get_itemsets.py

get_itemsets.py

import fileinput
import json
import urllib2

def main():
	submit_object = {
		"threshold": 12,
		"documents": []
	}

	lines = []
	for line in fileinput.input():
		lines.append(line)
		link, hand, generated = line.split("\t")
		submit_object["documents"].append(generated.strip().split(","))

	request = urllib2.Request('https://api.algorithmia.com/api/GrantMarshall/FrequentItemsets/0.1.1')
	request.add_header('Content-Type', 'application/json')
	request.add_header('Authorization', '###')  # Put you authorization here
	request.add_header('Accept', 'application/json')
	response = urllib2.urlopen(request, json.dumps(submit_object))
	response_string = response.read()
	print(response_string)
	response_object = json.loads(response_string)

	large_itemsets = []
	for itemset in response_object["result"]["itemsets"]:
		if len(itemset) > 1:
			large_itemsets.append(set(itemset))

	for line in lines:
		link, hand, generated = line.split("\t")
		new_tags = generated.split(",")
		generated_tags = set(generated.split(","))
		for itemset in large_itemsets:
			if set(itemset) <= generated_tags:
				new_tags.append(" ".join(itemset))
		print("{}\t{}\t{}".format(link, hand, ",".new_tags))


if __name__ == '__main__':
	main()