| 1 |
import unittest |
| 2 |
import unittest.mock |
| 3 |
import random |
| 4 |
import os |
| 5 |
import time |
| 6 |
import pickle |
| 7 |
import shlex |
| 8 |
import warnings |
| 9 |
import test.support |
| 10 |
|
| 11 |
from functools import partial |
| 12 |
from math import log, exp, pi, fsum, sin, factorial |
| 13 |
from test import support |
| 14 |
from fractions import Fraction |
| 15 |
from collections import abc, Counter |
| 16 |
|
| 17 |
|
| 18 |
class MyIndex: |
| 19 |
def __init__(self, value): |
| 20 |
self.value = value |
| 21 |
|
| 22 |
def __index__(self): |
| 23 |
return self.value |
| 24 |
|
| 25 |
|
| 26 |
class TestBasicOps: |
| 27 |
# Superclass with tests common to all generators. |
| 28 |
# Subclasses must arrange for self.gen to retrieve the Random instance |
| 29 |
# to be tested. |
| 30 |
|
| 31 |
def randomlist(self, n): |
| 32 |
"""Helper function to make a list of random numbers""" |
| 33 |
return [self.gen.random() for i in range(n)] |
| 34 |
|
| 35 |
def test_autoseed(self): |
| 36 |
self.gen.seed() |
| 37 |
state1 = self.gen.getstate() |
| 38 |
time.sleep(0.1) |
| 39 |
self.gen.seed() # different seeds at different times |
| 40 |
state2 = self.gen.getstate() |
| 41 |
self.assertNotEqual(state1, state2) |
| 42 |
|
| 43 |
def test_saverestore(self): |
| 44 |
N = 1000 |
| 45 |
self.gen.seed() |
| 46 |
state = self.gen.getstate() |
| 47 |
randseq = self.randomlist(N) |
| 48 |
self.gen.setstate(state) # should regenerate the same sequence |
| 49 |
self.assertEqual(randseq, self.randomlist(N)) |
| 50 |
|
| 51 |
def test_seedargs(self): |
| 52 |
# Seed value with a negative hash. |
| 53 |
class MySeed(object): |
| 54 |
def __hash__(self): |
| 55 |
return -1729 |
| 56 |
for arg in [None, 0, 1, -1, 10**20, -(10**20), |
| 57 |
False, True, 3.14, 'a']: |
| 58 |
self.gen.seed(arg) |
| 59 |
|
| 60 |
for arg in [1+2j, tuple('abc'), MySeed()]: |
| 61 |
with self.assertRaises(TypeError): |
| 62 |
self.gen.seed(arg) |
| 63 |
|
| 64 |
for arg in [list(range(3)), dict(one=1)]: |
| 65 |
self.assertRaises(TypeError, self.gen.seed, arg) |
| 66 |
self.assertRaises(TypeError, self.gen.seed, 1, 2, 3, 4) |
| 67 |
self.assertRaises(TypeError, type(self.gen), []) |
| 68 |
|
| 69 |
def test_seed_no_mutate_bug_44018(self): |
| 70 |
a = bytearray(b'1234') |
| 71 |
self.gen.seed(a) |
| 72 |
self.assertEqual(a, bytearray(b'1234')) |
| 73 |
|
| 74 |
@unittest.mock.patch('random._urandom') # os.urandom |
| 75 |
def test_seed_when_randomness_source_not_found(self, urandom_mock): |
| 76 |
# Random.seed() uses time.time() when an operating system specific |
| 77 |
# randomness source is not found. To test this on machines where it |
| 78 |
# exists, run the above test, test_seedargs(), again after mocking |
| 79 |
# os.urandom() so that it raises the exception expected when the |
| 80 |
# randomness source is not available. |
| 81 |
urandom_mock.side_effect = NotImplementedError |
| 82 |
self.test_seedargs() |
| 83 |
|
| 84 |
def test_shuffle(self): |
| 85 |
shuffle = self.gen.shuffle |
| 86 |
lst = [] |
| 87 |
shuffle(lst) |
| 88 |
self.assertEqual(lst, []) |
| 89 |
lst = [37] |
| 90 |
shuffle(lst) |
| 91 |
self.assertEqual(lst, [37]) |
| 92 |
seqs = [list(range(n)) for n in range(10)] |
| 93 |
shuffled_seqs = [list(range(n)) for n in range(10)] |
| 94 |
for shuffled_seq in shuffled_seqs: |
| 95 |
shuffle(shuffled_seq) |
| 96 |
for (seq, shuffled_seq) in zip(seqs, shuffled_seqs): |
| 97 |
self.assertEqual(len(seq), len(shuffled_seq)) |
| 98 |
self.assertEqual(set(seq), set(shuffled_seq)) |
| 99 |
# The above tests all would pass if the shuffle was a |
| 100 |
# no-op. The following non-deterministic test covers that. It |
| 101 |
# asserts that the shuffled sequence of 1000 distinct elements |
| 102 |
# must be different from the original one. Although there is |
| 103 |
# mathematically a non-zero probability that this could |
| 104 |
# actually happen in a genuinely random shuffle, it is |
| 105 |
# completely negligible, given that the number of possible |
| 106 |
# permutations of 1000 objects is 1000! (factorial of 1000), |
| 107 |
# which is considerably larger than the number of atoms in the |
| 108 |
# universe... |
| 109 |
lst = list(range(1000)) |
| 110 |
shuffled_lst = list(range(1000)) |
| 111 |
shuffle(shuffled_lst) |
| 112 |
self.assertTrue(lst != shuffled_lst) |
| 113 |
shuffle(lst) |
| 114 |
self.assertTrue(lst != shuffled_lst) |
| 115 |
self.assertRaises(TypeError, shuffle, (1, 2, 3)) |
| 116 |
|
| 117 |
def test_choice(self): |
| 118 |
choice = self.gen.choice |
| 119 |
with self.assertRaises(IndexError): |
| 120 |
choice([]) |
| 121 |
self.assertEqual(choice([50]), 50) |
| 122 |
self.assertIn(choice([25, 75]), [25, 75]) |
| 123 |
|
| 124 |
def test_choice_with_numpy(self): |
| 125 |
# Accommodation for NumPy arrays which have disabled __bool__(). |
| 126 |
# See: https://github.com/python/cpython/issues/100805 |
| 127 |
choice = self.gen.choice |
| 128 |
|
| 129 |
class NA(list): |
| 130 |
"Simulate numpy.array() behavior" |
| 131 |
def __bool__(self): |
| 132 |
raise RuntimeError |
| 133 |
|
| 134 |
with self.assertRaises(IndexError): |
| 135 |
choice(NA([])) |
| 136 |
self.assertEqual(choice(NA([50])), 50) |
| 137 |
self.assertIn(choice(NA([25, 75])), [25, 75]) |
| 138 |
|
| 139 |
def test_sample(self): |
| 140 |
# For the entire allowable range of 0 <= k <= N, validate that |
| 141 |
# the sample is of the correct length and contains only unique items |
| 142 |
N = 100 |
| 143 |
population = range(N) |
| 144 |
for k in range(N+1): |
| 145 |
s = self.gen.sample(population, k) |
| 146 |
self.assertEqual(len(s), k) |
| 147 |
uniq = set(s) |
| 148 |
self.assertEqual(len(uniq), k) |
| 149 |
self.assertTrue(uniq <= set(population)) |
| 150 |
self.assertEqual(self.gen.sample([], 0), []) # test edge case N==k==0 |
| 151 |
# Exception raised if size of sample exceeds that of population |
| 152 |
self.assertRaises(ValueError, self.gen.sample, population, N+1) |
| 153 |
self.assertRaises(ValueError, self.gen.sample, [], -1) |
| 154 |
self.assertRaises(TypeError, self.gen.sample, population, 1.0) |
| 155 |
|
| 156 |
def test_sample_distribution(self): |
| 157 |
# For the entire allowable range of 0 <= k <= N, validate that |
| 158 |
# sample generates all possible permutations |
| 159 |
n = 5 |
| 160 |
pop = range(n) |
| 161 |
trials = 10000 # large num prevents false negatives without slowing normal case |
| 162 |
for k in range(n): |
| 163 |
expected = factorial(n) // factorial(n-k) |
| 164 |
perms = {} |
| 165 |
for i in range(trials): |
| 166 |
perms[tuple(self.gen.sample(pop, k))] = None |
| 167 |
if len(perms) == expected: |
| 168 |
break |
| 169 |
else: |
| 170 |
self.fail() |
| 171 |
|
| 172 |
def test_sample_inputs(self): |
| 173 |
# SF bug #801342 -- population can be any iterable defining __len__() |
| 174 |
self.gen.sample(range(20), 2) |
| 175 |
self.gen.sample(range(20), 2) |
| 176 |
self.gen.sample(str('abcdefghijklmnopqrst'), 2) |
| 177 |
self.gen.sample(tuple('abcdefghijklmnopqrst'), 2) |
| 178 |
|
| 179 |
def test_sample_on_dicts(self): |
| 180 |
self.assertRaises(TypeError, self.gen.sample, dict.fromkeys('abcdef'), 2) |
| 181 |
|
| 182 |
def test_sample_on_sets(self): |
| 183 |
with self.assertRaises(TypeError): |
| 184 |
population = {10, 20, 30, 40, 50, 60, 70} |
| 185 |
self.gen.sample(population, k=5) |
| 186 |
|
| 187 |
def test_sample_on_seqsets(self): |
| 188 |
class SeqSet(abc.Sequence, abc.Set): |
| 189 |
def __init__(self, items): |
| 190 |
self._items = items |
| 191 |
|
| 192 |
def __len__(self): |
| 193 |
return len(self._items) |
| 194 |
|
| 195 |
def __getitem__(self, index): |
| 196 |
return self._items[index] |
| 197 |
|
| 198 |
population = SeqSet([2, 4, 1, 3]) |
| 199 |
with warnings.catch_warnings(): |
| 200 |
warnings.simplefilter("error", DeprecationWarning) |
| 201 |
self.gen.sample(population, k=2) |
| 202 |
|
| 203 |
def test_sample_with_counts(self): |
| 204 |
sample = self.gen.sample |
| 205 |
|
| 206 |
# General case |
| 207 |
colors = ['red', 'green', 'blue', 'orange', 'black', 'brown', 'amber'] |
| 208 |
counts = [500, 200, 20, 10, 5, 0, 1 ] |
| 209 |
k = 700 |
| 210 |
summary = Counter(sample(colors, counts=counts, k=k)) |
| 211 |
self.assertEqual(sum(summary.values()), k) |
| 212 |
for color, weight in zip(colors, counts): |
| 213 |
self.assertLessEqual(summary[color], weight) |
| 214 |
self.assertNotIn('brown', summary) |
| 215 |
|
| 216 |
# Case that exhausts the population |
| 217 |
k = sum(counts) |
| 218 |
summary = Counter(sample(colors, counts=counts, k=k)) |
| 219 |
self.assertEqual(sum(summary.values()), k) |
| 220 |
for color, weight in zip(colors, counts): |
| 221 |
self.assertLessEqual(summary[color], weight) |
| 222 |
self.assertNotIn('brown', summary) |
| 223 |
|
| 224 |
# Case with population size of 1 |
| 225 |
summary = Counter(sample(['x'], counts=[10], k=8)) |
| 226 |
self.assertEqual(summary, Counter(x=8)) |
| 227 |
|
| 228 |
# Case with all counts equal. |
| 229 |
nc = len(colors) |
| 230 |
summary = Counter(sample(colors, counts=[10]*nc, k=10*nc)) |
| 231 |
self.assertEqual(summary, Counter(10*colors)) |
| 232 |
|
| 233 |
# Test error handling |
| 234 |
with self.assertRaises(TypeError): |
| 235 |
sample(['red', 'green', 'blue'], counts=10, k=10) # counts not iterable |
| 236 |
with self.assertRaises(ValueError): |
| 237 |
sample(['red', 'green', 'blue'], counts=[-3, -7, -8], k=2) # counts are negative |
| 238 |
with self.assertRaises(ValueError): |
| 239 |
sample(['red', 'green'], counts=[10, 10], k=21) # population too small |
| 240 |
with self.assertRaises(ValueError): |
| 241 |
sample(['red', 'green', 'blue'], counts=[1, 2], k=2) # too few counts |
| 242 |
with self.assertRaises(ValueError): |
| 243 |
sample(['red', 'green', 'blue'], counts=[1, 2, 3, 4], k=2) # too many counts |
| 244 |
|
| 245 |
# Cases with zero counts match equivalents without counts (see gh-130285) |
| 246 |
self.assertEqual( |
| 247 |
sample('abc', k=0, counts=[0, 0, 0]), |
| 248 |
sample([], k=0), |
| 249 |
) |
| 250 |
self.assertEqual( |
| 251 |
sample([], 0, counts=[]), |
| 252 |
sample([], 0), |
| 253 |
) |
| 254 |
with self.assertRaises(ValueError): |
| 255 |
sample([], 1, counts=[]) |
| 256 |
with self.assertRaises(ValueError): |
| 257 |
sample('x', 1, counts=[0]) |
| 258 |
|
| 259 |
def test_choices(self): |
| 260 |
choices = self.gen.choices |
| 261 |
data = ['red', 'green', 'blue', 'yellow'] |
| 262 |
str_data = 'abcd' |
| 263 |
range_data = range(4) |
| 264 |
set_data = set(range(4)) |
| 265 |
|
| 266 |
# basic functionality |
| 267 |
for sample in [ |
| 268 |
choices(data, k=5), |
| 269 |
choices(data, range(4), k=5), |
| 270 |
choices(k=5, population=data, weights=range(4)), |
| 271 |
choices(k=5, population=data, cum_weights=range(4)), |
| 272 |
choices(data, k=MyIndex(5)), |
| 273 |
]: |
| 274 |
self.assertEqual(len(sample), 5) |
| 275 |
self.assertEqual(type(sample), list) |
| 276 |
self.assertTrue(set(sample) <= set(data)) |
| 277 |
|
| 278 |
# test argument handling |
| 279 |
with self.assertRaises(TypeError): # missing arguments |
| 280 |
choices(2) |
| 281 |
|
| 282 |
self.assertEqual(choices(data, k=0), []) # k == 0 |
| 283 |
self.assertEqual(choices(data, k=-1), []) # negative k behaves like ``[0] * -1`` |
| 284 |
with self.assertRaises(TypeError): |
| 285 |
choices(data, k=2.5) # k is a float |
| 286 |
|
| 287 |
self.assertTrue(set(choices(str_data, k=5)) <= set(str_data)) # population is a string sequence |
| 288 |
self.assertTrue(set(choices(range_data, k=5)) <= set(range_data)) # population is a range |
| 289 |
with self.assertRaises(TypeError): |
| 290 |
choices(set_data, k=2) # population is not a sequence |
| 291 |
|
| 292 |
self.assertTrue(set(choices(data, None, k=5)) <= set(data)) # weights is None |
| 293 |
self.assertTrue(set(choices(data, weights=None, k=5)) <= set(data)) |
| 294 |
with self.assertRaises(ValueError): |
| 295 |
choices(data, [1,2], k=5) # len(weights) != len(population) |
| 296 |
with self.assertRaises(TypeError): |
| 297 |
choices(data, 10, k=5) # non-iterable weights |
| 298 |
with self.assertRaises(TypeError): |
| 299 |
choices(data, [None]*4, k=5) # non-numeric weights |
| 300 |
for weights in [ |
| 301 |
[15, 10, 25, 30], # integer weights |
| 302 |
[15.1, 10.2, 25.2, 30.3], # float weights |
| 303 |
[Fraction(1, 3), Fraction(2, 6), Fraction(3, 6), Fraction(4, 6)], # fractional weights |
| 304 |
[True, False, True, False] # booleans (include / exclude) |
| 305 |
]: |
| 306 |
self.assertTrue(set(choices(data, weights, k=5)) <= set(data)) |
| 307 |
|
| 308 |
with self.assertRaises(ValueError): |
| 309 |
choices(data, cum_weights=[1,2], k=5) # len(weights) != len(population) |
| 310 |
with self.assertRaises(TypeError): |
| 311 |
choices(data, cum_weights=10, k=5) # non-iterable cum_weights |
| 312 |
with self.assertRaises(TypeError): |
| 313 |
choices(data, cum_weights=[None]*4, k=5) # non-numeric cum_weights |
| 314 |
with self.assertRaises(TypeError): |
| 315 |
choices(data, range(4), cum_weights=range(4), k=5) # both weights and cum_weights |
| 316 |
for weights in [ |
| 317 |
[15, 10, 25, 30], # integer cum_weights |
| 318 |
[15.1, 10.2, 25.2, 30.3], # float cum_weights |
| 319 |
[Fraction(1, 3), Fraction(2, 6), Fraction(3, 6), Fraction(4, 6)], # fractional cum_weights |
| 320 |
]: |
| 321 |
self.assertTrue(set(choices(data, cum_weights=weights, k=5)) <= set(data)) |
| 322 |
|
| 323 |
# Test weight focused on a single element of the population |
| 324 |
self.assertEqual(choices('abcd', [1, 0, 0, 0]), ['a']) |
| 325 |
self.assertEqual(choices('abcd', [0, 1, 0, 0]), ['b']) |
| 326 |
self.assertEqual(choices('abcd', [0, 0, 1, 0]), ['c']) |
| 327 |
self.assertEqual(choices('abcd', [0, 0, 0, 1]), ['d']) |
| 328 |
|
| 329 |
# Test consistency with random.choice() for empty population |
| 330 |
with self.assertRaises(IndexError): |
| 331 |
choices([], k=1) |
| 332 |
with self.assertRaises(IndexError): |
| 333 |
choices([], weights=[], k=1) |
| 334 |
with self.assertRaises(IndexError): |
| 335 |
choices([], cum_weights=[], k=5) |
| 336 |
|
| 337 |
def test_choices_subnormal(self): |
| 338 |
# Subnormal weights would occasionally trigger an IndexError |
| 339 |
# in choices() when the value returned by random() was large |
| 340 |
# enough to make `random() * total` round up to the total. |
| 341 |
# See https://bugs.python.org/msg275594 for more detail. |
| 342 |
choices = self.gen.choices |
| 343 |
choices(population=[1, 2], weights=[1e-323, 1e-323], k=5000) |
| 344 |
|
| 345 |
def test_choices_with_all_zero_weights(self): |
| 346 |
# See issue #38881 |
| 347 |
with self.assertRaises(ValueError): |
| 348 |
self.gen.choices('AB', [0.0, 0.0]) |
| 349 |
|
| 350 |
def test_choices_negative_total(self): |
| 351 |
with self.assertRaises(ValueError): |
| 352 |
self.gen.choices('ABC', [3, -5, 1]) |
| 353 |
|
| 354 |
def test_choices_infinite_total(self): |
| 355 |
with self.assertRaises(ValueError): |
| 356 |
self.gen.choices('A', [float('inf')]) |
| 357 |
with self.assertRaises(ValueError): |
| 358 |
self.gen.choices('AB', [0.0, float('inf')]) |
| 359 |
with self.assertRaises(ValueError): |
| 360 |
self.gen.choices('AB', [-float('inf'), 123]) |
| 361 |
with self.assertRaises(ValueError): |
| 362 |
self.gen.choices('AB', [0.0, float('nan')]) |
| 363 |
with self.assertRaises(ValueError): |
| 364 |
self.gen.choices('AB', [float('-inf'), float('inf')]) |
| 365 |
|
| 366 |
def test_gauss(self): |
| 367 |
# Ensure that the seed() method initializes all the hidden state. In |
| 368 |
# particular, through 2.2.1 it failed to reset a piece of state used |
| 369 |
# by (and only by) the .gauss() method. |
| 370 |
|
| 371 |
for seed in 1, 12, 123, 1234, 12345, 123456, 654321: |
| 372 |
self.gen.seed(seed) |
| 373 |
x1 = self.gen.random() |
| 374 |
y1 = self.gen.gauss(0, 1) |
| 375 |
|
| 376 |
self.gen.seed(seed) |
| 377 |
x2 = self.gen.random() |
| 378 |
y2 = self.gen.gauss(0, 1) |
| 379 |
|
| 380 |
self.assertEqual(x1, x2) |
| 381 |
self.assertEqual(y1, y2) |
| 382 |
|
| 383 |
@support.requires_IEEE_754 |
| 384 |
def test_53_bits_per_float(self): |
| 385 |
span = 2 ** 53 |
| 386 |
cum = 0 |
| 387 |
for i in range(100): |
| 388 |
cum |= int(self.gen.random() * span) |
| 389 |
self.assertEqual(cum, span-1) |
| 390 |
|
| 391 |
def test_getrandbits(self): |
| 392 |
getrandbits = self.gen.getrandbits |
| 393 |
# Verify ranges |
| 394 |
for k in range(1, 1000): |
| 395 |
self.assertTrue(0 <= getrandbits(k) < 2**k) |
| 396 |
self.assertEqual(getrandbits(0), 0) |
| 397 |
|
| 398 |
# Verify all bits active |
| 399 |
for span in [1, 2, 3, 4, 31, 32, 32, 52, 53, 54, 119, 127, 128, 129]: |
| 400 |
all_bits = 2**span-1 |
| 401 |
cum = 0 |
| 402 |
cpl_cum = 0 |
| 403 |
for i in range(100): |
| 404 |
v = getrandbits(span) |
| 405 |
cum |= v |
| 406 |
cpl_cum |= all_bits ^ v |
| 407 |
self.assertEqual(cum, all_bits) |
| 408 |
self.assertEqual(cpl_cum, all_bits) |
| 409 |
|
| 410 |
# Verify argument checking |
| 411 |
self.assertRaises(TypeError, getrandbits) |
| 412 |
self.assertRaises(TypeError, getrandbits, 1, 2) |
| 413 |
self.assertRaises(ValueError, getrandbits, -1) |
| 414 |
self.assertRaises(OverflowError, getrandbits, 1<<1000) |
| 415 |
self.assertRaises(ValueError, getrandbits, -1<<1000) |
| 416 |
self.assertRaises(TypeError, getrandbits, 10.1) |
| 417 |
|
| 418 |
def test_bigrand(self): |
| 419 |
# The randrange routine should build-up the required number of bits |
| 420 |
# in stages so that all bit positions are active. |
| 421 |
span = 2 ** 500 |
| 422 |
cum = 0 |
| 423 |
for i in range(100): |
| 424 |
r = self.gen.randrange(span) |
| 425 |
self.assertTrue(0 <= r < span) |
| 426 |
cum |= r |
| 427 |
self.assertEqual(cum, span-1) |
| 428 |
|
| 429 |
def test_bigrand_ranges(self): |
| 430 |
for i in [40,80, 160, 200, 211, 250, 375, 512, 550]: |
| 431 |
start = self.gen.randrange(2 ** (i-2)) |
| 432 |
stop = self.gen.randrange(2 ** i) |
| 433 |
if stop <= start: |
| 434 |
continue |
| 435 |
self.assertTrue(start <= self.gen.randrange(start, stop) < stop) |
| 436 |
|
| 437 |
def test_rangelimits(self): |
| 438 |
for start, stop in [(-2,0), (-(2**60)-2,-(2**60)), (2**60,2**60+2)]: |
| 439 |
self.assertEqual(set(range(start,stop)), |
| 440 |
set([self.gen.randrange(start,stop) for i in range(100)])) |
| 441 |
|
| 442 |
def test_randrange_nonunit_step(self): |
| 443 |
rint = self.gen.randrange(0, 10, 2) |
| 444 |
self.assertIn(rint, (0, 2, 4, 6, 8)) |
| 445 |
rint = self.gen.randrange(0, 2, 2) |
| 446 |
self.assertEqual(rint, 0) |
| 447 |
|
| 448 |
def test_randrange_errors(self): |
| 449 |
raises_value_error = partial(self.assertRaises, ValueError, self.gen.randrange) |
| 450 |
raises_type_error = partial(self.assertRaises, TypeError, self.gen.randrange) |
| 451 |
|
| 452 |
# Empty range |
| 453 |
raises_value_error(3, 3) |
| 454 |
raises_value_error(-721) |
| 455 |
raises_value_error(0, 100, -12) |
| 456 |
|
| 457 |
# Zero step |
| 458 |
raises_value_error(0, 42, 0) |
| 459 |
raises_type_error(0, 42, 0.0) |
| 460 |
raises_type_error(0, 0, 0.0) |
| 461 |
|
| 462 |
# Non-integer stop |
| 463 |
raises_type_error(3.14159) |
| 464 |
raises_type_error(3.0) |
| 465 |
raises_type_error(Fraction(3, 1)) |
| 466 |
raises_type_error('3') |
| 467 |
raises_type_error(0, 2.71827) |
| 468 |
raises_type_error(0, 2.0) |
| 469 |
raises_type_error(0, Fraction(2, 1)) |
| 470 |
raises_type_error(0, '2') |
| 471 |
raises_type_error(0, 2.71827, 2) |
| 472 |
|
| 473 |
# Non-integer start |
| 474 |
raises_type_error(2.71827, 5) |
| 475 |
raises_type_error(2.0, 5) |
| 476 |
raises_type_error(Fraction(2, 1), 5) |
| 477 |
raises_type_error('2', 5) |
| 478 |
raises_type_error(2.71827, 5, 2) |
| 479 |
|
| 480 |
# Non-integer step |
| 481 |
raises_type_error(0, 42, 3.14159) |
| 482 |
raises_type_error(0, 42, 3.0) |
| 483 |
raises_type_error(0, 42, Fraction(3, 1)) |
| 484 |
raises_type_error(0, 42, '3') |
| 485 |
raises_type_error(0, 42, 1.0) |
| 486 |
raises_type_error(0, 0, 1.0) |
| 487 |
|
| 488 |
def test_randrange_step(self): |
| 489 |
# bpo-42772: When stop is None, the step argument was being ignored. |
| 490 |
randrange = self.gen.randrange |
| 491 |
with self.assertRaises(TypeError): |
| 492 |
randrange(1000, step=100) |
| 493 |
with self.assertRaises(TypeError): |
| 494 |
randrange(1000, None, step=100) |
| 495 |
with self.assertRaises(TypeError): |
| 496 |
randrange(1000, step=MyIndex(1)) |
| 497 |
with self.assertRaises(TypeError): |
| 498 |
randrange(1000, None, step=MyIndex(1)) |
| 499 |
|
| 500 |
def test_randbelow_logic(self, _log=log, int=int): |
| 501 |
# check bitcount transition points: 2**i and 2**(i+1)-1 |
| 502 |
# show that: k = int(1.001 + _log(n, 2)) |
| 503 |
# is equal to or one greater than the number of bits in n |
| 504 |
for i in range(1, 1000): |
| 505 |
n = 1 << i # check an exact power of two |
| 506 |
numbits = i+1 |
| 507 |
k = int(1.00001 + _log(n, 2)) |
| 508 |
self.assertEqual(k, numbits) |
| 509 |
self.assertEqual(n, 2**(k-1)) |
| 510 |
|
| 511 |
n += n - 1 # check 1 below the next power of two |
| 512 |
k = int(1.00001 + _log(n, 2)) |
| 513 |
self.assertIn(k, [numbits, numbits+1]) |
| 514 |
self.assertTrue(2**k > n > 2**(k-2)) |
| 515 |
|
| 516 |
n -= n >> 15 # check a little farther below the next power of two |
| 517 |
k = int(1.00001 + _log(n, 2)) |
| 518 |
self.assertEqual(k, numbits) # note the stronger assertion |
| 519 |
self.assertTrue(2**k > n > 2**(k-1)) # note the stronger assertion |
| 520 |
|
| 521 |
def test_randrange_index(self): |
| 522 |
randrange = self.gen.randrange |
| 523 |
self.assertIn(randrange(MyIndex(5)), range(5)) |
| 524 |
self.assertIn(randrange(MyIndex(2), MyIndex(7)), range(2, 7)) |
| 525 |
self.assertIn(randrange(MyIndex(5), MyIndex(15), MyIndex(2)), range(5, 15, 2)) |
| 526 |
|
| 527 |
def test_randint(self): |
| 528 |
randint = self.gen.randint |
| 529 |
self.assertIn(randint(2, 5), (2, 3, 4, 5)) |
| 530 |
self.assertEqual(randint(2, 2), 2) |
| 531 |
self.assertIn(randint(MyIndex(2), MyIndex(5)), (2, 3, 4, 5)) |
| 532 |
self.assertEqual(randint(MyIndex(2), MyIndex(2)), 2) |
| 533 |
|
| 534 |
self.assertRaises(ValueError, randint, 5, 2) |
| 535 |
self.assertRaises(TypeError, randint) |
| 536 |
self.assertRaises(TypeError, randint, 2) |
| 537 |
self.assertRaises(TypeError, randint, 2, 5, 1) |
| 538 |
self.assertRaises(TypeError, randint, 2.0, 5) |
| 539 |
self.assertRaises(TypeError, randint, 2, 5.0) |
| 540 |
|
| 541 |
def test_pickling(self): |
| 542 |
for proto in range(pickle.HIGHEST_PROTOCOL + 1): |
| 543 |
state = pickle.dumps(self.gen, proto) |
| 544 |
origseq = [self.gen.random() for i in range(10)] |
| 545 |
newgen = pickle.loads(state) |
| 546 |
restoredseq = [newgen.random() for i in range(10)] |
| 547 |
self.assertEqual(origseq, restoredseq) |
| 548 |
|
| 549 |
def test_bug_1727780(self): |
| 550 |
# verify that version-2-pickles can be loaded |
| 551 |
# fine, whether they are created on 32-bit or 64-bit |
| 552 |
# platforms, and that version-3-pickles load fine. |
| 553 |
files = [("randv2_32.pck", 780), |
| 554 |
("randv2_64.pck", 866), |
| 555 |
("randv3.pck", 343)] |
| 556 |
for file, value in files: |
| 557 |
with open(support.findfile(file),"rb") as f: |
| 558 |
r = pickle.load(f) |
| 559 |
self.assertEqual(int(r.random()*1000), value) |
| 560 |
|
| 561 |
def test_bug_9025(self): |
| 562 |
# Had problem with an uneven distribution in int(n*random()) |
| 563 |
# Verify the fix by checking that distributions fall within expectations. |
| 564 |
n = 100000 |
| 565 |
randrange = self.gen.randrange |
| 566 |
k = sum(randrange(6755399441055744) % 3 == 2 for i in range(n)) |
| 567 |
self.assertTrue(0.30 < k/n < .37, (k/n)) |
| 568 |
|
| 569 |
def test_randrange_bug_1590891(self): |
| 570 |
start = 1000000000000 |
| 571 |
stop = -100000000000000000000 |
| 572 |
step = -200 |
| 573 |
x = self.gen.randrange(start, stop, step) |
| 574 |
self.assertTrue(stop < x <= start) |
| 575 |
self.assertEqual((x+stop)%step, 0) |
| 576 |
|
| 577 |
def test_randbytes(self): |
| 578 |
# Verify ranges |
| 579 |
for n in range(1, 10): |
| 580 |
data = self.gen.randbytes(n) |
| 581 |
self.assertEqual(type(data), bytes) |
| 582 |
self.assertEqual(len(data), n) |
| 583 |
|
| 584 |
self.assertEqual(self.gen.randbytes(0), b'') |
| 585 |
|
| 586 |
# Verify argument checking |
| 587 |
self.assertRaises(TypeError, self.gen.randbytes) |
| 588 |
self.assertRaises(TypeError, self.gen.randbytes, 1, 2) |
| 589 |
self.assertRaises(ValueError, self.gen.randbytes, -1) |
| 590 |
self.assertRaises(OverflowError, self.gen.randbytes, 1<<1000) |
| 591 |
self.assertRaises((ValueError, OverflowError), self.gen.randbytes, -1<<1000) |
| 592 |
self.assertRaises(TypeError, self.gen.randbytes, 1.0) |
| 593 |
|
| 594 |
def test_mu_sigma_default_args(self): |
| 595 |
self.assertIsInstance(self.gen.normalvariate(), float) |
| 596 |
self.assertIsInstance(self.gen.gauss(), float) |
| 597 |
|
| 598 |
|
| 599 |
try: |
| 600 |
random.SystemRandom().random() |
| 601 |
except NotImplementedError: |
| 602 |
SystemRandom_available = False |
| 603 |
else: |
| 604 |
SystemRandom_available = True |
| 605 |
|
| 606 |
@unittest.skipUnless(SystemRandom_available, "random.SystemRandom not available") |
| 607 |
class SystemRandom_TestBasicOps(TestBasicOps, unittest.TestCase): |
| 608 |
gen = random.SystemRandom() |
| 609 |
|
| 610 |
def test_autoseed(self): |
| 611 |
# Doesn't need to do anything except not fail |
| 612 |
self.gen.seed() |
| 613 |
|
| 614 |
def test_saverestore(self): |
| 615 |
self.assertRaises(NotImplementedError, self.gen.getstate) |
| 616 |
self.assertRaises(NotImplementedError, self.gen.setstate, None) |
| 617 |
|
| 618 |
def test_seedargs(self): |
| 619 |
# Doesn't need to do anything except not fail |
| 620 |
self.gen.seed(100) |
| 621 |
|
| 622 |
def test_gauss(self): |
| 623 |
self.gen.gauss_next = None |
| 624 |
self.gen.seed(100) |
| 625 |
self.assertEqual(self.gen.gauss_next, None) |
| 626 |
|
| 627 |
def test_pickling(self): |
| 628 |
for proto in range(pickle.HIGHEST_PROTOCOL + 1): |
| 629 |
self.assertRaises(NotImplementedError, pickle.dumps, self.gen, proto) |
| 630 |
|
| 631 |
|
| 632 |
class TestRawMersenneTwister(unittest.TestCase): |
| 633 |
@test.support.cpython_only |
| 634 |
def test_bug_41052(self): |
| 635 |
# _random.Random should not be allowed to serialization |
| 636 |
import _random |
| 637 |
for proto in range(pickle.HIGHEST_PROTOCOL + 1): |
| 638 |
r = _random.Random() |
| 639 |
self.assertRaises(TypeError, pickle.dumps, r, proto) |
| 640 |
|
| 641 |
@test.support.cpython_only |
| 642 |
def test_bug_42008(self): |
| 643 |
# _random.Random should call seed with first element of arg tuple |
| 644 |
import _random |
| 645 |
r1 = _random.Random() |
| 646 |
r1.seed(8675309) |
| 647 |
r2 = _random.Random(8675309) |
| 648 |
self.assertEqual(r1.random(), r2.random()) |
| 649 |
|
| 650 |
|
| 651 |
class MersenneTwister_TestBasicOps(TestBasicOps, unittest.TestCase): |
| 652 |
gen = random.Random() |
| 653 |
|
| 654 |
def test_guaranteed_stable(self): |
| 655 |
# These sequences are guaranteed to stay the same across versions of python |
| 656 |
self.gen.seed(3456147, version=1) |
| 657 |
self.assertEqual([self.gen.random().hex() for i in range(4)], |
| 658 |
['0x1.ac362300d90d2p-1', '0x1.9d16f74365005p-1', |
| 659 |
'0x1.1ebb4352e4c4dp-1', '0x1.1a7422abf9c11p-1']) |
| 660 |
self.gen.seed("the quick brown fox", version=2) |
| 661 |
self.assertEqual([self.gen.random().hex() for i in range(4)], |
| 662 |
['0x1.1239ddfb11b7cp-3', '0x1.b3cbb5c51b120p-4', |
| 663 |
'0x1.8c4f55116b60fp-1', '0x1.63eb525174a27p-1']) |
| 664 |
|
| 665 |
def test_bug_27706(self): |
| 666 |
# Verify that version 1 seeds are unaffected by hash randomization |
| 667 |
|
| 668 |
self.gen.seed('nofar', version=1) # hash('nofar') == 5990528763808513177 |
| 669 |
self.assertEqual([self.gen.random().hex() for i in range(4)], |
| 670 |
['0x1.8645314505ad7p-1', '0x1.afb1f82e40a40p-5', |
| 671 |
'0x1.2a59d2285e971p-1', '0x1.56977142a7880p-6']) |
| 672 |
|
| 673 |
self.gen.seed('rachel', version=1) # hash('rachel') == -9091735575445484789 |
| 674 |
self.assertEqual([self.gen.random().hex() for i in range(4)], |
| 675 |
['0x1.0b294cc856fcdp-1', '0x1.2ad22d79e77b8p-3', |
| 676 |
'0x1.3052b9c072678p-2', '0x1.578f332106574p-3']) |
| 677 |
|
| 678 |
self.gen.seed('', version=1) # hash('') == 0 |
| 679 |
self.assertEqual([self.gen.random().hex() for i in range(4)], |
| 680 |
['0x1.b0580f98a7dbep-1', '0x1.84129978f9c1ap-1', |
| 681 |
'0x1.aeaa51052e978p-2', '0x1.092178fb945a6p-2']) |
| 682 |
|
| 683 |
def test_bug_31478(self): |
| 684 |
# There shouldn't be an assertion failure in _random.Random.seed() in |
| 685 |
# case the argument has a bad __abs__() method. |
| 686 |
class BadInt(int): |
| 687 |
def __abs__(self): |
| 688 |
return None |
| 689 |
try: |
| 690 |
self.gen.seed(BadInt()) |
| 691 |
except TypeError: |
| 692 |
pass |
| 693 |
|
| 694 |
def test_bug_31482(self): |
| 695 |
# Verify that version 1 seeds are unaffected by hash randomization |
| 696 |
# when the seeds are expressed as bytes rather than strings. |
| 697 |
# The hash(b) values listed are the Python2.7 hash() values |
| 698 |
# which were used for seeding. |
| 699 |
|
| 700 |
self.gen.seed(b'nofar', version=1) # hash('nofar') == 5990528763808513177 |
| 701 |
self.assertEqual([self.gen.random().hex() for i in range(4)], |
| 702 |
['0x1.8645314505ad7p-1', '0x1.afb1f82e40a40p-5', |
| 703 |
'0x1.2a59d2285e971p-1', '0x1.56977142a7880p-6']) |
| 704 |
|
| 705 |
self.gen.seed(b'rachel', version=1) # hash('rachel') == -9091735575445484789 |
| 706 |
self.assertEqual([self.gen.random().hex() for i in range(4)], |
| 707 |
['0x1.0b294cc856fcdp-1', '0x1.2ad22d79e77b8p-3', |
| 708 |
'0x1.3052b9c072678p-2', '0x1.578f332106574p-3']) |
| 709 |
|
| 710 |
self.gen.seed(b'', version=1) # hash('') == 0 |
| 711 |
self.assertEqual([self.gen.random().hex() for i in range(4)], |
| 712 |
['0x1.b0580f98a7dbep-1', '0x1.84129978f9c1ap-1', |
| 713 |
'0x1.aeaa51052e978p-2', '0x1.092178fb945a6p-2']) |
| 714 |
|
| 715 |
b = b'\x00\x20\x40\x60\x80\xA0\xC0\xE0\xF0' |
| 716 |
self.gen.seed(b, version=1) # hash(b) == 5015594239749365497 |
| 717 |
self.assertEqual([self.gen.random().hex() for i in range(4)], |
| 718 |
['0x1.52c2fde444d23p-1', '0x1.875174f0daea4p-2', |
| 719 |
'0x1.9e9b2c50e5cd2p-1', '0x1.fa57768bd321cp-2']) |
| 720 |
|
| 721 |
def test_setstate_first_arg(self): |
| 722 |
self.assertRaises(ValueError, self.gen.setstate, (1, None, None)) |
| 723 |
|
| 724 |
def test_setstate_middle_arg(self): |
| 725 |
start_state = self.gen.getstate() |
| 726 |
# Wrong type, s/b tuple |
| 727 |
self.assertRaises(TypeError, self.gen.setstate, (2, None, None)) |
| 728 |
# Wrong length, s/b 625 |
| 729 |
self.assertRaises(ValueError, self.gen.setstate, (2, (1,2,3), None)) |
| 730 |
# Wrong type, s/b tuple of 625 ints |
| 731 |
self.assertRaises(TypeError, self.gen.setstate, (2, ('a',)*625, None)) |
| 732 |
# Last element s/b an int also |
| 733 |
self.assertRaises(TypeError, self.gen.setstate, (2, (0,)*624+('a',), None)) |
| 734 |
# Last element s/b between 0 and 624 |
| 735 |
with self.assertRaises((ValueError, OverflowError)): |
| 736 |
self.gen.setstate((2, (1,)*624+(625,), None)) |
| 737 |
with self.assertRaises((ValueError, OverflowError)): |
| 738 |
self.gen.setstate((2, (1,)*624+(-1,), None)) |
| 739 |
# Failed calls to setstate() should not have changed the state. |
| 740 |
bits100 = self.gen.getrandbits(100) |
| 741 |
self.gen.setstate(start_state) |
| 742 |
self.assertEqual(self.gen.getrandbits(100), bits100) |
| 743 |
|
| 744 |
# Little trick to make "tuple(x % (2**32) for x in internalstate)" |
| 745 |
# raise ValueError. I cannot think of a simple way to achieve this, so |
| 746 |
# I am opting for using a generator as the middle argument of setstate |
| 747 |
# which attempts to cast a NaN to integer. |
| 748 |
state_values = self.gen.getstate()[1] |
| 749 |
state_values = list(state_values) |
| 750 |
state_values[-1] = float('nan') |
| 751 |
state = (int(x) for x in state_values) |
| 752 |
self.assertRaises(TypeError, self.gen.setstate, (2, state, None)) |
| 753 |
|
| 754 |
def test_referenceImplementation(self): |
| 755 |
# Compare the python implementation with results from the original |
| 756 |
# code. Create 2000 53-bit precision random floats. Compare only |
| 757 |
# the last ten entries to show that the independent implementations |
| 758 |
# are tracking. Here is the main() function needed to create the |
| 759 |
# list of expected random numbers: |
| 760 |
# void main(void){ |
| 761 |
# int i; |
| 762 |
# unsigned long init[4]={61731, 24903, 614, 42143}, length=4; |
| 763 |
# init_by_array(init, length); |
| 764 |
# for (i=0; i<2000; i++) { |
| 765 |
# printf("%.15f ", genrand_res53()); |
| 766 |
# if (i%5==4) printf("\n"); |
| 767 |
# } |
| 768 |
# } |
| 769 |
expected = [0.45839803073713259, |
| 770 |
0.86057815201978782, |
| 771 |
0.92848331726782152, |
| 772 |
0.35932681119782461, |
| 773 |
0.081823493762449573, |
| 774 |
0.14332226470169329, |
| 775 |
0.084297823823520024, |
| 776 |
0.53814864671831453, |
| 777 |
0.089215024911993401, |
| 778 |
0.78486196105372907] |
| 779 |
|
| 780 |
self.gen.seed(61731 + (24903<<32) + (614<<64) + (42143<<96)) |
| 781 |
actual = self.randomlist(2000)[-10:] |
| 782 |
for a, e in zip(actual, expected): |
| 783 |
self.assertAlmostEqual(a,e,places=14) |
| 784 |
|
| 785 |
def test_strong_reference_implementation(self): |
| 786 |
# Like test_referenceImplementation, but checks for exact bit-level |
| 787 |
# equality. This should pass on any box where C double contains |
| 788 |
# at least 53 bits of precision (the underlying algorithm suffers |
| 789 |
# no rounding errors -- all results are exact). |
| 790 |
from math import ldexp |
| 791 |
|
| 792 |
expected = [0x0eab3258d2231f, |
| 793 |
0x1b89db315277a5, |
| 794 |
0x1db622a5518016, |
| 795 |
0x0b7f9af0d575bf, |
| 796 |
0x029e4c4db82240, |
| 797 |
0x04961892f5d673, |
| 798 |
0x02b291598e4589, |
| 799 |
0x11388382c15694, |
| 800 |
0x02dad977c9e1fe, |
| 801 |
0x191d96d4d334c6] |
| 802 |
self.gen.seed(61731 + (24903<<32) + (614<<64) + (42143<<96)) |
| 803 |
actual = self.randomlist(2000)[-10:] |
| 804 |
for a, e in zip(actual, expected): |
| 805 |
self.assertEqual(int(ldexp(a, 53)), e) |
| 806 |
|
| 807 |
def test_long_seed(self): |
| 808 |
# This is most interesting to run in debug mode, just to make sure |
| 809 |
# nothing blows up. Under the covers, a dynamically resized array |
| 810 |
# is allocated, consuming space proportional to the number of bits |
| 811 |
# in the seed. Unfortunately, that's a quadratic-time algorithm, |
| 812 |
# so don't make this horribly big. |
| 813 |
seed = (1 << (10000 * 8)) - 1 # about 10K bytes |
| 814 |
self.gen.seed(seed) |
| 815 |
|
| 816 |
def test_getrandbits(self): |
| 817 |
super().test_getrandbits() |
| 818 |
|
| 819 |
# Verify cross-platform repeatability |
| 820 |
self.gen.seed(1234567) |
| 821 |
self.assertEqual(self.gen.getrandbits(100), |
| 822 |
97904845777343510404718956115) |
| 823 |
self.gen.seed(1234567) |
| 824 |
self.assertEqual(self.gen.getrandbits(MyIndex(100)), |
| 825 |
97904845777343510404718956115) |
| 826 |
|
| 827 |
def test_getrandbits_2G_bits(self): |
| 828 |
size = 2**31 |
| 829 |
self.gen.seed(1234567) |
| 830 |
x = self.gen.getrandbits(size) |
| 831 |
self.assertEqual(x.bit_length(), size) |
| 832 |
self.assertEqual(x & (2**100-1), 890186470919986886340158459475) |
| 833 |
self.assertEqual(x >> (size-100), 1226514312032729439655761284440) |
| 834 |
|
| 835 |
@support.bigmemtest(size=2**32, memuse=1/8+2/15, dry_run=False) |
| 836 |
def test_getrandbits_4G_bits(self, size): |
| 837 |
self.gen.seed(1234568) |
| 838 |
x = self.gen.getrandbits(size) |
| 839 |
self.assertEqual(x.bit_length(), size) |
| 840 |
self.assertEqual(x & (2**100-1), 287241425661104632871036099814) |
| 841 |
self.assertEqual(x >> (size-100), 739728759900339699429794460738) |
| 842 |
|
| 843 |
def test_randrange_uses_getrandbits(self): |
| 844 |
# Verify use of getrandbits by randrange |
| 845 |
# Use same seed as in the cross-platform repeatability test |
| 846 |
# in test_getrandbits above. |
| 847 |
self.gen.seed(1234567) |
| 848 |
# If randrange uses getrandbits, it should pick getrandbits(100) |
| 849 |
# when called with a 100-bits stop argument. |
| 850 |
self.assertEqual(self.gen.randrange(2**99), |
| 851 |
97904845777343510404718956115) |
| 852 |
|
| 853 |
def test_randbelow_without_getrandbits(self): |
| 854 |
# Random._randbelow() can only use random() when the built-in one |
| 855 |
# has been overridden but no new getrandbits() method was supplied. |
| 856 |
maxsize = 1<<random.BPF |
| 857 |
with warnings.catch_warnings(): |
| 858 |
warnings.simplefilter("ignore", UserWarning) |
| 859 |
# Population range too large (n >= maxsize) |
| 860 |
self.gen._randbelow_without_getrandbits( |
| 861 |
maxsize+1, maxsize=maxsize |
| 862 |
) |
| 863 |
self.gen._randbelow_without_getrandbits(5640, maxsize=maxsize) |
| 864 |
|
| 865 |
# This might be going too far to test a single line, but because of our |
| 866 |
# noble aim of achieving 100% test coverage we need to write a case in |
| 867 |
# which the following line in Random._randbelow() gets executed: |
| 868 |
# |
| 869 |
# rem = maxsize % n |
| 870 |
# limit = (maxsize - rem) / maxsize |
| 871 |
# r = random() |
| 872 |
# while r >= limit: |
| 873 |
# r = random() # <== *This line* <==< |
| 874 |
# |
| 875 |
# Therefore, to guarantee that the while loop is executed at least |
| 876 |
# once, we need to mock random() so that it returns a number greater |
| 877 |
# than 'limit' the first time it gets called. |
| 878 |
|
| 879 |
n = 42 |
| 880 |
epsilon = 0.01 |
| 881 |
limit = (maxsize - (maxsize % n)) / maxsize |
| 882 |
with unittest.mock.patch.object(random.Random, 'random') as random_mock: |
| 883 |
random_mock.side_effect = [limit + epsilon, limit - epsilon] |
| 884 |
self.gen._randbelow_without_getrandbits(n, maxsize=maxsize) |
| 885 |
self.assertEqual(random_mock.call_count, 2) |
| 886 |
|
| 887 |
def test_choices_algorithms(self): |
| 888 |
# The various ways of specifying weights should produce the same results |
| 889 |
choices = self.gen.choices |
| 890 |
n = 104729 |
| 891 |
|
| 892 |
self.gen.seed(8675309) |
| 893 |
a = self.gen.choices(range(n), k=10000) |
| 894 |
|
| 895 |
self.gen.seed(8675309) |
| 896 |
b = self.gen.choices(range(n), [1]*n, k=10000) |
| 897 |
self.assertEqual(a, b) |
| 898 |
|
| 899 |
self.gen.seed(8675309) |
| 900 |
c = self.gen.choices(range(n), cum_weights=range(1, n+1), k=10000) |
| 901 |
self.assertEqual(a, c) |
| 902 |
|
| 903 |
# American Roulette |
| 904 |
population = ['Red', 'Black', 'Green'] |
| 905 |
weights = [18, 18, 2] |
| 906 |
cum_weights = [18, 36, 38] |
| 907 |
expanded_population = ['Red'] * 18 + ['Black'] * 18 + ['Green'] * 2 |
| 908 |
|
| 909 |
self.gen.seed(9035768) |
| 910 |
a = self.gen.choices(expanded_population, k=10000) |
| 911 |
|
| 912 |
self.gen.seed(9035768) |
| 913 |
b = self.gen.choices(population, weights, k=10000) |
| 914 |
self.assertEqual(a, b) |
| 915 |
|
| 916 |
self.gen.seed(9035768) |
| 917 |
c = self.gen.choices(population, cum_weights=cum_weights, k=10000) |
| 918 |
self.assertEqual(a, c) |
| 919 |
|
| 920 |
def test_randbytes(self): |
| 921 |
super().test_randbytes() |
| 922 |
|
| 923 |
# Mersenne Twister randbytes() is deterministic |
| 924 |
# and does not depend on the endian and bitness. |
| 925 |
seed = 8675309 |
| 926 |
expected = b'3\xa8\xf9f\xf4\xa4\xd06\x19\x8f\x9f\x82\x02oe\xf0' |
| 927 |
|
| 928 |
self.gen.seed(seed) |
| 929 |
self.assertEqual(self.gen.randbytes(16), expected) |
| 930 |
|
| 931 |
# randbytes(0) must not consume any entropy |
| 932 |
self.gen.seed(seed) |
| 933 |
self.assertEqual(self.gen.randbytes(0), b'') |
| 934 |
self.assertEqual(self.gen.randbytes(16), expected) |
| 935 |
|
| 936 |
# Four randbytes(4) calls give the same output than randbytes(16) |
| 937 |
self.gen.seed(seed) |
| 938 |
self.assertEqual(b''.join([self.gen.randbytes(4) for _ in range(4)]), |
| 939 |
expected) |
| 940 |
|
| 941 |
# Each randbytes(1), randbytes(2) or randbytes(3) call consumes |
| 942 |
# 4 bytes of entropy |
| 943 |
self.gen.seed(seed) |
| 944 |
expected1 = expected[3::4] |
| 945 |
self.assertEqual(b''.join(self.gen.randbytes(1) for _ in range(4)), |
| 946 |
expected1) |
| 947 |
|
| 948 |
self.gen.seed(seed) |
| 949 |
expected2 = b''.join(expected[i + 2: i + 4] |
| 950 |
for i in range(0, len(expected), 4)) |
| 951 |
self.assertEqual(b''.join(self.gen.randbytes(2) for _ in range(4)), |
| 952 |
expected2) |
| 953 |
|
| 954 |
self.gen.seed(seed) |
| 955 |
expected3 = b''.join(expected[i + 1: i + 4] |
| 956 |
for i in range(0, len(expected), 4)) |
| 957 |
self.assertEqual(b''.join(self.gen.randbytes(3) for _ in range(4)), |
| 958 |
expected3) |
| 959 |
|
| 960 |
def test_randbytes_getrandbits(self): |
| 961 |
# There is a simple relation between randbytes() and getrandbits() |
| 962 |
seed = 2849427419 |
| 963 |
gen2 = random.Random() |
| 964 |
self.gen.seed(seed) |
| 965 |
gen2.seed(seed) |
| 966 |
for n in range(9): |
| 967 |
self.assertEqual(self.gen.randbytes(n), |
| 968 |
gen2.getrandbits(n * 8).to_bytes(n, 'little')) |
| 969 |
|
| 970 |
@support.bigmemtest(size=2**29, memuse=1+16/15, dry_run=False) |
| 971 |
def test_randbytes_256M(self, size): |
| 972 |
self.gen.seed(2849427419) |
| 973 |
x = self.gen.randbytes(size) |
| 974 |
self.assertEqual(len(x), size) |
| 975 |
self.assertEqual(x[:12].hex(), 'f6fd9ae63855ab91ea238b4f') |
| 976 |
self.assertEqual(x[-12:].hex(), '0e7af69a84ee99bf4a11becc') |
| 977 |
|
| 978 |
def test_sample_counts_equivalence(self): |
| 979 |
# Test the documented strong equivalence to a sample with repeated elements. |
| 980 |
# We run this test on random.Random() which makes deterministic selections |
| 981 |
# for a given seed value. |
| 982 |
sample = self.gen.sample |
| 983 |
seed = self.gen.seed |
| 984 |
|
| 985 |
colors = ['red', 'green', 'blue', 'orange', 'black', 'amber'] |
| 986 |
counts = [500, 200, 20, 10, 5, 1 ] |
| 987 |
k = 700 |
| 988 |
seed(8675309) |
| 989 |
s1 = sample(colors, counts=counts, k=k) |
| 990 |
seed(8675309) |
| 991 |
expanded = [color for (color, count) in zip(colors, counts) for i in range(count)] |
| 992 |
self.assertEqual(len(expanded), sum(counts)) |
| 993 |
s2 = sample(expanded, k=k) |
| 994 |
self.assertEqual(s1, s2) |
| 995 |
|
| 996 |
pop = 'abcdefghi' |
| 997 |
counts = [10, 9, 8, 7, 6, 5, 4, 3, 2] |
| 998 |
seed(8675309) |
| 999 |
s1 = ''.join(sample(pop, counts=counts, k=30)) |
| 1000 |
expanded = ''.join([letter for (letter, count) in zip(pop, counts) for i in range(count)]) |
| 1001 |
seed(8675309) |
| 1002 |
s2 = ''.join(sample(expanded, k=30)) |
| 1003 |
self.assertEqual(s1, s2) |
| 1004 |
|
| 1005 |
|
| 1006 |
def gamma(z, sqrt2pi=(2.0*pi)**0.5): |
| 1007 |
# Reflection to right half of complex plane |
| 1008 |
if z < 0.5: |
| 1009 |
return pi / sin(pi*z) / gamma(1.0-z) |
| 1010 |
# Lanczos approximation with g=7 |
| 1011 |
az = z + (7.0 - 0.5) |
| 1012 |
return az ** (z-0.5) / exp(az) * sqrt2pi * fsum([ |
| 1013 |
0.9999999999995183, |
| 1014 |
676.5203681218835 / z, |
| 1015 |
-1259.139216722289 / (z+1.0), |
| 1016 |
771.3234287757674 / (z+2.0), |
| 1017 |
-176.6150291498386 / (z+3.0), |
| 1018 |
12.50734324009056 / (z+4.0), |
| 1019 |
-0.1385710331296526 / (z+5.0), |
| 1020 |
0.9934937113930748e-05 / (z+6.0), |
| 1021 |
0.1659470187408462e-06 / (z+7.0), |
| 1022 |
]) |
| 1023 |
|
| 1024 |
class TestDistributions(unittest.TestCase): |
| 1025 |
def test_zeroinputs(self): |
| 1026 |
# Verify that distributions can handle a series of zero inputs' |
| 1027 |
g = random.Random() |
| 1028 |
x = [g.random() for i in range(50)] + [0.0]*5 |
| 1029 |
g.random = x[:].pop; g.uniform(1,10) |
| 1030 |
g.random = x[:].pop; g.paretovariate(1.0) |
| 1031 |
g.random = x[:].pop; g.expovariate(1.0) |
| 1032 |
g.random = x[:].pop; g.expovariate() |
| 1033 |
g.random = x[:].pop; g.weibullvariate(1.0, 1.0) |
| 1034 |
g.random = x[:].pop; g.vonmisesvariate(1.0, 1.0) |
| 1035 |
g.random = x[:].pop; g.normalvariate(0.0, 1.0) |
| 1036 |
g.random = x[:].pop; g.gauss(0.0, 1.0) |
| 1037 |
g.random = x[:].pop; g.lognormvariate(0.0, 1.0) |
| 1038 |
g.random = x[:].pop; g.vonmisesvariate(0.0, 1.0) |
| 1039 |
g.random = x[:].pop; g.gammavariate(0.01, 1.0) |
| 1040 |
g.random = x[:].pop; g.gammavariate(1.0, 1.0) |
| 1041 |
g.random = x[:].pop; g.gammavariate(200.0, 1.0) |
| 1042 |
g.random = x[:].pop; g.betavariate(3.0, 3.0) |
| 1043 |
g.random = x[:].pop; g.triangular(0.0, 1.0, 1.0/3.0) |
| 1044 |
|
| 1045 |
def test_avg_std(self): |
| 1046 |
# Use integration to test distribution average and standard deviation. |
| 1047 |
# Only works for distributions which do not consume variates in pairs |
| 1048 |
g = random.Random() |
| 1049 |
N = 5000 |
| 1050 |
x = [i/float(N) for i in range(1,N)] |
| 1051 |
for variate, args, mu, sigmasqrd in [ |
| 1052 |
(g.uniform, (1.0,10.0), (10.0+1.0)/2, (10.0-1.0)**2/12), |
| 1053 |
(g.triangular, (0.0, 1.0, 1.0/3.0), 4.0/9.0, 7.0/9.0/18.0), |
| 1054 |
(g.expovariate, (1.5,), 1/1.5, 1/1.5**2), |
| 1055 |
(g.vonmisesvariate, (1.23, 0), pi, pi**2/3), |
| 1056 |
(g.paretovariate, (5.0,), 5.0/(5.0-1), |
| 1057 |
5.0/((5.0-1)**2*(5.0-2))), |
| 1058 |
(g.weibullvariate, (1.0, 3.0), gamma(1+1/3.0), |
| 1059 |
gamma(1+2/3.0)-gamma(1+1/3.0)**2) ]: |
| 1060 |
g.random = x[:].pop |
| 1061 |
y = [] |
| 1062 |
for i in range(len(x)): |
| 1063 |
try: |
| 1064 |
y.append(variate(*args)) |
| 1065 |
except IndexError: |
| 1066 |
pass |
| 1067 |
s1 = s2 = 0 |
| 1068 |
for e in y: |
| 1069 |
s1 += e |
| 1070 |
s2 += (e - mu) ** 2 |
| 1071 |
N = len(y) |
| 1072 |
self.assertAlmostEqual(s1/N, mu, places=2, |
| 1073 |
msg='%s%r' % (variate.__name__, args)) |
| 1074 |
self.assertAlmostEqual(s2/(N-1), sigmasqrd, places=2, |
| 1075 |
msg='%s%r' % (variate.__name__, args)) |
| 1076 |
|
| 1077 |
def test_constant(self): |
| 1078 |
g = random.Random() |
| 1079 |
N = 100 |
| 1080 |
for variate, args, expected in [ |
| 1081 |
(g.uniform, (10.0, 10.0), 10.0), |
| 1082 |
(g.triangular, (10.0, 10.0), 10.0), |
| 1083 |
(g.triangular, (10.0, 10.0, 10.0), 10.0), |
| 1084 |
(g.expovariate, (float('inf'),), 0.0), |
| 1085 |
(g.vonmisesvariate, (3.0, float('inf')), 3.0), |
| 1086 |
(g.gauss, (10.0, 0.0), 10.0), |
| 1087 |
(g.lognormvariate, (0.0, 0.0), 1.0), |
| 1088 |
(g.lognormvariate, (-float('inf'), 0.0), 0.0), |
| 1089 |
(g.normalvariate, (10.0, 0.0), 10.0), |
| 1090 |
(g.binomialvariate, (0, 0.5), 0), |
| 1091 |
(g.binomialvariate, (10, 0.0), 0), |
| 1092 |
(g.binomialvariate, (10, 1.0), 10), |
| 1093 |
(g.paretovariate, (float('inf'),), 1.0), |
| 1094 |
(g.weibullvariate, (10.0, float('inf')), 10.0), |
| 1095 |
(g.weibullvariate, (0.0, 10.0), 0.0), |
| 1096 |
]: |
| 1097 |
for i in range(N): |
| 1098 |
self.assertEqual(variate(*args), expected) |
| 1099 |
|
| 1100 |
def test_binomialvariate(self): |
| 1101 |
B = random.binomialvariate |
| 1102 |
|
| 1103 |
# Cover all the code paths |
| 1104 |
with self.assertRaises(ValueError): |
| 1105 |
B(n=-1) # Negative n |
| 1106 |
with self.assertRaises(ValueError): |
| 1107 |
B(n=1, p=-0.5) # Negative p |
| 1108 |
with self.assertRaises(ValueError): |
| 1109 |
B(n=1, p=1.5) # p > 1.0 |
| 1110 |
self.assertEqual(B(0, 0.5), 0) # n == 0 |
| 1111 |
self.assertEqual(B(10, 0.0), 0) # p == 0.0 |
| 1112 |
self.assertEqual(B(10, 1.0), 10) # p == 1.0 |
| 1113 |
self.assertTrue(B(1, 0.3) in {0, 1}) # n == 1 fast path |
| 1114 |
self.assertTrue(B(1, 0.9) in {0, 1}) # n == 1 fast path |
| 1115 |
self.assertTrue(B(1, 0.0) in {0}) # n == 1 fast path |
| 1116 |
self.assertTrue(B(1, 1.0) in {1}) # n == 1 fast path |
| 1117 |
|
| 1118 |
# BG method very small p |
| 1119 |
self.assertEqual(B(5, 1e-18), 0) |
| 1120 |
|
| 1121 |
# BG method p <= 0.5 and n*p=1.25 |
| 1122 |
self.assertTrue(B(5, 0.25) in set(range(6))) |
| 1123 |
|
| 1124 |
# BG method p >= 0.5 and n*(1-p)=1.25 |
| 1125 |
self.assertTrue(B(5, 0.75) in set(range(6))) |
| 1126 |
|
| 1127 |
# BTRS method p <= 0.5 and n*p=25 |
| 1128 |
self.assertTrue(B(100, 0.25) in set(range(101))) |
| 1129 |
|
| 1130 |
# BTRS method p > 0.5 and n*(1-p)=25 |
| 1131 |
self.assertTrue(B(100, 0.75) in set(range(101))) |
| 1132 |
|
| 1133 |
# Statistical tests chosen such that they are |
| 1134 |
# exceedingly unlikely to ever fail for correct code. |
| 1135 |
|
| 1136 |
# BG code path |
| 1137 |
# Expected dist: [31641, 42188, 21094, 4688, 391] |
| 1138 |
c = Counter(B(4, 0.25) for i in range(100_000)) |
| 1139 |
self.assertTrue(29_641 <= c[0] <= 33_641, c) |
| 1140 |
self.assertTrue(40_188 <= c[1] <= 44_188) |
| 1141 |
self.assertTrue(19_094 <= c[2] <= 23_094) |
| 1142 |
self.assertTrue(2_688 <= c[3] <= 6_688) |
| 1143 |
self.assertEqual(set(c), {0, 1, 2, 3, 4}) |
| 1144 |
|
| 1145 |
# BTRS code path |
| 1146 |
# Sum of c[20], c[21], c[22], c[23], c[24] expected to be 36,214 |
| 1147 |
c = Counter(B(100, 0.25) for i in range(100_000)) |
| 1148 |
self.assertTrue(34_214 <= c[20]+c[21]+c[22]+c[23]+c[24] <= 38_214) |
| 1149 |
self.assertTrue(set(c) <= set(range(101))) |
| 1150 |
self.assertEqual(c.total(), 100_000) |
| 1151 |
|
| 1152 |
# Demonstrate the BTRS works for huge values of n |
| 1153 |
self.assertTrue(19_000_000 <= B(100_000_000, 0.2) <= 21_000_000) |
| 1154 |
self.assertTrue(89_000_000 <= B(100_000_000, 0.9) <= 91_000_000) |
| 1155 |
|
| 1156 |
|
| 1157 |
def test_von_mises_range(self): |
| 1158 |
# Issue 17149: von mises variates were not consistently in the |
| 1159 |
# range [0, 2*PI]. |
| 1160 |
g = random.Random() |
| 1161 |
N = 100 |
| 1162 |
for mu in 0.0, 0.1, 3.1, 6.2: |
| 1163 |
for kappa in 0.0, 2.3, 500.0: |
| 1164 |
for _ in range(N): |
| 1165 |
sample = g.vonmisesvariate(mu, kappa) |
| 1166 |
self.assertTrue( |
| 1167 |
0 <= sample <= random.TWOPI, |
| 1168 |
msg=("vonmisesvariate({}, {}) produced a result {} out" |
| 1169 |
" of range [0, 2*pi]").format(mu, kappa, sample)) |
| 1170 |
|
| 1171 |
def test_von_mises_large_kappa(self): |
| 1172 |
# Issue #17141: vonmisesvariate() was hang for large kappas |
| 1173 |
random.vonmisesvariate(0, 1e15) |
| 1174 |
random.vonmisesvariate(0, 1e100) |
| 1175 |
|
| 1176 |
def test_gammavariate_errors(self): |
| 1177 |
# Both alpha and beta must be > 0.0 |
| 1178 |
self.assertRaises(ValueError, random.gammavariate, -1, 3) |
| 1179 |
self.assertRaises(ValueError, random.gammavariate, 0, 2) |
| 1180 |
self.assertRaises(ValueError, random.gammavariate, 2, 0) |
| 1181 |
self.assertRaises(ValueError, random.gammavariate, 1, -3) |
| 1182 |
|
| 1183 |
# There are three different possibilities in the current implementation |
| 1184 |
# of random.gammavariate(), depending on the value of 'alpha'. What we |
| 1185 |
# are going to do here is to fix the values returned by random() to |
| 1186 |
# generate test cases that provide 100% line coverage of the method. |
| 1187 |
@unittest.mock.patch('random.Random.random') |
| 1188 |
def test_gammavariate_alpha_greater_one(self, random_mock): |
| 1189 |
|
| 1190 |
# #1: alpha > 1.0. |
| 1191 |
# We want the first random number to be outside the |
| 1192 |
# [1e-7, .9999999] range, so that the continue statement executes |
| 1193 |
# once. The values of u1 and u2 will be 0.5 and 0.3, respectively. |
| 1194 |
random_mock.side_effect = [1e-8, 0.5, 0.3] |
| 1195 |
returned_value = random.gammavariate(1.1, 2.3) |
| 1196 |
self.assertAlmostEqual(returned_value, 2.53) |
| 1197 |
|
| 1198 |
@unittest.mock.patch('random.Random.random') |
| 1199 |
def test_gammavariate_alpha_equal_one(self, random_mock): |
| 1200 |
|
| 1201 |
# #2.a: alpha == 1. |
| 1202 |
# The execution body of the while loop executes once. |
| 1203 |
# Then random.random() returns 0.45, |
| 1204 |
# which causes while to stop looping and the algorithm to terminate. |
| 1205 |
random_mock.side_effect = [0.45] |
| 1206 |
returned_value = random.gammavariate(1.0, 3.14) |
| 1207 |
self.assertAlmostEqual(returned_value, 1.877208182372648) |
| 1208 |
|
| 1209 |
@unittest.mock.patch('random.Random.random') |
| 1210 |
def test_gammavariate_alpha_equal_one_equals_expovariate(self, random_mock): |
| 1211 |
|
| 1212 |
# #2.b: alpha == 1. |
| 1213 |
# It must be equivalent of calling expovariate(1.0 / beta). |
| 1214 |
beta = 3.14 |
| 1215 |
random_mock.side_effect = [1e-8, 1e-8] |
| 1216 |
gammavariate_returned_value = random.gammavariate(1.0, beta) |
| 1217 |
expovariate_returned_value = random.expovariate(1.0 / beta) |
| 1218 |
self.assertAlmostEqual(gammavariate_returned_value, expovariate_returned_value) |
| 1219 |
|
| 1220 |
@unittest.mock.patch('random.Random.random') |
| 1221 |
def test_gammavariate_alpha_between_zero_and_one(self, random_mock): |
| 1222 |
|
| 1223 |
# #3: 0 < alpha < 1. |
| 1224 |
# This is the most complex region of code to cover, |
| 1225 |
# as there are multiple if-else statements. Let's take a look at the |
| 1226 |
# source code, and determine the values that we need accordingly: |
| 1227 |
# |
| 1228 |
# while 1: |
| 1229 |
# u = random() |
| 1230 |
# b = (_e + alpha)/_e |
| 1231 |
# p = b*u |
| 1232 |
# if p <= 1.0: # <=== (A) |
| 1233 |
# x = p ** (1.0/alpha) |
| 1234 |
# else: # <=== (B) |
| 1235 |
# x = -_log((b-p)/alpha) |
| 1236 |
# u1 = random() |
| 1237 |
# if p > 1.0: # <=== (C) |
| 1238 |
# if u1 <= x ** (alpha - 1.0): # <=== (D) |
| 1239 |
# break |
| 1240 |
# elif u1 <= _exp(-x): # <=== (E) |
| 1241 |
# break |
| 1242 |
# return x * beta |
| 1243 |
# |
| 1244 |
# First, we want (A) to be True. For that we need that: |
| 1245 |
# b*random() <= 1.0 |
| 1246 |
# r1 = random() <= 1.0 / b |
| 1247 |
# |
| 1248 |
# We now get to the second if-else branch, and here, since p <= 1.0, |
| 1249 |
# (C) is False and we take the elif branch, (E). For it to be True, |
| 1250 |
# so that the break is executed, we need that: |
| 1251 |
# r2 = random() <= _exp(-x) |
| 1252 |
# r2 <= _exp(-(p ** (1.0/alpha))) |
| 1253 |
# r2 <= _exp(-((b*r1) ** (1.0/alpha))) |
| 1254 |
|
| 1255 |
_e = random._e |
| 1256 |
_exp = random._exp |
| 1257 |
_log = random._log |
| 1258 |
alpha = 0.35 |
| 1259 |
beta = 1.45 |
| 1260 |
b = (_e + alpha)/_e |
| 1261 |
epsilon = 0.01 |
| 1262 |
|
| 1263 |
r1 = 0.8859296441566 # 1.0 / b |
| 1264 |
r2 = 0.3678794411714 # _exp(-((b*r1) ** (1.0/alpha))) |
| 1265 |
|
| 1266 |
# These four "random" values result in the following trace: |
| 1267 |
# (A) True, (E) False --> [next iteration of while] |
| 1268 |
# (A) True, (E) True --> [while loop breaks] |
| 1269 |
random_mock.side_effect = [r1, r2 + epsilon, r1, r2] |
| 1270 |
returned_value = random.gammavariate(alpha, beta) |
| 1271 |
self.assertAlmostEqual(returned_value, 1.4499999999997544) |
| 1272 |
|
| 1273 |
# Let's now make (A) be False. If this is the case, when we get to the |
| 1274 |
# second if-else 'p' is greater than 1, so (C) evaluates to True. We |
| 1275 |
# now encounter a second if statement, (D), which in order to execute |
| 1276 |
# must satisfy the following condition: |
| 1277 |
# r2 <= x ** (alpha - 1.0) |
| 1278 |
# r2 <= (-_log((b-p)/alpha)) ** (alpha - 1.0) |
| 1279 |
# r2 <= (-_log((b-(b*r1))/alpha)) ** (alpha - 1.0) |
| 1280 |
r1 = 0.8959296441566 # (1.0 / b) + epsilon -- so that (A) is False |
| 1281 |
r2 = 0.9445400408898141 |
| 1282 |
|
| 1283 |
# And these four values result in the following trace: |
| 1284 |
# (B) and (C) True, (D) False --> [next iteration of while] |
| 1285 |
# (B) and (C) True, (D) True [while loop breaks] |
| 1286 |
random_mock.side_effect = [r1, r2 + epsilon, r1, r2] |
| 1287 |
returned_value = random.gammavariate(alpha, beta) |
| 1288 |
self.assertAlmostEqual(returned_value, 1.5830349561760781) |
| 1289 |
|
| 1290 |
@unittest.mock.patch('random.Random.gammavariate') |
| 1291 |
def test_betavariate_return_zero(self, gammavariate_mock): |
| 1292 |
# betavariate() returns zero when the Gamma distribution |
| 1293 |
# that it uses internally returns this same value. |
| 1294 |
gammavariate_mock.return_value = 0.0 |
| 1295 |
self.assertEqual(0.0, random.betavariate(2.71828, 3.14159)) |
| 1296 |
|
| 1297 |
|
| 1298 |
class TestRandomSubclassing(unittest.TestCase): |
| 1299 |
def test_random_subclass_with_kwargs(self): |
| 1300 |
# SF bug #1486663 -- this used to erroneously raise a TypeError |
| 1301 |
class Subclass(random.Random): |
| 1302 |
def __init__(self, newarg=None): |
| 1303 |
random.Random.__init__(self) |
| 1304 |
Subclass(newarg=1) |
| 1305 |
|
| 1306 |
def test_subclasses_overriding_methods(self): |
| 1307 |
# Subclasses with an overridden random, but only the original |
| 1308 |
# getrandbits method should not rely on getrandbits in for randrange, |
| 1309 |
# but should use a getrandbits-independent implementation instead. |
| 1310 |
|
| 1311 |
# subclass providing its own random **and** getrandbits methods |
| 1312 |
# like random.SystemRandom does => keep relying on getrandbits for |
| 1313 |
# randrange |
| 1314 |
class SubClass1(random.Random): |
| 1315 |
def random(self): |
| 1316 |
called.add('SubClass1.random') |
| 1317 |
return random.Random.random(self) |
| 1318 |
|
| 1319 |
def getrandbits(self, n): |
| 1320 |
called.add('SubClass1.getrandbits') |
| 1321 |
return random.Random.getrandbits(self, n) |
| 1322 |
called = set() |
| 1323 |
SubClass1().randrange(42) |
| 1324 |
self.assertEqual(called, {'SubClass1.getrandbits'}) |
| 1325 |
|
| 1326 |
# subclass providing only random => can only use random for randrange |
| 1327 |
class SubClass2(random.Random): |
| 1328 |
def random(self): |
| 1329 |
called.add('SubClass2.random') |
| 1330 |
return random.Random.random(self) |
| 1331 |
called = set() |
| 1332 |
SubClass2().randrange(42) |
| 1333 |
self.assertEqual(called, {'SubClass2.random'}) |
| 1334 |
|
| 1335 |
# subclass defining getrandbits to complement its inherited random |
| 1336 |
# => can now rely on getrandbits for randrange again |
| 1337 |
class SubClass3(SubClass2): |
| 1338 |
def getrandbits(self, n): |
| 1339 |
called.add('SubClass3.getrandbits') |
| 1340 |
return random.Random.getrandbits(self, n) |
| 1341 |
called = set() |
| 1342 |
SubClass3().randrange(42) |
| 1343 |
self.assertEqual(called, {'SubClass3.getrandbits'}) |
| 1344 |
|
| 1345 |
# subclass providing only random and inherited getrandbits |
| 1346 |
# => random takes precedence |
| 1347 |
class SubClass4(SubClass3): |
| 1348 |
def random(self): |
| 1349 |
called.add('SubClass4.random') |
| 1350 |
return random.Random.random(self) |
| 1351 |
called = set() |
| 1352 |
SubClass4().randrange(42) |
| 1353 |
self.assertEqual(called, {'SubClass4.random'}) |
| 1354 |
|
| 1355 |
# Following subclasses don't define random or getrandbits directly, |
| 1356 |
# but inherit them from classes which are not subclasses of Random |
| 1357 |
class Mixin1: |
| 1358 |
def random(self): |
| 1359 |
called.add('Mixin1.random') |
| 1360 |
return random.Random.random(self) |
| 1361 |
class Mixin2: |
| 1362 |
def getrandbits(self, n): |
| 1363 |
called.add('Mixin2.getrandbits') |
| 1364 |
return random.Random.getrandbits(self, n) |
| 1365 |
|
| 1366 |
class SubClass5(Mixin1, random.Random): |
| 1367 |
pass |
| 1368 |
called = set() |
| 1369 |
SubClass5().randrange(42) |
| 1370 |
self.assertEqual(called, {'Mixin1.random'}) |
| 1371 |
|
| 1372 |
class SubClass6(Mixin2, random.Random): |
| 1373 |
pass |
| 1374 |
called = set() |
| 1375 |
SubClass6().randrange(42) |
| 1376 |
self.assertEqual(called, {'Mixin2.getrandbits'}) |
| 1377 |
|
| 1378 |
class SubClass7(Mixin1, Mixin2, random.Random): |
| 1379 |
pass |
| 1380 |
called = set() |
| 1381 |
SubClass7().randrange(42) |
| 1382 |
self.assertEqual(called, {'Mixin1.random'}) |
| 1383 |
|
| 1384 |
class SubClass8(Mixin2, Mixin1, random.Random): |
| 1385 |
pass |
| 1386 |
called = set() |
| 1387 |
SubClass8().randrange(42) |
| 1388 |
self.assertEqual(called, {'Mixin2.getrandbits'}) |
| 1389 |
|
| 1390 |
|
| 1391 |
class TestModule(unittest.TestCase): |
| 1392 |
def testMagicConstants(self): |
| 1393 |
self.assertAlmostEqual(random.NV_MAGICCONST, 1.71552776992141) |
| 1394 |
self.assertAlmostEqual(random.TWOPI, 6.28318530718) |
| 1395 |
self.assertAlmostEqual(random.LOG4, 1.38629436111989) |
| 1396 |
self.assertAlmostEqual(random.SG_MAGICCONST, 2.50407739677627) |
| 1397 |
|
| 1398 |
def test__all__(self): |
| 1399 |
# tests validity but not completeness of the __all__ list |
| 1400 |
self.assertTrue(set(random.__all__) <= set(dir(random))) |
| 1401 |
|
| 1402 |
@test.support.requires_fork() |
| 1403 |
def test_after_fork(self): |
| 1404 |
# Test the global Random instance gets reseeded in child |
| 1405 |
r, w = os.pipe() |
| 1406 |
pid = os.fork() |
| 1407 |
if pid == 0: |
| 1408 |
# child process |
| 1409 |
try: |
| 1410 |
val = random.getrandbits(128) |
| 1411 |
with open(w, "w") as f: |
| 1412 |
f.write(str(val)) |
| 1413 |
finally: |
| 1414 |
os._exit(0) |
| 1415 |
else: |
| 1416 |
# parent process |
| 1417 |
os.close(w) |
| 1418 |
val = random.getrandbits(128) |
| 1419 |
with open(r, "r") as f: |
| 1420 |
child_val = eval(f.read()) |
| 1421 |
self.assertNotEqual(val, child_val) |
| 1422 |
|
| 1423 |
support.wait_process(pid, exitcode=0) |
| 1424 |
|
| 1425 |
|
| 1426 |
class CommandLineTest(unittest.TestCase): |
| 1427 |
@support.force_not_colorized |
| 1428 |
def test_parse_args(self): |
| 1429 |
args, help_text = random._parse_args(shlex.split("--choice a b c")) |
| 1430 |
self.assertEqual(args.choice, ["a", "b", "c"]) |
| 1431 |
self.assertStartsWith(help_text, "usage: ") |
| 1432 |
|
| 1433 |
args, help_text = random._parse_args(shlex.split("--integer 5")) |
| 1434 |
self.assertEqual(args.integer, 5) |
| 1435 |
self.assertStartsWith(help_text, "usage: ") |
| 1436 |
|
| 1437 |
args, help_text = random._parse_args(shlex.split("--float 2.5")) |
| 1438 |
self.assertEqual(args.float, 2.5) |
| 1439 |
self.assertStartsWith(help_text, "usage: ") |
| 1440 |
|
| 1441 |
args, help_text = random._parse_args(shlex.split("a b c")) |
| 1442 |
self.assertEqual(args.input, ["a", "b", "c"]) |
| 1443 |
self.assertStartsWith(help_text, "usage: ") |
| 1444 |
|
| 1445 |
args, help_text = random._parse_args(shlex.split("5")) |
| 1446 |
self.assertEqual(args.input, ["5"]) |
| 1447 |
self.assertStartsWith(help_text, "usage: ") |
| 1448 |
|
| 1449 |
args, help_text = random._parse_args(shlex.split("2.5")) |
| 1450 |
self.assertEqual(args.input, ["2.5"]) |
| 1451 |
self.assertStartsWith(help_text, "usage: ") |
| 1452 |
|
| 1453 |
def test_main(self): |
| 1454 |
for command, expected in [ |
| 1455 |
("--choice a b c", "b"), |
| 1456 |
('"a b c"', "b"), |
| 1457 |
("a b c", "b"), |
| 1458 |
("--choice 'a a' 'b b' 'c c'", "b b"), |
| 1459 |
("'a a' 'b b' 'c c'", "b b"), |
| 1460 |
("--integer 5", 4), |
| 1461 |
("5", 4), |
| 1462 |
("--float 2.5", 2.1110546288126204), |
| 1463 |
("2.5", 2.1110546288126204), |
| 1464 |
]: |
| 1465 |
random.seed(0) |
| 1466 |
self.assertEqual(random.main(shlex.split(command)), expected) |
| 1467 |
|
| 1468 |
|
| 1469 |
if __name__ == "__main__": |
| 1470 |
unittest.main() |