#!/usr/bin/env python3
"""
Example utility: How to download Creative Commons videos using yt-dlp
from the metadata.jsonl / metadata.csv dataset.

Links and metadata only, no video files included. Each video keeps its original
Creative Commons license; credit the creators.
"""

import json
import argparse
import subprocess
import os

def download_videos(jsonl_path: str, output_dir: str = "downloads", max_count: int = 5, category: str = None):
    os.makedirs(output_dir, exist_ok=True)
    count = 0

    with open(jsonl_path, "r", encoding="utf-8") as f:
        for line in f:
            item = json.loads(line)
            if category and item.get("category") != category:
                continue

            url = item["url"]
            cat = item["category"]
            title = item["title"]
            vid_id = item["video_id"]

            print(f"\n[{count + 1}/{max_count}] Downloading CC video: {title}")
            print(f"URL: {url} | Category: {cat}")

            dest_folder = os.path.join(output_dir, cat)
            os.makedirs(dest_folder, exist_ok=True)
            output_template = os.path.join(dest_folder, "%(id)s.%(ext)s")

            # yt-dlp command: 720p max to save bandwidth
            cmd = [
                "yt-dlp",
                "-f", "bestvideo[height<=720][ext=mp4]+bestaudio[ext=m4a]/best[height<=720]/best",
                "-o", output_template,
                "--max-filesize", "150M",
                "--no-playlist",
                url
            ]

            try:
                subprocess.run(cmd, check=True)
                count += 1
            except subprocess.CalledProcessError as e:
                print(f"Error downloading {vid_id}: {e}")

            if count >= max_count:
                break

    print(f"\nDone! Downloaded {count} sample videos to '{output_dir}'.")

if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Download sample CC videos using yt-dlp")
    parser.add_argument("--jsonl", type=str, default="metadata.jsonl", help="Path to metadata.jsonl")
    parser.add_argument("--count", type=int, default=3, help="Number of videos to download")
    parser.add_argument("--category", type=str, default="nepal_content", help="Category filter (optional)")
    parser.add_argument("--out", type=str, default="downloads", help="Output directory")
    args = parser.parse_args()

    download_videos(args.jsonl, args.out, args.count, args.category)