Graph edge lookup specifically implemented for relation lookup on phone numbers

map_relation.py

from json import loads
from random import choice, randint, sample
import timeit



# With this model it is imperative to use integer form of phone number instead of string
# We will use median contact frequency distribution from customer data
# Maximum contacts: 14551
# Minmum: 233
# Median: 393



def create_single_phone_number(s=1000000000, e=9999999999):
	return randint(s, e)


def create_phone_number(n):
	for v in range(n):
		yield create_single_phone_number()

def create_contacts(contacts, _min, _max, _median):
	#We will simplify by using median for now
	corpus = {}
	for c in contacts:
		c1 = sample(contacts, _median) # It should remove self-reference but that does not effect performance
		corpus[c] = c1
	return corpus


def convert_to_long(contact, pair):
	return(contact * 10000000000) + pair # the phone number is ten digits
	
def convert_contacts_to_long(contacts_dict):
	for k, v in contacts_dict.items():
		for j in v:
			yield convert_to_long(k, j)


def lookup_in_dictionary(contacts_dict, test_sample):
	l = []
	for t, v in test_sample:
		if t in contacts_dict:
			if contacts_dict[t][0] == v[0]:
				l.append(0)
			else:
				continue
	return len(l)

def lookup_in_set(contacts_set, test_sample_set):
	return len(contacts_set.intersection(test_sample_set))

l = list(create_phone_number(10000))
print("Created the phone number pool")
contacts_dict = create_contacts(l, 233, 14551, 393)
print("Created the phone contacts datastructure")
test_sample = sample(contacts_dict.items(), 10)
contacts_set = set(convert_contacts_to_long(contacts_dict))
test_sample_set = set([convert_to_long(k, v[0]) for k, v in test_sample])

if __name__ == '__main__':
	print("Executed using integer dictionary/string dictionary will be slower than this")
	print(timeit.timeit("lookup_in_dictionary(contacts_dict, test_sample)", setup="from __main__ import lookup_in_dictionary, contacts_dict, test_sample"))
	print("Executed using set intersection")
	print(timeit.timeit("lookup_in_set(contacts_set, test_sample_set)", setup="from __main__ import lookup_in_set, contacts_set, test_sample_set"))