|
| 1 | +# Copyright (c) 2018, Oracle and/or its affiliates. |
| 2 | +# Copyright (C) 1996-2017 Python Software Foundation |
| 3 | +# |
| 4 | +# Licensed under the PYTHON SOFTWARE FOUNDATION LICENSE VERSION 2 |
| 5 | + |
| 6 | +import unittest |
| 7 | +import random |
| 8 | +import time |
| 9 | + |
| 10 | + |
| 11 | +class TestBasicOps: |
| 12 | + def __init__(self): |
| 13 | + super().__init__() |
| 14 | + self.gen = random.Random() |
| 15 | + |
| 16 | + def randomlist(self, n): |
| 17 | + """Helper function to make a list of random numbers""" |
| 18 | + return [self.gen.random() for i in range(n)] |
| 19 | + |
| 20 | + def test_autoseed(self): |
| 21 | + self.gen.seed() |
| 22 | + state1 = self.gen.getstate() |
| 23 | + time.sleep(0.1) |
| 24 | + self.gen.seed() # diffent seeds at different times |
| 25 | + state2 = self.gen.getstate() |
| 26 | + self.assertNotEqual(state1, state2) |
| 27 | + |
| 28 | + def test_saverestore(self): |
| 29 | + N = 1000 |
| 30 | + self.gen.seed() |
| 31 | + state = self.gen.getstate() |
| 32 | + randseq = self.randomlist(N) |
| 33 | + self.gen.setstate(state) # should regenerate the same sequence |
| 34 | + self.assertEqual(randseq, self.randomlist(N)) |
| 35 | + |
| 36 | + def test_seedargs(self): |
| 37 | + # Seed value with a negative hash. |
| 38 | + class MySeed(object): |
| 39 | + def __hash__(self): |
| 40 | + return -1729 |
| 41 | + for arg in [None, 0, 0, 1, 1, -1, -1, 10**20, -(10**20), |
| 42 | + 3.14, 1+2j, 'a', tuple('abc'), MySeed()]: |
| 43 | + self.gen.seed(arg) |
| 44 | + for arg in [list(range(3)), dict(one=1)]: |
| 45 | + self.assertRaises(TypeError, self.gen.seed, arg) |
| 46 | + self.assertRaises(TypeError, self.gen.seed, 1, 2, 3, 4) |
| 47 | + self.assertRaises(TypeError, type(self.gen), []) |
| 48 | + |
| 49 | + def test_shuffle(self): |
| 50 | + shuffle = self.gen.shuffle |
| 51 | + lst = [] |
| 52 | + shuffle(lst) |
| 53 | + self.assertEqual(lst, []) |
| 54 | + lst = [37] |
| 55 | + shuffle(lst) |
| 56 | + self.assertEqual(lst, [37]) |
| 57 | + seqs = [list(range(n)) for n in range(10)] |
| 58 | + shuffled_seqs = [list(range(n)) for n in range(10)] |
| 59 | + for shuffled_seq in shuffled_seqs: |
| 60 | + shuffle(shuffled_seq) |
| 61 | + for (seq, shuffled_seq) in zip(seqs, shuffled_seqs): |
| 62 | + self.assertEqual(len(seq), len(shuffled_seq)) |
| 63 | + self.assertEqual(set(seq), set(shuffled_seq)) |
| 64 | + # The above tests all would pass if the shuffle was a |
| 65 | + # no-op. The following non-deterministic test covers that. It |
| 66 | + # asserts that the shuffled sequence of 1000 distinct elements |
| 67 | + # must be different from the original one. Although there is |
| 68 | + # mathematically a non-zero probability that this could |
| 69 | + # actually happen in a genuinely random shuffle, it is |
| 70 | + # completely negligible, given that the number of possible |
| 71 | + # permutations of 1000 objects is 1000! (factorial of 1000), |
| 72 | + # which is considerably larger than the number of atoms in the |
| 73 | + # universe... |
| 74 | + lst = list(range(1000)) |
| 75 | + shuffled_lst = list(range(1000)) |
| 76 | + shuffle(shuffled_lst) |
| 77 | + self.assertTrue(lst != shuffled_lst) |
| 78 | + shuffle(lst) |
| 79 | + self.assertTrue(lst != shuffled_lst) |
| 80 | + |
| 81 | + # def test_choice(self): |
| 82 | + # choice = self.gen.choice |
| 83 | + # with self.assertRaises(IndexError): |
| 84 | + # choice([]) |
| 85 | + # self.assertEqual(choice([50]), 50) |
| 86 | + # self.assertIn(choice([25, 75]), [25, 75]) |
| 87 | + |
| 88 | + def test_sample(self): |
| 89 | + # For the entire allowable range of 0 <= k <= N, validate that |
| 90 | + # the sample is of the correct length and contains only unique items |
| 91 | + N = 100 |
| 92 | + population = range(N) |
| 93 | + for k in range(N+1): |
| 94 | + s = self.gen.sample(population, k) |
| 95 | + self.assertEqual(len(s), k) |
| 96 | + uniq = set(s) |
| 97 | + self.assertEqual(len(uniq), k) |
| 98 | + self.assertTrue(uniq <= set(population)) |
| 99 | + self.assertEqual(self.gen.sample([], 0), []) # test edge case N==k==0 |
| 100 | + # Exception raised if size of sample exceeds that of population |
| 101 | + self.assertRaises(ValueError, self.gen.sample, population, N+1) |
| 102 | + self.assertRaises(ValueError, self.gen.sample, [], -1) |
| 103 | + |
| 104 | + def test_sample_distribution(self): |
| 105 | + # For the entire allowable range of 0 <= k <= N, validate that |
| 106 | + # sample generates all possible permutations |
| 107 | + n = 5 |
| 108 | + pop = range(n) |
| 109 | + trials = 10000 # large num prevents false negatives without slowing normal case |
| 110 | + def factorial(n): |
| 111 | + if n == 0: |
| 112 | + return 1 |
| 113 | + return n * factorial(n - 1) |
| 114 | + for k in range(n): |
| 115 | + expected = factorial(n) // factorial(n-k) |
| 116 | + perms = {} |
| 117 | + for i in range(trials): |
| 118 | + perms[tuple(self.gen.sample(pop, k))] = None |
| 119 | + if len(perms) == expected: |
| 120 | + break |
| 121 | + else: |
| 122 | + self.fail() |
| 123 | + |
| 124 | + def test_sample_inputs(self): |
| 125 | + # SF bug #801342 -- population can be any iterable defining __len__() |
| 126 | + self.gen.sample(set(range(20)), 2) |
| 127 | + self.gen.sample(range(20), 2) |
| 128 | + self.gen.sample(range(20), 2) |
| 129 | + self.gen.sample(str('abcdefghijklmnopqrst'), 2) |
| 130 | + self.gen.sample(tuple('abcdefghijklmnopqrst'), 2) |
| 131 | + |
| 132 | + def test_sample_on_dicts(self): |
| 133 | + self.assertRaises(TypeError, self.gen.sample, dict.fromkeys('abcdef'), 2) |
| 134 | + |
| 135 | + def test_choices(self): |
| 136 | + choices = self.gen.choices |
| 137 | + data = ['red', 'green', 'blue', 'yellow'] |
| 138 | + str_data = 'abcd' |
| 139 | + range_data = range(4) |
| 140 | + set_data = set(range(4)) |
| 141 | + |
| 142 | + # basic functionality |
| 143 | + for sample in [ |
| 144 | + choices(data, k=5), |
| 145 | + choices(data, range(4), k=5), |
| 146 | + choices(k=5, population=data, weights=range(4)), |
| 147 | + ]: |
| 148 | + self.assertEqual(len(sample), 5) |
| 149 | + self.assertEqual(type(sample), list) |
| 150 | + self.assertTrue(set(sample) <= set(data)) |
| 151 | + |
| 152 | + def test_gauss(self): |
| 153 | + # Ensure that the seed() method initializes all the hidden state. In |
| 154 | + # particular, through 2.2.1 it failed to reset a piece of state used |
| 155 | + # by (and only by) the .gauss() method. |
| 156 | + |
| 157 | + for seed in 1, 12, 123, 1234, 12345, 123456, 654321: |
| 158 | + self.gen.seed(seed) |
| 159 | + x1 = self.gen.random() |
| 160 | + y1 = self.gen.gauss(0, 1) |
| 161 | + |
| 162 | + self.gen.seed(seed) |
| 163 | + x2 = self.gen.random() |
| 164 | + y2 = self.gen.gauss(0, 1) |
| 165 | + |
| 166 | + self.assertEqual(x1, x2) |
| 167 | + self.assertEqual(y1, y2) |
| 168 | + |
| 169 | + def test_bug_9025(self): |
| 170 | + # Had problem with an uneven distribution in int(n*random()) |
| 171 | + # Verify the fix by checking that distributions fall within expectations. |
| 172 | + n = 100000 |
| 173 | + randrange = self.gen.randrange |
| 174 | + k = sum(randrange(6755399441055744) % 3 == 2 for i in range(n)) |
| 175 | + self.assertTrue(0.30 < k/n < .37, (k/n)) |
0 commit comments