ReadMeABook/documentation/phase3/qbittorrent.md

qBittorrent Integration

Status: ✅ Implemented

Free, open-source BitTorrent client with comprehensive Web API.

Enterprise Torrent Addition

Challenge: /api/v2/torrents/add returns only "Ok." without torrent hash.

Solution (Professional):

Magnet Links:

1. Extract info_hash from magnet URI (deterministic)
2. Upload via urls parameter
3. Return extracted hash immediately

Torrent Files:

1. Download .torrent file to memory
2. Parse with parse-torrent (bencode decoder)
3. Extract info_hash (SHA-1 of info dict)
4. Upload file content via torrents parameter (multipart/form-data)
5. Return extracted hash immediately

Benefits: Deterministic, no race conditions, works with Docker networking, handles expired URLs

API Endpoints

Base: http://qbittorrent:8080/api/v2 Auth: Cookie-based (login required)

POST /auth/login - Get session cookie POST /torrents/add - Add torrent (supports urls and torrents params, savepath override) GET /torrents/info?hashes={hash} - Get status/progress POST /torrents/pause - Pause torrent POST /torrents/resume - Resume POST /torrents/delete - Delete torrent GET /torrents/files - Get file list POST /torrents/createCategory - Create category with save path POST /torrents/editCategory - Update category save path POST /torrents/setCategory - Set category for torrent

Config

Required (database only, no env fallbacks):

• download_client_url - qBittorrent Web UI URL
• download_client_username - qBittorrent username
• download_client_password - qBittorrent password
• download_dir - Download save path (passed to qBittorrent for all torrents)

Validation: All fields checked before service initialization.

Category Management

Category: readmeabook (auto-created for all torrents)

Save Path Synchronization:

• Category created/updated on every torrent addition
• Category save path always synced with download_dir config
• Handles config changes: if user changes download_dir, category updates automatically
• Uses both createCategory and editCategory APIs for reliability

Why Both Create and Edit:

1. Create: Ensures category exists (idempotent, won't fail if exists)
2. Edit: Updates save path to match current config (handles user changing settings)

This prevents issues where category retains old save path after user changes download_dir setting.

Data Models

interface TorrentInfo {
  hash: string;
  name: string;
  size: number;
  progress: number; // 0.0-1.0
  dlspeed: number; // bytes/s
  upspeed: number;
  eta: number; // seconds
  state: TorrentState;
  category: string;
  savePath: string;
  completionDate: number;
}

type TorrentState = 'downloading' | 'uploading' | 'stalledDL' |
  'pausedDL' | 'queuedDL' | 'checkingDL' | 'error' | 'missingFiles';

Fixed Issues ✅

1. Naive torrent identification - Fixed with deterministic hash extraction 2. Docker networking issues - Fixed by downloading .torrent ourselves 3. Duplicate detection - Check if hash exists before adding 4. Config fallbacks to env - Removed, database only 5. Unclear error messages - List missing fields explicitly 6. Race condition on torrent availability - Fixed with 3s initial delay + exponential backoff retry (500ms, 1s, 2s) 7. Error logging during duplicate check - Removed console.error in getTorrent() during expected "not found" cases (duplicate checking) 8. Prowlarr magnet link redirects - Some indexers return HTTP URLs that redirect to magnet: links. Fixed by intercepting 3xx redirects before axios follows them, extracting the Location header, and routing to magnet flow if target is a magnet: link 9. Category save path not updating - When user changes download_dir setting, category keeps old path. Fixed by calling both createCategory and editCategory on every torrent addition to ensure category path matches current config

Tech Stack

• axios (HTTP + cookie mgmt)
• parse-torrent (bencode + hash extraction)
• form-data (multipart uploads)

Related

• See File Organization for seeding support