util.py 1.5 KB

12345678910111213141516171819202122232425262728293031323334353637383940
  1. from random import random
  2. chars = [str(d) for d in range(1, 10)]
  3. ps = [2. for _ in chars]
  4. letter_dist = [("E", 21912, 12.02), ("T", 16587, 9.1), ("A", 14810, 8.12), ("O", 14003, 7.68), ("I", 13318, 7.31),
  5. ("N", 12666, 6.95), ("S", 11450, 6.28), ("R", 10977, 6.02), ("H", 10795, 5.92), ("D", 7874, 4.32),
  6. ("L", 7253, 3.98), ("U", 5246, 2.88), ("C", 4943, 2.71), ("M", 4761, 2.61), ("F", 4200, 2.3),
  7. ("Y", 3853, 2.11), ("W", 3819, 2.09), ("G", 3693, 2.03), ("P", 3316, 1.82), ("B", 2715, 1.49),
  8. ("V", 2019, 1.11), ("K", 1257, 0.69), ("X", 315, 0.17), ("Q", 205, 0.11), ("J", 188, 0.1),
  9. ("Z", 128, 0.07), ]
  10. sp = sum(ps)
  11. for row in letter_dist:
  12. chars.append(row[0])
  13. ps.append(float(row[2]))
  14. ps = [p / sum(ps) for p in ps]
  15. def choice(sequence, probabilities):
  16. # if sum(probabilities) != 1:
  17. # raise AssertionError('Probabilities must sum up to 1')
  18. r = random()
  19. for idx, c in enumerate(sequence):
  20. r -= probabilities[idx]
  21. if r < 0:
  22. return c
  23. raise AssertionError('Probabilities must sum up to 1')
  24. def random_chars(count):
  25. return ''.join(choice(chars, probabilities=ps) for _ in range(count))
  26. def str2bool(v):
  27. v = str(v).strip().lower()
  28. if v in ["yes", 'y' "true", "t", "1"]:
  29. return True
  30. if v in ["no", 'n' "false", "f", "0", '', 'null', 'none']:
  31. return False
  32. raise ValueError('Can not convert `' + v + '` to bool')