more conditions in automations

remove actions in peer
offline libs
2026-05-06 21:58:49 +02:00 · 2026-05-06 13:57:26 +02:00 · 2026-05-06 11:25:41 +02:00 · 2026-05-06 11:06:08 +02:00 · 2026-05-06 08:50:40 +02:00 · 2026-05-06 08:47:20 +02:00
34 changed files with 4042 additions and 719 deletions
--- a/.env.example
+++ b/.env.example
@@ -6,6 +6,19 @@ PYTORRENT_DEBUG=0
 PYTORRENT_POLL_INTERVAL=1.0
 PYTORRENT_WORKERS=16
 PYTORRENT_GEOIP_DB=data/GeoLite2-City.mmdb
+PYTORRENT_ALLOW_UNSAFE_WERKZEUG=0
+PYTORRENT_SCGI_RETRIES=8
+
+# css/js libs
+PYTORRENT_USE_OFFLINE_LIBS=false
+
+# python -m pytorrent.cli reset-password admin new_Pass
+PYTORRENT_AUTH_ENABLE=false
+
+# Reverse proxy / HTTPS
+PYTORRENT_PROXY_FIX_ENABLE=false
+PYTORRENT_SESSION_COOKIE_SECURE=false
+# PYTORRENT_SOCKETIO_CORS_ALLOWED_ORIGINS=https://your-domain.com

 # Retention / Smart Queue
 PYTORRENT_TRAFFIC_HISTORY_RETENTION_DAYS=90
@@ -13,4 +26,3 @@ PYTORRENT_JOBS_RETENTION_DAYS=30
 PYTORRENT_SMART_QUEUE_HISTORY_RETENTION_DAYS=30
 PYTORRENT_LOG_RETENTION_DAYS=30
 PYTORRENT_SMART_QUEUE_LABEL="Smart Queue Paused"
-
--- a/.gitignore
+++ b/.gitignore
@@ -37,3 +37,4 @@ data/*
 logs/*

 todo.txt
+pytorrent/static/libs/*
--- a/README.md
+++ b/README.md
@@ -1,33 +1,33 @@
 # pyTorrent

-Monopage web UI dla rTorrent inspirowany workflow ruTorrent.
+Single-page web UI for rTorrent inspired by the ruTorrent workflow.

-## Funkcje
+## Features

 - Flask + Flask-SocketIO.
- SQLite na preferencje, profile SCGI, motyw Bootstrapa i font UI.
- Dowolna liczba profili rTorrent per user.
- Profile można dodawać i edytować z UI; flaga zdalnej lokalizacji ukrywa CPU/RAM hosta aplikacji, żeby nie mylić ich z zasobami zdalnego rTorrenta; publiczny IP dla port check jest dalej sprawdzany zdalnie, jeśli rTorrent to obsługuje.
- Przełączanie aktywnego rTorrent z UI.
- Live lista torrentów przez WebSocket.
- Cache aplikacyjny i wysyłanie patchy bez przeładowywania całej tabeli.
- Operacje usera wykonywane w ThreadPoolExecutor.
- Akcje `move` i `remove` są wykonywane per profil w kolejności zlecenia, więc późniejsze usunięcie poczeka na wcześniejsze przenoszenia.
- Log jobsów pokazuje krótką datę i godzinę w tabeli oraz pełny timestamp w tooltipie.
- Masowe start/pause/stop/resume/recheck/remove/move.
- Move obsługuje `move_data=true`, który fizycznie przenosi dane po stronie rTorrent w tle i odpytuje plik statusu, dzięki czemu długie `mv` nie kończy się timeoutem SCGI; jeśli cel już istnieje, jest nadpisywany (`force`), a timeouty z `mkdir`/startu/pollingu move nie przerywają operacji. Potem aktualizuje katalog torrenta, a `recheck` domyślnie włącza się przy fizycznym przenoszeniu.
- Modal dodawania wielu magnetów.
- Dolny status bar: CPU, RAM, wersja rTorrent, prędkości, limity, total DL/UP oraz status portu, gdy port check jest włączony.
- Prawoklik na torrentach.
- Skróty klawiaturowe.
- Szczegóły: General, Files, Peers, Trackers, Log.
- Smart Queue pokazuje domyślnie 10 ostatnich operacji; można rozwinąć historię do 100 wpisów.
- GeoIP peerów z MaxMind GeoLite2-City.mmdb, z cache IP.
- Cache-busting statyków przez MD5 i nagłówki cache.
- Preferencje wyglądu: domyślny Bootstrap albo Bootswatch: Flatly, Litera, Lumen, Minty, Sketchy, Solar, Spacelab, United, Zephyr.
- Preferencje fontu: domyślny font motywu, Adwaita Mono oraz dodatkowe pasujące fonty.
+- SQLite storage for preferences, SCGI profiles, Bootstrap theme and UI font.
+- Multiple rTorrent profiles per user.
+- Profiles can be added and edited from the UI; the remote profile flag hides local CPU/RAM usage to avoid confusing it with remote rTorrent host resources.
+- Active rTorrent profile switching from the UI.
+- Live torrent list over WebSocket.
+- Application-side cache with patch updates instead of full table reloads.
+- User operations executed through ThreadPoolExecutor.
+- `move` and `remove` actions are executed per profile in request order, so later deletes wait for earlier moves.
+- Job log shows a short date/time in the table and the full timestamp in the tooltip.
+- Bulk start, pause, stop, resume, recheck, remove and move.
+- Move supports `move_data=true`; data is physically moved on the rTorrent side in the background and status is polled from a marker file, so long `mv` operations do not hit the SCGI timeout.
+- Multi-magnet add modal.
+- Bottom status bar with CPU, RAM, rTorrent version, speeds, limits, total DL/UP and port-check status when enabled.
+- Torrent context menu.
+- Keyboard shortcuts.
+- Details tabs: General, Files, Peers, Trackers and Log.
+- Smart Queue shows the last 10 operations by default and can expand history to 100 rows.
+- Peer GeoIP with MaxMind GeoLite2-City.mmdb and IP cache.
+- Static cache busting with MD5 and cache headers.
+- Appearance preferences: default Bootstrap or Bootswatch themes Flatly, Litera, Lumen, Minty, Sketchy, Solar, Spacelab, United and Zephyr.
+- Font preferences: default theme font, Adwaita Mono and additional matching fonts.

-## Uruchomienie
+## Run locally

 ```bash
 ./install.sh
@@ -35,17 +35,54 @@ Monopage web UI dla rTorrent inspirowany workflow ruTorrent.
 python app.py
 ```

-Domyślnie: `http://127.0.0.1:8090`.
+Default URL: `http://127.0.0.1:8090`.

-## Profil SCGI
+## Production run

-Przykład:
+Preferred mode without development Werkzeug:
+
+```bash
+. venv/bin/activate
+gunicorn --worker-class gthread --workers 1 --threads 32 --bind 0.0.0.0:8090 --access-logfile - --error-logfile - wsgi:app
+```
+
+Note: the app keeps `async_mode="threading"`, so WebSocket, `start_background_task`, operation queues and the poller run in the same model as before.
+
+Alternatives reviewed but not enabled by default:
+
+- Gunicorn with `eventlet`: works with Flask-SocketIO, but requires green threads and monkey patching, which increases regression risk for file and SCGI operations.
+- Gunicorn with `gevent`: a valid production option, but it needs extra dependencies and compatibility testing.
+- Multiple Gunicorn workers: requires Redis, RabbitMQ or Kafka as the Socket.IO message queue, so it is not a drop-in replacement.
+
+## Reverse proxy
+
+When pyTorrent is served behind a reverse proxy, enable proxy header handling only when the proxy is trusted:
+
+```env
+PYTORRENT_PROXY_FIX_ENABLE=true
+PYTORRENT_SESSION_COOKIE_SECURE=true
+```
+
+The proxy should forward at least:
+
+```txt
+X-Forwarded-For
+X-Forwarded-Proto
+X-Forwarded-Host
+X-Forwarded-Port
+```
+
+This keeps login redirects, session cookies and same-origin API checks correct when HTTPS is terminated by the proxy. If pyTorrent is mounted under a sub-path, also forward `X-Forwarded-Prefix`.
+
+## SCGI profile
+
+Example:

 ```txt
 scgi://127.0.0.1:5000/RPC2
 ```

-Po stronie rTorrent:
+On the rTorrent side:

 ```txt
 network.scgi.open_port = 127.0.0.1:5000
@@ -53,22 +90,39 @@ network.scgi.open_port = 127.0.0.1:5000

 ## GeoIP

-Instalator pobiera bazę GeoLite2-City jednorazowo do:
+The installer downloads GeoLite2-City once to:

 ```txt
 data/GeoLite2-City.mmdb
 ```

-Można też uruchomić ręcznie:
+Manual download:

 ```bash
 ./scripts/download_geoip.sh
 ```

-Skrypt używa głównego źródła `https://git.io/GeoLite2-City.mmdb`, a przy błędzie fallbacku `https://github.com/P3TERX/GeoLite.mmdb/raw/download/GeoLite2-City.mmdb`. Katalog `data` ma uprawnienia `755`, a plik bazy `644`.
+The script uses `https://git.io/GeoLite2-City.mmdb` as the primary source and `https://github.com/P3TERX/GeoLite.mmdb/raw/download/GeoLite2-City.mmdb` as fallback. The `data` directory is set to `755`, and the database file is set to `644`.

 ## API docs

-Dokumentacja OpenAPI jest dostępna pod `/docs`. Endpoint `/api/profiles` obsługuje `max_parallel_jobs` z domyślną wartością `5` oraz `is_remote`; `PUT /api/profiles/{profile_id}` edytuje istniejący profil. Endpoint `/api/preferences` obsługuje m.in. `theme`, `bootstrap_theme`, `font_family`, `table_columns_json`, `peers_refresh_seconds` i `port_check_enabled`. Endpoint `/api/port-check` zwraca status portu wraz z `checked_at`; dla zdalnego profilu publiczny IP jest pobierany przez rTorrent z fallbackami `ifconfig.co`, `ifconfig.me` i `ipapi.linuxiarz.pl`, jeśli dana konfiguracja rTorrenta wspiera zdalne polecenia, a metoda `POST` wymusza ponowny check z pominięciem cache. Endpoint `/api/system/status` dla zdalnego profilu zwraca `usage_available=false` i nie odczytuje CPU/RAM.
+OpenAPI documentation is available at `/docs`. `/api/profiles` supports `max_parallel_jobs` with default value `5` and `is_remote`; `PUT /api/profiles/{profile_id}` edits an existing profile. `/api/preferences` supports fields including `theme`, `bootstrap_theme`, `font_family`, `table_columns_json`, `peers_refresh_seconds` and `port_check_enabled`. `/api/port-check` returns port status with `checked_at`; for remote profiles the public IP is read through rTorrent with fallbacks when supported. `/api/system/status` returns `usage_available=false` for remote profiles and does not read local CPU/RAM.

-`/api/openapi.json` zawiera reusable schemas dla głównych odpowiedzi API, w tym `TorrentListResponse`, `TorrentSummary`, `TorrentFilterSummary`, `CleanupSummary` i `AppStatus`. `GET /api/torrents` dokumentuje teraz pole `summary` używane przez sidebar filters.
+`/api/openapi.json` includes reusable schemas for main API responses, including `TorrentListResponse`, `TorrentSummary`, `TorrentFilterSummary`, `CleanupSummary` and `AppStatus`. `GET /api/torrents` documents the `summary` field used by sidebar filters.
+
+## Admin CLI
+
+Reset an existing user's password:
+
+```bash
+. venv/bin/activate
+python -m pytorrent.cli reset-password admin new_password
+```
+
+Without the password argument, the CLI asks for it interactively:
+
+```bash
+python -m pytorrent.cli reset-password admin
+```
+
+The command uses the same database as the app and respects `PYTORRENT_DB_PATH` from `.env`. The reset changes only the password hash and leaves role and permissions unchanged.
--- a/app.py
+++ b/app.py
@@ -1,7 +1,14 @@
 from pytorrent import create_app, socketio
-from pytorrent.config import HOST, PORT, DEBUG
+from pytorrent.config import ALLOW_UNSAFE_WERKZEUG, DEBUG, HOST, PORT

 app = create_app()

 if __name__ == "__main__":
-    socketio.run(app, host=HOST, port=PORT, debug=DEBUG, allow_unsafe_werkzeug=True)
+    # Note: This entrypoint is kept for local development; production should use gunicorn via wsgi:app.
+    socketio.run(
+        app,
+        host=HOST,
+        port=PORT,
+        debug=DEBUG,
+        allow_unsafe_werkzeug=ALLOW_UNSAFE_WERKZEUG,
+    )
--- a/deploy/pytorrent.service
+++ b/deploy/pytorrent.service
@@ -8,20 +8,16 @@ Wants=network-online.target

 [Service]
 Type=simple
-#User=root
-#Group=root
 User=pytorrent
 Group=pytorrent
 WorkingDirectory=/opt/pyTorrent
 Environment="PYTHONUNBUFFERED=1"
 EnvironmentFile=/opt/pyTorrent/.env
-ExecStart=/opt/pyTorrent/venv/bin/python /opt/pyTorrent/app.py
+ExecStart=/opt/pyTorrent/venv/bin/gunicorn -c /opt/pyTorrent/gunicorn.conf.py --worker-class gthread --workers 1 --threads 32 --bind ${PYTORRENT_HOST}:${PYTORRENT_PORT} --access-logfile - --error-logfile - wsgi:app
 Restart=always
 RestartSec=3
 KillSignal=SIGINT
 TimeoutStopSec=20
-
-# opcjonalnie
 NoNewPrivileges=true
 PrivateTmp=true

--- a/gunicorn.conf.py
+++ b/gunicorn.conf.py
@@ -0,0 +1,3 @@
+import gunicorn.http.wsgi
+
+gunicorn.http.wsgi.SERVER = "pyTorrent"
--- a/install.sh
+++ b/install.sh
@@ -5,6 +5,8 @@ python3 -m venv venv
 pip install --upgrade pip
 pip install -r requirements.txt
 cp -n .env.example .env || true
+grep -q '^PYTORRENT_USE_OFFLINE_LIBS=' .env || echo 'PYTORRENT_USE_OFFLINE_LIBS=true' >> .env
+./scripts/download_frontend_libs.py
 mkdir -p data
 chmod 755 data
 ./scripts/download_geoip.sh data/GeoLite2-City.mmdb
--- a/make_zip.py
+++ b/make_zip.py
@@ -46,7 +46,7 @@ def make_zip(repo_path: Path, output_zip: Path) -> None:
            zf.write(abs_path, arcname=rel_path)

    print(f"Utworzono archiwum: {output_zip}")
-    print(f"Dodano plików: {len(files)}")
+    print(f"Added files: {len(files)}")


 def main():
@@ -60,7 +60,7 @@ def main():
    try:
        run_git_command(["rev-parse", "--show-toplevel"], repo_path)
    except subprocess.CalledProcessError:
-        print("Błąd: ten katalog nie jest repozytorium Git.", file=sys.stderr)
+        print("Error: this directory is not a Git repository.", file=sys.stderr)
        sys.exit(1)

    make_zip(repo_path, output_zip)
--- a/pytorrent/init.py
+++ b/pytorrent/init.py
@@ -1,19 +1,79 @@
 from __future__ import annotations

 from pathlib import Path
-from flask import Flask, request, url_for
+from flask import Flask, jsonify, render_template, request, url_for
 from flask_socketio import SocketIO
-from .config import SECRET_KEY
+from werkzeug.middleware.proxy_fix import ProxyFix
+from .config import (
+    SECRET_KEY,
+    SESSION_COOKIE_SECURE,
+    PROXY_FIX_ENABLE,
+    PROXY_FIX_X_FOR,
+    PROXY_FIX_X_PROTO,
+    PROXY_FIX_X_HOST,
+    PROXY_FIX_X_PORT,
+    PROXY_FIX_X_PREFIX,
+    SOCKETIO_CORS_ALLOWED_ORIGINS,
+)
 from .db import init_db
+from .services.frontend_assets import asset_path, bootstrap_css_path, validate_offline_assets
 from .utils import file_md5

-socketio = SocketIO(cors_allowed_origins="*", ping_timeout=30, async_mode="threading")
+socketio = SocketIO(cors_allowed_origins=SOCKETIO_CORS_ALLOWED_ORIGINS, ping_timeout=30, async_mode="threading")
 _static_md5_cache: dict[tuple, str] = {}


+def _wants_json_response() -> bool:
+    """Return true for API/error clients that should not receive an HTML page."""
+    best = request.accept_mimetypes.best_match(["application/json", "text/html"])
+    return request.path.startswith("/api/") or best == "application/json"
+
+
+def register_error_pages(app: Flask) -> None:
+    # Notatka: własne strony błędów zastępują generyczne 404/500 i zachowują JSON dla API.
+    @app.errorhandler(404)
+    def not_found(error):
+        if _wants_json_response():
+            return jsonify({"ok": False, "error": "Not found"}), 404
+        return render_template(
+            "error.html",
+            code=404,
+            title="Page not found",
+            message="The requested pyTorrent view does not exist or is not available.",
+            icon="fa-compass-drafting",
+        ), 404
+
+    @app.errorhandler(500)
+    def server_error(error):
+        if _wants_json_response():
+            return jsonify({"ok": False, "error": "Internal server error"}), 500
+        return render_template(
+            "error.html",
+            code=500,
+            title="Application error",
+            message="pyTorrent hit an internal error while handling this request.",
+            icon="fa-bug",
+        ), 500
+
+
 def create_app() -> Flask:
+    validate_offline_assets()
    app = Flask(__name__)
+    if PROXY_FIX_ENABLE:
+        app.wsgi_app = ProxyFix(
+            app.wsgi_app,
+            x_for=PROXY_FIX_X_FOR,
+            x_proto=PROXY_FIX_X_PROTO,
+            x_host=PROXY_FIX_X_HOST,
+            x_port=PROXY_FIX_X_PORT,
+            x_prefix=PROXY_FIX_X_PREFIX,
+        )
    app.secret_key = SECRET_KEY
+    app.config.update(
+        SESSION_COOKIE_HTTPONLY=True,
+        SESSION_COOKIE_SAMESITE="Lax",
+        SESSION_COOKIE_SECURE=SESSION_COOKIE_SECURE,
+    )

    @app.context_processor
    def static_helpers():
@@ -30,7 +90,21 @@ def create_app() -> Flask:
                return url_for("static", filename=filename, v=version)
            except OSError:
                return url_for("static", filename=filename)
-        return {"static_url": static_url}
+
+        def frontend_asset_url(key: str) -> str:
+            # Notatka: helper przełącza szablony między CDN i lokalnymi plikami bez duplikowania logiki.
+            path = asset_path(key)
+            return path if path.startswith("http") else static_url(path)
+
+        def bootstrap_theme_url(theme: str | None = None) -> str:
+            path = bootstrap_css_path(theme)
+            return path if path.startswith("http") else static_url(path)
+
+        return {
+            "static_url": static_url,
+            "frontend_asset_url": frontend_asset_url,
+            "bootstrap_theme_url": bootstrap_theme_url,
+        }

    @app.after_request
    def cache_headers(response):
@@ -39,16 +113,17 @@ def create_app() -> Flask:
        if request.endpoint == "static":
            response.headers["Cache-Control"] = "public, max-age=31536000, immutable"
        else:
-            response.headers["Cache-Control"] = "no-cache, no-store, must-revalidate"
-            response.headers["Pragma"] = "no-cache"
-            response.headers["Expires"] = "0"
+            response.headers["Cache-Control"] = "no-store, private"
        return response

    from .routes.main import bp as main_bp
    from .routes.api import bp as api_bp
    app.register_blueprint(main_bp)
    app.register_blueprint(api_bp)
+    register_error_pages(app)
    init_db()
+    from .services.auth import install_guards
+    install_guards(app)

    socketio.init_app(app)
    from .services.workers import set_socketio
--- a/pytorrent/cli.py
+++ b/pytorrent/cli.py
@@ -0,0 +1,79 @@
+from __future__ import annotations
+
+import argparse
+import getpass
+import sys
+
+from .db import connect, init_db, utcnow
+from .services.auth import password_hash
+
+
+def reset_password(username: str, password: str) -> bool:
+    """Note: Reset the selected user password hash without changing role or permissions."""
+    username = (username or "").strip()
+    if not username:
+        raise ValueError("Username is required")
+    if password is None or password == "":
+        raise ValueError("Password cannot be empty")
+
+    init_db()
+    now = utcnow()
+    hashed = password_hash(password)
+    with connect() as conn:
+        row = conn.execute("SELECT id FROM users WHERE username=?", (username,)).fetchone()
+        if not row:
+            return False
+        conn.execute(
+            "UPDATE users SET password_hash=?, updated_at=? WHERE username=?",
+            (hashed, now, username),
+        )
+    return True
+
+
+def _password_from_args(args: argparse.Namespace) -> str:
+    """Note: Allow the password to be passed as an argument or entered securely in interactive mode."""
+    if args.password is not None:
+        return args.password
+    first = getpass.getpass("New password: ")
+    second = getpass.getpass("Repeat password: ")
+    if first != second:
+        raise ValueError("Passwords do not match")
+    return first
+
+
+def build_parser() -> argparse.ArgumentParser:
+    """Note: Define simple administrative commands launched with python -m pytorrent.cli."""
+    parser = argparse.ArgumentParser(description="pyTorrent CLI")
+    sub = parser.add_subparsers(dest="command", required=True)
+
+    reset = sub.add_parser("reset-password", help="Reset password for an existing user")
+    reset.add_argument("username", help="User login")
+    reset.add_argument("password", nargs="?", help="New password; omit to type it interactively")
+    reset.set_defaults(func=_cmd_reset_password)
+
+    return parser
+
+
+def _cmd_reset_password(args: argparse.Namespace) -> int:
+    """Note: Run the password reset and return a readable terminal status."""
+    password = _password_from_args(args)
+    if reset_password(args.username, password):
+        print(f"Password reset for user: {args.username}")
+        return 0
+    print(f"User not found: {args.username}", file=sys.stderr)
+    return 1
+
+
+def main(argv: list[str] | None = None) -> int:
+    """Note: Main CLI entrypoint with error handling and without starting the web app."""
+    parser = build_parser()
+    args = parser.parse_args(argv)
+    try:
+        return int(args.func(args) or 0)
+    except Exception as exc:
+        print(f"Error: {exc}", file=sys.stderr)
+        return 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
--- a/pytorrent/config.py
+++ b/pytorrent/config.py
@@ -1,20 +1,46 @@
 from __future__ import annotations

 import os
+import secrets
 from pathlib import Path
 from dotenv import load_dotenv

 BASE_DIR = Path(__file__).resolve().parent.parent
 load_dotenv(BASE_DIR / ".env")

-SECRET_KEY = os.getenv("PYTORRENT_SECRET_KEY", "dev-change-me")
+
+def _env_bool(name: str, default: bool = False) -> bool:
+    value = os.getenv(name)
+    if value is None:
+        return default
+    return value.strip().lower() in {"1", "true", "yes", "on"}
+
+
+_SECRET_KEY_ENV = os.getenv("PYTORRENT_SECRET_KEY")
+SECRET_KEY = _SECRET_KEY_ENV or "dev-change-me"
 DB_PATH = Path(os.getenv("PYTORRENT_DB_PATH", str(BASE_DIR / "data" / "pytorrent.sqlite3")))
 if not DB_PATH.is_absolute():
    DB_PATH = BASE_DIR / DB_PATH

 HOST = os.getenv("PYTORRENT_HOST", "0.0.0.0")
 PORT = int(os.getenv("PYTORRENT_PORT", "8090"))
-DEBUG = os.getenv("PYTORRENT_DEBUG", "0") == "1"
+DEBUG = _env_bool("PYTORRENT_DEBUG", False)
+# Notatka: tryb offline wymusza lokalne JS/CSS i wyłącza zależność od CDN.
+USE_OFFLINE_LIBS = _env_bool("PYTORRENT_USE_OFFLINE_LIBS", False)
+# Note: Optional authentication remains disabled unless explicitly enabled in .env.
+AUTH_ENABLE = _env_bool("PYTORRENT_AUTH_ENABLE", False)
+if AUTH_ENABLE and (not _SECRET_KEY_ENV or SECRET_KEY == "dev-change-me"):
+    # Note: Auth mode cannot use Flask's development secret; persist a local random session key instead.
+    _secret_file = BASE_DIR / "data" / ".session_secret"
+    _secret_file.parent.mkdir(parents=True, exist_ok=True)
+    if _secret_file.exists():
+        SECRET_KEY = _secret_file.read_text(encoding="utf-8").strip()
+    if not SECRET_KEY or SECRET_KEY == "dev-change-me":
+        SECRET_KEY = secrets.token_urlsafe(48)
+        _secret_file.write_text(SECRET_KEY, encoding="utf-8")
+SESSION_COOKIE_SECURE = _env_bool("PYTORRENT_SESSION_COOKIE_SECURE", False)
+# Note: Keep Werkzeug opt-in only for explicit local/dev use, never by default in services.
+ALLOW_UNSAFE_WERKZEUG = _env_bool("PYTORRENT_ALLOW_UNSAFE_WERKZEUG", DEBUG)
 POLL_INTERVAL = float(os.getenv("PYTORRENT_POLL_INTERVAL", "1.0"))
 WORKERS = int(os.getenv("PYTORRENT_WORKERS", "16"))
 GEOIP_DB = Path(os.getenv("PYTORRENT_GEOIP_DB", str(BASE_DIR / "data" / "GeoLite2-City.mmdb")))
@@ -29,6 +55,17 @@ def _env_int(name: str, default: int, minimum: int = 0) -> int:
        return default


+
+PROXY_FIX_ENABLE = _env_bool("PYTORRENT_PROXY_FIX_ENABLE", False)
+PROXY_FIX_X_FOR = _env_int("PYTORRENT_PROXY_FIX_X_FOR", 1, 0)
+PROXY_FIX_X_PROTO = _env_int("PYTORRENT_PROXY_FIX_X_PROTO", 1, 0)
+PROXY_FIX_X_HOST = _env_int("PYTORRENT_PROXY_FIX_X_HOST", 1, 0)
+PROXY_FIX_X_PORT = _env_int("PYTORRENT_PROXY_FIX_X_PORT", 1, 0)
+PROXY_FIX_X_PREFIX = _env_int("PYTORRENT_PROXY_FIX_X_PREFIX", 1, 0)
+
+_SOCKETIO_CORS = os.getenv("PYTORRENT_SOCKETIO_CORS_ALLOWED_ORIGINS", "").strip()
+SOCKETIO_CORS_ALLOWED_ORIGINS = None if not _SOCKETIO_CORS else [item.strip() for item in _SOCKETIO_CORS.split(",") if item.strip()]
+
 TRAFFIC_HISTORY_RETENTION_DAYS = _env_int("PYTORRENT_TRAFFIC_HISTORY_RETENTION_DAYS", 90, 1)
 JOBS_RETENTION_DAYS = _env_int("PYTORRENT_JOBS_RETENTION_DAYS", 30, 1)
 SMART_QUEUE_HISTORY_RETENTION_DAYS = _env_int("PYTORRENT_SMART_QUEUE_HISTORY_RETENTION_DAYS", 30, 1)
--- a/pytorrent/db.py
+++ b/pytorrent/db.py
@@ -10,7 +10,20 @@ CREATE TABLE IF NOT EXISTS users (
  id INTEGER PRIMARY KEY AUTOINCREMENT,
  username TEXT UNIQUE NOT NULL,
  password_hash TEXT,
-  created_at TEXT NOT NULL
+  role TEXT DEFAULT 'user',
+  is_active INTEGER DEFAULT 1,
+  created_at TEXT NOT NULL,
+  updated_at TEXT
+);
+
+CREATE TABLE IF NOT EXISTS user_profile_permissions (
+  user_id INTEGER NOT NULL,
+  profile_id INTEGER NOT NULL DEFAULT 0,
+  access_level TEXT NOT NULL DEFAULT 'ro',
+  created_at TEXT NOT NULL,
+  updated_at TEXT NOT NULL,
+  PRIMARY KEY(user_id, profile_id),
+  FOREIGN KEY(user_id) REFERENCES users(id)
 );

 CREATE TABLE IF NOT EXISTS user_preferences (
@@ -126,6 +139,7 @@ CREATE TABLE IF NOT EXISTS smart_queue_settings (
  stalled_seconds INTEGER DEFAULT 300,
  min_speed_bytes INTEGER DEFAULT 1024,
  min_seeds INTEGER DEFAULT 1,
+  manage_stopped INTEGER DEFAULT 0,
  updated_at TEXT NOT NULL,
  PRIMARY KEY(user_id, profile_id)
 );
@@ -234,9 +248,20 @@ CREATE TABLE IF NOT EXISTS app_settings (
  key TEXT PRIMARY KEY,
  value TEXT
 );
+
+CREATE TABLE IF NOT EXISTS torrent_stats_cache (
+  profile_id INTEGER PRIMARY KEY,
+  payload_json TEXT NOT NULL,
+  created_at TEXT NOT NULL,
+  updated_at TEXT NOT NULL,
+  updated_epoch REAL DEFAULT 0
+);
 """

 MIGRATIONS = [
+    "ALTER TABLE users ADD COLUMN role TEXT DEFAULT 'user'",
+    "ALTER TABLE users ADD COLUMN is_active INTEGER DEFAULT 1",
+    "ALTER TABLE users ADD COLUMN updated_at TEXT",
    "ALTER TABLE user_preferences ADD COLUMN mobile_mode INTEGER DEFAULT 0",
    "ALTER TABLE user_preferences ADD COLUMN peers_refresh_seconds INTEGER DEFAULT 0",
    "ALTER TABLE user_preferences ADD COLUMN port_check_enabled INTEGER DEFAULT 0",
@@ -253,6 +278,8 @@ MIGRATIONS = [
    "ALTER TABLE automation_rules ADD COLUMN cooldown_minutes INTEGER DEFAULT 60",
    "ALTER TABLE rtorrent_config_overrides ADD COLUMN apply_on_start INTEGER DEFAULT 0",
    "ALTER TABLE rtorrent_config_overrides ADD COLUMN baseline_value TEXT",
+    "ALTER TABLE torrent_stats_cache ADD COLUMN updated_epoch REAL DEFAULT 0",
+    "ALTER TABLE smart_queue_settings ADD COLUMN manage_stopped INTEGER DEFAULT 0",
 ]


@@ -288,15 +315,21 @@ def init_db():
                pass
        now = utcnow()
        conn.execute(
-            "INSERT OR IGNORE INTO users(id, username, password_hash, created_at) VALUES(1, 'default', NULL, ?)",
-            (now,),
+            "INSERT OR IGNORE INTO users(id, username, password_hash, role, is_active, created_at, updated_at) VALUES(1, 'default', NULL, 'admin', 1, ?, ?)",
+            (now, now),
        )
+        conn.execute("UPDATE users SET role=COALESCE(role, 'admin'), is_active=COALESCE(is_active, 1), updated_at=COALESCE(updated_at, ?) WHERE id=1", (now,))
        pref = conn.execute("SELECT id FROM user_preferences WHERE user_id=1").fetchone()
        if not pref:
            conn.execute(
                "INSERT INTO user_preferences(user_id, theme, created_at, updated_at) VALUES(1, 'dark', ?, ?)",
                (now, now),
            )
+    try:
+        from .services.auth import ensure_admin_user
+        ensure_admin_user()
+    except Exception:
+        pass


 def default_user_id() -> int:
--- a/pytorrent/routes/api.py
+++ b/pytorrent/routes/api.py
@@ -13,17 +13,91 @@ import socket
 import json
 import psutil
 import xml.etree.ElementTree as ET
-from flask import Blueprint, jsonify, request
+from flask import Blueprint, jsonify, request, abort
 from ..config import DB_PATH, JOBS_RETENTION_DAYS, SMART_QUEUE_HISTORY_RETENTION_DAYS, WORKERS
-from ..db import default_user_id, connect, utcnow
-from ..services import preferences, rtorrent
+from ..db import connect, utcnow
+from ..services.auth import current_user_id as default_user_id, current_user, list_users, save_user, delete_user, login_user, logout_user, enabled as auth_enabled, require_profile_write
+from ..services import preferences, rtorrent, torrent_stats
 from ..services.torrent_cache import torrent_cache
 from ..services.torrent_summary import cached_summary
-from ..services.workers import enqueue, list_jobs, cancel_job, retry_job, clear_jobs
+from ..services.workers import enqueue, list_jobs, cancel_job, retry_job, clear_jobs, emergency_clear_jobs
 from ..services.geoip import lookup_ip

 bp = Blueprint("api", __name__, url_prefix="/api")

+MOVE_BULK_MAX_HASHES = 100
+
+
+@bp.post("/auth/login")
+def auth_login():
+    # Note: Auth API is hidden when optional authentication is disabled.
+    if not auth_enabled():
+        abort(404)
+    data = request.get_json(silent=True) or {}
+    user = login_user(str(data.get("username") or ""), str(data.get("password") or ""))
+    if not user:
+        return jsonify({"ok": False, "error": "Invalid username or password"}), 401
+    return ok({"user": user, "auth_enabled": auth_enabled()})
+
+
+@bp.get("/auth/me")
+def auth_me():
+    if not auth_enabled():
+        abort(404)
+    return ok({"user": current_user(), "auth_enabled": auth_enabled()})
+
+
+@bp.post("/auth/logout")
+def auth_logout():
+    if not auth_enabled():
+        abort(404)
+    logout_user()
+    return ok()
+
+
+@bp.get("/auth/users")
+def auth_users_list():
+    if not auth_enabled():
+        abort(404)
+    return ok({"users": list_users()})
+
+
+@bp.post("/auth/users")
+def auth_users_create():
+    if not auth_enabled():
+        abort(404)
+    try:
+        return ok({"user": save_user(request.get_json(silent=True) or {})})
+    except Exception as exc:
+        return jsonify({"ok": False, "error": str(exc)}), 400
+
+
+@bp.put("/auth/users/<int:user_id>")
+def auth_users_update(user_id: int):
+    if not auth_enabled():
+        abort(404)
+    try:
+        return ok({"user": save_user(request.get_json(silent=True) or {}, user_id)})
+    except Exception as exc:
+        return jsonify({"ok": False, "error": str(exc)}), 400
+
+
+@bp.delete("/auth/users/<int:user_id>")
+def auth_users_delete(user_id: int):
+    if not auth_enabled():
+        abort(404)
+    try:
+        delete_user(user_id)
+        return ok({"users": list_users()})
+    except Exception as exc:
+        return jsonify({"ok": False, "error": str(exc)}), 400
+
+
+
+def _job_profile_id(job_id: str) -> int | None:
+    with connect() as conn:
+        row = conn.execute("SELECT profile_id FROM jobs WHERE id=?", (job_id,)).fetchone()
+    return int(row.get("profile_id") or 0) if row else None

 def ok(payload=None):
    data = {"ok": True}
@@ -303,6 +377,52 @@ def enrich_bulk_payload(profile: dict, action_name: str, data: dict) -> dict:
    return payload


+def _chunk_hashes(hashes: list[str], size: int = MOVE_BULK_MAX_HASHES) -> list[list[str]]:
+    # Note: Splits very large torrent selections into predictable chunks so each queued job stays small and recoverable.
+    safe_size = max(1, int(size or MOVE_BULK_MAX_HASHES))
+    return [hashes[index:index + safe_size] for index in range(0, len(hashes), safe_size)]
+
+
+def enqueue_bulk_parts(profile: dict, action_name: str, data: dict) -> list[dict]:
+    # Note: One shared helper splits large move/remove operations into small ordered parts without changing other actions.
+    base_payload = enrich_bulk_payload(profile, action_name, data)
+    hashes = base_payload.get("hashes") or []
+    chunks = _chunk_hashes(hashes)
+    if len(chunks) <= 1:
+        job_id = enqueue(action_name, profile["id"], base_payload)
+        return [{"job_id": job_id, "label": "bulk-1", "part": 1, "parts": 1, "hashes": hashes, "hash_count": len(hashes)}]
+
+    jobs = []
+    items_by_hash = {str(item.get("hash")): item for item in (base_payload.get("job_context") or {}).get("items") or []}
+    for index, chunk in enumerate(chunks, start=1):
+        payload = dict(base_payload)
+        payload["hashes"] = chunk
+        context = dict(base_payload.get("job_context") or {})
+        context.update({
+            "bulk": True,
+            "bulk_label": f"bulk-{index}",
+            "bulk_part": index,
+            "bulk_parts": len(chunks),
+            "hash_count": len(chunk),
+            "parent_hash_count": len(hashes),
+            "items": [items_by_hash[h] for h in chunk if h in items_by_hash],
+        })
+        payload["job_context"] = context
+        job_id = enqueue(action_name, profile["id"], payload)
+        jobs.append({"job_id": job_id, "label": context["bulk_label"], "part": index, "parts": len(chunks), "hashes": chunk, "hash_count": len(chunk)})
+    return jobs
+
+
+def enqueue_move_bulk_parts(profile: dict, data: dict) -> list[dict]:
+    # Note: Keep the old public move helper while using the same partitioning logic.
+    return enqueue_bulk_parts(profile, "move", data)
+
+
+def enqueue_remove_bulk_parts(profile: dict, data: dict) -> list[dict]:
+    # Note: Remove/rm uses the same partitioning as move, which lowers rTorrent load.
+    return enqueue_bulk_parts(profile, "remove", data)
+
+
@bp.get("/profiles")
 def profiles_list():
    return ok({"profiles": preferences.list_profiles(), "active": preferences.active_profile()})
@@ -362,6 +482,19 @@ def torrents():
    })


+@bp.get("/torrent-stats")
+def torrent_stats_get():
+    profile = preferences.active_profile()
+    if not profile:
+        return ok({"stats": {}, "error": "No profile"})
+    force = str(request.args.get("force") or "").lower() in {"1", "true", "yes"}
+    try:
+        # Note: Heavy file metadata is served from a 15-minute DB cache unless the user explicitly refreshes it.
+        return ok({"stats": torrent_stats.get(profile, force=force)})
+    except Exception as exc:
+        return jsonify({"ok": False, "error": str(exc)}), 500
+
+
@bp.get("/torrents/<torrent_hash>/files")
 def torrent_files(torrent_hash: str):
    profile = preferences.active_profile()
@@ -395,19 +528,6 @@ def torrent_peers(torrent_hash: str):
    return ok({"peers": peers})


-@bp.post("/torrents/<torrent_hash>/peers/action")
-def torrent_peer_action(torrent_hash: str):
-    profile = preferences.active_profile()
-    if not profile:
-        return jsonify({"ok": False, "error": "No profile"}), 400
-    data = request.get_json(silent=True) or {}
-    try:
-        result = rtorrent.peer_action(profile, torrent_hash, int(data.get("peer_index")), str(data.get("action") or ""))
-        return ok({"result": result, "message": f"Peer {result['action']} via {result['method']}"})
-    except Exception as exc:
-        return jsonify({"ok": False, "error": str(exc)}), 400
-
-
@bp.get("/torrents/<torrent_hash>/trackers")
 def torrent_trackers(torrent_hash: str):
    profile = preferences.active_profile()
@@ -434,9 +554,23 @@ def torrent_action(action_name: str):
    if not profile:
        return jsonify({"ok": False, "error": "No profile"}), 400
    data = request.get_json(silent=True) or {}
-    allowed = {"start", "pause", "stop", "resume", "recheck", "reannounce", "remove", "move", "set_label", "set_ratio_group"}
+    allowed = {"start", "pause", "unpause", "stop", "resume", "recheck", "reannounce", "remove", "move", "set_label", "set_ratio_group"}
    if action_name not in allowed:
        return jsonify({"ok": False, "error": "Unknown action"}), 400
+    if action_name in {"move", "remove"}:
+        # Note: Large move/remove requests are split into ordered bulk parts; smaller requests keep the old single-job response shape.
+        jobs = enqueue_bulk_parts(profile, action_name, data)
+        first_job_id = jobs[0]["job_id"] if jobs else None
+        total_hashes = sum(int(job.get("hash_count") or 0) for job in jobs)
+        return ok({
+            "job_id": first_job_id,
+            "job_ids": [job["job_id"] for job in jobs],
+            "jobs": jobs,
+            "hash_count": total_hashes,
+            "bulk": total_hashes > 1,
+            "bulk_parts": len(jobs),
+            "chunk_size": MOVE_BULK_MAX_HASHES,
+        })
    payload = enrich_bulk_payload(profile, action_name, data)
    job_id = enqueue(action_name, profile["id"], payload)
    return ok({"job_id": job_id, "hash_count": len(payload.get("hashes") or []), "bulk": len(payload.get("hashes") or []) > 1})
@@ -566,8 +700,12 @@ def jobs_list():

@bp.post("/jobs/clear")
 def jobs_clear():
+    if str(request.args.get("force") or "").lower() in {"1", "true", "yes"}:
+        # Note: Emergency cleanup keeps the endpoint behavior unchanged, while force=1 enables rescue mode.
+        deleted = emergency_clear_jobs()
+        return ok({"deleted": deleted, "emergency": True})
    deleted = clear_jobs()
-    return ok({"deleted": deleted})
+    return ok({"deleted": deleted, "emergency": False})


@bp.get("/cleanup/summary")
@@ -608,13 +746,15 @@ def cleanup_all():

@bp.post("/jobs/<job_id>/cancel")
 def jobs_cancel(job_id: str):
+    require_profile_write(_job_profile_id(job_id))
    if not cancel_job(job_id):
-        return jsonify({"ok": False, "error": "Only pending or failed jobs can be cancelled"}), 400
-    return ok()
+        return jsonify({"ok": False, "error": "Only unfinished jobs can be cancelled"}), 400
+    return ok({"emergency": True})


@bp.post("/jobs/<job_id>/retry")
 def jobs_retry(job_id: str):
+    require_profile_write(_job_profile_id(job_id))
    if not retry_job(job_id):
        return jsonify({"ok": False, "error": "Only failed or cancelled jobs can be retried"}), 400
    return ok()
@@ -832,7 +972,11 @@ def smart_queue_check():
    if not profile:
        return ok({'result': {'ok': False, 'error': 'No profile'}})
    try:
-        return ok({'result': smart_queue.check(profile, force=True)})
+        result = smart_queue.check(profile, force=True)
+        # Note: Manual check immediately returns a fresh snapshot so the UI shows the real Downloading count after the action.
+        diff = torrent_cache.refresh(profile)
+        rows = torrent_cache.snapshot(profile['id'])
+        return ok({'result': result, 'torrent_patch': {**diff, 'summary': cached_summary(profile['id'], rows, force=True)}})
    except Exception as exc:
        return jsonify({'ok': False, 'error': str(exc)}), 500

--- a/pytorrent/routes/main.py
+++ b/pytorrent/routes/main.py
@@ -1,11 +1,42 @@
 from __future__ import annotations

-from flask import Blueprint, render_template, jsonify, Response
-from ..services.preferences import get_preferences, list_profiles, active_profile, BOOTSTRAP_THEMES, FONT_FAMILIES, bootstrap_css_url
+from flask import Blueprint, render_template, jsonify, Response, request, redirect, url_for, abort
+from ..services.preferences import get_preferences, list_profiles, active_profile, BOOTSTRAP_THEMES, FONT_FAMILIES
+from ..services import auth
+from ..services.frontend_assets import asset_path

 bp = Blueprint("main", __name__)


+def _asset_url(key: str) -> str:
+    # Notatka: API docs korzysta z tego samego przełącznika CDN/offline co reszta aplikacji.
+    path = asset_path(key)
+    return path if path.startswith("http") else url_for("static", filename=path)
+
+
+
+@bp.route("/login", methods=["GET", "POST"])
+def login():
+    # Note: When optional authentication is disabled, /login is intentionally unavailable.
+    if not auth.enabled():
+        abort(404)
+    error = ""
+    if request.method == "POST":
+        user = auth.login_user(request.form.get("username", ""), request.form.get("password", ""))
+        if user:
+            return redirect(request.args.get("next") or url_for("main.index"))
+        error = "Invalid username or password"
+    return render_template("login.html", error=error)
+
+
+@bp.get("/logout")
+def logout():
+    auth.logout_user()
+    if not auth.enabled():
+        return redirect(url_for("main.index"))
+    return redirect(url_for("main.login"))
+
+
@bp.get("/")
 def index():
    prefs = get_preferences()
@@ -16,13 +47,14 @@ def index():
        active_profile=active_profile(),
        bootstrap_themes=BOOTSTRAP_THEMES,
        font_families=FONT_FAMILIES,
-        bootstrap_css_url=bootstrap_css_url((prefs or {}).get("bootstrap_theme")),
+        auth_enabled=auth.enabled(),
+        current_user=auth.current_user(),
    )


@bp.get("/docs")
 def docs():
-    html = """<!doctype html><html lang=\"en\"><head><meta charset=\"utf-8\"><meta name=\"viewport\" content=\"width=device-width, initial-scale=1\"><title>pyTorrent API Docs</title><link rel=\"stylesheet\" href=\"https://cdn.jsdelivr.net/npm/swagger-ui-dist@5/swagger-ui.css\"></head><body><div id=\"swagger-ui\"></div><script src=\"https://cdn.jsdelivr.net/npm/swagger-ui-dist@5/swagger-ui-bundle.js\"></script><script>window.onload=()=>SwaggerUIBundle({url:'/api/openapi.json',dom_id:'#swagger-ui',deepLinking:true,persistAuthorization:true});</script></body></html>"""
+    html = f"""<!doctype html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1"><title>pyTorrent API Docs</title><link rel="stylesheet" href="{_asset_url('swagger_css')}"></head><body><div id="swagger-ui"></div><script src="{_asset_url('swagger_js')}"></script><script>window.onload=()=>SwaggerUIBundle({{url:'/api/openapi.json',dom_id:'#swagger-ui',deepLinking:true,persistAuthorization:true}});</script></body></html>"""
    return Response(html, mimetype="text/html")


@@ -55,7 +87,7 @@ def openapi():
            },
        },
        "/api/torrents": {"get": {"summary": "Get cached torrent snapshot", "responses": {"200": {"description": "Torrent list"}}}},
-        "/api/torrents/{action_name}": {"post": {"summary": "Queue torrent action", "description": "For move, path is the target directory; move_data=true physically moves data on the rTorrent host using a detached shell move with status polling, force-overwrites an existing destination, tolerates rTorrent execute timeouts around mkdir/start/polling, handles retries after a partially completed move, avoids SCGI timeout on long mv operations, and recheck defaults to move_data. Move and remove jobs are ordered per profile, so a later remove waits for earlier move/remove jobs to finish.", "parameters": [{"name": "action_name", "in": "path", "required": True, "schema": {"type": "string", "enum": ["start", "pause", "stop", "resume", "recheck", "remove", "move", "set_label", "set_ratio_group"]}}], "requestBody": {"content": {"application/json": {"schema": {"type": "object", "properties": {"hashes": {"type": "array", "items": {"type": "string"}}, "path": {"type": "string", "description": "Target directory for move"}, "move_data": {"type": "boolean", "description": "Physically move data before setting torrent directory"}, "recheck": {"type": "boolean", "description": "Run hash check after physical move; defaults to move_data"}, "label": {"type": "string"}, "ratio_group": {"type": "string"}, "remove_data": {"type": "boolean"}}}}}}, "responses": {"200": {"description": "Job queued"}}}},
+        "/api/torrents/{action_name}": {"post": {"summary": "Queue torrent action", "description": "For move, path is the target directory; move_data=true physically moves data on the rTorrent host using a detached shell move with status polling, force-overwrites an existing destination, tolerates rTorrent execute timeouts around mkdir/start/polling, handles retries after a partially completed move, avoids SCGI timeout on long mv operations, and recheck defaults to move_data. Large move selections are split into ordered bulk parts of up to 100 hashes. Move and remove jobs are ordered per profile, so a later remove waits for earlier move/remove jobs to finish.", "parameters": [{"name": "action_name", "in": "path", "required": True, "schema": {"type": "string", "enum": ["start", "pause", "stop", "resume", "recheck", "remove", "move", "set_label", "set_ratio_group"]}}], "requestBody": {"content": {"application/json": {"schema": {"type": "object", "properties": {"hashes": {"type": "array", "items": {"type": "string"}}, "path": {"type": "string", "description": "Target directory for move"}, "move_data": {"type": "boolean", "description": "Physically move data before setting torrent directory"}, "recheck": {"type": "boolean", "description": "Run hash check after physical move; defaults to move_data"}, "label": {"type": "string"}, "ratio_group": {"type": "string"}, "remove_data": {"type": "boolean"}}}}}}, "responses": {"200": {"description": "Job queued"}}}},
        "/api/torrents/add": {"post": {"summary": "Add magnet links or torrent files", "requestBody": {"content": {"multipart/form-data": {"schema": {"type": "object", "properties": {"uris": {"type": "string"}, "directory": {"type": "string"}, "label": {"type": "string"}, "start": {"type": "boolean"}, "files": {"type": "array", "items": {"type": "string", "format": "binary"}}}}}, "application/json": {"schema": {"type": "object"}}}}, "responses": {"200": {"description": "Jobs queued"}}}},
        "/api/torrents/{torrent_hash}/files": {"get": {"summary": "Torrent files", "parameters": [{"name": "torrent_hash", "in": "path", "required": True, "schema": {"type": "string"}}], "responses": {"200": {"description": "Files"}}}},
        "/api/torrents/{torrent_hash}/peers": {"get": {"summary": "Torrent peers with GeoIP", "parameters": [{"name": "torrent_hash", "in": "path", "required": True, "schema": {"type": "string"}}], "responses": {"200": {"description": "Peers"}}}},
@@ -80,10 +112,15 @@ def openapi():
        "/api/traffic/history": {"get": {"summary": "Transfer history for charts", "parameters": [{"name": "range", "in": "query", "schema": {"type": "string", "enum": ["15m", "1h", "3h", "6h", "24h", "7d", "30d", "90d"]}}], "responses": {"200": {"description": "Aggregated traffic history"}}}}
    }
    paths.update({
+        "/api/auth/login": {"post": {"summary": "Log in with username and password when authentication is enabled", "requestBody": {"content": {"application/json": {"schema": {"type": "object", "properties": {"username": {"type": "string"}, "password": {"type": "string", "format": "password"}}, "required": ["username", "password"]}}}}, "responses": {"200": {"description": "Logged in"}, "401": {"description": "Invalid credentials"}, "404": {"description": "Authentication disabled"}}}},
+        "/api/auth/me": {"get": {"summary": "Return current authenticated user", "responses": {"200": {"description": "Current user"}, "404": {"description": "Authentication disabled"}}}},
+        "/api/auth/logout": {"post": {"summary": "Log out current user", "responses": {"200": {"description": "Logged out"}, "404": {"description": "Authentication disabled"}}}},
+        "/api/auth/users": {"get": {"summary": "List users, admin only", "responses": {"200": {"description": "Users"}, "403": {"description": "Admin only"}}}, "post": {"summary": "Create user, admin only", "requestBody": {"content": {"application/json": {"schema": {"$ref": "#/components/schemas/AuthUserInput"}}}}, "responses": {"200": {"description": "User created"}, "403": {"description": "Admin only"}}}},
+        "/api/auth/users/{user_id}": {"put": {"summary": "Update user, admin only", "parameters": [{"name": "user_id", "in": "path", "required": True, "schema": {"type": "integer"}}], "requestBody": {"content": {"application/json": {"schema": {"$ref": "#/components/schemas/AuthUserInput"}}}}, "responses": {"200": {"description": "User updated"}, "403": {"description": "Admin only"}}}, "delete": {"summary": "Delete user, admin only", "parameters": [{"name": "user_id", "in": "path", "required": True, "schema": {"type": "integer"}}], "responses": {"200": {"description": "User deleted"}, "403": {"description": "Admin only"}}}},
        "/api/profiles/{profile_id}": {"delete": {"summary": "Delete rTorrent profile", "parameters": [{"name": "profile_id", "in": "path", "required": True, "schema": {"type": "integer"}}], "responses": {"200": {"description": "Deleted"}}}},
+        "/api/torrent-stats": {"get": {"summary": "Torrent statistics and cached file metadata", "parameters": [{"name": "force", "in": "query", "schema": {"type": "boolean", "default": False}}], "responses": {"200": {"description": "Torrent statistics"}}}},
        "/api/path/default": {"get": {"summary": "Read active rTorrent default download path", "responses": {"200": {"description": "Default path"}}}},
        "/api/torrents/{torrent_hash}/files/priority": {"post": {"summary": "Set file priorities", "parameters": [{"name": "torrent_hash", "in": "path", "required": True, "schema": {"type": "string"}}], "requestBody": {"content": {"application/json": {"schema": {"type": "object", "properties": {"files": {"type": "array", "items": {"type": "object", "properties": {"index": {"type": "integer"}, "priority": {"type": "integer", "enum": [0, 1, 2]}}}}}}}}}, "responses": {"200": {"description": "Updated priorities"}, "207": {"description": "Partial update"}}}},
-        "/api/torrents/{torrent_hash}/peers/action": {"post": {"summary": "Run peer action", "parameters": [{"name": "torrent_hash", "in": "path", "required": True, "schema": {"type": "string"}}], "requestBody": {"content": {"application/json": {"schema": {"type": "object", "properties": {"peer_index": {"type": "integer"}, "action": {"type": "string", "enum": ["disconnect", "kick", "snub", "unsnub", "ban"]}}}}}}, "responses": {"200": {"description": "Peer action result"}}}},
        "/api/labels/{label_id}": {"delete": {"summary": "Delete saved label", "parameters": [{"name": "label_id", "in": "path", "required": True, "schema": {"type": "integer"}}], "responses": {"200": {"description": "Labels"}}}},
        "/api/rtorrent-config": {"get": {"summary": "Read supported rTorrent config fields", "responses": {"200": {"description": "Config fields"}}}, "post": {"summary": "Save supported rTorrent config fields", "requestBody": {"content": {"application/json": {"schema": {"type": "object", "properties": {"values": {"type": "object"}}}}}}, "responses": {"200": {"description": "Save result"}}}},
        "/api/rtorrent-config/generate": {"post": {"summary": "Generate rTorrent config text from provided values", "requestBody": {"content": {"application/json": {"schema": {"type": "object", "properties": {"values": {"type": "object"}}}}}}, "responses": {"200": {"description": "Generated config text"}}}},
@@ -98,6 +135,17 @@ def openapi():
                "properties": {"ok": {"type": "boolean"}},
                "required": ["ok"],
            },
+            "AuthUserInput": {
+                "type": "object",
+                "properties": {
+                    "username": {"type": "string"},
+                    "password": {"type": "string", "format": "password", "description": "Optional on update"},
+                    "role": {"type": "string", "enum": ["admin", "user"]},
+                    "is_active": {"type": "boolean"},
+                    "permissions": {"type": "array", "items": {"type": "object", "properties": {"profile_id": {"type": "integer", "description": "0 means all profiles"}, "access_level": {"type": "string", "enum": ["ro", "full"]}}}},
+                },
+                "required": ["username"],
+            },
            "Profile": {
                "type": "object",
                "additionalProperties": True,
@@ -278,4 +326,9 @@ def openapi():
        },
    })

-    return jsonify({"openapi": "3.0.3", "info": {"title": "pyTorrent API", "version": "0.2.0"}, "paths": paths, "components": components})
+    components.setdefault("securitySchemes", {})["sessionCookie"] = {"type": "apiKey", "in": "cookie", "name": "session"}
+    for path, methods in paths.items():
+        if path != "/api/auth/login":
+            for operation in methods.values():
+                operation.setdefault("security", [{"sessionCookie": []}])
+    return jsonify({"openapi": "3.0.3", "info": {"title": "pyTorrent API", "version": "0.0.1"}, "paths": paths, "components": components})
--- a/pytorrent/services/auth.py
+++ b/pytorrent/services/auth.py
@@ -0,0 +1,344 @@
+from __future__ import annotations
+
+from functools import wraps
+from typing import Any
+from urllib.parse import urlparse
+
+from flask import abort, jsonify, redirect, request, session, url_for
+from werkzeug.security import check_password_hash, generate_password_hash
+
+from ..config import AUTH_ENABLE
+from ..db import connect, default_user_id, utcnow
+
+PUBLIC_ENDPOINTS = {"main.login", "main.logout", "api.auth_login", "api.auth_me", "static"}
+RTORRENT_WRITE_PREFIXES = (
+    "/api/torrents/",
+    "/api/speed/limits",
+    "/api/labels",
+    "/api/ratio-groups",
+    "/api/rss",
+    "/api/smart-queue",
+    "/api/automations",
+    "/api/jobs",
+)
+RTORRENT_CONFIG_PREFIXES = ("/api/rtorrent-config",)
+ADMIN_PREFIXES = ("/api/auth/users", "/api/profiles")
+# Note: API reads that expose rTorrent/profile data must also respect profile permissions.
+PROFILE_READ_PREFIXES = (
+    "/api/torrents",
+    "/api/torrent-stats",
+    "/api/system/status",
+    "/api/app/status",
+    "/api/port-check",
+    "/api/path",
+    "/api/labels",
+    "/api/ratio-groups",
+    "/api/rss",
+    "/api/rtorrent-config",
+    "/api/smart-queue",
+    "/api/traffic/history",
+    "/api/automations",
+)
+
+
+def enabled() -> bool:
+    return bool(AUTH_ENABLE)
+
+
+def password_hash(password: str) -> str:
+    return generate_password_hash(password or "")
+
+
+def current_user_id() -> int:
+    if not enabled():
+        return default_user_id()
+    try:
+        return int(session.get("user_id") or 0)
+    except Exception:
+        return 0
+
+
+def current_user() -> dict[str, Any] | None:
+    uid = current_user_id()
+    if not uid:
+        return None
+    with connect() as conn:
+        return conn.execute(
+            "SELECT id, username, role, is_active, created_at, updated_at FROM users WHERE id=?",
+            (uid,),
+        ).fetchone()
+
+
+def is_admin(user: dict[str, Any] | None = None) -> bool:
+    if not enabled():
+        return True
+    user = user or current_user()
+    return bool(user and user.get("role") == "admin" and int(user.get("is_active") or 0))
+
+
+def _permissions(user_id: int | None = None) -> list[dict[str, Any]]:
+    if not enabled():
+        return [{"profile_id": 0, "access_level": "full"}]
+    uid = user_id or current_user_id()
+    if not uid:
+        return []
+    with connect() as conn:
+        return conn.execute(
+            "SELECT profile_id, access_level FROM user_profile_permissions WHERE user_id=?",
+            (uid,),
+        ).fetchall()
+
+
+def can_access_profile(profile_id: int | None, user_id: int | None = None) -> bool:
+    if not enabled():
+        return True
+    uid = user_id or current_user_id()
+    if not uid:
+        return False
+    with connect() as conn:
+        user = conn.execute("SELECT role, is_active FROM users WHERE id=?", (uid,)).fetchone()
+        if not user or not int(user.get("is_active") or 0):
+            return False
+        if user.get("role") == "admin":
+            return True
+        pid = int(profile_id or 0)
+        row = conn.execute(
+            "SELECT 1 FROM user_profile_permissions WHERE user_id=? AND (profile_id=0 OR profile_id=?) LIMIT 1",
+            (uid, pid),
+        ).fetchone()
+        return bool(row)
+
+
+def can_write_profile(profile_id: int | None, user_id: int | None = None) -> bool:
+    if not enabled():
+        return True
+    uid = user_id or current_user_id()
+    if not uid:
+        return False
+    with connect() as conn:
+        user = conn.execute("SELECT role, is_active FROM users WHERE id=?", (uid,)).fetchone()
+        if not user or not int(user.get("is_active") or 0):
+            return False
+        if user.get("role") == "admin":
+            return True
+        pid = int(profile_id or 0)
+        row = conn.execute(
+            "SELECT access_level FROM user_profile_permissions WHERE user_id=? AND (profile_id=0 OR profile_id=?) ORDER BY profile_id DESC LIMIT 1",
+            (uid, pid),
+        ).fetchone()
+        return bool(row and row.get("access_level") == "full")
+
+
+def visible_profile_ids(user_id: int | None = None) -> set[int] | None:
+    if not enabled():
+        return None
+    uid = user_id or current_user_id()
+    if not uid:
+        return set()
+    with connect() as conn:
+        user = conn.execute("SELECT role, is_active FROM users WHERE id=?", (uid,)).fetchone()
+        if not user or not int(user.get("is_active") or 0):
+            return set()
+        if user.get("role") == "admin":
+            return None
+        rows = conn.execute("SELECT profile_id FROM user_profile_permissions WHERE user_id=?", (uid,)).fetchall()
+    if any(int(row.get("profile_id") or 0) == 0 for row in rows):
+        return None
+    return {int(row.get("profile_id") or 0) for row in rows}
+
+
+
+def same_origin_request() -> bool:
+    """Return False only when an unsafe request clearly comes from another origin."""
+    origin = request.headers.get("Origin") or request.headers.get("Referer")
+    if not origin:
+        return True
+    try:
+        parsed = urlparse(origin)
+        return parsed.scheme == request.scheme and parsed.netloc == request.host
+    except Exception:
+        return False
+
+
+def writable_profile_ids(user_id: int | None = None) -> set[int] | None:
+    if not enabled():
+        return None
+    uid = user_id or current_user_id()
+    if not uid:
+        return set()
+    with connect() as conn:
+        user = conn.execute("SELECT role, is_active FROM users WHERE id=?", (uid,)).fetchone()
+        if not user or not int(user.get("is_active") or 0):
+            return set()
+        if user.get("role") == "admin":
+            return None
+        rows = conn.execute("SELECT profile_id FROM user_profile_permissions WHERE user_id=? AND access_level='full'", (uid,)).fetchall()
+    if any(int(row.get("profile_id") or 0) == 0 for row in rows):
+        return None
+    return {int(row.get("profile_id") or 0) for row in rows}
+
+def require_admin() -> None:
+    if enabled() and not is_admin():
+        abort(403)
+
+
+def require_profile_read(profile_id: int | None) -> None:
+    if enabled() and not can_access_profile(profile_id):
+        abort(403)
+
+
+def require_profile_write(profile_id: int | None) -> None:
+    if enabled() and not can_write_profile(profile_id):
+        abort(403)
+
+
+def login_user(username: str, password: str) -> dict[str, Any] | None:
+    if not enabled():
+        return {"id": default_user_id(), "username": "default", "role": "admin", "is_active": 1}
+    with connect() as conn:
+        user = conn.execute("SELECT * FROM users WHERE username=?", (username.strip(),)).fetchone()
+    if not user or not int(user.get("is_active") or 0):
+        return None
+    if not user.get("password_hash") or not check_password_hash(user.get("password_hash"), password or ""):
+        return None
+    session.clear()
+    session["user_id"] = int(user["id"])
+    session["username"] = user["username"]
+    session["role"] = user.get("role") or "user"
+    return current_user()
+
+
+def logout_user() -> None:
+    session.clear()
+
+
+def ensure_admin_user() -> None:
+    if not enabled():
+        return
+    now = utcnow()
+    with connect() as conn:
+        row = conn.execute("SELECT id FROM users WHERE username='admin'").fetchone()
+        if not row:
+            conn.execute(
+                "INSERT INTO users(username,password_hash,role,is_active,created_at,updated_at) VALUES(?,?,?,?,?,?)",
+                ("admin", password_hash("admin"), "admin", 1, now, now),
+            )
+        else:
+            conn.execute("UPDATE users SET role='admin', is_active=1, updated_at=? WHERE username='admin'", (now,))
+
+
+def list_users() -> list[dict[str, Any]]:
+    require_admin()
+    with connect() as conn:
+        users = conn.execute(
+            "SELECT id, username, role, is_active, created_at, updated_at FROM users ORDER BY username COLLATE NOCASE"
+        ).fetchall()
+        perms = conn.execute(
+            "SELECT user_id, profile_id, access_level FROM user_profile_permissions ORDER BY user_id, profile_id"
+        ).fetchall()
+    by_user: dict[int, list[dict[str, Any]]] = {}
+    for perm in perms:
+        by_user.setdefault(int(perm["user_id"]), []).append({
+            "profile_id": int(perm.get("profile_id") or 0),
+            "access_level": perm.get("access_level") or "ro",
+        })
+    for user in users:
+        user["permissions"] = by_user.get(int(user["id"]), [])
+    return users
+
+
+def save_user(data: dict[str, Any], user_id: int | None = None) -> dict[str, Any]:
+    require_admin()
+    now = utcnow()
+    username = str(data.get("username") or "").strip()
+    role = "admin" if data.get("role") == "admin" else "user"
+    is_active = 1 if data.get("is_active", True) else 0
+    if not username:
+        raise ValueError("Username is required")
+    with connect() as conn:
+        if user_id:
+            row = conn.execute("SELECT id FROM users WHERE id=?", (user_id,)).fetchone()
+            if not row:
+                raise ValueError("User does not exist")
+            conn.execute(
+                "UPDATE users SET username=?, role=?, is_active=?, updated_at=? WHERE id=?",
+                (username, role, is_active, now, user_id),
+            )
+        else:
+            cur = conn.execute(
+                "INSERT INTO users(username,password_hash,role,is_active,created_at,updated_at) VALUES(?,?,?,?,?,?)",
+                (username, password_hash(str(data.get("password") or username)), role, is_active, now, now),
+            )
+            user_id = int(cur.lastrowid)
+        if data.get("password"):
+            conn.execute("UPDATE users SET password_hash=?, updated_at=? WHERE id=?", (password_hash(str(data.get("password"))), now, user_id))
+        if role != "admin":
+            conn.execute("DELETE FROM user_profile_permissions WHERE user_id=?", (user_id,))
+            for item in data.get("permissions") or []:
+                profile_id = int(item.get("profile_id") or 0)
+                access = "full" if item.get("access_level") == "full" else "ro"
+                conn.execute(
+                    "INSERT OR REPLACE INTO user_profile_permissions(user_id,profile_id,access_level,created_at,updated_at) VALUES(?,?,?,?,?)",
+                    (user_id, profile_id, access, now, now),
+                )
+        else:
+            conn.execute("DELETE FROM user_profile_permissions WHERE user_id=?", (user_id,))
+        return conn.execute("SELECT id, username, role, is_active, created_at, updated_at FROM users WHERE id=?", (user_id,)).fetchone()
+
+
+def delete_user(user_id: int) -> None:
+    require_admin()
+    if int(user_id) == current_user_id():
+        raise ValueError("Cannot delete current user")
+    with connect() as conn:
+        conn.execute("DELETE FROM user_profile_permissions WHERE user_id=?", (user_id,))
+        conn.execute("DELETE FROM users WHERE id=? AND username <> 'admin'", (user_id,))
+
+
+def install_guards(app) -> None:
+    @app.before_request
+    def _auth_guard():
+        if not enabled():
+            return None
+        endpoint = request.endpoint or ""
+        if endpoint in PUBLIC_ENDPOINTS or endpoint.startswith("static"):
+            return None
+        if not current_user_id():
+            if request.path.startswith("/api/"):
+                return jsonify({"ok": False, "error": "Authentication required"}), 401
+            return redirect(url_for("main.login", next=request.full_path if request.query_string else request.path))
+        user = current_user()
+        if not user or not int(user.get("is_active") or 0):
+            logout_user()
+            return jsonify({"ok": False, "error": "Authentication required"}), 401 if request.path.startswith("/api/") else redirect(url_for("main.login"))
+        if request.path.startswith("/api/auth/users") and not is_admin(user):
+            return jsonify({"ok": False, "error": "Admin only"}), 403
+        if request.path.startswith(PROFILE_READ_PREFIXES):
+            profile_id = _request_profile_id()
+            if profile_id and not can_access_profile(profile_id):
+                return jsonify({"ok": False, "error": "Profile access denied"}), 403
+        if request.method not in {"GET", "HEAD", "OPTIONS"}:
+            if request.path.startswith("/api/") and not same_origin_request():
+                return jsonify({"ok": False, "error": "Cross-origin API request blocked"}), 403
+            if request.path.startswith("/api/profiles") and not request.path.endswith("/activate") and not is_admin(user):
+                return jsonify({"ok": False, "error": "Admin only"}), 403
+            profile_id = _request_profile_id()
+            if request.path.startswith(RTORRENT_CONFIG_PREFIXES) and not can_write_profile(profile_id):
+                return jsonify({"ok": False, "error": "Read-only profile access"}), 403
+            if request.path.startswith(RTORRENT_WRITE_PREFIXES) and not can_write_profile(profile_id):
+                return jsonify({"ok": False, "error": "Read-only profile access"}), 403
+        return None
+
+
+def _request_profile_id() -> int | None:
+    if request.view_args and request.view_args.get("profile_id"):
+        return int(request.view_args["profile_id"])
+    try:
+        payload = request.get_json(silent=True) or {}
+        if payload.get("profile_id"):
+            return int(payload.get("profile_id"))
+    except Exception:
+        pass
+    from . import preferences
+    profile = preferences.active_profile()
+    return int(profile["id"]) if profile else None
--- a/pytorrent/services/automation_rules.py
+++ b/pytorrent/services/automation_rules.py
@@ -137,8 +137,11 @@ def _apply_effects(c: Any, profile: dict[str, Any], torrent: dict[str, Any], eff
    for eff in effects:
        typ = str(eff.get('type') or '')
        if typ == 'move':
+            # Note: Automation move-to-path now uses the same move implementation as the main app action.
            path = str(eff.get('path') or '').strip() or rtorrent.default_download_path(profile)
-            if path: c.call('d.directory.set', h, path); applied.append({'type': 'move', 'path': path})
+            move_payload = {'path': path, 'move_data': bool(eff.get('move_data')), 'recheck': bool(eff.get('recheck', eff.get('move_data'))), 'keep_seeding': bool(eff.get('keep_seeding'))}
+            result = rtorrent.move_torrents(profile, [h], move_payload) if path else None
+            if path: applied.append({'type': 'move', 'path': path, 'move_data': bool(eff.get('move_data')), 'recheck': bool(move_payload['recheck']), 'keep_seeding': bool(eff.get('keep_seeding')), 'result': result})
        elif typ == 'add_label':
            label = str(eff.get('label') or '').strip()
            if label and label not in labels: labels.append(label); c.call('d.custom1.set', h, _label_value(labels))
--- a/pytorrent/services/frontend_assets.py
+++ b/pytorrent/services/frontend_assets.py
@@ -0,0 +1,111 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+from ..config import BASE_DIR, USE_OFFLINE_LIBS
+
+# Notatka: jeden manifest utrzymuje spójne adresy CDN i ścieżki lokalne dla trybu offline.
+LIBS_STATIC_DIR = "libs"
+LIBS_DIR = BASE_DIR / "pytorrent" / "static" / LIBS_STATIC_DIR
+BOOTSTRAP_VERSION = "5.3.3"
+BOOTSWATCH_VERSION = "5.3.3"
+FONTAWESOME_VERSION = "6.5.2"
+FLAG_ICONS_VERSION = "7.2.3"
+SWAGGER_UI_VERSION = "5"
+SOCKET_IO_VERSION = "4.7.5"
+
+BOOTSTRAP_THEMES = (
+    "default",
+    "flatly",
+    "litera",
+    "lumen",
+    "minty",
+    "sketchy",
+    "solar",
+    "spacelab",
+    "united",
+    "zephyr",
+)
+
+STATIC_ASSETS = {
+    "bootstrap_js": {
+        "local": f"{LIBS_STATIC_DIR}/bootstrap/{BOOTSTRAP_VERSION}/js/bootstrap.bundle.min.js",
+        "cdn": f"https://cdn.jsdelivr.net/npm/bootstrap@{BOOTSTRAP_VERSION}/dist/js/bootstrap.bundle.min.js",
+    },
+    "fontawesome_css": {
+        "local": f"{LIBS_STATIC_DIR}/fontawesome/{FONTAWESOME_VERSION}/css/all.min.css",
+        "cdn": f"https://cdnjs.cloudflare.com/ajax/libs/font-awesome/{FONTAWESOME_VERSION}/css/all.min.css",
+    },
+    "flag_icons_css": {
+        "local": f"{LIBS_STATIC_DIR}/flag-icons/{FLAG_ICONS_VERSION}/css/flag-icons.min.css",
+        "cdn": f"https://cdn.jsdelivr.net/gh/lipis/flag-icons@{FLAG_ICONS_VERSION}/css/flag-icons.min.css",
+    },
+    "socket_io_js": {
+        "local": f"{LIBS_STATIC_DIR}/socket.io/{SOCKET_IO_VERSION}/socket.io.min.js",
+        "cdn": f"https://cdn.socket.io/{SOCKET_IO_VERSION}/socket.io.min.js",
+    },
+    "swagger_css": {
+        "local": f"{LIBS_STATIC_DIR}/swagger-ui/{SWAGGER_UI_VERSION}/swagger-ui.css",
+        "cdn": f"https://cdn.jsdelivr.net/npm/swagger-ui-dist@{SWAGGER_UI_VERSION}/swagger-ui.css",
+    },
+    "swagger_js": {
+        "local": f"{LIBS_STATIC_DIR}/swagger-ui/{SWAGGER_UI_VERSION}/swagger-ui-bundle.js",
+        "cdn": f"https://cdn.jsdelivr.net/npm/swagger-ui-dist@{SWAGGER_UI_VERSION}/swagger-ui-bundle.js",
+    },
+}
+
+
+def bootstrap_css_asset(theme: str | None = None) -> dict[str, str]:
+    theme = theme if theme in BOOTSTRAP_THEMES else "default"
+    if theme == "default":
+        return {
+            "local": f"{LIBS_STATIC_DIR}/bootstrap/{BOOTSTRAP_VERSION}/css/bootstrap.min.css",
+            "cdn": f"https://cdn.jsdelivr.net/npm/bootstrap@{BOOTSTRAP_VERSION}/dist/css/bootstrap.min.css",
+        }
+    return {
+        "local": f"{LIBS_STATIC_DIR}/bootswatch/{BOOTSWATCH_VERSION}/{theme}/bootstrap.min.css",
+        "cdn": f"https://cdn.jsdelivr.net/npm/bootswatch@{BOOTSWATCH_VERSION}/dist/{theme}/bootstrap.min.css",
+    }
+
+
+def asset_path(key: str) -> str:
+    return STATIC_ASSETS[key]["local" if USE_OFFLINE_LIBS else "cdn"]
+
+
+def bootstrap_css_path(theme: str | None = None) -> str:
+    return bootstrap_css_asset(theme)["local" if USE_OFFLINE_LIBS else "cdn"]
+
+
+def required_offline_paths() -> list[Path]:
+    paths = [LIBS_DIR.parent / item["local"] for item in STATIC_ASSETS.values()]
+    paths.extend(LIBS_DIR.parent / bootstrap_css_asset(theme)["local"] for theme in BOOTSTRAP_THEMES)
+    return paths
+
+
+def missing_offline_paths() -> list[Path]:
+    missing = [path for path in required_offline_paths() if not path.is_file() or path.stat().st_size <= 0]
+    # Notatka: sprawdzane są też zasoby referencjonowane przez CSS, np. fonty ikon i pliki flag.
+    required_dirs = [
+        LIBS_DIR / f"fontawesome/{FONTAWESOME_VERSION}/webfonts",
+        LIBS_DIR / f"flag-icons/{FLAG_ICONS_VERSION}/flags/4x3",
+        LIBS_DIR / f"flag-icons/{FLAG_ICONS_VERSION}/flags/1x1",
+    ]
+    for directory in required_dirs:
+        if not directory.is_dir() or not any(directory.iterdir()):
+            missing.append(directory)
+    return missing
+
+
+def validate_offline_assets() -> None:
+    # Notatka: aplikacja zatrzymuje start, gdy tryb offline jest aktywny, a pliki nie są zainstalowane.
+    if not USE_OFFLINE_LIBS:
+        return
+    missing = missing_offline_paths()
+    if missing:
+        preview = "\n".join(f"- {path.relative_to(BASE_DIR)}" for path in missing[:20])
+        extra = "" if len(missing) <= 20 else f"\n- ... and {len(missing) - 20} more"
+        raise RuntimeError(
+            "PYTORRENT_USE_OFFLINE_LIBS=true, but frontend libraries are missing. "
+            "Run: ./scripts/download_frontend_libs.py or ./install.sh\n"
+            f"Missing files:\n{preview}{extra}"
+        )
--- a/pytorrent/services/preferences.py
+++ b/pytorrent/services/preferences.py
@@ -3,6 +3,7 @@ from __future__ import annotations
 import json

 from ..db import connect, utcnow, default_user_id
+from . import auth

 BOOTSTRAP_THEMES = {
    "default": "Default Bootstrap",
@@ -27,50 +28,50 @@ FONT_FAMILIES = {
 }

 def bootstrap_css_url(theme: str | None) -> str:
-    theme = theme if theme in BOOTSTRAP_THEMES else "default"
-    if theme == "default":
-        return "https://cdn.jsdelivr.net/npm/bootstrap@5.3.3/dist/css/bootstrap.min.css"
-    return f"https://cdn.jsdelivr.net/npm/bootswatch@5.3.3/dist/{theme}/bootstrap.min.css"
+    # Notatka: zachowana funkcja zwraca aktualny adres motywu, ale źródło wybiera konfiguracja offline.
+    from .frontend_assets import bootstrap_css_path

+    return bootstrap_css_path(theme)

 def list_profiles(user_id: int | None = None):
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
+    visible = auth.visible_profile_ids(user_id)
    with connect() as conn:
+        if visible is None:
+            return conn.execute(
+                "SELECT * FROM rtorrent_profiles ORDER BY is_default DESC, name COLLATE NOCASE"
+            ).fetchall()
+        if not visible:
+            return []
+        placeholders = ",".join("?" for _ in visible)
        return conn.execute(
-            "SELECT * FROM rtorrent_profiles WHERE user_id=? ORDER BY is_default DESC, name COLLATE NOCASE",
-            (user_id,),
+            f"SELECT * FROM rtorrent_profiles WHERE id IN ({placeholders}) ORDER BY is_default DESC, name COLLATE NOCASE",
+            tuple(visible),
        ).fetchall()


 def get_profile(profile_id: int, user_id: int | None = None):
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
+    if not auth.can_access_profile(profile_id, user_id):
+        return None
    with connect() as conn:
-        return conn.execute(
-            "SELECT * FROM rtorrent_profiles WHERE id=? AND user_id=?",
-            (profile_id, user_id),
-        ).fetchone()
+        return conn.execute("SELECT * FROM rtorrent_profiles WHERE id=?", (profile_id,)).fetchone()


 def active_profile(user_id: int | None = None):
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
    with connect() as conn:
        pref = conn.execute("SELECT active_rtorrent_id FROM user_preferences WHERE user_id=?", (user_id,)).fetchone()
-        if pref and pref.get("active_rtorrent_id"):
-            row = conn.execute(
-                "SELECT * FROM rtorrent_profiles WHERE id=? AND user_id=?",
-                (pref["active_rtorrent_id"], user_id),
-            ).fetchone()
+        if pref and pref.get("active_rtorrent_id") and auth.can_access_profile(int(pref["active_rtorrent_id"]), user_id):
+            row = conn.execute("SELECT * FROM rtorrent_profiles WHERE id=?", (pref["active_rtorrent_id"],)).fetchone()
            if row:
                return row
-        row = conn.execute(
-            "SELECT * FROM rtorrent_profiles WHERE user_id=? ORDER BY is_default DESC, id ASC LIMIT 1",
-            (user_id,),
-        ).fetchone()
-        return row
+        profiles = list_profiles(user_id)
+        return profiles[0] if profiles else None


 def save_profile(data: dict, user_id: int | None = None):
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
    now = utcnow()
    name = str(data.get("name") or "rTorrent").strip()
    scgi_url = str(data.get("scgi_url") or "").strip()
@@ -79,7 +80,7 @@ def save_profile(data: dict, user_id: int | None = None):
    is_remote = 1 if data.get("is_remote") else 0
    is_default = 1 if data.get("is_default") else 0
    if not scgi_url.startswith("scgi://"):
-        raise ValueError("SCGI URL musi zaczynać się od scgi://")
+        raise ValueError("SCGI URL must start with scgi://")
    with connect() as conn:
        if is_default:
            conn.execute("UPDATE rtorrent_profiles SET is_default=0 WHERE user_id=?", (user_id,))
@@ -94,11 +95,11 @@ def save_profile(data: dict, user_id: int | None = None):
                "UPDATE user_preferences SET active_rtorrent_id=?, updated_at=? WHERE user_id=?",
                (profile_id, now, user_id),
            )
-        return conn.execute("SELECT * FROM rtorrent_profiles WHERE id=? AND user_id=?", (profile_id, user_id)).fetchone()
+        return conn.execute("SELECT * FROM rtorrent_profiles WHERE id=?", (profile_id,)).fetchone()


 def update_profile(profile_id: int, data: dict, user_id: int | None = None):
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
    now = utcnow()
    name = str(data.get("name") or "rTorrent").strip()
    scgi_url = str(data.get("scgi_url") or "").strip()
@@ -107,24 +108,25 @@ def update_profile(profile_id: int, data: dict, user_id: int | None = None):
    is_remote = 1 if data.get("is_remote") else 0
    is_default = 1 if data.get("is_default") else 0
    if not scgi_url.startswith("scgi://"):
-        raise ValueError("SCGI URL musi zaczynać się od scgi://")
+        raise ValueError("SCGI URL must start with scgi://")
    with connect() as conn:
-        row = conn.execute("SELECT id FROM rtorrent_profiles WHERE id=? AND user_id=?", (profile_id, user_id)).fetchone()
-        if not row:
+        row = conn.execute("SELECT id FROM rtorrent_profiles WHERE id=?", (profile_id,)).fetchone()
+        if not row or not auth.can_write_profile(profile_id, user_id):
            raise ValueError("Profil nie istnieje")
        if is_default:
            conn.execute("UPDATE rtorrent_profiles SET is_default=0 WHERE user_id=?", (user_id,))
        conn.execute(
-            "UPDATE rtorrent_profiles SET name=?, scgi_url=?, is_default=?, timeout_seconds=?, max_parallel_jobs=?, is_remote=?, updated_at=? WHERE id=? AND user_id=?",
-            (name, scgi_url, is_default, timeout, max_parallel, is_remote, now, profile_id, user_id),
+            "UPDATE rtorrent_profiles SET name=?, scgi_url=?, is_default=?, timeout_seconds=?, max_parallel_jobs=?, is_remote=?, updated_at=? WHERE id=?",
+            (name, scgi_url, is_default, timeout, max_parallel, is_remote, now, profile_id),
        )
-        return conn.execute("SELECT * FROM rtorrent_profiles WHERE id=? AND user_id=?", (profile_id, user_id)).fetchone()
+        return conn.execute("SELECT * FROM rtorrent_profiles WHERE id=?", (profile_id,)).fetchone()


 def delete_profile(profile_id: int, user_id: int | None = None):
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
+    auth.require_profile_write(profile_id)
    with connect() as conn:
-        conn.execute("DELETE FROM rtorrent_profiles WHERE id=? AND user_id=?", (profile_id, user_id))
+        conn.execute("DELETE FROM rtorrent_profiles WHERE id=?", (profile_id,))
        active = active_profile(user_id)
        conn.execute(
            "UPDATE user_preferences SET active_rtorrent_id=?, updated_at=? WHERE user_id=?",
@@ -133,10 +135,10 @@ def delete_profile(profile_id: int, user_id: int | None = None):


 def activate_profile(profile_id: int, user_id: int | None = None):
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
    with connect() as conn:
-        row = conn.execute("SELECT id FROM rtorrent_profiles WHERE id=? AND user_id=?", (profile_id, user_id)).fetchone()
-        if not row:
+        row = conn.execute("SELECT id FROM rtorrent_profiles WHERE id=?", (profile_id,)).fetchone()
+        if not row or not auth.can_access_profile(profile_id, user_id):
            raise ValueError("Profil nie istnieje")
        conn.execute(
            "UPDATE user_preferences SET active_rtorrent_id=?, updated_at=? WHERE user_id=?",
@@ -146,13 +148,18 @@ def activate_profile(profile_id: int, user_id: int | None = None):


 def get_preferences(user_id: int | None = None):
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
    with connect() as conn:
-        return conn.execute("SELECT * FROM user_preferences WHERE user_id=?", (user_id,)).fetchone()
+        pref = conn.execute("SELECT * FROM user_preferences WHERE user_id=?", (user_id,)).fetchone()
+        if not pref:
+            now = utcnow()
+            conn.execute("INSERT INTO user_preferences(user_id, theme, created_at, updated_at) VALUES(?, 'dark', ?, ?)", (user_id, now, now))
+            pref = conn.execute("SELECT * FROM user_preferences WHERE user_id=?", (user_id,)).fetchone()
+        return pref


 def save_preferences(data: dict, user_id: int | None = None):
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
    allowed_theme = data.get("theme") if data.get("theme") in {"light", "dark"} else None
    bootstrap_theme = data.get("bootstrap_theme") if data.get("bootstrap_theme") in BOOTSTRAP_THEMES else None
    font_family = data.get("font_family") if data.get("font_family") in FONT_FAMILIES else None
--- a/pytorrent/services/rtorrent.py
+++ b/pytorrent/services/rtorrent.py
@@ -1,5 +1,6 @@
 from __future__ import annotations

+import errno
 import os
 import posixpath
 import socket
@@ -53,24 +54,57 @@ class ScgiRtorrentClient:
        }
        header_blob = b"".join(k.encode() + b"\0" + v.encode() + b"\0" for k, v in headers.items())
        payload = str(len(header_blob)).encode("ascii") + b":" + header_blob + b"," + body
-        with socket.create_connection((self.host, self.port), timeout=self.timeout) as sock:
-            sock.settimeout(self.timeout)
-            sock.sendall(payload)
-            chunks: list[bytes] = []
-            while True:
-                chunk = sock.recv(65536)
-                if not chunk:
-                    break
-                chunks.append(chunk)
-        response = b"".join(chunks)
-        if not response:
-            raise ConnectionError("Empty response from rTorrent SCGI")
-        if b"\r\n\r\n" in response:
-            response = response.split(b"\r\n\r\n", 1)[1]
-        elif b"\n\n" in response:
-            response = response.split(b"\n\n", 1)[1]
-        result, _ = loads(response)
-        return result[0] if len(result) == 1 else result
+        attempts = _scgi_retry_attempts()
+        last_exc = None
+        for attempt in range(1, attempts + 1):
+            try:
+                with socket.create_connection((self.host, self.port), timeout=self.timeout) as sock:
+                    sock.settimeout(self.timeout)
+                    sock.sendall(payload)
+                    chunks: list[bytes] = []
+                    while True:
+                        chunk = sock.recv(65536)
+                        if not chunk:
+                            break
+                        chunks.append(chunk)
+                response = b"".join(chunks)
+                if not response:
+                    raise ConnectionError("Empty response from rTorrent SCGI")
+                if b"\r\n\r\n" in response:
+                    response = response.split(b"\r\n\r\n", 1)[1]
+                elif b"\n\n" in response:
+                    response = response.split(b"\n\n", 1)[1]
+                result, _ = loads(response)
+                return result[0] if len(result) == 1 else result
+            except Exception as exc:
+                last_exc = exc
+                if attempt >= attempts or not _is_transient_scgi_error(exc):
+                    raise
+                time.sleep(_scgi_retry_delay(attempt))
+        raise last_exc or ConnectionError("rTorrent SCGI call failed")
+
+
+def _scgi_retry_attempts() -> int:
+    # Note: Short retry/backoff protects bulk operations from temporary Errno 111 during high rTorrent load.
+    try:
+        return max(1, min(10, int(os.environ.get("PYTORRENT_SCGI_RETRIES", "5"))))
+    except Exception:
+        return 5
+
+
+def _scgi_retry_delay(attempt: int) -> float:
+    return min(5.0, 0.35 * (2 ** max(0, attempt - 1)))
+
+
+def _is_transient_scgi_error(exc: Exception) -> bool:
+    # Note: Retry covers common temporary SCGI/socket errors but does not hide semantic XML-RPC errors.
+    if isinstance(exc, (ConnectionRefusedError, ConnectionResetError, TimeoutError, socket.timeout)):
+        return True
+    err_no = getattr(exc, "errno", None)
+    if err_no in {errno.ECONNREFUSED, errno.ECONNRESET, errno.ETIMEDOUT, errno.EHOSTUNREACH, errno.ENETUNREACH}:
+        return True
+    msg = str(exc).lower()
+    return any(text in msg for text in ("connection refused", "connection reset", "timed out", "timeout", "empty response", "pipe creation failed", "resource temporarily unavailable", "try again", "temporarily unavailable"))


 def client_for(profile: dict) -> ScgiRtorrentClient:
@@ -78,32 +112,78 @@ def client_for(profile: dict) -> ScgiRtorrentClient:


 _UNSUPPORTED_EXEC_METHODS: set[str] = set()
+_EXEC_TARGET_STYLE: dict[str, int] = {}
+
+def _rt_execute_preview(method_name: str, call_args: tuple) -> str:
+    # Note: The compact RPC summary removes long scripts from error messages while keeping the method and first arguments for diagnostics.
+    preview = ", ".join(repr(x) for x in call_args[:3])
+    if len(call_args) > 3:
+        preview += ", ..."
+    return f"{method_name}({preview})"
+
+
+def _rt_execute_target_variants(method: str, args: tuple) -> list[tuple]:
+    # Note: Depending on version, rTorrent XML-RPC either requires or rejects an empty target; cache the working variant per method.
+    variants = [("", *args), args]
+    preferred = _EXEC_TARGET_STYLE.get(method)
+    if preferred is not None and 0 <= preferred < len(variants):
+        return [variants[preferred]] + [v for i, v in enumerate(variants) if i != preferred]
+    return variants
+
+
+def _is_rt_method_missing(exc: Exception) -> bool:
+    msg = str(exc).lower()
+    return "not defined" in msg or "no such method" in msg or "unknown method" in msg
+
+
+def _rt_execute_methods(method: str) -> list[str]:
+    # Note: execute2.* is tried only when the base execute.* method does not exist to avoid false retry errors.
+    methods = [method]
+    if method.startswith("execute."):
+        fallback = method.replace("execute.", "execute2.", 1)
+        if fallback not in _UNSUPPORTED_EXEC_METHODS:
+            methods.append(fallback)
+    return methods
+

 def _rt_execute(c: ScgiRtorrentClient, method: str, *args):
    """Run rTorrent execute.* as the rTorrent user across XML-RPC variants."""
-    method_names = [method]
-    if method.startswith("execute."):
-        execute2 = method.replace("execute.", "execute2.", 1)
-        if execute2 not in _UNSUPPORTED_EXEC_METHODS:
-            method_names.append(execute2)
-    errors = []
-    for method_name in method_names:
-        for call_args in (("", *args), args):
-            try:
-                return c.call(method_name, *call_args)
-            except Exception as exc:
-                message = str(exc)
-                if "not defined" in message.lower():
-                    _UNSUPPORTED_EXEC_METHODS.add(method_name)
-                preview = ", ".join(repr(x) for x in call_args[:3])
-                if len(call_args) > 3:
-                    preview += ", ..."
-                errors.append(f"{method_name}({preview}): {exc}")
+    errors: list[str] = []
+    attempts = _scgi_retry_attempts()
+    for attempt in range(1, attempts + 1):
+        errors.clear()
+        transient_seen = False
+        primary_missing = False
+        for method_index, method_name in enumerate(_rt_execute_methods(method)):
+            if method_name in _UNSUPPORTED_EXEC_METHODS:
+                continue
+            if method_index > 0 and not primary_missing:
+                continue
+            for call_args in _rt_execute_target_variants(method_name, args):
+                try:
+                    result = c.call(method_name, *call_args)
+                    if method_name == method:
+                        _EXEC_TARGET_STYLE[method_name] = 0 if call_args and call_args[0] == "" else 1
+                    return result
+                except Exception as exc:
+                    if _is_rt_method_missing(exc):
+                        _UNSUPPORTED_EXEC_METHODS.add(method_name)
+                        if method_name == method:
+                            primary_missing = True
+                        errors.append(f"{method_name}: method not defined")
+                        break
+                    transient_seen = transient_seen or _is_transient_scgi_error(exc)
+                    errors.append(f"{_rt_execute_preview(method_name, call_args)}: {exc}")
+        if transient_seen and attempt < attempts:
+            time.sleep(_scgi_retry_delay(attempt))
+            continue
+        break
    raise RuntimeError("rTorrent execute failed: " + "; ".join(errors))


 def _is_rt_timeout_error(exc: Exception) -> bool:
-    return isinstance(exc, (TimeoutError, socket.timeout)) or "timed out" in str(exc).lower()
+    msg = str(exc).lower()
+    return isinstance(exc, (TimeoutError, socket.timeout)) or "timed out" in msg or "timeout" in msg


 def _rt_execute_allow_timeout(c: ScgiRtorrentClient, method: str, *args):
@@ -159,7 +239,8 @@ def _run_remote_move(c: ScgiRtorrentClient, src: str, dst: str, poll_interval: f
        try:
            output = str(_rt_execute(c, "execute.capture", "sh", "-c", poll_script, "pytorrent-move-poll", status_path) or "").strip()
        except Exception as exc:
-            if _is_rt_timeout_error(exc):
+            # Note: During bulk moves, rTorrent may briefly not create the execute.capture pipe; polling waits and retries.
+            if _is_rt_timeout_error(exc) or _is_transient_scgi_error(exc):
                continue
            raise
        if not output:
@@ -207,6 +288,47 @@ def _safe_rm_rf_path(path: str) -> str:
    return path


+def _run_remote_rm(c: ScgiRtorrentClient, path: str, poll_interval: float = 2.0) -> None:
+    # Note: rm -rf runs in the background on the rTorrent side, so long deletes do not hold a single SCGI connection.
+    token = uuid.uuid4().hex
+    status_path = f"/tmp/pytorrent-rm-{token}.status"
+    script = (
+        'target=$1; status=$2; tmp=${status}.tmp; '
+        'rm -f "$status" "$tmp"; '
+        '( rc=0; '
+        'if [ -z "$target" ] || [ "$target" = "/" ] || [ "$target" = "." ]; then echo "unsafe remove target: $target" >&2; rc=5; '
+        'else rm -rf -- "$target" || rc=$?; fi; '
+        'if [ $rc -eq 0 ]; then printf "OK\n" > "$status"; else printf "ERR %s\n" "$rc" > "$status"; fi; '
+        'if [ -s "$tmp" ]; then cat "$tmp" >> "$status"; fi; '
+        'rm -f "$tmp" ) > "$tmp" 2>&1 &'
+    )
+    poll_script = 'status=$1; [ -f "$status" ] && cat "$status" || true'
+    cleanup_script = 'rm -f "$1"'
+    _rt_execute_allow_timeout(c, "execute.throw", "sh", "-c", script, "pytorrent-rm-start", path, status_path)
+    while True:
+        time.sleep(max(0.25, poll_interval))
+        try:
+            output = str(_rt_execute(c, "execute.capture", "sh", "-c", poll_script, "pytorrent-rm-poll", status_path) or "").strip()
+        except Exception as exc:
+            # Note: Remove uses the same safe polling as move, so a temporary missing pipe does not fail the whole queue.
+            if _is_rt_timeout_error(exc) or _is_transient_scgi_error(exc):
+                continue
+            raise
+        if not output:
+            continue
+        try:
+            _rt_execute(c, "execute.throw", "sh", "-c", cleanup_script, "pytorrent-rm-clean", status_path)
+        except Exception:
+            pass
+        first_line = output.splitlines()[0].strip()
+        if first_line == "OK":
+            return
+        if first_line.startswith("ERR"):
+            details = "\n".join(output.splitlines()[1:]).strip()
+            raise RuntimeError(details or first_line)
+        raise RuntimeError(output)
+
+
 def _remove_torrent_data(c: ScgiRtorrentClient, torrent_hash: str) -> dict:
    data_path = _safe_rm_rf_path(_torrent_data_path(c, torrent_hash))
    try:
@@ -217,7 +339,7 @@ def _remove_torrent_data(c: ScgiRtorrentClient, torrent_hash: str) -> dict:
        c.call("d.close", torrent_hash)
    except Exception:
        pass
-    _rt_execute(c, "execute.throw", "rm", "-rf", data_path)
+    _run_remote_rm(c, data_path)
    return {"hash": torrent_hash, "removed_path": data_path}


@@ -249,6 +371,86 @@ def browse_path(profile: dict, path: str | None = None) -> dict:
    return {"path": base, "parent": parent, "dirs": dirs[:300], "source": "rtorrent"}


+POST_CHECK_DOWNLOAD_LABEL = "To download after check"
+
+
+def _label_names(value: str) -> list[str]:
+    names: list[str] = []
+    for part in str(value or "").replace(";", ",").replace("|", ",").split(","):
+        label = part.strip()
+        if label and label not in names:
+            names.append(label)
+    return names
+
+
+def _label_value(labels: list[str]) -> str:
+    return ", ".join([label for label in labels if str(label or "").strip()])
+
+
+def _row_progress_complete(row: dict) -> bool:
+    size = int(row.get("size") or 0)
+    completed = int(row.get("completed_bytes") or 0)
+    return bool(row.get("complete")) or (size > 0 and completed >= size) or float(row.get("progress") or 0) >= 100.0
+
+
+def _remove_post_check_label_if_finished(c: ScgiRtorrentClient, row: dict) -> bool:
+    labels = _label_names(str(row.get("label") or ""))
+    if POST_CHECK_DOWNLOAD_LABEL not in labels:
+        return False
+    status = str(row.get("status") or "").lower()
+    if not (_row_progress_complete(row) or status == "seeding"):
+        return False
+    labels = [label for label in labels if label != POST_CHECK_DOWNLOAD_LABEL]
+    value = _label_value(labels)
+    # Note: Clean the temporary label after reaching 100% or entering seeding, even when the state no longer comes directly from recheck.
+    c.call("d.custom1.set", str(row.get("hash") or ""), value)
+    row["label"] = value
+    return True
+
+
+def apply_post_check_policy(profile: dict, rows: list[dict], previous_rows: dict[str, dict] | None = None) -> list[dict]:
+    """Start complete torrents after check; pause and label incomplete ones."""
+    previous_rows = previous_rows or {}
+    c = client_for(profile)
+    changes: list[dict] = []
+    for row in rows:
+        h = str(row.get("hash") or "")
+        prev = previous_rows.get(h) or {}
+        try:
+            if h and _remove_post_check_label_if_finished(c, row):
+                changes.append({"hash": h, "action": "remove_post_check_label", "complete": True})
+        except Exception as exc:
+            changes.append({"hash": h, "action": "remove_post_check_label_failed", "error": str(exc)})
+        was_checking = str(prev.get("status") or "") == "Checking" or int(prev.get("hashing") or 0) > 0
+        is_checking = str(row.get("status") or "") == "Checking" or int(row.get("hashing") or 0) > 0
+        if not h or not was_checking or is_checking:
+            continue
+        complete = _row_progress_complete(row)
+        try:
+            if complete:
+                # Note: After a completed check, a complete torrent is started automatically so it can seed immediately.
+                c.call("d.start", h)
+                labels = [label for label in _label_names(str(row.get("label") or "")) if label != POST_CHECK_DOWNLOAD_LABEL]
+                if _label_value(labels) != str(row.get("label") or ""):
+                    c.call("d.custom1.set", h, _label_value(labels))
+                    row["label"] = _label_value(labels)
+                row.update({"state": 1, "active": 1, "paused": False, "status": "Seeding"})
+                changes.append({"hash": h, "action": "start", "complete": True})
+            else:
+                # Note: After check, an incomplete torrent is paused and labeled to show that it needs more downloading.
+                c.call("d.start", h)
+                c.call("d.pause", h)
+                labels = _label_names(str(row.get("label") or ""))
+                if POST_CHECK_DOWNLOAD_LABEL not in labels:
+                    labels.append(POST_CHECK_DOWNLOAD_LABEL)
+                    c.call("d.custom1.set", h, _label_value(labels))
+                row.update({"state": 1, "active": 0, "paused": True, "status": "Paused", "label": _label_value(labels)})
+                changes.append({"hash": h, "action": "pause_and_label", "complete": False, "label": POST_CHECK_DOWNLOAD_LABEL})
+        except Exception as exc:
+            changes.append({"hash": h, "action": "post_check_policy_failed", "error": str(exc)})
+    return changes
+
+
 TORRENT_FIELDS = [
    "d.hash=", "d.name=", "d.state=", "d.complete=", "d.size_bytes=", "d.completed_bytes=",
    "d.ratio=", "d.up.rate=", "d.down.rate=", "d.up.total=", "d.down.total=", "d.peers_connected=",
@@ -646,31 +848,6 @@ def torrent_peers(profile: dict, torrent_hash: str) -> list[dict]:
    return peers


-def peer_action(profile: dict, torrent_hash: str, peer_index: int, action_name: str) -> dict:
-    if peer_index < 0:
-        raise ValueError("Invalid peer index")
-    methods = {
-        "disconnect": ["p.disconnect", "p.close"],
-        "kick": ["p.disconnect", "p.close"],
-        "snub": ["p.snub"],
-        "unsnub": ["p.unsnub"],
-        "ban": ["p.ban", "p.disconnect"],
-    }
-    candidates = methods.get(action_name)
-    if not candidates:
-        raise ValueError(f"Unknown peer action: {action_name}")
-    c = client_for(profile)
-    target = f"{torrent_hash}:p{int(peer_index)}"
-    errors = []
-    for method in candidates:
-        try:
-            c.call(method, target)
-            return {"ok": True, "action": action_name, "method": method, "peer_index": peer_index}
-        except Exception as exc:
-            errors.append(f"{method}: {exc}")
-    raise RuntimeError("; ".join(errors))
-
-


 def _call_first(c: ScgiRtorrentClient, candidates: list[tuple[str, tuple]]) -> dict:
@@ -984,14 +1161,191 @@ def apply_startup_overrides(profile: dict) -> dict:
        return {"ok": True, "updated": [], "errors": [], "skipped": True}
    return set_config(profile, values, apply_now=True, apply_on_start=True)

+
+def _int_rpc(c: ScgiRtorrentClient, method: str, h: str, default: int = 0) -> int:
+    try:
+        return int(c.call(method, h) or 0)
+    except Exception:
+        return default
+
+
+def _str_rpc(c: ScgiRtorrentClient, method: str, h: str, default: str = '') -> str:
+    try:
+        return str(c.call(method, h) or '')
+    except Exception:
+        return default
+
+
+def _download_runtime_state(c: ScgiRtorrentClient, h: str) -> dict:
+    """Read rTorrent state using the native pause model: stopped, paused or active."""
+    state = _int_rpc(c, 'd.state', h)
+    active = _int_rpc(c, 'd.is_active', h)
+    opened = _int_rpc(c, 'd.is_open', h)
+    # Note: In rTorrent, pause does not change d.state. Paused means state=1, open=1, active=0.
+    return {
+        'state': state,
+        'open': opened,
+        'active': active,
+        'paused': bool(state and opened and not active),
+        'stopped': not bool(state),
+        'message': _str_rpc(c, 'd.message', h),
+    }
+
+
+def pause_hash(c: ScgiRtorrentClient, torrent_hash: str) -> dict:
+    """Pause an active rTorrent item without stopping or closing it."""
+    h = str(torrent_hash or '')
+    if not h:
+        return {'hash': h, 'ok': False, 'error': 'missing hash'}
+    before = _download_runtime_state(c, h)
+    result = {'hash': h, 'before': before, 'commands': []}
+    try:
+        # Note: Smart Queue frees a slot with d.pause, not d.stop, so later d.resume behaves like ruTorrent.
+        c.call('d.pause', h)
+        result['commands'].append('d.pause')
+        result['after'] = _download_runtime_state(c, h)
+        result['ok'] = True
+    except Exception as exc:
+        result.update({'ok': False, 'error': str(exc), 'after': _download_runtime_state(c, h)})
+    return result
+
+
+def resume_paused_hash(c: ScgiRtorrentClient, torrent_hash: str) -> dict:
+    """Resume only a paused rTorrent item; never convert it through stop/start."""
+    h = str(torrent_hash or '')
+    if not h:
+        return {'hash': h, 'ok': False, 'error': 'missing hash'}
+    before = _download_runtime_state(c, h)
+    result: dict = {'hash': h, 'before': before, 'commands': []}
+    if before.get('stopped'):
+        result.update({'ok': False, 'skipped': 'stopped_not_paused', 'after': before})
+        return result
+    if before.get('active'):
+        result.update({'ok': True, 'skipped': 'already_active', 'after': before})
+        return result
+    try:
+        # Note: ruTorrent unpauses with the equivalent of d.resume. Do not add d.start/d.open,
+        # because those commands belong to Stopped/Open state, not a clean Paused state.
+        c.call('d.resume', h)
+        result['commands'].append('d.resume')
+        result['after'] = _download_runtime_state(c, h)
+        result['ok'] = True
+    except Exception as exc:
+        result.update({'ok': False, 'error': str(exc), 'after': _download_runtime_state(c, h)})
+    return result
+
+
+def start_or_resume_hash(c: ScgiRtorrentClient, torrent_hash: str) -> dict:
+    """Start stopped torrents or resume torrents paused with d.pause, without mixing both paths."""
+    h = str(torrent_hash or '')
+    if not h:
+        return {'hash': h, 'ok': False, 'error': 'missing hash'}
+    before = _download_runtime_state(c, h)
+    result: dict = {'hash': h, 'before': before, 'commands': []}
+
+    if before.get('active'):
+        result.update({'ok': True, 'skipped': 'already_active', 'after': before})
+        return result
+
+    if before.get('paused') or (before.get('state') and not before.get('active')):
+        # Note: Paused rTorrent items are resumed only with d.resume; d.start is intentionally skipped here.
+        resumed = resume_paused_hash(c, h)
+        resumed['mode'] = 'resume_paused'
+        return resumed
+
+    try:
+        # Note: d.start remains only for Stopped/closed items, not for the pause-to-resume path.
+        c.call('d.open', h)
+        result['commands'].append('d.open')
+    except Exception as exc:
+        result.setdefault('ignored_errors', []).append(f'd.open: {exc}')
+    try:
+        c.call('d.start', h)
+        result['commands'].append('d.start')
+    except Exception as exc:
+        result.setdefault('ignored_errors', []).append(f'd.start: {exc}')
+        try:
+            c.call('d.try_start', h)
+            result['commands'].append('d.try_start')
+        except Exception as exc2:
+            result.setdefault('ignored_errors', []).append(f'd.try_start: {exc2}')
+            result['ok'] = False
+    result['after'] = _download_runtime_state(c, h)
+    result['ok'] = result.get('ok', True)
+    return result
+
+
+def move_torrents(profile: dict, torrent_hashes: list[str], payload: dict | None = None) -> dict:
+    # Note: Shared move implementation keeps API move and automation move-to-path identical.
+    payload = payload or {}
+    c = client_for(profile)
+    path = _remote_clean_path(payload.get("path") or "")
+    move_data = bool(payload.get("move_data"))
+    recheck = bool(payload.get("recheck", move_data))
+    keep_seeding = bool(payload.get("keep_seeding"))
+    # Note: keep_seeding lets automation move completed data to another path and force the torrent back into seeding.
+    if not path:
+        raise ValueError("Missing path")
+    results = []
+    if move_data:
+        _rt_execute_allow_timeout(c, "execute.throw", "mkdir", "-p", path)
+    for h in torrent_hashes:
+        item = {"hash": h, "path": path, "move_data": move_data, "keep_seeding": keep_seeding}
+        try:
+            was_state = int(c.call("d.state", h) or 0)
+        except Exception:
+            was_state = 0
+        try:
+            was_active = int(c.call("d.is_active", h) or 0)
+        except Exception:
+            was_active = was_state
+        if move_data:
+            src = _remote_clean_path(_torrent_data_path(c, h))
+            if not src:
+                raise ValueError(f"Cannot determine source path for {h}")
+            dst = _remote_join(path, posixpath.basename(src.rstrip("/")))
+            if src != dst:
+                try:
+                    c.call("d.stop", h)
+                except Exception:
+                    pass
+                try:
+                    c.call("d.close", h)
+                except Exception:
+                    pass
+                _run_remote_move(c, src, dst)
+                item["moved_from"] = src
+                item["moved_to"] = dst
+            else:
+                item["skipped"] = "source and destination are the same"
+            c.call("d.directory.set", h, path)
+            if recheck:
+                try:
+                    c.call("d.check_hash", h)
+                except Exception as exc:
+                    item["recheck_error"] = str(exc)
+            if keep_seeding or was_state or was_active:
+                try:
+                    c.call("d.start", h)
+                    item["started_after_move"] = True
+                except Exception as exc:
+                    item["start_error"] = str(exc)
+        else:
+            c.call("d.directory.set", h, path)
+            if keep_seeding:
+                try:
+                    c.call("d.start", h)
+                    item["started_after_path_change"] = True
+                except Exception as exc:
+                    item["start_error"] = str(exc)
+        results.append(item)
+    return {"ok": True, "count": len(torrent_hashes), "move_data": move_data, "keep_seeding": keep_seeding, "results": results}
+
 def action(profile: dict, torrent_hashes: list[str], name: str, payload: dict | None = None) -> dict:
    payload = payload or {}
    c = client_for(profile)
    methods = {
-        "start": "d.start",
-        "pause": "d.pause",
        "stop": "d.stop",
-        "resume": "d.resume",
        "recheck": "d.check_hash",
        "reannounce": "d.tracker_announce",
        "remove": "d.erase",
@@ -1007,58 +1361,21 @@ def action(profile: dict, torrent_hashes: list[str], name: str, payload: dict |
            c.call("d.custom.set", h, "py_ratio_group", group)
        return {"ok": True, "count": len(torrent_hashes), "ratio_group": group}
    if name == "move":
-        path = _remote_clean_path(payload.get("path") or "")
-        move_data = bool(payload.get("move_data"))
-        recheck = bool(payload.get("recheck", move_data))
-        if not path:
-            raise ValueError("Missing path")
-        results = []
-        if move_data:
-            _rt_execute_allow_timeout(c, "execute.throw", "mkdir", "-p", path)
-        for h in torrent_hashes:
-            item = {"hash": h, "path": path, "move_data": move_data}
-            try:
-                was_state = int(c.call("d.state", h) or 0)
-            except Exception:
-                was_state = 0
-            try:
-                was_active = int(c.call("d.is_active", h) or 0)
-            except Exception:
-                was_active = was_state
-            if move_data:
-                src = _remote_clean_path(_torrent_data_path(c, h))
-                if not src:
-                    raise ValueError(f"Cannot determine source path for {h}")
-                dst = _remote_join(path, posixpath.basename(src.rstrip("/")))
-                if src != dst:
-                    try:
-                        c.call("d.stop", h)
-                    except Exception:
-                        pass
-                    try:
-                        c.call("d.close", h)
-                    except Exception:
-                        pass
-                    _run_remote_move(c, src, dst)
-                    item["moved_from"] = src
-                    item["moved_to"] = dst
-                else:
-                    item["skipped"] = "source and destination are the same"
-                c.call("d.directory.set", h, path)
-                if recheck:
-                    try:
-                        c.call("d.check_hash", h)
-                    except Exception as exc:
-                        item["recheck_error"] = str(exc)
-                if was_state or was_active:
-                    try:
-                        c.call("d.start", h)
-                    except Exception:
-                        pass
-            else:
-                c.call("d.directory.set", h, path)
-            results.append(item)
-        return {"ok": True, "count": len(torrent_hashes), "move_data": move_data, "results": results}
+        # Note: Main move delegates to the shared helper used by automations.
+        return move_torrents(profile, torrent_hashes, payload)
+    if name == "pause":
+        # Note: The app pause action is now a pure d.pause so later resume works without stop/start.
+        results = [pause_hash(c, h) for h in torrent_hashes]
+        return {"ok": True, "count": len(torrent_hashes), "remove_data": False, "results": results}
+    if name in {"resume", "unpause"}:
+        # Note: Resume/Unpause uses only d.resume for Paused state.
+        results = [resume_paused_hash(c, h) for h in torrent_hashes]
+        return {"ok": True, "count": len(torrent_hashes), "remove_data": False, "results": results}
+    if name == "start":
+        # Note: Start separates Stopped from Paused; paused items go through d.resume, stopped items through d.start.
+        results = [start_or_resume_hash(c, h) for h in torrent_hashes]
+        return {"ok": True, "count": len(torrent_hashes), "remove_data": False, "results": results}
+
    method = methods.get(name)
    if not method:
        raise ValueError(f"Unknown action: {name}")
--- a/pytorrent/services/smart_queue.py
+++ b/pytorrent/services/smart_queue.py
@@ -29,6 +29,7 @@ def _default_settings(user_id: int, profile_id: int) -> dict[str, Any]:
        'stalled_seconds': 300,
        'min_speed_bytes': 1024,
        'min_seeds': 1,
+        'manage_stopped': 0,
        'updated_at': utcnow(),
    }

@@ -52,20 +53,23 @@ def save_settings(profile_id: int, data: dict[str, Any], user_id: int | None = N
        'stalled_seconds': max(30, int(data.get('stalled_seconds') or current.get('stalled_seconds') or 300)),
        'min_speed_bytes': max(0, int(data.get('min_speed_bytes') or current.get('min_speed_bytes') or 0)),
        'min_seeds': max(0, int(data.get('min_seeds') or current.get('min_seeds') or 0)),
+        # Note: This switch protects fully stopped torrents from automatic starts; by default Smart Queue manages only paused items.
+        'manage_stopped': 1 if data.get('manage_stopped', current.get('manage_stopped')) else 0,
    }
    now = utcnow()
    with connect() as conn:
        conn.execute(
-            '''INSERT INTO smart_queue_settings(user_id,profile_id,enabled,max_active_downloads,stalled_seconds,min_speed_bytes,min_seeds,updated_at)
-               VALUES(?,?,?,?,?,?,?,?)
+            '''INSERT INTO smart_queue_settings(user_id,profile_id,enabled,max_active_downloads,stalled_seconds,min_speed_bytes,min_seeds,manage_stopped,updated_at)
+               VALUES(?,?,?,?,?,?,?,?,?)
               ON CONFLICT(user_id, profile_id) DO UPDATE SET
               enabled=excluded.enabled,
               max_active_downloads=excluded.max_active_downloads,
               stalled_seconds=excluded.stalled_seconds,
               min_speed_bytes=excluded.min_speed_bytes,
               min_seeds=excluded.min_seeds,
+               manage_stopped=excluded.manage_stopped,
               updated_at=excluded.updated_at''',
-            (user_id, profile_id, settings['enabled'], settings['max_active_downloads'], settings['stalled_seconds'], settings['min_speed_bytes'], settings['min_seeds'], now),
+            (user_id, profile_id, settings['enabled'], settings['max_active_downloads'], settings['stalled_seconds'], settings['min_speed_bytes'], settings['min_seeds'], settings['manage_stopped'], now),
        )
    return get_settings(profile_id, user_id)

@@ -147,17 +151,33 @@ def _remember_auto_label(profile_id: int, torrent_hash: str, previous_label: str
            )


+def _read_label(client: Any, torrent_hash: str, fallback: str = '') -> str:
+    try:
+        return str(client.call('d.custom1', torrent_hash) or '')
+    except Exception:
+        return fallback
+
+
 def _restore_auto_label(client: Any, profile_id: int, torrent_hash: str, current_label: str | None = None) -> bool:
    with connect() as conn:
        row = conn.execute(
            'SELECT previous_label FROM smart_queue_auto_labels WHERE profile_id=? AND torrent_hash=?',
            (profile_id, torrent_hash),
        ).fetchone()
+        live_label = _read_label(client, torrent_hash, current_label or '')
        if not row:
-            return False
+            if live_label != SMART_QUEUE_LABEL:
+                return False
+            try:
+                # Note: Clear the Smart Queue label even when the torrent was marked earlier but no previous-label entry remains.
+                client.call('d.custom1.set', torrent_hash, '')
+                return True
+            except Exception:
+                return False
        previous = row.get('previous_label') or ''
        try:
-            if current_label is None or current_label == SMART_QUEUE_LABEL:
+            # Note: On resume, Smart Queue restores the previous label only while it still sees its own technical label.
+            if live_label == SMART_QUEUE_LABEL or current_label is None:
                client.call('d.custom1.set', torrent_hash, previous)
            conn.execute('DELETE FROM smart_queue_auto_labels WHERE profile_id=? AND torrent_hash=?', (profile_id, torrent_hash))
            return True
@@ -165,6 +185,103 @@ def _restore_auto_label(client: Any, profile_id: int, torrent_hash: str, current
            return False


+
+
+
+def _call_rtorrent_setter(client: Any, method: str, value: int) -> bool:
+    """Set a scalar rTorrent setting while tolerating XMLRPC signature differences."""
+    for args in ((int(value),), ('', int(value))):
+        try:
+            client.call(method, *args)
+            return True
+        except Exception:
+            continue
+    return False
+
+
+def _ensure_rtorrent_download_cap(client: Any, max_active: int) -> dict[str, Any]:
+    """Raise rTorrent download caps that can silently limit Smart Queue to one item."""
+    result: dict[str, Any] = {'checked': False, 'updated': False, 'items': []}
+    # Note: rTorrent may have separate global and per-throttle limits. When div=1,
+    # starts can effectively stop at one active torrent even when the target is 100.
+    for key in ('throttle.max_downloads.global', 'throttle.max_downloads.div'):
+        item: dict[str, Any] = {'key': key, 'checked': False, 'updated': False}
+        try:
+            current = int(client.call(key) or 0)
+            item.update({'checked': True, 'current': current, 'target': int(max_active)})
+            result['checked'] = True
+            # Note: 0 means unlimited; raise only positive limits lower than the target.
+            if 0 < current < max_active:
+                ok = _call_rtorrent_setter(client, f'{key}.set', int(max_active))
+                item['updated'] = ok
+                if ok:
+                    result['updated'] = True
+                    item['new'] = int(max_active)
+                    result.setdefault('current', current)
+                    result['new'] = int(max_active)
+        except Exception as exc:
+            item.update({'error': str(exc)})
+        result['items'].append(item)
+    return result
+
+
+def _start_download(client: Any, torrent: dict[str, Any]) -> dict[str, Any]:
+    """Resume paused torrents through rTorrent's pause model."""
+    h = str(torrent.get('hash') or '')
+    if not h:
+        return {'hash': h, 'ok': False, 'error': 'missing hash'}
+    if bool(torrent.get('paused')) or str(torrent.get('status') or '').lower() == 'paused' or int(torrent.get('state') or 0):
+        # Note: Smart Queue candidates paused with d.pause must be resumed with d.resume, without d.start/d.stop.
+        return rtorrent.resume_paused_hash(client, h)
+    # Note: Only optional manage_stopped uses the start path for fully stopped torrents.
+    return rtorrent.start_or_resume_hash(client, h)
+
+
+def _verify_started_downloads(client: Any, hashes: list[str], attempts: int = 10, delay: float = 0.5) -> tuple[list[str], list[dict[str, Any]]]:
+    """Verify starts after rTorrent has time to process resume/start commands."""
+    pending = [h for h in hashes if h]
+    started: list[str] = []
+    no_effect: list[dict[str, Any]] = []
+    seen_started: set[str] = set()
+    last_state: dict[str, dict[str, Any]] = {}
+
+    for attempt in range(max(1, attempts)):
+        if attempt:
+            time.sleep(delay)
+        for h in list(pending):
+            live = _read_live_start_state(client, h)
+            last_state[h] = live
+            if live.get('started'):
+                seen_started.add(h)
+                pending.remove(h)
+        if not pending:
+            break
+
+    started = [h for h in hashes if h in seen_started]
+    no_effect = [last_state.get(h, {'hash': h, 'started': False}) for h in hashes if h and h not in seen_started]
+    return started, no_effect
+
+def _read_live_start_state(client: Any, torrent_hash: str) -> dict[str, Any]:
+    result: dict[str, Any] = {'hash': torrent_hash}
+    fields = (
+        ('state', 'd.state'),
+        ('active', 'd.is_active'),
+        ('open', 'd.is_open'),
+        ('priority', 'd.priority'),
+        ('message', 'd.message'),
+        ('label', 'd.custom1'),
+    )
+    for key, method in fields:
+        try:
+            value = client.call(method, torrent_hash)
+            result[key] = int(value or 0) if key in {'state', 'active', 'open', 'priority'} else str(value or '')
+        except Exception as exc:
+            result[f'{key}_error'] = str(exc)
+    # Note: Do not treat d.is_open or state=1 as resumed; Paused can also have those values.
+    # Smart Queue counts a start only after d.is_active=1, meaning the pause was actually removed.
+    result['started'] = bool(int(result.get('active') or 0))
+    return result
+
 def _set_smart_queue_label(client: Any, torrent_hash: str, attempts: int = 3) -> bool:
    for attempt in range(max(1, attempts)):
        try:
@@ -186,31 +303,87 @@ def _mark_auto_paused(client: Any, profile_id: int, torrent: dict[str, Any]) ->
    return _set_smart_queue_label(client, torrent_hash)


-def _cleanup_auto_labels(client: Any, profile_id: int, torrents: list[dict[str, Any]], keep_hashes: set[str]) -> list[str]:
+def _is_smart_queue_hold(torrent: dict[str, Any] | None, manage_stopped: bool = True) -> bool:
+    if not torrent or int(torrent.get('complete') or 0):
+        return False
+    if str(torrent.get('label') or '') == SMART_QUEUE_LABEL:
+        return True
+    # Note: Paused in rTorrent usually has state=1 and active=0, so state=0 must not be required.
+    # This lets Smart Queue treat paused torrents as pending and fill the queue target later.
+    if bool(torrent.get('paused')):
+        return True
+    # Note: Fully stopped items are managed only when Use stopped torrents is enabled.
+    if not manage_stopped:
+        return False
+    return not int(torrent.get('state') or 0)
+
+
+def _clear_untracked_smart_queue_label(client: Any, torrent_hash: str, current_label: str) -> bool:
+    if current_label != SMART_QUEUE_LABEL:
+        return False
+    try:
+        # Note: Clear an orphaned Smart Queue label when no previous-label entry exists in the database.
+        client.call('d.custom1.set', torrent_hash, '')
+        return True
+    except Exception:
+        return False
+
+
+def _cleanup_auto_labels(client: Any, profile_id: int, torrents: list[dict[str, Any]], keep_hashes: set[str], manage_stopped: bool = True) -> list[str]:
    by_hash = {str(t.get('hash') or ''): t for t in torrents}
    restored: list[str] = []
    with connect() as conn:
        rows = conn.execute('SELECT torrent_hash FROM smart_queue_auto_labels WHERE profile_id=?', (profile_id,)).fetchall()
+    tracked_hashes = {str(row.get('torrent_hash') or '') for row in rows if row.get('torrent_hash')}
+
    for row in rows:
        h = str(row.get('torrent_hash') or '')
        t = by_hash.get(h)
        if not h or h in keep_hashes:
            continue
-        if t is None or int(t.get('complete') or 0):
-            if _restore_auto_label(client, profile_id, h, None if t is None else str(t.get('label') or '')):
+        current_label = '' if t is None else str(t.get('label') or '')
+        if not _is_smart_queue_hold(t, manage_stopped):
+            if _restore_auto_label(client, profile_id, h, None if t is None else current_label):
                restored.append(h)
            continue
-        is_paused_or_stopped = bool(t.get('paused')) or not int(t.get('active') or 0) or not int(t.get('state') or 0)
-        current_label = str(t.get('label') or '')
-        if is_paused_or_stopped:
-            if current_label != SMART_QUEUE_LABEL:
-                _set_smart_queue_label(client, h)
+        if current_label != SMART_QUEUE_LABEL:
+            _set_smart_queue_label(client, h)
+
+    for h, t in by_hash.items():
+        if not h or h in keep_hashes or h in tracked_hashes or _is_smart_queue_hold(t, manage_stopped):
            continue
-        if _restore_auto_label(client, profile_id, h, current_label):
+        if _clear_untracked_smart_queue_label(client, h, str(t.get('label') or '')):
            restored.append(h)
    return restored


+def _is_running_download_slot(t: dict[str, Any]) -> bool:
+    """Return True for incomplete torrents that already occupy a Smart Queue slot."""
+    # Note: The Smart Queue limit means the target number of actually active slots.
+    # Paused can have state=1/open=1, so a slot is counted only after d.is_active=1.
+    if int(t.get('complete') or 0):
+        return False
+    if str(t.get('label') or '') == SMART_QUEUE_LABEL:
+        return False
+    status = str(t.get('status') or '').lower()
+    if status == 'checking' or status == 'paused' or bool(t.get('paused')):
+        return False
+    return bool(int(t.get('active') or 0))
+
+
+def _is_waiting_download_candidate(t: dict[str, Any], manage_stopped: bool) -> bool:
+    """Return True for paused/held torrents Smart Queue may resume later."""
+    if int(t.get('complete') or 0):
+        return False
+    if str(t.get('label') or '') == SMART_QUEUE_LABEL:
+        return True
+    # Note: Paused items are the primary source for filling the queue, regardless of manage_stopped.
+    if bool(t.get('paused')) or str(t.get('status') or '').lower() == 'paused':
+        return True
+    # Note: Stopped items are added only when the user enabled Use stopped torrents.
+    return bool(manage_stopped) and not int(t.get('state') or 0)
+
+
 def check(profile: dict | None = None, user_id: int | None = None, force: bool = False) -> dict[str, Any]:
    profile = profile or active_profile()
    if not profile:
@@ -219,13 +392,37 @@ def check(profile: dict | None = None, user_id: int | None = None, force: bool =
    profile_id = int(profile['id'])
    settings = get_settings(profile_id, user_id)
    if not force and not int(settings.get('enabled') or 0):
-        add_history(profile_id, 'skipped_disabled', [], [], 0, {'enabled': False}, user_id)
-        return {'ok': True, 'enabled': False, 'paused': [], 'resumed': [], 'message': 'Smart Queue disabled'}
+        restored: list[str] = []
+        try:
+            # Note: When Smart Queue is disabled, only technical labels are cleaned up, without starting or pausing torrents.
+            torrents = rtorrent.list_torrents(profile)
+            restored = _cleanup_auto_labels(rtorrent.client_for(profile), profile_id, torrents, set(), bool(settings.get('manage_stopped')))
+        except Exception:
+            restored = []
+        add_history(profile_id, 'skipped_disabled', [], [], 0, {'enabled': False, 'labels_restored': restored}, user_id)
+        return {'ok': True, 'enabled': False, 'paused': [], 'resumed': [], 'labels_restored': restored, 'message': 'Smart Queue disabled'}

    torrents = rtorrent.list_torrents(profile)
    excluded = _excluded_hashes(profile_id, user_id)
-    downloading = [t for t in torrents if not int(t.get('complete') or 0) and int(t.get('state') or 0) and not t.get('paused') and t.get('hash') not in excluded]
-    stopped = [t for t in torrents if not int(t.get('complete') or 0) and (not int(t.get('state') or 0) or t.get('paused')) and t.get('hash') not in excluded]
+    manage_stopped = bool(settings.get('manage_stopped'))
+    def is_managed_hold(t: dict[str, Any]) -> bool:
+        return str(t.get('label') or '') == SMART_QUEUE_LABEL
+
+    # Note: Count Smart Queue slots by d.is_active because Paused can have state=1/open=1 and must not occupy the limit.
+    downloading = [
+        t for t in torrents
+        if _is_running_download_slot(t)
+        and not is_managed_hold(t)
+        and t.get('hash') not in excluded
+    ]
+    # Note: Candidates also include regular Paused items without a label. Otherwise the queue sees only one or two items
+    # and cannot fill the configured target of 100.
+    stopped = [
+        t for t in torrents
+        if t.get('hash') not in excluded
+        and _is_waiting_download_candidate(t, manage_stopped)
+        and not _is_running_download_slot(t)
+    ]
    min_speed = int(settings.get('min_speed_bytes') or 0)
    min_seeds = int(settings.get('min_seeds') or 0)
    stalled_seconds = int(settings.get('stalled_seconds') or 300)
@@ -275,9 +472,9 @@ def check(profile: dict | None = None, user_id: int | None = None, force: bool =
    to_pause: list[dict[str, Any]] = pause_rank[:max(0, len(downloading) - max_active)]
    pause_hashes = {str(t.get('hash') or '') for t in to_pause}

-    # When the cap is not exceeded, stalled downloads can still be rotated out
-    # one-for-one with better stopped candidates while staying within max_active.
-    if candidates:
+    # Note: Stalled rotation runs only when the queue is full. When slots are missing, Smart Queue should
+    # first add missing items instead of pausing existing or incorrectly detected stalled items.
+    if candidates and len(downloading) >= max_active:
        replaceable_stalled = [t for t in stalled if str(t.get('hash') or '') not in pause_hashes]
        for t in replaceable_stalled[:max(0, len(candidates) - len(to_pause))]:
            to_pause.append(t)
@@ -286,27 +483,64 @@ def check(profile: dict | None = None, user_id: int | None = None, force: bool =
    active_after_pause = max(0, len(downloading) - len(to_pause))
    available_slots = max(0, max_active - active_after_pause)
    to_resume = candidates[:available_slots]
+    # Note: Items outside the current start batch are explicitly marked as pending Smart Queue items.
+    to_label_waiting = candidates[available_slots:]

    c = rtorrent.client_for(profile)
+    rtorrent_cap = _ensure_rtorrent_download_cap(c, max_active)
    paused: list[str] = []
    resumed: list[str] = []
    label_failed: list[str] = []
+    start_failed: list[dict[str, str]] = []
+    start_no_effect: list[dict[str, Any]] = []
+    resume_requested: list[str] = []
+    start_results: list[dict[str, Any]] = []
+
    for t in to_pause:
        try:
-            c.call('d.pause', t['hash'])
+            pause_result = rtorrent.pause_hash(c, t['hash'])
+            if not pause_result.get('ok'):
+                raise RuntimeError(pause_result.get('error') or 'pause failed')
            if not _mark_auto_paused(c, profile_id, t):
                label_failed.append(t['hash'])
            paused.append(t['hash'])
        except Exception:
            pass
-    for t in to_resume:
+
+    for t in to_label_waiting:
+        h = str(t.get('hash') or '')
+        if not h or h in pause_hashes:
+            continue
        try:
-            _restore_auto_label(c, profile_id, t['hash'], str(t.get('label') or ''))
-            c.call('d.resume', t['hash'])
-            c.call('d.start', t['hash'])
-            resumed.append(t['hash'])
+            if not _mark_auto_paused(c, profile_id, t):
+                label_failed.append(h)
        except Exception:
-            pass
-    restored = _cleanup_auto_labels(c, profile_id, torrents, set(paused))
-    add_history(profile_id, 'force_check' if force else 'auto_check', paused, resumed, len(torrents), {'excluded': len(excluded), 'enabled': bool(settings.get('enabled')), 'auto_label': SMART_QUEUE_LABEL, 'labels_restored': restored, 'labels_failed': label_failed, 'max_active_downloads': max_active, 'active_before': len(downloading), 'active_after': active_after_pause + len(resumed)}, user_id)
-    return {'ok': True, 'enabled': bool(settings.get('enabled')), 'paused': paused, 'resumed': resumed, 'labels_restored': restored, 'labels_failed': label_failed, 'checked': len(torrents), 'excluded': len(excluded), 'settings': settings}
+            label_failed.append(h)
+
+    # Note: Start the whole candidate batch in one round. Remove the label after an accepted RPC,
+    # because rTorrent may keep some items in its own queue with active=0 despite a valid d.start/d.resume.
+    for t in to_resume:
+        h = str(t.get('hash') or '')
+        if not h:
+            continue
+        try:
+            result = _start_download(c, t)
+            start_results.append(result)
+            resume_requested.append(h)
+        except Exception as exc:
+            start_failed.append({'hash': h, 'error': str(exc)})
+
+    active_verified, start_no_effect = _verify_started_downloads(c, resume_requested)
+    for h in active_verified:
+        _restore_auto_label(c, profile_id, h, None)
+    # Note: History shows only torrents actually unpaused, not just the number of sent commands.
+    resumed = list(active_verified)
+    keep_labels = (
+        set(paused)
+        | {str(t.get('hash') or '') for t in to_label_waiting}
+        | {str(t.get('hash') or '') for t in stopped if str(t.get('label') or '') == SMART_QUEUE_LABEL and str(t.get('hash') or '') not in set(resumed)}
+    )
+    restored = _cleanup_auto_labels(c, profile_id, torrents, keep_labels, manage_stopped)
+    details = {'excluded': len(excluded), 'enabled': bool(settings.get('enabled')), 'auto_label': SMART_QUEUE_LABEL, 'labels_restored': restored, 'labels_failed': label_failed, 'start_failed': start_failed, 'start_no_effect': start_no_effect, 'start_results': start_results, 'resume_requested': resume_requested, 'active_verified': active_verified, 'waiting_labeled': len(to_label_waiting), 'manage_stopped': manage_stopped, 'max_active_downloads': max_active, 'active_before': len(downloading), 'active_after_expected': active_after_pause + len(resumed), 'paused_planned': len(to_pause), 'resumed_planned': len(to_resume), 'rtorrent_cap': rtorrent_cap}
+    add_history(profile_id, 'force_check' if force else 'auto_check', paused, resumed, len(torrents), details, user_id)
+    return {'ok': True, 'enabled': bool(settings.get('enabled')), 'paused': paused, 'resumed': resumed, 'resume_requested': resume_requested, 'waiting_labeled': len(to_label_waiting), 'labels_restored': restored, 'labels_failed': label_failed, 'start_failed': start_failed, 'start_no_effect': start_no_effect, 'active_verified': active_verified, 'rtorrent_cap': rtorrent_cap, 'checked': len(torrents), 'excluded': len(excluded), 'settings': settings}
--- a/pytorrent/services/torrent_cache.py
+++ b/pytorrent/services/torrent_cache.py
@@ -26,9 +26,11 @@ class TorrentCache:
        profile_id = int(profile["id"])
        try:
            rows = rtorrent.list_torrents(profile)
+            with self._lock:
+                old = dict(self._data.get(profile_id, {}))
+            post_check_changes = rtorrent.apply_post_check_policy(profile, rows, old)
            fresh = {t["hash"]: t for t in rows}
            with self._lock:
-                old = self._data.get(profile_id, {})
                added = [v for h, v in fresh.items() if h not in old]
                removed = [h for h in old.keys() if h not in fresh]
                updated = []
@@ -45,7 +47,7 @@ class TorrentCache:
                self._data[profile_id] = fresh
                self._errors[profile_id] = ""
                self._updated_at[profile_id] = time()
-            return {"ok": True, "profile_id": profile_id, "added": added, "updated": updated, "removed": removed}
+            return {"ok": True, "profile_id": profile_id, "added": added, "updated": updated, "removed": removed, "post_check_changes": post_check_changes}
        except Exception as exc:
            with self._lock:
                self._errors[profile_id] = str(exc)
--- a/pytorrent/services/torrent_stats.py
+++ b/pytorrent/services/torrent_stats.py
@@ -0,0 +1,209 @@
+from __future__ import annotations
+
+import json
+import threading
+import time
+from typing import Any
+
+from ..db import connect, utcnow
+from . import rtorrent
+from .torrent_cache import torrent_cache
+
+CACHE_SECONDS = 15 * 60
+_STARTUP_DELAY_SECONDS = 3 * 60
+_STARTED_AT = time.monotonic()
+_LOCK = threading.Lock()
+_BACKGROUND_LOCK = threading.Lock()
+_BACKGROUND_PROFILE_IDS: set[int] = set()
+
+
+def _human_size(value: int | float) -> str:
+    size = float(value or 0)
+    for unit in ("B", "KiB", "MiB", "GiB", "TiB", "PiB"):
+        if abs(size) < 1024 or unit == "PiB":
+            return f"{size:.1f} {unit}" if unit != "B" else f"{int(size)} B"
+        size /= 1024
+    return f"{size:.1f} PiB"
+
+
+def _empty(profile_id: int, error: str = "") -> dict[str, Any]:
+    now = utcnow()
+    return {
+        "profile_id": profile_id,
+        "torrent_count": 0,
+        "complete_count": 0,
+        "incomplete_count": 0,
+        "total_torrent_size": 0,
+        "total_torrent_size_h": _human_size(0),
+        "total_file_size": 0,
+        "total_file_size_h": _human_size(0),
+        "file_count": 0,
+        "seeds_total": 0,
+        "peers_total": 0,
+        "down_rate_total": 0,
+        "up_rate_total": 0,
+        "down_rate_total_h": "0 B/s",
+        "up_rate_total_h": "0 B/s",
+        "sampled_torrents": 0,
+        "errors": [],
+        "error": error,
+        "created_at": now,
+        "updated_at": now,
+        "age_seconds": 0,
+        "stale": True,
+    }
+
+
+def _load_cached(profile_id: int) -> dict[str, Any] | None:
+    with connect() as conn:
+        row = conn.execute("SELECT * FROM torrent_stats_cache WHERE profile_id=?", (profile_id,)).fetchone()
+    if not row:
+        return None
+    payload = json.loads(row.get("payload_json") or "{}")
+    payload["created_at"] = row.get("created_at")
+    payload["updated_at"] = row.get("updated_at")
+    try:
+        payload["age_seconds"] = max(0, int(time.time() - float(row.get("updated_epoch") or 0)))
+    except Exception:
+        payload["age_seconds"] = 0
+    payload["stale"] = payload["age_seconds"] >= CACHE_SECONDS
+    return payload
+
+
+def _save(profile_id: int, payload: dict[str, Any]) -> dict[str, Any]:
+    now = utcnow()
+    payload = dict(payload)
+    payload["updated_at"] = now
+    payload["age_seconds"] = 0
+    payload["stale"] = False
+    with connect() as conn:
+        conn.execute(
+            """
+            INSERT INTO torrent_stats_cache(profile_id,payload_json,created_at,updated_at,updated_epoch)
+            VALUES(?,?,?,?,?)
+            ON CONFLICT(profile_id) DO UPDATE SET
+              payload_json=excluded.payload_json,
+              updated_at=excluded.updated_at,
+              updated_epoch=excluded.updated_epoch
+            """,
+            (profile_id, json.dumps(payload), now, now, time.time()),
+        )
+    return payload
+
+
+def collect(profile: dict) -> dict[str, Any]:
+    """Collect heavier torrent/file statistics on demand or every cache window."""
+    profile_id = int(profile.get("id") or 0)
+    torrents = rtorrent.list_torrents(profile)
+    total_torrent_size = sum(int(t.get("size") or 0) for t in torrents)
+    seeds_total = sum(int(t.get("seeds") or 0) for t in torrents)
+    peers_total = sum(int(t.get("peers") or 0) for t in torrents)
+    down_rate_total = sum(int(t.get("down_rate") or 0) for t in torrents)
+    up_rate_total = sum(int(t.get("up_rate") or 0) for t in torrents)
+    total_file_size = 0
+    file_count = 0
+    errors: list[dict[str, str]] = []
+
+    # Note: File metadata is queried per torrent only during cached statistics refresh, not during every UI poll.
+    for torrent in torrents:
+        h = str(torrent.get("hash") or "")
+        if not h:
+            continue
+        try:
+            files = rtorrent.torrent_files(profile, h)
+            file_count += len(files)
+            total_file_size += sum(int(f.get("size") or 0) for f in files)
+        except Exception as exc:
+            errors.append({"hash": h, "name": str(torrent.get("name") or ""), "error": str(exc)})
+
+    torrent_cache.refresh(profile)
+    payload = {
+        "profile_id": profile_id,
+        "torrent_count": len(torrents),
+        "complete_count": sum(1 for t in torrents if int(t.get("complete") or 0)),
+        "incomplete_count": sum(1 for t in torrents if not int(t.get("complete") or 0)),
+        "total_torrent_size": total_torrent_size,
+        "total_torrent_size_h": _human_size(total_torrent_size),
+        "total_file_size": total_file_size,
+        "total_file_size_h": _human_size(total_file_size),
+        "file_count": file_count,
+        "seeds_total": seeds_total,
+        "peers_total": peers_total,
+        "down_rate_total": down_rate_total,
+        "up_rate_total": up_rate_total,
+        "down_rate_total_h": rtorrent.human_rate(down_rate_total),
+        "up_rate_total_h": rtorrent.human_rate(up_rate_total),
+        "sampled_torrents": len(torrents),
+        "errors": errors[:25],
+        "error": "" if not errors else f"File metadata failed for {len(errors)} torrent(s)",
+        "created_at": utcnow(),
+    }
+    return _save(profile_id, payload)
+
+
+def get(profile: dict | None, force: bool = False) -> dict[str, Any]:
+    if not profile:
+        return _empty(0, "No active rTorrent profile")
+    profile_id = int(profile.get("id") or 0)
+    cached = _load_cached(profile_id)
+    if cached and not force and not cached.get("stale"):
+        return cached
+    if cached and not force:
+        return cached
+    with _LOCK:
+        cached = _load_cached(profile_id)
+        if cached and not force and not cached.get("stale"):
+            return cached
+        return collect(profile)
+
+
+def maybe_refresh(profile: dict | None, force: bool = False) -> dict[str, Any] | None:
+    if not profile:
+        return None
+    if not force and time.monotonic() - _STARTED_AT < _STARTUP_DELAY_SECONDS:
+        return None
+    cached = _load_cached(int(profile.get("id") or 0))
+    if cached and not cached.get("stale") and not force:
+        return cached
+    try:
+        return get(profile, force=True)
+    except Exception:
+        return cached
+
+
+def queue_refresh(socketio, profile: dict | None, force: bool = False, emit_update: bool = True, room: str | None = None) -> dict[str, Any] | None:
+    """Schedule heavier statistics refresh outside the main WebSocket/system poller."""
+    if not profile:
+        return None
+    if not force and time.monotonic() - _STARTED_AT < _STARTUP_DELAY_SECONDS:
+        return _load_cached(int(profile.get("id") or 0))
+
+    profile_id = int(profile.get("id") or 0)
+    cached = _load_cached(profile_id)
+    if cached and not cached.get("stale") and not force:
+        return cached
+
+    with _BACKGROUND_LOCK:
+        if profile_id in _BACKGROUND_PROFILE_IDS:
+            return cached
+        _BACKGROUND_PROFILE_IDS.add(profile_id)
+
+    profile_snapshot = dict(profile)
+
+    def runner():
+        try:
+            # Note: This can query file metadata per torrent, so it never runs inside the fast CPU/RAM/disk poller.
+            stats = get(profile_snapshot, force=True)
+            if emit_update and stats:
+                payload = {"profile_id": profile_id, "stats": stats}
+                socketio.emit("torrent_stats_update", payload, to=room) if room else socketio.emit("torrent_stats_update", payload)
+        except Exception as exc:
+            if emit_update:
+                payload = {"profile_id": profile_id, "ok": False, "error": str(exc)}
+                socketio.emit("torrent_stats_update", payload, to=room) if room else socketio.emit("torrent_stats_update", payload)
+        finally:
+            with _BACKGROUND_LOCK:
+                _BACKGROUND_PROFILE_IDS.discard(profile_id)
+
+    socketio.start_background_task(runner)
+    return cached
--- a/pytorrent/services/torrent_summary.py
+++ b/pytorrent/services/torrent_summary.py
@@ -36,22 +36,28 @@ def _has_error(row: dict) -> bool:
    return bool(message and any(pattern in message for pattern in _ERROR_PATTERNS))


+def _is_checking(row: dict) -> bool:
+    return str(row.get("status") or "") == "Checking" or _number(row, "hashing") > 0
+
+
 def _matches(row: dict, summary_type: str) -> bool:
    status = str(row.get("status") or "")
+    checking = _is_checking(row)
    if summary_type == "all":
        return True
    if summary_type == "downloading":
-        return not bool(row.get("complete")) and bool(row.get("state")) and not bool(row.get("paused"))
+        return not checking and not bool(row.get("complete")) and bool(row.get("state")) and not bool(row.get("paused"))
    if summary_type == "seeding":
-        return status != "Checking" and bool(row.get("complete")) and bool(row.get("state")) and not bool(row.get("paused"))
+        return not checking and bool(row.get("complete")) and bool(row.get("state")) and not bool(row.get("paused"))
    if summary_type == "paused":
-        return bool(row.get("paused")) or status == "Paused"
+        return not checking and (bool(row.get("paused")) or status == "Paused")
    if summary_type == "checking":
-        return status == "Checking" or _number(row, "hashing") > 0
+        return checking
    if summary_type == "error":
        return _has_error(row)
    if summary_type == "stopped":
-        return not bool(row.get("state"))
+        # Note: Stopped count follows the UI filter exactly, so torrents being hash-checked do not inflate an empty Stopped list.
+        return not checking and not bool(row.get("state"))
    return False


--- a/pytorrent/services/websocket.py
+++ b/pytorrent/services/websocket.py
@@ -1,30 +1,52 @@
 from __future__ import annotations

+import threading
 import psutil
-from flask_socketio import emit
+from flask_socketio import emit, join_room, leave_room, disconnect
 from ..config import POLL_INTERVAL
 from .preferences import active_profile, get_profile
 from .torrent_cache import torrent_cache
 from .torrent_summary import cached_summary
-from . import rtorrent, smart_queue, traffic_history, automation_rules
+from . import rtorrent, smart_queue, traffic_history, automation_rules, torrent_stats, auth
+
+
+def _profile_room(profile_id: int) -> str:
+    return f"profile:{int(profile_id)}"
+
+
+def _poller_profiles() -> list[dict]:
+    # Note: Background polling has no browser session, so auth-enabled mode refreshes all profiles and emits only to per-profile rooms.
+    if not auth.enabled():
+        profile = active_profile()
+        return [profile] if profile else []
+    from ..db import connect
+    with connect() as conn:
+        return conn.execute("SELECT * FROM rtorrent_profiles ORDER BY id").fetchall()
+
+
+def _emit_profile(socketio, event: str, payload: dict, profile_id: int) -> None:
+    target = _profile_room(profile_id) if auth.enabled() else None
+    socketio.emit(event, payload, to=target) if target else socketio.emit(event, payload)

 _started = False
+_start_lock = threading.Lock()


 def register_socketio_handlers(socketio):
-    global _started

    def poller():
        tick = 0
        while True:
-            profile = active_profile()
-            if profile:
+            for profile in _poller_profiles():
+                if not profile:
+                    continue
+                pid = int(profile["id"])
                diff = torrent_cache.refresh(profile)
-                heartbeat = {"ok": bool(diff.get("ok")), "profile_id": profile["id"], "tick": tick, "error": diff.get("error", "")}
+                heartbeat = {"ok": bool(diff.get("ok")), "profile_id": pid, "tick": tick, "error": diff.get("error", "")}
                if diff.get("ok") and (diff["added"] or diff["updated"] or diff["removed"]):
-                    socketio.emit("torrent_patch", {**diff, "summary": cached_summary(profile["id"], torrent_cache.snapshot(profile["id"]), force=True)})
+                    _emit_profile(socketio, "torrent_patch", {**diff, "summary": cached_summary(pid, torrent_cache.snapshot(pid), force=True)}, pid)
                elif not diff.get("ok"):
-                    socketio.emit("rtorrent_error", diff)
+                    _emit_profile(socketio, "rtorrent_error", diff, pid)
                try:
                    status = rtorrent.system_status(profile)
                    if bool(profile.get("is_remote")):
@@ -35,41 +57,62 @@ def register_socketio_handlers(socketio):
                        status["ram"] = psutil.virtual_memory().percent
                        status["usage_source"] = "local"
                        status["usage_available"] = True
-                    status["profile_id"] = profile["id"]
-                    traffic_history.record(profile["id"], status.get("down_rate", 0), status.get("up_rate", 0), status.get("total_down", 0), status.get("total_up", 0))
-                    socketio.emit("system_stats", status)
+                    status["profile_id"] = pid
+                    traffic_history.record(pid, status.get("down_rate", 0), status.get("up_rate", 0), status.get("total_down", 0), status.get("total_up", 0))
+                    _emit_profile(socketio, "system_stats", status, pid)
                    heartbeat["ok"] = True
                except Exception as exc:
                    heartbeat["ok"] = False
                    heartbeat["error"] = str(exc)
-                    socketio.emit("rtorrent_error", {"profile_id": profile["id"], "error": str(exc)})
+                    _emit_profile(socketio, "rtorrent_error", {"profile_id": pid, "error": str(exc)}, pid)
+                if tick % max(1, int(15 * 60 / POLL_INTERVAL)) == 0:
+                    # Note: Queue heavier torrent statistics outside the fast system_stats poller.
+                    torrent_stats.queue_refresh(socketio, profile, force=False, room=_profile_room(pid) if auth.enabled() else None)
                if tick % max(1, int(30 / POLL_INTERVAL)) == 0:
                    try:
                        result = smart_queue.check(profile, force=False)
                        if result.get("enabled"):
-                            socketio.emit("smart_queue_update", result)
+                            _emit_profile(socketio, "smart_queue_update", result, pid)
+                            if result.get("paused") or result.get("resumed") or result.get("resume_requested"):
+                                # Note: After Smart Queue changes, refresh cache immediately so the Downloading list does not wait for the next poller cycle.
+                                queue_diff = torrent_cache.refresh(profile)
+                                if queue_diff.get("ok"):
+                                    _emit_profile(socketio, "torrent_patch", {**queue_diff, "summary": cached_summary(pid, torrent_cache.snapshot(pid), force=True)}, pid)
                    except Exception as exc:
-                        socketio.emit("smart_queue_update", {"ok": False, "error": str(exc)})
+                        _emit_profile(socketio, "smart_queue_update", {"ok": False, "error": str(exc)}, pid)
                    try:
                        auto_result = automation_rules.check(profile, force=False)
                        if auto_result.get("applied"):
-                            socketio.emit("automation_update", auto_result)
+                            _emit_profile(socketio, "automation_update", auto_result, pid)
                    except Exception as exc:
-                        socketio.emit("automation_update", {"ok": False, "error": str(exc)})
-                socketio.emit("heartbeat", heartbeat)
+                        _emit_profile(socketio, "automation_update", {"ok": False, "error": str(exc)}, pid)
+                _emit_profile(socketio, "heartbeat", heartbeat, pid)
            tick += 1
            socketio.sleep(POLL_INTERVAL)

+    def ensure_poller_started():
+        global _started
+        with _start_lock:
+            if not _started:
+                # Note: The poller starts with the app, so Smart Queue and automations work without an open UI.
+                socketio.start_background_task(poller)
+                _started = True
+
+    ensure_poller_started()
+
    @socketio.on("connect")
    def handle_connect():
-        global _started
-        if not _started:
-            socketio.start_background_task(poller)
-            _started = True
+        ensure_poller_started()
+        if auth.enabled() and not auth.current_user_id():
+            # Note: Socket.IO uses the same session auth as REST API; unauthenticated clients are disconnected.
+            disconnect()
+            return False
        profile = active_profile()
+        if profile:
+            join_room(_profile_room(profile["id"]))
        emit("connected", {"ok": True, "profile": profile})
        if not profile:
-            # Note: Fresh installs have no rTorrent yet; tell the client to show setup instead of waiting for a snapshot.
+            # Note: Fresh installs or users without profile access get setup state, not another user's snapshot.
            emit("profile_required", {"ok": True, "profiles": []})
            return
        rows = torrent_cache.snapshot(profile["id"])
@@ -77,6 +120,12 @@ def register_socketio_handlers(socketio):

    @socketio.on("select_profile")
    def handle_select_profile(data):
+        if auth.enabled() and not auth.current_user_id():
+            disconnect()
+            return
+        old_profile = active_profile()
+        if old_profile:
+            leave_room(_profile_room(old_profile["id"]))
        profile_id = int((data or {}).get("profile_id") or 0)
        if not profile_id:
            # Note: Ignore empty profile selections created before the first rTorrent profile exists.
@@ -84,8 +133,9 @@ def register_socketio_handlers(socketio):
            return
        profile = get_profile(profile_id)
        if not profile:
-            emit("rtorrent_error", {"error": "Profile does not exist"})
+            emit("rtorrent_error", {"error": "Profile access denied or profile does not exist"})
            return
+        join_room(_profile_room(profile_id))
        diff = torrent_cache.refresh(profile)
        rows = torrent_cache.snapshot(profile_id)
        emit("torrent_snapshot", {"profile_id": profile_id, "torrents": rows, "summary": cached_summary(profile_id, rows, force=True), "error": diff.get("error", "")})
--- a/pytorrent/services/workers.py
+++ b/pytorrent/services/workers.py
@@ -5,7 +5,7 @@ import threading
 import time
 import uuid
 from concurrent.futures import ThreadPoolExecutor
-from . import rtorrent
+from . import rtorrent, auth
 from .preferences import get_profile
 from ..config import WORKERS
 from ..db import connect, utcnow, default_user_id
@@ -23,7 +23,13 @@ def set_socketio(socketio):


 def _emit(name: str, payload: dict):
-    if _socketio:
+    if not _socketio:
+        return
+    profile_id = payload.get("profile_id")
+    if auth.enabled() and profile_id:
+        # Note: Job/socket events are sent only to clients joined to the affected profile room.
+        _socketio.emit(name, payload, to=f"profile:{int(profile_id)}")
+    else:
        _socketio.emit(name, payload)


@@ -97,7 +103,7 @@ def _set_job(job_id: str, status: str, error: str = "", result: dict | None = No


 def enqueue(action_name: str, profile_id: int, payload: dict, user_id: int | None = None, max_attempts: int = 2) -> str:
-    user_id = user_id or default_user_id()
+    user_id = user_id or auth.current_user_id() or default_user_id()
    job_id = uuid.uuid4().hex
    now = utcnow()
    with connect() as conn:
@@ -130,7 +136,7 @@ def _run(job_id: str):
    profile = get_profile(int(job["profile_id"]), int(job["user_id"]))
    if not profile:
        _set_job(job_id, "failed", "rTorrent profile does not exist", finished=True)
-        _emit("job_update", {"id": job_id, "status": "failed", "error": "profile not found"})
+        _emit("job_update", {"id": job_id, "profile_id": job.get("profile_id"), "status": "failed", "error": "profile not found"})
        return
    profile_id = int(profile["id"])
    ordered_lock = None
@@ -150,19 +156,26 @@ def _run(job_id: str):
        with connect() as conn:
            conn.execute("UPDATE jobs SET status='running', attempts=?, started_at=COALESCE(started_at, ?), updated_at=? WHERE id=?", (attempts, utcnow(), utcnow(), job_id))
        _emit("operation_started", {"job_id": job_id, "action": job["action"], "profile_id": profile["id"], "hashes": payload.get("hashes") or [], "hash_count": len(payload.get("hashes") or []), "bulk": len(payload.get("hashes") or []) > 1})
-        _emit("job_update", {"id": job_id, "status": "running", "attempts": attempts})
+        _emit("job_update", {"id": job_id, "profile_id": profile["id"], "status": "running", "attempts": attempts})
        result = _execute(profile, job["action"], payload)
+        fresh = _job_row(job_id)
+        # Note: Emergency cancel keeps a cancelled job from being overwritten when work finishes later.
+        if fresh and fresh["status"] == "cancelled":
+            return
        _set_job(job_id, "done", result=result, finished=True)
        _emit("operation_finished", {"job_id": job_id, "action": job["action"], "profile_id": profile["id"], "hashes": payload.get("hashes") or [], "hash_count": len(payload.get("hashes") or []), "bulk": len(payload.get("hashes") or []) > 1, "result": result})
-        _emit("job_update", {"id": job_id, "status": "done", "result": result})
+        _emit("job_update", {"id": job_id, "profile_id": profile["id"], "status": "done", "result": result})
    except Exception as exc:
        fresh = _job_row(job_id) or {}
        attempts = int(fresh.get("attempts") or 1)
        max_attempts = int(fresh.get("max_attempts") or 2)
+        # Note: Emergency cancel keeps an exception from a cancelled job from moving it back to retry or failed.
+        if fresh and fresh.get("status") == "cancelled":
+            return
        status = "pending" if attempts < max_attempts else "failed"
        _set_job(job_id, status, str(exc), finished=(status == "failed"))
        _emit("operation_failed", {"job_id": job_id, "action": job.get("action"), "profile_id": job.get("profile_id"), "hashes": payload.get("hashes") or [], "error": str(exc)})
-        _emit("job_update", {"id": job_id, "status": status, "error": str(exc), "attempts": attempts})
+        _emit("job_update", {"id": job_id, "profile_id": job.get("profile_id"), "status": status, "error": str(exc), "attempts": attempts})
        if status == "pending":
            _executor.submit(_run, job_id)
    finally:
@@ -182,6 +195,9 @@ def _job_summary(row: dict, payload: dict, result: dict) -> str:
    ctx = payload.get("job_context") or {}
    count = int(ctx.get("hash_count") or len(payload.get("hashes") or []) or result.get("count") or 0)
    parts = []
+    if ctx.get("bulk_label"):
+        # Note: Shows which generated bulk part is being displayed in the job queue.
+        parts.append(f"{ctx.get('bulk_label')} of {ctx.get('bulk_parts')}")
    if count:
        parts.append(("bulk " if count > 1 else "single ") + f"{count} torrent(s)")
    if ctx.get("target_path"):
@@ -215,36 +231,65 @@ def _public_job(row) -> dict:
    return d


+def _job_scope_sql(writable: bool = False) -> tuple[str, tuple]:
+    visible = auth.writable_profile_ids() if writable else auth.visible_profile_ids()
+    if visible is None:
+        return "", ()
+    if not visible:
+        return " WHERE 1=0", ()
+    placeholders = ",".join("?" for _ in visible)
+    return f" WHERE profile_id IN ({placeholders})", tuple(visible)
+
+
 def list_jobs(limit: int = 200, offset: int = 0):
    limit = max(1, min(int(limit or 50), 500))
    offset = max(0, int(offset or 0))
+    where, params = _job_scope_sql()
    with connect() as conn:
-        rows = conn.execute("SELECT * FROM jobs ORDER BY created_at DESC LIMIT ? OFFSET ?", (limit, offset)).fetchall()
-        total = conn.execute("SELECT COUNT(*) AS n FROM jobs").fetchone()["n"]
+        rows = conn.execute(f"SELECT * FROM jobs{where} ORDER BY created_at DESC LIMIT ? OFFSET ?", (*params, limit, offset)).fetchall()
+        total = conn.execute(f"SELECT COUNT(*) AS n FROM jobs{where}", params).fetchone()["n"]
    return {"rows": [_public_job(r) for r in rows], "total": total, "limit": limit, "offset": offset}


 def cancel_job(job_id: str) -> bool:
    row = _job_row(job_id)
-    if not row or row["status"] not in {"pending", "failed"}:
+    if not row or row["status"] not in {"pending", "running"}:
        return False
+    # Note: Emergency cancel is useful only for unfinished jobs; failed/done entries stay available for retry or log cleanup.
    _set_job(job_id, "cancelled", finished=True)
-    _emit("job_update", {"id": job_id, "status": "cancelled"})
+    _emit("job_update", {"id": job_id, "profile_id": row.get("profile_id"), "status": "cancelled"})
    return True


 def clear_jobs() -> int:
+    where, params = _job_scope_sql(writable=True)
+    status_clause = "status NOT IN ('pending', 'running')"
+    sql = f"DELETE FROM jobs{where} AND {status_clause}" if where else f"DELETE FROM jobs WHERE {status_clause}"
    with connect() as conn:
-        cur = conn.execute("DELETE FROM jobs WHERE status NOT IN ('pending', 'running')")
+        cur = conn.execute(sql, params)
        return int(cur.rowcount or 0)


+def emergency_clear_jobs() -> int:
+    # Note: Emergency cleanup first marks active jobs as cancelled, then clears the whole job log list.
+    now = utcnow()
+    where, params = _job_scope_sql(writable=True)
+    status_clause = "status IN ('pending', 'running')"
+    update_sql = f"UPDATE jobs SET status='cancelled', error='Emergency cancelled by user', finished_at=COALESCE(finished_at, ?), updated_at=?{where} AND {status_clause}" if where else "UPDATE jobs SET status='cancelled', error='Emergency cancelled by user', finished_at=COALESCE(finished_at, ?), updated_at=? WHERE status IN ('pending', 'running')"
+    with connect() as conn:
+        conn.execute(update_sql, (now, now, *params) if where else (now, now))
+        cur = conn.execute(f"DELETE FROM jobs{where}", params) if where else conn.execute("DELETE FROM jobs")
+        deleted = int(cur.rowcount or 0)
+    _emit("job_update", {"status": "cleared", "emergency": True})
+    return deleted
+
+
 def retry_job(job_id: str) -> bool:
    row = _job_row(job_id)
    if not row or row["status"] not in {"failed", "cancelled"}:
        return False
    with connect() as conn:
        conn.execute("UPDATE jobs SET status='pending', error='', finished_at=NULL, updated_at=? WHERE id=?", (utcnow(), job_id))
-    _emit("job_update", {"id": job_id, "status": "pending"})
+    _emit("job_update", {"id": job_id, "profile_id": row.get("profile_id"), "status": "pending"})
    _executor.submit(_run, job_id)
    return True
--- a/pytorrent/static/app.js
+++ b/pytorrent/static/app.js
--- a/pytorrent/static/styles.css
+++ b/pytorrent/static/styles.css
--- a/pytorrent/templates/error.html
+++ b/pytorrent/templates/error.html
@@ -0,0 +1,24 @@
+<!doctype html>
+<html lang="en" data-bs-theme="dark">
+<head>
+  <meta charset="utf-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1">
+  <title>pyTorrent {{ code }}</title>
+  <link href="{{ bootstrap_theme_url('default') }}" rel="stylesheet">
+  <link href="{{ frontend_asset_url('fontawesome_css') }}" rel="stylesheet">
+  <link href="{{ static_url('styles.css') }}" rel="stylesheet">
+</head>
+<body class="error-page">
+  <main class="error-card" role="alert">
+    <div class="error-brand"><i class="fa-solid fa-robot"></i> pyTorrent</div>
+    <div class="error-icon" aria-hidden="true"><i class="fa-solid {{ icon }}"></i></div>
+    <p class="error-code">{{ code }}</p>
+    <h1>{{ title }}</h1>
+    <p>{{ message }}</p>
+    <div class="error-actions">
+      <a class="btn btn-primary" href="{{ url_for('main.index') }}"><i class="fa-solid fa-house"></i> Back to dashboard</a>
+      <a class="btn btn-outline-secondary" href="{{ url_for('main.docs') }}"><i class="fa-solid fa-book"></i> API docs</a>
+    </div>
+  </main>
+</body>
+</html>
--- a/pytorrent/templates/index.html
+++ b/pytorrent/templates/index.html
--- a/pytorrent/templates/login.html
+++ b/pytorrent/templates/login.html
@@ -0,0 +1,27 @@
+<!doctype html>
+<html lang="en" data-bs-theme="dark">
+<head>
+  <meta charset="utf-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1">
+  <title>pyTorrent login</title>
+  <link href="{{ bootstrap_theme_url('default') }}" rel="stylesheet">
+  <link href="{{ frontend_asset_url('fontawesome_css') }}" rel="stylesheet">
+  <link href="{{ static_url('styles.css') }}" rel="stylesheet">
+</head>
+<body class="auth-page">
+  <main class="initial-loader-card auth-card">
+    <div class="initial-loader-brand"><i class="fa-solid fa-robot"></i> pyTorrent</div>
+    <div class="auth-lock" aria-hidden="true"><i class="fa-solid fa-lock"></i></div>
+    <h1 class="initial-loader-title">Sign in</h1>
+    <p class="initial-loader-text">Authentication is enabled for this pyTorrent instance.</p>
+    {% if error %}<div class="alert alert-danger auth-alert">{{ error }}</div>{% endif %}
+    <form class="auth-form" method="post">
+      <label class="form-label" for="username">User</label>
+      <input id="username" class="form-control" name="username" autocomplete="username" autofocus>
+      <label class="form-label" for="password">Password</label>
+      <input id="password" class="form-control" name="password" type="password" autocomplete="current-password">
+      <button class="btn btn-primary w-100" type="submit"><i class="fa-solid fa-right-to-bracket"></i> Log in</button>
+    </form>
+  </main>
+</body>
+</html>
--- a/requirements.txt
+++ b/requirements.txt
@@ -4,3 +4,4 @@ python-dotenv>=1.0
 geoip2>=4.8
 psutil>=5.9
 simple-websocket>=1.0
+gunicorn>=22.0
--- a/scripts/download_frontend_libs.py
+++ b/scripts/download_frontend_libs.py
@@ -0,0 +1,113 @@
+#!/usr/bin/env python3
+from __future__ import annotations
+
+import re
+from pathlib import Path
+from urllib.parse import urljoin
+from urllib.request import Request, urlopen
+
+ROOT = Path(__file__).resolve().parents[1]
+LIBS_STATIC_DIR = "libs"
+BOOTSTRAP_VERSION = "5.3.3"
+BOOTSWATCH_VERSION = "5.3.3"
+FONTAWESOME_VERSION = "6.5.2"
+FLAG_ICONS_VERSION = "7.2.3"
+SWAGGER_UI_VERSION = "5"
+SOCKET_IO_VERSION = "4.7.5"
+BOOTSTRAP_THEMES = (
+    "default",
+    "flatly",
+    "litera",
+    "lumen",
+    "minty",
+    "sketchy",
+    "solar",
+    "spacelab",
+    "united",
+    "zephyr",
+)
+STATIC_ASSETS = {
+    "bootstrap_js": {
+        "local": f"{LIBS_STATIC_DIR}/bootstrap/{BOOTSTRAP_VERSION}/js/bootstrap.bundle.min.js",
+        "cdn": f"https://cdn.jsdelivr.net/npm/bootstrap@{BOOTSTRAP_VERSION}/dist/js/bootstrap.bundle.min.js",
+    },
+    "socket_io_js": {
+        "local": f"{LIBS_STATIC_DIR}/socket.io/{SOCKET_IO_VERSION}/socket.io.min.js",
+        "cdn": f"https://cdn.socket.io/{SOCKET_IO_VERSION}/socket.io.min.js",
+    },
+    "fontawesome_css": {
+        "local": f"{LIBS_STATIC_DIR}/fontawesome/{FONTAWESOME_VERSION}/css/all.min.css",
+        "cdn": f"https://cdnjs.cloudflare.com/ajax/libs/font-awesome/{FONTAWESOME_VERSION}/css/all.min.css",
+    },
+    "flag_icons_css": {
+        "local": f"{LIBS_STATIC_DIR}/flag-icons/{FLAG_ICONS_VERSION}/css/flag-icons.min.css",
+        "cdn": f"https://cdn.jsdelivr.net/gh/lipis/flag-icons@{FLAG_ICONS_VERSION}/css/flag-icons.min.css",
+    },
+    "swagger_css": {
+        "local": f"{LIBS_STATIC_DIR}/swagger-ui/{SWAGGER_UI_VERSION}/swagger-ui.css",
+        "cdn": f"https://cdn.jsdelivr.net/npm/swagger-ui-dist@{SWAGGER_UI_VERSION}/swagger-ui.css",
+    },
+    "swagger_js": {
+        "local": f"{LIBS_STATIC_DIR}/swagger-ui/{SWAGGER_UI_VERSION}/swagger-ui-bundle.js",
+        "cdn": f"https://cdn.jsdelivr.net/npm/swagger-ui-dist@{SWAGGER_UI_VERSION}/swagger-ui-bundle.js",
+    },
+}
+URL_RE = re.compile(r"url\((['\"]?)(?!data:)(?!https?:)([^)'\"]+)\1\)")
+
+
+def bootstrap_css_asset(theme: str) -> dict[str, str]:
+    if theme == "default":
+        return {
+            "local": f"{LIBS_STATIC_DIR}/bootstrap/{BOOTSTRAP_VERSION}/css/bootstrap.min.css",
+            "cdn": f"https://cdn.jsdelivr.net/npm/bootstrap@{BOOTSTRAP_VERSION}/dist/css/bootstrap.min.css",
+        }
+    return {
+        "local": f"{LIBS_STATIC_DIR}/bootswatch/{BOOTSWATCH_VERSION}/{theme}/bootstrap.min.css",
+        "cdn": f"https://cdn.jsdelivr.net/npm/bootswatch@{BOOTSWATCH_VERSION}/dist/{theme}/bootstrap.min.css",
+    }
+
+
+def download(url: str, dest: Path) -> None:
+    dest.parent.mkdir(parents=True, exist_ok=True)
+    req = Request(url, headers={"User-Agent": "pyTorrent installer"})
+    with urlopen(req, timeout=60) as response:
+        data = response.read()
+    if not data:
+        raise RuntimeError(f"Empty response for {url}")
+    tmp = dest.with_suffix(dest.suffix + ".tmp")
+    tmp.write_bytes(data)
+    tmp.replace(dest)
+    print(f"OK {dest.relative_to(ROOT)}")
+
+
+def download_css_with_assets(url: str, dest: Path) -> None:
+    download(url, dest)
+    text = dest.read_text(encoding="utf-8", errors="ignore")
+    for match in URL_RE.finditer(text):
+        rel = match.group(2).split("#", 1)[0].split("?", 1)[0]
+        if not rel:
+            continue
+        asset_url = urljoin(url, rel)
+        asset_dest = (dest.parent / rel).resolve()
+        try:
+            asset_dest.relative_to(ROOT)
+        except ValueError:
+            continue
+        if not asset_dest.exists():
+            download(asset_url, asset_dest)
+
+
+def main() -> None:
+    items = list(STATIC_ASSETS.values())
+    items.extend(bootstrap_css_asset(theme) for theme in BOOTSTRAP_THEMES)
+    for item in items:
+        url = item["cdn"]
+        dest = ROOT / "pytorrent" / "static" / item["local"]
+        if dest.suffix == ".css":
+            download_css_with_assets(url, dest)
+        else:
+            download(url, dest)
+
+
+if __name__ == "__main__":
+    main()
--- a/systemd/pytorrent.service
+++ b/systemd/pytorrent.service
@@ -1,16 +1,25 @@
 [Unit]
-Description=pyTorrent web UI for rTorrent
-After=network.target
+Description=pyTorrent Web UI
+After=network-online.target
+Wants=network-online.target

 [Service]
 Type=simple
-WorkingDirectory=/opt/pytorrent
-EnvironmentFile=/opt/pytorrent/.env
-ExecStart=/opt/pytorrent/venv/bin/python /opt/pytorrent/app.py
+#User=root
+#Group=root
+User=pytorrent
+Group=pytorrent
+WorkingDirectory=/opt/pyTorrent
+Environment="PYTHONUNBUFFERED=1"
+EnvironmentFile=/opt/pyTorrent/.env
+# Note: threaded Gunicorn preserves Flask-SocketIO background tasks without running Werkzeug in production.
+ExecStart=/opt/pyTorrent/venv/bin/gunicorn -c /opt/pyTorrent/gunicorn.conf.py --worker-class gthread --workers 1 --threads 32 --bind ${PYTORRENT_HOST}:${PYTORRENT_PORT} --access-logfile - --error-logfile - wsgi:app
 Restart=always
 RestartSec=3
-User=www-data
-Group=www-data
+KillSignal=SIGINT
+TimeoutStopSec=20
+NoNewPrivileges=true
+PrivateTmp=true

 [Install]
 WantedBy=multi-user.target
--- a/wsgi.py
+++ b/wsgi.py
@@ -0,0 +1,4 @@
+from pytorrent import create_app
+
+# Note: Gunicorn imports this object; background Socket.IO tasks still start through create_app().
+app = create_app()
Author	SHA1	Message	Date
Mateusz Gruszczyński	e2017b8344	more conditions in automations	2026-05-06 21:58:49 +02:00
Mateusz Gruszczyński	21cbd5baa6	remove actions in peer	2026-05-06 13:57:26 +02:00
Mateusz Gruszczyński	c19ff17134	offline libs	2026-05-06 11:25:41 +02:00
Mateusz Gruszczyński	6587e74892	about and errorpages	2026-05-06 11:06:08 +02:00
Mateusz Gruszczyński	1baf4a8495	headers	2026-05-06 08:50:40 +02:00
Mateusz Gruszczyński	890e3d1564	headers	2026-05-06 08:47:20 +02:00
Mateusz Gruszczyński	dc1cac4e6f	add auth support	2026-05-06 08:38:07 +02:00
Mateusz Gruszczyński	aea3c92830	fix queue	2026-05-05 19:19:47 +02:00
Mateusz Gruszczyński	45cb6cbb3a	fix queue	2026-05-05 18:52:34 +02:00
Mateusz Gruszczyński	eedfce7207	fix queue	2026-05-05 17:48:21 +02:00
Mateusz Gruszczyński	fc5fedbde2	fix queue	2026-05-05 17:29:45 +02:00
Mateusz Gruszczyński	0f6f9d740c	smart queue fix	2026-05-05 16:01:00 +02:00
Mateusz Gruszczyński	0e0c3359ee	smart queue fix	2026-05-05 15:43:56 +02:00
Mateusz Gruszczyński	912d89abba	smart queue fix	2026-05-05 15:39:32 +02:00
Mateusz Gruszczyński	48bf3ca209	smart queue fix	2026-05-05 15:35:41 +02:00
Mateusz Gruszczyński	c3d12bde46	smart queue fix	2026-05-05 15:25:33 +02:00
Mateusz Gruszczyński	904f36e06f	smart queue fix	2026-05-05 14:43:58 +02:00
Mateusz Gruszczyński	df8750bacb	smart queue fix	2026-05-05 14:30:19 +02:00
Mateusz Gruszczyński	b74b32dd21	smart queue fix	2026-05-05 14:14:47 +02:00
Mateusz Gruszczyński	2d19481c4c	queue changes	2026-05-05 14:03:31 +02:00
Mateusz Gruszczyński	dc78f8fd38	fix in queue	2026-05-05 11:07:54 +02:00
Mateusz Gruszczyński	eb3d743500	fix in queue	2026-05-05 10:50:01 +02:00
Mateusz Gruszczyński	5874f8669d	fix in queue	2026-05-05 10:47:54 +02:00
Mateusz Gruszczyński	08772ddda5	smart queue fix	2026-05-05 09:53:14 +02:00
Mateusz Gruszczyński	3c14a6f510	new functions	2026-05-05 08:21:36 +02:00
Mateusz Gruszczyński	0b580f590e	new functions	2026-05-05 08:08:48 +02:00
Mateusz Gruszczyński	fa0d2f13fe	new functions	2026-05-05 08:00:32 +02:00
gru	4b236c21f8	Merge pull request 'gunicorn' (#1 ) from gunicorn into master Reviewed-on: #1	2026-05-05 07:23:23 +02:00
Mateusz Gruszczyński	0dcdf0e22b	filters and jobs finished date	2026-05-04 22:57:39 +02:00
Mateusz Gruszczyński	879c60d563	execute.capture fix	2026-05-04 21:37:02 +02:00
Mateusz Gruszczyński	d5b7d97528	badge colors	2026-05-04 21:14:17 +02:00
Mateusz Gruszczyński	1ff1525f0b	bulk-part-jobs, and scgi retries	2026-05-04 21:08:30 +02:00
Mateusz Gruszczyński	d55533d78a	bulk-part-jobs	2026-05-04 20:12:26 +02:00
Mateusz Gruszczyński	e9940bf16c	force kill job	2026-05-04 19:21:09 +02:00
Mateusz Gruszczyński	dca7389a1a	gunicorn	2026-05-04 10:57:03 +02:00
Mateusz Gruszczyński	31bba1269d	gunicorn	2026-05-04 10:43:31 +02:00