droid/scripts/create_chunks.py

#!/usr/bin/env python3
"""
Divide valid episodes into N equal chunks for distributed processing.
"""

import json
from pathlib import Path
import argparse

def main():
    parser = argparse.ArgumentParser()
    parser.add_argument('--input', type=str, required=True,
                        help='Input JSON file with valid episodes')
    parser.add_argument('--num-chunks', type=int, default=16,
                        help='Number of chunks to create')
    parser.add_argument('--output-dir', type=str, default='/mnt/kevin/data/droid_chunks',
                        help='Output directory for chunk files')
    args = parser.parse_args()

    # Load valid episodes
    with open(args.input) as f:
        data = json.load(f)

    episodes = data['episodes']
    stats = data['stats']

    print(f"\nLoaded {len(episodes)} valid episodes")
    print(f"Dividing into {args.num_chunks} chunks...")

    # Calculate chunk size
    chunk_size = len(episodes) // args.num_chunks
    remainder = len(episodes) % args.num_chunks

    # Create output directory
    output_dir = Path(args.output_dir)
    output_dir.mkdir(parents=True, exist_ok=True)

    # Divide into chunks
    chunks = []
    start_idx = 0
    for chunk_id in range(args.num_chunks):
        # Distribute remainder across first chunks
        current_chunk_size = chunk_size + (1 if chunk_id < remainder else 0)
        end_idx = start_idx + current_chunk_size

        chunk_episodes = episodes[start_idx:end_idx]

        # Calculate total frames for this chunk
        total_frames = sum(ep['num_frames'] for ep in chunk_episodes)

        chunk_data = {
            'chunk_id': chunk_id,
            'episodes': chunk_episodes,
            'num_episodes': len(chunk_episodes),
            'total_frames': total_frames,
            'episode_indices': [ep['episode_idx'] for ep in chunk_episodes]
        }

        # Save chunk file
        chunk_file = output_dir / f'chunk_{chunk_id:02d}.json'
        with open(chunk_file, 'w') as f:
            json.dump(chunk_data, f, indent=2)

        chunks.append(chunk_data)
        start_idx = end_idx

        print(f"  Chunk {chunk_id:2d}: {len(chunk_episodes):5d} episodes, "
              f"{total_frames:7d} frames ({chunk_file})")

    # Save summary
    summary = {
        'num_chunks': args.num_chunks,
        'total_episodes': len(episodes),
        'total_frames': sum(ep['num_frames'] for ep in episodes),
        'original_stats': stats,
        'chunks': [
            {
                'chunk_id': c['chunk_id'],
                'num_episodes': c['num_episodes'],
                'total_frames': c['total_frames']
            }
            for c in chunks
        ]
    }

    summary_file = output_dir / 'summary.json'
    with open(summary_file, 'w') as f:
        json.dump(summary, f, indent=2)

    print(f"\n{'='*80}")
    print(f"Created {args.num_chunks} chunks")
    print(f"  Total episodes: {len(episodes)}")
    print(f"  Avg episodes per chunk: {len(episodes) / args.num_chunks:.1f}")
    print(f"  Total frames: {sum(ep['num_frames'] for ep in episodes):,}")
    print(f"  Avg frames per chunk: {sum(ep['num_frames'] for ep in episodes) / args.num_chunks:,.0f}")
    print(f"\nChunk files saved to: {output_dir}/")
    print(f"Summary: {summary_file}")
    print(f"{'='*80}")

if __name__ == '__main__':
    main()