java2-extention 1.0.8

Async video downloader with Cloudflare bypass, parallel chunked transfers, and multi-source HTML extraction. Termux / ARMv7 optimised.

java2-extention

Async video downloader with Cloudflare bypass, parallel chunked transfers, and multi-source HTML extraction.
Optimised for Termux / ARMv7 / 2 GB RAM.

---

Install

pip install java2-extention

With faster optional deps (lxml, aiodns, brotli):

pip install "java2-extention[fast]"

Termux one-liner

pkg update && pkg install python python-pip
pip install java2-extention

---

CLI usage

# Download from a video page (auto-extracts best quality)
java2dl "https://example.com/some-video-page" ./downloads/

# Download a known direct URL
java2dl "https://cdn.example.com/video.mp4" ./video.mp4

# Skip Cloudflare warmup (clean CDN, no challenge)
java2dl "https://cdn.example.com/video.mp4" . --no-warm

# Verbose debug output
java2dl "https://example.com/page" . --debug

---

Python API

import asyncio
from java2_extention import (
    VideoExtractor,
    pick_best,
    rank_qualities,
    fetch_html,
    warm_download_url,
    download_to_file,
    ProgressTracker,
)
from java2_extention.downloader import make_session

async def main():
    page_url = "https://example.com/video-page"

    # 1. Fetch the page HTML (Cloudflare-aware, sync → thread)
    html = await asyncio.to_thread(fetch_html, page_url)

    # 2. Extract all video URLs from the page
    extractor = VideoExtractor(base_url=page_url)
    entries = extractor.extract(html)

    # 3. Pick the best direct download
    pick = pick_best(entries)
    if not pick:
        print("No downloadable video found")
        return

    print(f"Best: {pick.url}  quality={pick.quality}")

    # 4. Warm the CDN (solve Cloudflare, harvest cookies)
    final_url, cookies = await warm_download_url(pick.url, page_url)

    # 5. Download with a progress callback
    async def on_progress(downloaded: int, total):
        if total:
            print(f"\r{downloaded/total*100:.1f}%", end="", flush=True)

    tracker = ProgressTracker(callback=on_progress)

    async with make_session(cookies) as session:
        size = await download_to_file(
            session,
            final_url,
            dest=__import__("pathlib").Path("video.mp4"),
            progress=tracker,
            referer=page_url,
            warm=False,   # already warmed above
        )
    print(f"\nDone — {size / 1024 / 1024:.1f} MB")

asyncio.run(main())

---

How it works

Step	What happens
Extract	VideoExtractor tries 5 strategies in order: OG video meta, Twitter player meta, <video>/<source> tags, inline <script> JSON scan, <link rel> tags.
Rank	Each URL is scored by format (mp4 > webm > m3u8), resolution label, and penalises preview/thumbnail URLs.
Warm	warm_download_url runs cloudscraper (HEAD then range probe) to solve Cloudflare challenges and harvest cookies.
Download	If the server supports byte ranges and the file is > 2 MB, splits into N parallel chunks. Worker count is capped by available RAM. Falls back to streaming if ranges are unsupported.

---

Environment variables

Variable	Default	Purpose
J2_CHUNK_SIZE	524288 (512 KB)	Bytes per read chunk
J2_DL_WORKERS	8	Max parallel range workers
J2_PROGRESS_SEC	2	Progress callback interval (seconds)

Lower J2_DL_WORKERS to 2–4 on a 2 GB ARMv7 device if you see OOM.

---

License

MIT