Scripts/extract_track23.py

import webdataset as wds
import os
import json
import glob
import argparse

def extract_webdataset(webdataset_path, output_dir):
    """
    Extract all WebDataset tar files back to original directory structure
    
    Args:
        webdataset_path: Path to the directory containing WebDataset tar files
        output_dir: Directory where to extract the dataset
    """
    os.makedirs(output_dir, exist_ok=True)
    
    # Get all tar files in the directory
    tar_files = sorted(glob.glob(os.path.join(webdataset_path, "*.tar")))
    if not tar_files:
        print(f"No tar files found in {webdataset_path}")
        return
    
    print(f"Found {len(tar_files)} tar files")
    total_sample_count = 0
    
    for tar_file in tar_files:
        print(f"\nProcessing {os.path.basename(tar_file)}")
        
        # Initialize WebDataset reader for current tar file
        dataset = wds.WebDataset(
            tar_file,
            shardshuffle=False,
            empty_check=False
        )
        
        sample_count = 0
        
        for sample in dataset:
            try:
                # Get path information
                path_info = json.loads(sample['path_info.json'])
                category = path_info['category']
                scene = path_info['scene']
                style = path_info['style']
                resolution = path_info['resolution']
                view = path_info['view']
                
                # Create base directory structure
                base_dir = os.path.join(output_dir, category, scene, style, resolution)
                
                # Create image type directories
                depth_dir = os.path.join(base_dir, "Depth_images")
                id_dir = os.path.join(base_dir, "ID_images")
                rgb_dir = os.path.join(base_dir, "RGB_images")
                
                # Create view subdirectories under each image type
                depth_view_dir = os.path.join(depth_dir, view)
                id_view_dir = os.path.join(id_dir, view)
                rgb_view_dir = os.path.join(rgb_dir, view)
                
                # Create all necessary directories
                os.makedirs(depth_view_dir, exist_ok=True)
                os.makedirs(id_view_dir, exist_ok=True)
                os.makedirs(rgb_view_dir, exist_ok=True)
                
                # Get base filename from key (excluding view)
                base_name = sample['__key__'].split('/')[-2]  # Changed to get correct part of key
                
                # Save Depth files
                depth_exr_path = os.path.join(depth_view_dir, f"{base_name}.exr")
                depth_json_path = os.path.join(depth_view_dir, f"{base_name}.json")
                
                with open(depth_exr_path, 'wb') as f:
                    f.write(sample['depth.exr'])
                with open(depth_json_path, 'w') as f:
                    json.dump(json.loads(sample['depth.json']), f, indent=4)
                
                # Save ID files
                id_png_path = os.path.join(id_view_dir, f"{base_name}.png")
                id_json_path = os.path.join(id_view_dir, f"{base_name}.json")
                
                with open(id_png_path, 'wb') as f:
                    f.write(sample['id.png'])
                with open(id_json_path, 'w') as f:
                    json.dump(json.loads(sample['id.json']), f, indent=4)
                
                # Save RGB files
                rgb_png_path = os.path.join(rgb_view_dir, f"{base_name}.png")
                rgb_json_path = os.path.join(rgb_view_dir, f"{base_name}.json")
                
                with open(rgb_png_path, 'wb') as f:
                    f.write(sample['rgb.png'])
                with open(rgb_json_path, 'w') as f:
                    json.dump(json.loads(sample['rgb.json']), f, indent=4)
                
                sample_count += 1
                if sample_count % 100 == 0:
                    print(f"Extracted {sample_count} samples from current tar...")
                    
            except Exception as e:
                print(f"Error extracting sample {sample['__key__']}: {str(e)}")
                continue
        
        total_sample_count += sample_count
        print(f"Completed {os.path.basename(tar_file)}: {sample_count} samples extracted")
    
    print(f"\nExtraction complete. Total samples extracted: {total_sample_count}")

def verify_extraction(extracted_dir):
    """
    Verify the extraction by counting files
    """
    print("\nVerifying extraction...")
    
    def count_files(directory):
        exr_count = 0
        png_count = 0
        json_count = 0
        
        for root, _, files in os.walk(directory):
            for file in files:
                if file.endswith('.exr'):
                    exr_count += 1
                elif file.endswith('.png'):
                    png_count += 1
                elif file.endswith('.json'):
                    json_count += 1
        
        return {
            'exr': exr_count,
            'png': png_count,
            'json': json_count
        }
    
    counts = count_files(extracted_dir)
    
    print("\nExtracted file counts:")
    print(f"  EXR files: {counts['exr']}")
    print(f"  PNG files: {counts['png']}")
    print(f"  JSON files: {counts['json']}")
    
    if all(count > 0 for count in counts.values()):
        print("\nAll file types present in extraction.")
    else:
        print("\nWARNING: Some file types are missing!")

def parse_arguments():
    parser = argparse.ArgumentParser(description='Extract WebDataset tar files')
    parser.add_argument('--input_path', '-i', required=True,
                        help='Input directory containing WebDataset tar files')
    parser.add_argument('--output_path', '-o', required=True,
                        help='Output directory for extracted files')
    parser.add_argument('--verify', '-v', action='store_true',
                        help='Verify the extraction after completion')
    return parser.parse_args()

if __name__ == "__main__":
    args = parse_arguments()
    
    print(f"Input directory: {args.input_path}")
    print(f"Output directory: {args.output_path}")
    
    print("\nStarting extraction...")
    extract_webdataset(args.input_path, args.output_path)
    
    if args.verify:
        verify_extraction(args.output_path)