Module: PropCheck::Generators

Defined in:
lib/prop_check/generators.rb

Overview

Contains common generators. Use this module by including it in the class (e.g. in your test suite) where you want to use them.

Constant Summary collapse

@@special_floats =
[Float::NAN, Float::INFINITY, -Float::INFINITY, Float::MAX, Float::MIN, 0.0.next_float, 0.0.prev_float]
@@alphanumeric_chars =
[('a'..'z'), ('A'..'Z'), ('0'..'9')].flat_map(&:to_a).freeze
@@printable_ascii_chars =
(' '..'~').to_a.freeze
@@ascii_chars =
[
  @@printable_ascii_chars,
  [
    "\n",
    "\r",
    "\t",
    "\v",
    "\b",
    "\f",
    "\e",
    "\d",
    "\a"
  ]
].flat_map(&:to_a).freeze
@@printable_chars =
[
  @@ascii_chars,
  "\u{A0}".."\u{D7FF}",
  "\u{E000}".."\u{FFFD}",
  "\u{10000}".."\u{10FFFF}"
].flat_map(&:to_a).freeze

Class Method Summary collapse

Class Method Details

.alphanumeric_charObject

Generates a single-character string containing one of a..z, A..Z, 0..9

Shrinks towards lowercase 'a'.

>> Generators.alphanumeric_char.sample(5, size: 10, rng: Random.new(42))
=> ["M", "Z", "C", "o", "Q"]

361
362
363
# File 'lib/prop_check/generators.rb', line 361

def alphanumeric_char
  one_of(*@@alphanumeric_chars.map(&method(:constant)))
end

.alphanumeric_string(**kwargs) ⇒ Object

Generates a string containing only the characters a..z, A..Z, 0..9

Shrinks towards fewer characters, and towards lowercase 'a'.

>> Generators.alphanumeric_string.sample(5, size: 10, rng: Random.new(42))
=> ["ZCoQ", "8uM", "wkkx0JNx", "v0bxRDLb", "Gl5v8RyWA6"]

Accepts the same options as `array`


375
376
377
# File 'lib/prop_check/generators.rb', line 375

def alphanumeric_string(**kwargs)
  array(alphanumeric_char, **kwargs).map(&:join)
end

.array(element_generator, min: 0, max: nil, empty: true, uniq: false) ⇒ Object

Generates an array of elements, where each of the elements is generated by `element_generator`.

Shrinks to shorter arrays (with shrunken elements). Accepted keyword arguments:

`empty:` When false, behaves the same as `min: 1` `min:` Ensures at least this many elements are generated. (default: 0) `max:` Ensures at most this many elements are generated. When nil, an arbitrary count is used instead. (default: nil) `uniq:` When `true`, ensures that all elements in the array are unique.

     When given a proc, uses the result of this proc to check for uniqueness.
     (matching the behaviour of `Array#uniq`)
     If it is not possible to generate another unique value after the configured `max_consecutive_attempts`
     an `PropCheck::Errors::GeneratorExhaustedError` will be raised.
     (default: `false`)

>> Generators.array(Generators.positive_integer).sample(5, size: 1, rng: Random.new(42))
=>  [[2], [2], [2], [1], [2]]
>> Generators.array(Generators.positive_integer).sample(5, size: 10, rng: Random.new(42))
=> [[10, 5, 1, 4], [5, 9, 1, 1, 11, 8, 4, 9, 11, 10], [6], [11, 11, 2, 2, 7, 2, 6, 5, 5], [2, 10, 9, 7, 9, 5, 11, 3]]

>> Generators.array(Generators.positive_integer, empty: true).sample(5, size: 1, rng: Random.new(1))
=>  [[], [2], [], [], [2]]
>> Generators.array(Generators.positive_integer, empty: false).sample(5, size: 1, rng: Random.new(1))
=>  [[2], [1], [2], [1], [1]]

>> Generators.array(Generators.boolean, uniq: true).sample(5, rng: Random.new(1))
=> [[true, false], [false, true], [true, false], [false, true], [false, true]]

265
266
267
268
269
270
271
272
273
274
# File 'lib/prop_check/generators.rb', line 265

def array(element_generator, min: 0, max: nil, empty: true, uniq: false)
  min = 1 if min.zero? && !empty
  uniq = proc { |x| x } if uniq == true

  if max.nil?
    nonnegative_integer.bind { |count| make_array(element_generator, min, count, uniq) }
  else
    make_array(element_generator, min, max, uniq)
  end
end

.ascii_charObject

Generates a single-character string from the printable ASCII character set.

Shrinks towards 'n'.

>> Generators.ascii_char.sample(size: 10, rng: Random.new(42))
=> ["d", "S", "|", ".", "g", "\\", "4", "d", "r", "v"]

430
431
432
# File 'lib/prop_check/generators.rb', line 430

def ascii_char
  one_of(*@@ascii_chars.map(&method(:constant)))
end

.ascii_string(**kwargs) ⇒ Object

Generates strings from the printable ASCII character set.

Shrinks towards fewer characters, and towards 'n'.

>> Generators.ascii_string.sample(5, size: 10, rng: Random.new(42))
=> ["S|.g", "drvjjw\b\a7\"", "!w=E!_[[email protected]", "x", "zZI{[o"]

Accepts the same options as `array`


444
445
446
# File 'lib/prop_check/generators.rb', line 444

def ascii_string(**kwargs)
  array(ascii_char, **kwargs).map(&:join)
end

.booleanObject

Generates either `true` or `false`

Shrinks towards `false`

>> Generators.boolean.sample(5, size: 10, rng: Random.new(42))
=> [false, true, false, false, false]

516
517
518
# File 'lib/prop_check/generators.rb', line 516

def boolean
  one_of(constant(false), constant(true))
end

.charObject

Generates a single unicode character (both printable and non-printable).

Shrinks towards characters with lower codepoints, e.g. ASCII

>> Generators.printable_char.sample(size: 10, rng: Random.new(42))
=> ["吏", "", "", "", "", "", "", "", "", "Ȍ"]

489
490
491
492
493
# File 'lib/prop_check/generators.rb', line 489

def char
  choose(0..0x10FFFF).map do |num|
    [num].pack('U')
  end
end

.choose(range) ⇒ Object

Returns a random integer in the given range (if a range is given) or between 0..num (if a single integer is given).

Does not scale when `size` changes. This means `choose` is useful for e.g. picking an element out of multiple possibilities, but for other purposes you probably want to use `integer` et co.

Shrinks to integers closer to zero.

>> r = Random.new(42); Generators.choose(0..5).sample(size: 10, rng: r)
=> [3, 4, 2, 4, 4, 1, 2, 2, 2, 4]
>> r = Random.new(42); Generators.choose(0..5).sample(size: 20000, rng: r)
=> [3, 4, 2, 4, 4, 1, 2, 2, 2, 4]

62
63
64
65
66
67
# File 'lib/prop_check/generators.rb', line 62

def choose(range)
  Generator.new do |rng:, **|
    val = rng.rand(range)
    LazyTree.new(val, integer_shrink(val))
  end
end

.constant(val) ⇒ Object

Always returns the same value, regardless of `size` or `rng` (random number generator state)

No shrinking (only considers the current single value `val`).

>> Generators.constant("pie").sample(5, size: 10, rng: Random.new(42))
=> ["pie", "pie", "pie", "pie", "pie"]

21
22
23
# File 'lib/prop_check/generators.rb', line 21

def constant(val)
  Generator.wrap(val)
end

.falseyObject

Generates `nil` or `false`.

Shrinks towards `nil`.

>> Generators.falsey.sample(5, size: 10, rng: Random.new(42))
=> [nil, false, nil, nil, nil]

538
539
540
# File 'lib/prop_check/generators.rb', line 538

def falsey
  one_of(constant(nil), constant(false))
end

.fixed_hash(hash) ⇒ Object

Given a `hash` where the values are generators, creates a generator that returns hashes with the same keys, and their corresponding values from their corresponding generators.

Shrinks element generators.

>> Generators.fixed_hash(a: Generators.integer(), b: Generators.real_float(), c: Generators.integer()).call(size: 10, rng: Random.new(42))
=> {:a=>-4, :b=>13.0, :c=>-3}

224
225
226
227
228
229
230
231
232
# File 'lib/prop_check/generators.rb', line 224

def fixed_hash(hash)
  keypair_generators =
    hash.map do |key, generator|
      generator.map { |val| [key, val] }
    end

  tuple(*keypair_generators)
    .map(&:to_h)
end

.floatObject

Generates floating-point numbers Will generate NaN, Infinity, -Infinity, as well as Float::EPSILON, Float::MAX, Float::MIN, 0.0.next_float, 0.0.prev_float, to test the handling of floating-point edge cases. Approx. 1/100 generated numbers is a special one.

Shrinks to smaller, real floats.

>> Generators.float().sample(10, size: 10, rng: Random.new(42))
=> [4.0, 9.555555555555555, 0.0, -Float::INFINITY, 5.5, -5.818181818181818, 1.1428571428571428, 0.0, 8.0, 7.857142857142858]

161
162
163
# File 'lib/prop_check/generators.rb', line 161

def float
  frequency(99 => real_float, 1 => one_of(*@@special_floats.map(&method(:constant))))
end

.frequency(frequencies) ⇒ Object

Picks one of the choices given in `frequencies` at random every time. `frequencies` expects keys to be numbers (representing the relative frequency of this generator) and values to be generators.

Side note: If you want to use the same frequency number for multiple generators, Ruby syntax requires you to send an array of two-element arrays instead of a hash.

Shrinks to arbitrary elements (since hashes are not ordered).

>> Generators.frequency(5 => Generators.integer, 1 => Generators.printable_ascii_char).sample(size: 10, rng: Random.new(42))
=> [4, -3, 10, 8, 0, -7, 10, 1, "E", 10]

191
192
193
194
195
196
197
# File 'lib/prop_check/generators.rb', line 191

def frequency(frequencies)
  choices = frequencies.reduce([]) do |acc, elem|
    freq, val = elem
    acc + ([val] * freq)
  end
  one_of(*choices)
end

.hash(*args, **kwargs) ⇒ Object

Generates a hash of key->values, where each of the keys is made using the `key_generator` and each of the values using the `value_generator`.

Shrinks to hashes with less key/value pairs.

>> Generators.hash(Generators.printable_ascii_string, Generators.positive_integer).sample(5, size: 3, rng: Random.new(42))
=> [{""=>2, "g\\4"=>4, "rv"=>2}, {"7"=>2}, {"!"=>1, "E!"=>1}, {"kY5"=>2}, {}]

335
336
337
338
339
340
341
# File 'lib/prop_check/generators.rb', line 335

def hash(*args, **kwargs)
  if args.length == 2
    hash_of(*args, **kwargs)
  else
    super
  end
end

.hash_of(key_generator, value_generator, **kwargs) ⇒ Object

Alias for `#hash` that does not conflict with a possibly overriden `Object#hash`.


347
348
349
350
# File 'lib/prop_check/generators.rb', line 347

def hash_of(key_generator, value_generator, **kwargs)
  array(tuple(key_generator, value_generator), **kwargs)
    .map(&:to_h)
end

.instance(klass, *args, **kwargs) ⇒ Object

Generates an instance of `klass` using `args` and/or `kwargs` as generators for the arguments that are passed to `klass.new`

## Example:

Given a class like this:

 class User
   attr_accessor :name, :age
   def initialize(name: , age: )
     @name = name
     @age = age
   end

   def inspect
     "<User name: #{@name.inspect}, age: #{@age.inspect}>"
   end
 end

>> user_gen = Generators.instance(User, name: Generators.printable_ascii_string, age: Generators.nonnegative_integer)
>> user_gen.sample(3, rng: Random.new(42)).inspect
=> "[<User name: \"S|.g\", age: 10>, <User name: \"rvjj\", age: 10>, <User name: \"7\\\"5T!w=\", age: 5>]"

616
617
618
619
620
621
622
623
624
625
626
627
628
# File 'lib/prop_check/generators.rb', line 616

def instance(klass, *args, **kwargs)
  tuple(*args).bind do |vals|
    fixed_hash(**kwargs).map do |kwvals|
      if kwvals == {}
        klass.new(*vals)
      elsif vals == []
        klass.new(**kwvals)
      else
        klass.new(*vals, **kwvals)
      end
    end
  end
end

.integerObject

A random integer which scales with `size`. Integers start small (around 0) and become more extreme (both higher and lower, negative) when `size` increases.

Shrinks to integers closer to zero.

>> Generators.integer.call(size: 2, rng: Random.new(42))
=> 1
>> Generators.integer.call(size: 10000, rng: Random.new(42))
=> 5795
>> r = Random.new(42); Generators.integer.sample(size: 20000, rng: r)
=> [-4205, -19140, 18158, -8716, -13735, -3150, 17194, 1962, -3977, -18315]

83
84
85
86
87
88
89
90
# File 'lib/prop_check/generators.rb', line 83

def integer
  Generator.new do |size:, rng:, **|
    ensure_proper_size!(size)

    val = rng.rand(-size..size)
    LazyTree.new(val, integer_shrink(val))
  end
end

.negative_integerObject

Only returns integers that are smaller than zero. See `integer` for more information.


122
123
124
# File 'lib/prop_check/generators.rb', line 122

def negative_integer
  positive_integer.map(&:[email protected])
end

.nilObject

Generates always `nil`.

Does not shrink.

>> Generators.nil.sample(5, size: 10, rng: Random.new(42))
=> [nil, nil, nil, nil, nil]

527
528
529
# File 'lib/prop_check/generators.rb', line 527

def nil
  constant(nil)
end

.nillable(other_generator) ⇒ Object

Generates whatever `other_generator` generates but sometimes instead `nil`.`

>> Generators.nillable(Generators.integer).sample(20, size: 10, rng: Random.new(42))
=> [9, 10, 8, 0, 10, -3, -8, 10, 1, -9, -10, nil, 1, 6, nil, 1, 9, -8, 8, 10]

587
588
589
# File 'lib/prop_check/generators.rb', line 587

def nillable(other_generator)
  frequency(9 => other_generator, 1 => constant(nil))
end

.nonnegative_integerObject

Only returns integers that are zero or larger. See `integer` for more information.


101
102
103
# File 'lib/prop_check/generators.rb', line 101

def nonnegative_integer
  integer.map(&:abs)
end

.nonpositive_integerObject

Only returns integers that are zero or smaller. See `integer` for more information.


115
116
117
# File 'lib/prop_check/generators.rb', line 115

def nonpositive_integer
  nonnegative_integer.map(&:[email protected])
end

.one_of(*choices) ⇒ Object

Picks one of the given generators in `choices` at random uniformly every time.

Shrinks to values earlier in the list of `choices`.

>> Generators.one_of(Generators.constant(true), Generators.constant(false)).sample(5, size: 10, rng: Random.new(42))
=> [true, false, true, true, true]

172
173
174
175
176
# File 'lib/prop_check/generators.rb', line 172

def one_of(*choices)
  choose(choices.length).bind do |index|
    choices[index]
  end
end

.positive_integerObject

Only returns integers that are larger than zero. See `integer` for more information.


108
109
110
# File 'lib/prop_check/generators.rb', line 108

def positive_integer
  nonnegative_integer.map { |x| x + 1 }
end

.printable_ascii_charObject

Generates a single-character string from the printable ASCII character set.

Shrinks towards ' '.

>> Generators.printable_ascii_char.sample(size: 10, rng: Random.new(42))
=> ["S", "|", ".", "g", "\\", "4", "r", "v", "j", "j"]

389
390
391
# File 'lib/prop_check/generators.rb', line 389

def printable_ascii_char
  one_of(*@@printable_ascii_chars.map(&method(:constant)))
end

.printable_ascii_string(**kwargs) ⇒ Object

Generates strings from the printable ASCII character set.

Shrinks towards fewer characters, and towards ' '.

>> Generators.printable_ascii_string.sample(5, size: 10, rng: Random.new(42))
=> ["S|.g", "rvjjw7\"5T!", "=", "!_[[email protected]", "Y"]

Accepts the same options as `array`


403
404
405
# File 'lib/prop_check/generators.rb', line 403

def printable_ascii_string(**kwargs)
  array(printable_ascii_char, **kwargs).map(&:join)
end

.printable_charObject

Generates a single-character printable string both ASCII characters and Unicode.

Shrinks towards characters with lower codepoints, e.g. ASCII

>> Generators.printable_char.sample(size: 10, rng: Random.new(42))
=> ["吏", "", "", "", "", "", "", "", "", "Ȍ"]

463
464
465
# File 'lib/prop_check/generators.rb', line 463

def printable_char
  one_of(*@@printable_chars.map(&method(:constant)))
end

.printable_string(**kwargs) ⇒ Object

Generates a printable string both ASCII characters and Unicode.

Shrinks towards shorter strings, and towards characters with lower codepoints, e.g. ASCII

>> Generators.printable_string.sample(5, size: 10, rng: Random.new(42))
=> ["", "Ȍ", "𐁂", "Ȕ", ""]

Accepts the same options as `array`


477
478
479
# File 'lib/prop_check/generators.rb', line 477

def printable_string(**kwargs)
  array(printable_char, **kwargs).map(&:join)
end

.real_floatObject

Generates floating-point numbers These start small (around 0) and become more extreme (large positive and large negative numbers)

Will only generate 'reals', that is: no infinity, no NaN, no numbers testing the limits of floating-point arithmetic.

Shrinks to numbers closer to zero.

>> Generators.real_float().sample(10, size: 10, rng: Random.new(42))
=> [-2.2, -0.2727272727272727, 4.0, 1.25, -3.7272727272727275, -8.833333333333334, -8.090909090909092, 1.1428571428571428, 0.0, 8.0]

143
144
145
146
147
# File 'lib/prop_check/generators.rb', line 143

def real_float
  tuple(integer, integer, integer).map do |a, b, c|
    fraction(a, b, c)
  end
end

.simple_symbolObject

Generates symbols consisting of lowercase letters and potentially underscores.

Shrinks towards shorter symbols and the letter 'a'.

>> Generators.simple_symbol.sample(5, size: 10, rng: Random.new(42))
=> [:tokh, :gzswkkxudh, :vubxlfbu, :lzvlyq__jp, :oslw]

549
550
551
552
553
554
555
# File 'lib/prop_check/generators.rb', line 549

def simple_symbol
  alphabet = ('a'..'z').to_a
  alphabet << '_'
  array(one_of(*alphabet.map(&method(:constant))))
    .map(&:join)
    .map(&:to_sym)
end

.string(**kwargs) ⇒ Object

Generates a string of unicode characters (which might contain both printable and non-printable characters).

Shrinks towards characters with lower codepoints, e.g. ASCII

>> Generators.string.sample(5, size: 10, rng: Random.new(42))
=> ["\u{A3DB3}𠍜\u{3F46A}\u{1AEBC}", "􍙦𡡹󴇒\u{DED74}𪱣\u{43E97}ꂂ\u{50695}􏴴\u{C0301}", "\u{4FD9D}", "\u{C14BF}\u{193BB}𭇋󱣼\u{76B58}", "𦐺\u{9FDDB}\u{80ABB}\u{9E3CF}𐂽\u{14AAE}"]

Accepts the same options as `array`


505
506
507
# File 'lib/prop_check/generators.rb', line 505

def string(**kwargs)
  array(char, **kwargs).map(&:join)
end

.truthyObject

Generates common terms that are not `nil` or `false`.

Shrinks towards simpler terms, like `true`, an empty array, a single character or an integer.

>> Generators.truthy.sample(5, size: 10, rng: Random.new(42))
=> [[4, 0, -3, 10, -4, 8, 0, 0, 10], -3, [5.5, -5.818181818181818, 1.1428571428571428, 0.0, 8.0, 7.857142857142858, -0.6666666666666665, 5.25], [], ["\u{9E553}\u{DD56E}\u{A5BBB}\u{8BDAB}\u{3E9FC}\u{C4307}\u{DAFAE}\u{1A022}\u{938CD}\u{70631}", "\u{C4C01}\u{32D85}\u{425DC}"]]

564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
# File 'lib/prop_check/generators.rb', line 564

def truthy
  one_of(constant(true),
         constant([]),
         char,
         integer,
         float,
         string,
         array(integer),
         array(float),
         array(char),
         array(string),
         hash(simple_symbol, integer),
         hash(string, integer),
         hash(string, string)
        )
end

.tuple(*generators) ⇒ Object

Generates an array containing always exactly one value from each of the passed generators, in the same order as specified:

Shrinks element generators, one at a time (trying last one first).

>> Generators.tuple(Generators.integer, Generators.real_float).call(size: 10, rng: Random.new(42))
=> [-4, 13.0]

207
208
209
210
211
212
213
# File 'lib/prop_check/generators.rb', line 207

def tuple(*generators)
  Generator.new do |**kwargs|
    LazyTree.zip(generators.map do |generator|
      generator.generate(**kwargs)
    end)
  end
end