irs.py

#!/usr/bin/env python
import sys
import ujson

# Script/Library created by Mike McCabe to do multiple
# metadata fetches in parallel
# Available here: https://gist.github.com/3784845
from parallel_md_get import metadata_record_iterator


def search_for_foundation(query, manifest):
    for form in manifest:
        if query.lower() in form['foundation'].lower():
            yield form

if __name__ == '__main__':

    ids = open(sys.argv[1])
    results = metadata_record_iterator(ids, workers=20)

    for i, id, md_json in results:
        metadata = ujson.loads(md_json)
        for result in search_for_foundation('cancer', metadata['manifest']):
            if result:
                print id, result