Download all images from an Internet Archive collection and write the caption to a text file

ia_image_downloader.py

import internetarchive as ia
from pathlib import Path
import argparse
import time

config = dict(general=dict(secure=False))

def download_collection(collection_name, output_dir):
    search = ia.search_items(f'collection:{collection_name}', config=config)
    dir = Path(output_dir) / collection_name
    dir.mkdir(exist_ok=True)
    for result in search:
        download_item(result['identifier'], dir)
        time.sleep(0.1)

def download_item(item_id, dir):
    item = ia.get_item(item_id, config=config)
    meta = item.metadata
    title = meta['title']
    if isinstance(title, list): # not sure why a list is sometimes returned, so just move on
        return
    description = meta.get('description', None)
    formats = ['JPEG']
    for file in item.files:
        if file['format'] in formats:
            filename = file['name']
            try:
                item.download(files=filename, \
                    formats=file['format'], \
                    destdir=dir,
                    no_directory=True, \
                    verbose=True, \
                    ignore_existing=True, \
                    retries=2)
            except Exception as e:
                print(f'Download failed: {e}')
                return
            caption = dir / Path(filename).with_suffix('.txt')
            with open(caption, 'w', encoding='utf-8') as f:
                if description and not isinstance(description, list):
                    f.write(description)
                else:
                    f.write(title)
            return
    print(f'no image for {item_id}')

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('collection', help='Collection name')
    parser.add_argument('output_dir', help='Output directory')
    args = parser.parse_args()

    print(f'Downloading collection {args.collection} to {args.output_dir}')

    download_collection(args.collection, args.output_dir)