- Handle both cookie formats: flat array and {"cookies": [...]} object
(scrapers page saves the latter, code only accepted the former)
- Update scraper status to 'ok' after successful authenticated API call
(previously only set to 'failed' on 401, never cleared on success)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Raw filesystem paths were exposed in browser URLs, dev tools, and proxy logs.
Now all file-serving endpoints accept an opaque encrypted token (t= param)
derived from the session secret via HKDF, with a 4-hour TTL.
Backend:
- Add core/path_tokens.py with Fernet encrypt/decrypt (HKDF from .session_secret)
- Add file_token to all list/gallery/feed/search responses across 7 routers
- Accept optional t= param on all file-serving endpoints (backward compatible)
Frontend:
- Update 4 URL helpers in api.ts to prefer token when available
- Add 4 new helpers for paid-content/embedded-metadata URLs
- Update all 14 page/component files to pass file_token to URL builders
- Add file_token to all relevant TypeScript interfaces
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- scheduler.py: Use full path for scheduler_state.db instead of relative name
- recycle.py: Use full path for thumbnails.db instead of relative name
- cloud_backup.py, maintenance.py, stats.py: Require admin for config/cleanup/settings endpoints
- press.py: Add auth to press image serving endpoint
- private_gallery.py: Fix _create_pg_job call and add missing secrets import
- appearances.py: Use sync httpx instead of asyncio.run for background thread HTTP call
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
