bloom-filter/test-bloom-filter

#!/usr/bin/python

# pylint: disable=W0402
# W0402: We want the deprecated string module, for a use that isn't deprecated

'''Unit tests for bloom_filter_mod'''

import sys
import random

import bloom_filter_mod

CHARACTERS = 'abcdefghijklmnopqrstuvwxyz1234567890'

def my_range(maximum):
	'''A range function with consistent semantics on 2.x and 3.x'''
	value = 0
	while True:
		if value >= maximum:
			break
		yield value
		value += 1

def test(description, values, trials, error_rate):
	'''Some quick automatic tests for the bloom filter class'''

	all_good = True

	bloom_filter = bloom_filter_mod.Bloom_filter(ideal_num_elements_n=trials * 2, error_rate_p=error_rate)
	#print(repr(bloom_filter))

	print '\n', description, bloom_filter.num_words, bloom_filter.num_probes_k

	print 'adding'
	for include in values.generator():
		bloom_filter.add(include)

	print 'testing all known members'''
	include_in_count = sum(include in bloom_filter for include in values.generator())
	if include_in_count == values.length():
		# Good
		pass
	else:
		sys.stderr.write('Include count bad: %s, %d\n' % (include_in_count, values.length()))
		all_good = False

	print 'testing random non-members'''
	false_positives = 0
	for trialno in my_range(trials):
		if trialno % 100000 == 0:
			sys.stderr.write('trialno countdown: %d\n' % (trials-trialno))
		#dummy = trialno
		while True:
			candidate = ''.join(random.sample(CHARACTERS, 5))
			# If we accidentally found a member, try again
			if values.within(candidate):
				continue
			if candidate in bloom_filter:
				#print 'We erroneously think %s is in the filter' % candidate
				false_positives += 1
			break

	actual_error_rate = float(false_positives) / trials

	if actual_error_rate > error_rate:
		sys.stderr.write('%s: Too many false positives: actual: %s, expected: %s\n' % (
			sys.argv[0],
			actual_error_rate,
			error_rate,
			))
		all_good = False

	return all_good

class States:
	'''Generate the USA's state names'''

	def __init__(self):
		pass

	states = '''Alabama Alaska Arizona Arkansas California Colorado Connecticut
		Delaware Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas
		Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota
		Mississippi Missouri Montana Nebraska Nevada NewHampshire NewJersey
		NewMexico NewYork NorthCarolina NorthDakota Ohio Oklahoma Oregon
		Pennsylvania RhodeIsland SouthCarolina SouthDakota Tennessee Texas Utah
		Vermont Virginia Washington WestVirginia Wisconsin Wyoming'''.split()

	@staticmethod
	def generator():
		'''Generate the states'''
		for state in States.states:
			yield state

	@staticmethod
	def within(value):
		'''Is the vaoue in our list of states?'''
		return value in States.states

	@staticmethod
	def length():
		'''What is the length of our contained values?'''
		return len(States.states)

def random_string():
	'''Generate a random, 10 character string - for testing purposes'''
	list_ = []
	for chrno in range(10):
		dummy = chrno
		character = CHARACTERS[int(random.random() * len(CHARACTERS))]
		list_.append(character)
	return ''.join(list_)

class Random_content:
	'''Generated a bunch of random strings in sorted order'''

	random_content = [ random_string() for dummy in range(1000) ]

	def __init__(self):
		pass

	@staticmethod
	def generator():
		'''Generate all values'''
		for item in Random_content.random_content:
			yield item

	@staticmethod
	def within(value):
		'''Test for membership'''
		return value in Random_content.random_content

	@staticmethod
	def length():
		'''How many members?'''
		return len(Random_content.random_content)

class Evens:
	'''Generate a bunch of even numbers'''

	def __init__(self, maximum):
		self.maximum = maximum

	def generator(self):
		'''Generate all values'''
		for value in my_range(self.maximum):
			if value % 2 == 0:
				yield str(value)

	def within(self, value):
		'''Test for membership'''
		try:
			int_value = int(value)
		except ValueError:
			return False

		if int_value >= 0 and int_value < self.maximum and int_value % 2 == 0:
			return True
		else:
			return False

	def length(self):
		'''How many members?'''
		return self.maximum // 2

def and_test():
	'''Test the & operator'''

	all_good = True

	abc = bloom_filter_mod.Bloom_filter(ideal_num_elements_n=100, error_rate_p=0.01)
	for character in [ 'a', 'b', 'c' ]:
		abc += character

	bcd = bloom_filter_mod.Bloom_filter(ideal_num_elements_n=100, error_rate_p=0.01)
	for character in [ 'b', 'c', 'd' ]:
		bcd += character

	abc_and_bcd = abc
	abc_and_bcd &= bcd

	if 'a' in abc_and_bcd:
		sys.stderr.write('a in abc_and_bcd, but should not be')
		all_good = False
	if not 'b' in abc_and_bcd:
		sys.stderr.write('b not in abc_and_bcd, but should be')
		all_good = False
	if not 'c' in abc_and_bcd:
		sys.stderr.write('c not in abc_and_bcd, but should be')
		all_good = False
	if 'd' in abc_and_bcd:
		sys.stderr.write('d in abc_and_bcd, but should not be')
		all_good = False

	return all_good

def or_test():
	'''Test the | operator'''

	all_good = True

	abc = bloom_filter_mod.Bloom_filter(ideal_num_elements_n=100, error_rate_p=0.01)
	for character in [ 'a', 'b', 'c' ]:
		abc += character

	bcd = bloom_filter_mod.Bloom_filter(ideal_num_elements_n=100, error_rate_p=0.01)
	for character in [ 'b', 'c', 'd' ]:
		bcd += character

	abc_and_bcd = abc
	abc_and_bcd |= bcd

	if not 'a' in abc_and_bcd:
		sys.stderr.write('a not in abc_and_bcd, but should be')
		all_good = False
	if not 'b' in abc_and_bcd:
		sys.stderr.write('b not in abc_and_bcd, but should be')
		all_good = False
	if not 'c' in abc_and_bcd:
		sys.stderr.write('c not in abc_and_bcd, but should be')
		all_good = False
	if not 'd' in abc_and_bcd:
		sys.stderr.write('d not in abc_and_bcd, but should be')
		all_good = False
	if 'e' in abc_and_bcd:
		sys.stderr.write('e in abc_and_bcd, but should be')
		all_good = False

	return all_good
	
def main():
	'''Unit tests for Bloom_filter class'''

	all_good = True

	all_good &= test('states', States(), trials=100000, error_rate=0.01)

	all_good &= test('random', Random_content(), trials=10000, error_rate=0.1)

	#for elements in [ 1, 10, 100, 1000 ]:
	for elements in [ 1, 10, 100, 1000, 10000, 100000, 1000000 ]:
		all_good &= test('evens %d' % elements, Evens(elements), trials=elements, error_rate=1e-12)

	all_good &= and_test()

	all_good &= or_test()

	if all_good:
		sys.stderr.write('%s: All tests passed\n' % sys.argv[0])
		sys.exit(0)
	else:
		sys.stderr.write('%s: One or more tests failed\n' % sys.argv[0])
		sys.exit(1)

main()