naeb-page-counts.py

#!/usr/bin/env python3

import internetarchive

ia = internetarchive.get_session()

print("ids,pages")

for result in ia.search_items('collection:mediahistory creator:National Association of Educational Broadcasters'):
    ia_id = result['identifier']
    item = ia.get_item(ia_id)
    metadata = item.metadata

    # i'm not sure why, but IA's imagecount always seems to be off by one 
    # greater than the actual number?
    cols = [metadata['identifier'], str(int(metadata.get('imagecount', 1)) - 1)]
    print(','.join(cols))