''' Time to build, resize, and deallocate sets.

* turn-off hash randomization for better comparability
  PYTHONHASHSEED=3385876387

* need case that has many redundant, non-identical strings
  (because specialization only applies to unicode, inexact matches)
  and vary the set sizes to explore cache effects
'''

from string import ascii_letters as letters
from random import choice, seed, shuffle, randrange
from pprint import pprint
from timeit import Timer
import sys

seed('the tale of two cities it was the best of times')

def makeuniq(words):
    words = '~'.join(words).split('~')
    assert len(set(map(id, words))) == len(words)
    return words

def makeuniqints(ints):
    ints = [i * 1 for i in ints]
    assert len(set(map(id, ints))) == len(ints)
    return ints

def makewordlist(size, redundancy=5, wsize=5, identical=True):
    npool = size // redundancy
    pool = [''.join(choice(letters) for i in range(wsize)) for j in range(npool)]
    words = pool * redundancy
    if not identical:
        words = makeuniq(words)
    shuffle(words)
    return words

def makeintlist(size, redundancy=5, wsize=9, identical=True):
    irange = 10 ** wsize
    npool = size // redundancy
    pool = [randrange(1000, irange) for j in range(npool)]
    ints = pool * redundancy
    if not identical:
        ints = makeuniqints(ints)
    shuffle(ints)
    return ints

trials = 7
n = 10000000
print('  size      dup words   dup ints   uniq words   uniq ints')
print('---------- ----------- ----------- ----------- -----------')
for size in [100, 1000, 10000, 100000, 1000000]:
    setup = 'from __main__ import makewordlist, makeintlist\n'
    setup += 's = makewordlist(%d, identical=True)\n' % size
    setup += 't = makeintlist(%d, identical=True)\n' % size
    setup += 'u = makewordlist(%d, identical=False)\n' % size
    setup += 'v = makeintlist(%d, identical=False)\n' % size
    setup += '_set = set\n'
    reps = n // size
    t1 = min(Timer('_set(s)', setup).repeat(trials, reps))
    t2 = min(Timer('_set(t)', setup).repeat(trials, reps))
    t3 = min(Timer('_set(u)', setup).repeat(trials, reps))
    t4 = min(Timer('_set(v)', setup).repeat(trials, reps))
    print('%-10d   %7.05f     %7.05f     %7.05f     %7.05f' % (size, t1, t2, t3, t4))