Changelog

All notable changes to this project will be documented in this file.

v1.0.0 — First stable release

Release date: 2026-02-19

First stable release of XFlix — a self-hosted media browser for local video and photo collections with a Netflix-like UI, admin panel, and CLI tools.

Performer-based browsing (one folder = one performer).
Video streaming with HTTP Range requests (seek, pause/resume).
Photo lightbox with full-screen view and keyboard navigation.
Asynchronous, incremental file scanner (non-blocking, batch DB inserts).
Automatic thumbnail generation (photos via sharp, videos via ffmpeg).
Video metadata enrichment (duration extraction via ffprobe).
Public REST API for performers, media, search, stats and discovery.
JWT authentication: register / login / profile / password reset flows.
Per-media comments and like/dislike reactions.
Personal favourites (per-user) and global favourites (admin-set).
Admin panel: scan (SSE), batch thumbnail generation (SSE), duplicate detection, cleanup tools, purge short videos, user management, SMTP settings.
CLI utilities: node cli.js scan, node cli.js clear.
install.sh for one-shot provisioning (nvm/Node, MariaDB, ffmpeg, PM2).
Comprehensive README.md and improved .gitignore + .gitkeep to track data/thumbs directory.

Database access optimized: batch inserts, appropriate indexes for fast sorting/filters.
random_cover_id stored per performer to avoid ORDER BY RAND() at runtime.
Selective gzip compression (skip media & SSE) to avoid double compression.
Frontend improvements: global IntersectionObserver for thumbnails, client-side retry/backoff for thumbnail requests.

Thumbnail generation flood fixed: server-side semaphore limits concurrent ffmpeg/sharp jobs (max 3), with dedup and fast-fail 503 when saturated to keep browser connections free.
Replaced an aggressive inline onerror strategy that opened too many video streams and caused "Failed to fetch" errors.
Various admin & scanner fixes to handle deep directories and large collections reliably.

Scanner designed to handle very large libraries (50k+ files) without blocking the event loop.
Background enrichment/generation runs with bounded concurrency to avoid resource exhaustion.

Password hashing via bcrypt (bcryptjs used in code for portability).
Secrets & credentials live in .env — change JWT_SECRET and DB passwords before exposing publicly.
Recommended to run behind a reverse proxy (nginx/Caddy) with HTTPS.

Requirements: Node >= 18, MariaDB (or MySQL compatible), ffmpeg, build-essential for sharp native build.
After cloning: copy .env.example → .env, set at least MEDIA_DIR, DB_PASS, JWT_SECRET.
Run bash install.sh for automatic provisioning, or follow manual install steps in README.md.
Run an initial scan: node cli.js scan or trigger from Admin → Scan.