media-dlp

Automated media downloader that uses Playwright (browser automation) to load streaming sites, detect video streams via network interception, and download them with yt-dlp.

Why?

Streaming sites load video players that fetch HLS/DASH manifests from CDNs. media-dlp:

1. Opens the site in a real Firefox window (cookies, referrer, fingerprint – the whole deal)
2. Watches all network traffic to find the video stream URL
3. Hands that URL to yt-dlp for the actual download (progress bars, resume support, quality selection)

No sneaky DOM scraping, no brittle CSS selectors against popups – just raw network-level video detection.

Features

• ✅ Network-level detection – catches HLS, MP4, DASH, WebM streams
• ✅ TMDB integration – provide a TMDB movie ID instead of a full URL
• ✅ Source chaining – tries multiple providers (mapple → cineby → …)
• ✅ yt-dlp download – progress output, quality selection, resume
• ✅ Tab muting – no audio from the automated browser window
• ✅ CLI + Python API – use from the terminal or import as a library
• ✅ Browser cookies – passes session cookies to yt-dlp for authenticated streams

Installation

pip install media-dlp

You also need the Playwright Firefox browser:

playwright install firefox

TMDB API Key (for --tmdb-id)

Set the environment variable:

export TMDB_API_KEY="your_key_here"

Get a free key at themoviedb.org/settings/api.

CLI Usage

media-dlp --url https://mapple.club/watch/movie/1034716

media-dlp --tmdb-id 1034716 --quality 1080p --output ~/Movies

media-dlp --tmdb-id 1034716 --providers mapple,cineby --filename "Dune 2"

Options

Python API

from media_dlp import download_movie, list_providers

# List available source providers
print(list_providers())

# Download by URL
result = download_movie(
    url="https://mapple.club/watch/movie/1034716",
    quality="1080p",
    output_dir="./movies",
    mute=True,
    verbose=True,
)
print(result)  # e.g. ./movies/movie_12345.mp4

# Download by TMDB ID
result = download_movie(
    tmdb_id=1034716,
    quality="best",
    providers="mapple,cineby",
    filename="Dune Part Two",
)

How It Works

┌─────────────────────────────────────────────────────┐
│  1. Playwright launches Firefox (persistent profile)│
│  2. Navigate to movie URL / resolve via TMDB        │
│  3. Click play button if needed                     │
│  4. Intercept ALL network requests/responses        │
│  5. Filter for HLS manifests, MP4 URLs, DASH MPDs   │
│  6. Pick best matching quality                      │
│  7. Spawn yt-dlp with: detected URL, cookies,       │
│     referrer, user-agent, origin headers             │
│  8. Stream progress to console                      │
│  9. File appears in output directory                │
└─────────────────────────────────────────────────────┘

The detection works at the network level, not the DOM level. This means it doesn't matter how the video player constructs its URLs – if a HTTP response carries video content, media-dlp sees it.

Source Providers

Providers are configured in media_dlp/sources.py. Each maps a TMDB ID to a streaming-site URL:

When you pass --providers mapple,cineby, it tries mapple first; if that fails to produce a video it moves to cineby (future enhancement – currently it takes the first working URL).

Development

git clone https://github.com/jeevan/media-dlp
cd media-dlp
pip install -e .
playwright install firefox

Testing

media-dlp --url https://mapple.club/watch/movie/1034716 --verbose

FAQ

Q: Does this work with DRM-protected content?
A: No. Widevine-level DRM cannot be bypassed. The tool relies on plain HLS/DASH/HTTP streams.

Q: The download is slow
A: Increase the --timeout value. Feature-length movies can take 10-30 minutes.

Q: Can I run this headless?
A: Not currently – most streaming sites require a visible browser to serve video.

Q: The browser opens but no video is detected
A: Run with --verbose to see network traffic. The video player may need manual interaction (e.g. a "Play" click) which the tool attempts automatically via several common CSS selectors.

License

MIT