
• Llamadas LLM externas: 0
• Tokens de prompt: N/A
• Tokens de respuesta: N/A
• Tokens totales: N/A
• Generaciones de imágenes/audio/vídeo: 0
• Tiempo transcurrido: N/A
• Proveedores/modelos: N/A

Notes and limitations

Fix PR for #245. Merge this into grok/agente2-story-audio-06ffec07 before landing the picker work. Catalog non-mp3 rows remain rejected. Cover still prefers uploads/audio/ when the same basename exists there and in the workspace.

Follow-up work

None.

Checklist

• The executive summary is understandable without reading the code.
• The detailed Summary/Overview has not been removed or shortened.
• Tests and their actual results are recorded.
• Generated assets, secrets and local-only files are not committed.
• Required CI and Cursor/Bugbot review are complete, or the PR is clearly marked as waiting for them.

  

---

Note

Low Risk
Narrow UI guard and confined media-path resolution reuse existing security patterns; no auth or data-model changes.

Overview
Fixes two Story audio picker regressions: custom MP3 from disk and cover generation from catalog/workspace tracks.

On the UI, isCustomMp3Output now treats artifacts under uploads/audio (including the WAV returned after /upload-audio transcodes mp3/m4a/aac) as valid custom imports when requireMp3 is set, while catalog rows served via /api/v1/file/ still must look like MP3.

On the backend, cover jobs no longer join only uploads/audio/{basename}. New resolve_story_cover_audio prefers a confined file in uploads/audio, then falls back to resolve_permitted_media_path in the active workspace. Both start_story_music_candidates_job and generate_story_music_candidates use _story_cover_reference_path for that logic. Tests cover upload-vs-workspace precedence and AST checks that both endpoints call the shared helper.

^{Reviewed by Cursor Bugbot for commit 2a50c15. Configure here.}

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_e25bafd4-a0e9-4797-ac35-9beba4a7f77e)

[github-actions]

PR Review — Loreframe Studio

Risk: low
Scope: 5 file(s); +141/-10; React UI, backend services

Automated review from scripts/analyze_pr.py. This is a heuristic pass (no LLM) so humans still own the merge decision.

Findings

• low — UI changed — rebuild before merge
  Run cd ui && npm run build (CI already does this). Pinokio Update rebuilds for end users; keep ui/dist untracked.

Changed files

• modified: app/_launch_runtime.py, app/services/media_paths.py, tests/test_media_path_security.py, ui/src/features/stories/storyAudioPick.ts, ui/tests/storyAudioPick.test.mjs

CONTRIBUTING checklist

• python scripts/verify_clean_repo.py
• python -m compileall -q app/services app/launch.py scripts
• cd ui && npm run build if the UI changed
• No weights, CivitAI sidecars, or generated guides
• Stays local-first (no required accounts / telemetry)

Posted by the repo PR review workflow. Re-runs on each push to the PR.