import json from pyquery import PyQuery as pq def do(): data = json.loads(open("radia.json").read()) outfile = "padma.json" ret = {} for d in data: # print get_padma_ids(data[d]['link']) ret[d] = get_padma_ids(data[d]['link']) f = open(outfile, "w") f.write(json.dumps(ret, indent=2)) f.close() print "generated padma.json from radia.json - currently returns empty if link points to a single video. this should be fixed soon" def get_padma_ids(link): ids = [] if link.startswith("http://powertapes.pad.ma/"): jq = pq(url=link) elems = jq.find('.iconText') for e in elems: id = e.get("id").replace("iconText", "") ids.append(id) return ids