def do_draw(self, data): size = 16 sign_mask = 2 ** (size * 8 - 1) def distribution(random, n): assert n == size if random.randint(0, 2) > 0: k = 1 elif random.randint(0, 2) > 0: k = 2 else: k = random.randint(1, size) r = random.getrandbits(k * 8) if random.randint(0, 1): r |= sign_mask else: r &= ~sign_mask return int_to_bytes(r, n) byt = data.draw_bytes(size, distribution=distribution) r = int_from_bytes(byt) negative = r & sign_mask r &= ~sign_mask if negative: r = -r return int(r)
def do_draw(self, data): size = 16 sign_mask = 2 ** (size * 8 - 1) def distribution(random, n): assert n == size k = min( random.randint(0, n * 8 - 1), random.randint(0, n * 8 - 1), ) if k > 0: r = random.getrandbits(k) else: r = 0 if random.randint(0, 1): r |= sign_mask else: r &= (~sign_mask) return int_to_bytes(r, n) byt = data.draw_bytes(size, distribution=distribution) r = int_from_bytes(byt) negative = r & sign_mask r &= (~sign_mask) if negative: r = -r return int(r)
def problem(draw): b = hbytes(draw(st.binary(min_size=1, max_size=8))) m = int_from_bytes(b) * 256 assume(m > 0) marker = draw(st.binary(max_size=8)) bound = draw(st.integers(0, m - 1)) return (b, marker, bound)
def do_draw(self, data): size = 16 sign_mask = 2**(size * 8 - 1) def distribution(random, n): assert n == size k = min( random.randint(0, n * 8 - 1), random.randint(0, n * 8 - 1), ) if k > 0: r = random.getrandbits(k) else: r = 0 if random.randint(0, 1): r |= sign_mask else: r &= (~sign_mask) return int_to_bytes(r, n) byt = data.draw_bytes(size, distribution=distribution) r = int_from_bytes(byt) negative = r & sign_mask r &= (~sign_mask) if negative: r = -r return int(r)
def write(self, string): """Write ``string`` to the output buffer.""" self.__assert_not_frozen("write") string = hbytes(string) if not string: return self.draw_bits(len(string) * 8, forced=int_from_bytes(string)) return self.buffer[-len(string) :]
def write(self, string): """Write ``string`` to the output buffer.""" self.__assert_not_frozen("write") string = hbytes(string) if not string: return self.draw_bits(len(string) * 8, forced=int_from_bytes(string)) return self.buffer[-len(string):]
def draw_bits(self, n): self.__assert_not_frozen('draw_bits') if n == 0: result = 0 elif n % 8 == 0: return int_from_bytes(self.draw_bytes(n // 8)) else: n_bytes = (n // 8) + 1 self.__check_capacity(n_bytes) buf = bytearray(self._draw_bytes(self, n_bytes)) assert len(buf) == n_bytes mask = (1 << (n % 8)) - 1 buf[0] &= mask self.capped_indices[self.index] = mask buf = hbytes(buf) self.__write(buf) result = int_from_bytes(buf) assert bit_length(result) <= n return result
def do_draw(self, data): size = 16 sign_mask = 2 ** (size * 8 - 1) byt = data.draw_bytes(size) r = int_from_bytes(byt) negative = r & sign_mask r &= (~sign_mask) if negative: r = -r return int(r)
def do_draw(self, data): size = 16 sign_mask = 2**(size * 8 - 1) byt = data.draw_bytes(size) r = int_from_bytes(byt) negative = r & sign_mask r &= (~sign_mask) if negative: r = -r return int(r)
def integer_range(data, lower, upper, center=None, distribution=None): assert lower <= upper if lower == upper: return int(lower) if center is None: center = lower center = min(max(center, lower), upper) if distribution is None: if lower < center < upper: def distribution(random): if random.randint(0, 1): return random.randint(center, upper) else: return random.randint(lower, center) else: def distribution(random): return random.randint(lower, upper) gap = upper - lower bits = bit_length(gap) nbytes = bits // 8 + int(bits % 8 != 0) mask = saturate(gap) def byte_distribution(random, n): assert n == nbytes v = distribution(random) if v >= center: probe = v - center else: probe = upper - v return int_to_bytes(probe, n) probe = gap + 1 while probe > gap: probe = int_from_bytes(data.draw_bytes(nbytes, byte_distribution)) & mask if center == upper: result = upper - probe elif center == lower: result = lower + probe else: if center + probe <= upper: result = center + probe else: result = upper - probe assert lower <= result <= upper return int(result)
def integer_range(data, lower, upper, center=None, distribution=None): assert lower <= upper if lower == upper: return int(lower) if center is None: center = lower center = min(max(center, lower), upper) if distribution is None: if lower < center < upper: def distribution(random): if random.randint(0, 1): return random.randint(center, upper) else: return random.randint(lower, center) else: def distribution(random): return random.randint(lower, upper) gap = upper - lower bits = bit_length(gap) nbytes = bits // 8 + int(bits % 8 != 0) mask = saturate(gap) def byte_distribution(random, n): assert n == nbytes v = distribution(random) if v >= center: probe = v - center else: probe = upper - v return int_to_bytes(probe, n) probe = gap + 1 while probe > gap: probe = int_from_bytes( data.draw_bytes(nbytes, byte_distribution) ) & mask if center == upper: result = upper - probe elif center == lower: result = lower + probe else: if center + probe <= upper: result = center + probe else: result = upper - probe assert lower <= result <= upper return int(result)
def get_random_for_wrapped_test(test, wrapped_test): settings = wrapped_test._hypothesis_internal_use_settings wrapped_test._hypothesis_internal_use_generated_seed = None if wrapped_test._hypothesis_internal_use_seed is not None: return Random(wrapped_test._hypothesis_internal_use_seed) elif settings.derandomize: return Random(int_from_bytes(function_digest(test))) elif global_force_seed is not None: return Random(global_force_seed) else: seed = rnd_module.getrandbits(128) wrapped_test._hypothesis_internal_use_generated_seed = seed return Random(seed)
def test_regression_1(): # This is a really hard to reproduce bug that previously triggered a very # specific exception inside one of the shrink passes. It's unclear how # useful this regression test really is, but nothing else caught the # problem. @run_to_buffer def x(data): data.write(hbytes(b'\x01\x02')) data.write(hbytes(b'\x01\x00')) v = data.draw_bits(41) if v >= 512 or v == 254: data.mark_interesting() assert list(x)[:-2] == [1, 2, 1, 0, 0, 0, 0, 0] assert int_from_bytes(x[-2:]) in (254, 512)
def go(ex): if ex.length == 0: return if len(ex.children) == 0: stack[-1].append(int_from_bytes(data.buffer[ex.start:ex.end])) else: node = [] stack.append(node) for v in ex.children: go(v) stack.pop() if len(node) == 1: stack[-1].extend(node) else: stack[-1].append(node)
def test_regression_1(): # This is a really hard to reproduce bug that previously triggered a very # specific exception inside one of the shrink passes. It's unclear how # useful this regression test really is, but nothing else caught the # problem. @run_to_buffer def x(data): data.write(b"\x01\x02") data.write(b"\x01\x00") v = data.draw_bits(41) if v >= 512 or v == 254: data.mark_interesting() assert list(x)[:-2] == [1, 2, 1, 0, 0, 0, 0, 0] assert int_from_bytes(x[-2:]) in (254, 512)
def go(ex): if ex.length == 0: return if len(ex.children) == 0: stack[-1].append(int_from_bytes(data.buffer[ex.start : ex.end])) else: node = [] stack.append(node) for v in ex.children: go(v) stack.pop() if len(node) == 1: stack[-1].extend(node) else: stack[-1].append(node)
def draw_bits(self, n, *, forced=None): """Return an ``n``-bit integer from the underlying source of bytes. If ``forced`` is set to an integer will instead ignore the underlying source and simulate a draw as if it had returned that integer.""" self.__assert_not_frozen("draw_bits") if n == 0: return 0 assert n > 0 n_bytes = bits_to_bytes(n) self.__check_capacity(n_bytes) if forced is not None: buf = int_to_bytes(forced, n_bytes) elif self.__bytes_drawn < len(self.__prefix): index = self.__bytes_drawn buf = self.__prefix[index:index + n_bytes] if len(buf) < n_bytes: buf += uniform(self.__random, n_bytes - len(buf)) else: buf = uniform(self.__random, n_bytes) buf = bytearray(buf) self.__bytes_drawn += n_bytes assert len(buf) == n_bytes # If we have a number of bits that is not a multiple of 8 # we have to mask off the high bits. buf[0] &= BYTE_MASKS[n % 8] buf = bytes(buf) result = int_from_bytes(buf) self.observer.draw_bits(n, forced is not None, result) self.__example_record.draw_bits(n, forced) initial = self.index self.buffer.extend(buf) self.index = len(self.buffer) if forced is not None: self.forced_indices.update(range(initial, self.index)) self.blocks.add_endpoint(self.index) assert bit_length(result) <= n return result
def geometric(data, p): denom = math.log1p(-p) n_bytes = 8 def distribution(random, n): assert n == n_bytes for _ in range(100): try: return int_to_bytes(int( math.log1p(-random.random()) / denom), n) # This is basically impossible to hit but is required for # correctness except OverflowError: # pragma: no cover pass # We got a one in a million chance 100 times in a row. Something is up. assert False # pragma: no cover return int_from_bytes(data.draw_bytes(n_bytes, distribution))
def float_hack(self): """Our encoding of floating point numbers does the right thing when you lexically shrink it, but there are some highly non-obvious lexical shrinks corresponding to natural floating point operations. We can't actually tell when the floating point encoding is being used (that would break the assumptions that Hypothesis doesn't inspect the generated values), but we can cheat: We just guess when it might be being used and perform shrinks that are valid regardless of our guess is correct. So that's what this method does. It's a cheat to give us good shrinking of floating at low cost in runtime and only moderate cost in elegance. """ # If the block is of the wrong size then we're certainly not using the # float encoding. if self.size != 8: return # If the high bit is zero then we're in the integer representation of # floats so we don't need these hacks because it will shrink normally. if self.current[0] >> 7 == 0: return i = self.current_int f = lex_to_float(i) # This floating point number can be represented in our simple format. # So we try converting it to that (which will give the same float, but # a different encoding of it). If that doesn't work then the float # value of this doesn't unambiguously give the desired predicate, so # this approach isn't useful. If it *does* work, then we're now in a # situation where we don't need it, so either way we return here. if is_simple(f): self.incorporate_float(f) return self.delegate( Float, convert_to=lambda b: lex_to_float(int_from_bytes(b)), convert_from=lambda f: int_to_bytes(float_to_lex(f), self.size), )
def draw_bits(self, n, forced=None): """Return an ``n``-bit integer from the underlying source of bytes. If ``forced`` is set to an integer will instead ignore the underlying source and simulate a draw as if it had returned that integer.""" self.__assert_not_frozen("draw_bits") if n == 0: return 0 assert n > 0 n_bytes = bits_to_bytes(n) self.__check_capacity(n_bytes) if forced is not None: buf = bytearray(int_to_bytes(forced, n_bytes)) else: buf = bytearray(self._draw_bytes(self, n_bytes)) assert len(buf) == n_bytes # If we have a number of bits that is not a multiple of 8 # we have to mask off the high bits. buf[0] &= BYTE_MASKS[n % 8] buf = hbytes(buf) result = int_from_bytes(buf) self.observer.draw_bits(n, forced is not None, result) self.start_example(DRAW_BYTES_LABEL) initial = self.index self.buffer.extend(buf) self.index = len(self.buffer) if forced is not None: self.forced_indices.update(hrange(initial, self.index)) self.blocks.add_endpoint(self.index) self.stop_example() assert bit_length(result) <= n return result
def integer_range(data, lower, upper, center=None): assert lower <= upper if lower == upper: return int(lower) if center is None: center = lower center = min(max(center, lower), upper) bits = bit_length(max(upper - center, center - lower)) nbytes = bits // 8 + int(bits % 8 != 0) if center == upper: above = False elif center == lower: above = True else: above = boolean(data) if above: gap = upper - center else: gap = center - lower assert gap > 0 mask = saturate(gap) probe = gap + 1 while probe > gap: probe = int_from_bytes(data.draw_bytes(nbytes)) & mask if above: result = center + probe else: result = center - probe assert lower <= result <= upper return int(result)
def draw_bits(self, n, forced=None): """Return an ``n``-bit integer from the underlying source of bytes. If ``forced`` is set to an integer will instead ignore the underlying source and simulate a draw as if it had returned that integer.""" self.__assert_not_frozen("draw_bits") if n == 0: return 0 assert n > 0 n_bytes = bits_to_bytes(n) self.__check_capacity(n_bytes) if forced is not None: buf = bytearray(int_to_bytes(forced, n_bytes)) else: buf = bytearray(self._draw_bytes(self, n_bytes)) assert len(buf) == n_bytes # If we have a number of bits that is not a multiple of 8 # we have to mask off the high bits. buf[0] &= BYTE_MASKS[n % 8] buf = hbytes(buf) result = int_from_bytes(buf) self.observer.draw_bits(n, forced is not None, result) self.__example_record.draw_bits(n, forced) initial = self.index self.buffer.extend(buf) self.index = len(self.buffer) if forced is not None: self.forced_indices.update(hrange(initial, self.index)) self.blocks.add_endpoint(self.index) assert bit_length(result) <= n return result
def f(data): i = int_from_bytes(data.draw_bytes(2)) data.draw_bytes(i) data.mark_interesting()
def calc_label_from_name(name: str) -> int: hashed = hashlib.sha384(name.encode()).digest() return int_from_bytes(hashed[:8])
def current_int(self): return int_from_bytes(self.current)
def test_to_int_in_big_endian_order(x, y): x, y = sorted((x, y)) assert 0 <= int_from_bytes(x) <= int_from_bytes(y)
def test_convert_back(bs): bs = bytearray(bs) assert int_to_bytes(int_from_bytes(bs), len(bs)) == bs
def getrandbits(data, n): n_bytes = n // 8 if n % 8 != 0: n_bytes += 1 return int_from_bytes(data.draw_bytes(n_bytes)) & ((1 << n) - 1)
def float_hack(self): """Our encoding of floating point numbers does the right thing when you lexically shrink it, but there are some highly non-obvious lexical shrinks corresponding to natural floating point operations. We can't actually tell when the floating point encoding is being used (that would break the assumptions that Hypothesis doesn't inspect the generated values), but we can cheat: We just guess when it might be being used and perform shrinks that are valid regardless of our guess is correct. So that's what this method does. It's a cheat to give us good shrinking of floating at low cost in runtime and only moderate cost in elegance. """ # If the block is of the wrong size then we're certainly not using the # float encoding. if self.size != 8: return # If the high bit is zero then we're in the integer representation of # floats so we don't need these hacks because it will shrink normally. if self.current[0] >> 7 == 0: return i = int_from_bytes(self.current) f = lex_to_float(i) # This floating point number can be represented in our simple format. # So we try converting it to that (which will give the same float, but # a different encoding of it). If that doesn't work then the float # value of this doesn't unambiguously give the desired predicate, so # this approach isn't useful. If it *does* work, then we're now in a # situation where we don't need it, so either way we return here. if is_simple(f): self.incorporate_float(f) return # We check for a bunch of standard "large" floats. If we're currently # worse than them and the shrink downwards doesn't help, abort early # because there's not much useful we can do here. for g in [ float('nan'), float('inf'), sys.float_info.max, ]: j = float_to_lex(g) if j < i: if self.incorporate_int(j): f = g i = j if math.isinf(f) or math.isnan(f): return # Finally we get to the important bit: Each of these is a small change # to the floating point number that corresponds to a large change in # the lexical representation. Trying these ensures that our floating # point shrink can always move past these obstacles. In particular it # ensures we can always move to integer boundaries and shrink past a # change that would require shifting the exponent while not changing # the float value much. for g in [ floor(f), ceil(f), ]: if self.incorporate_float(g): return if f > 2: self.incorporate_float(f - 1)
def calc_label(cls): name = str_to_bytes(qualname(cls)) hashed = hashlib.md5(name).digest() return int_from_bytes(hashed[:8])
def calc_label_from_name(name): hashed = hashlib.md5(str_to_bytes(name)).digest() return int_from_bytes(hashed[:8])
def n_byte_unsigned(data, n): return int_from_bytes(data.draw_bytes(n))
def attempt_to_improve(self, example_index, upwards): """Part of our hill climbing implementation. Attempts to improve a given score by regenerating an example.""" data = self.current_data self.current_score ex = data.examples[example_index] assert ex.length > 0 prefix = data.buffer[:ex.start] suffix = data.buffer[ex.end:] existing = data.buffer[ex.start:ex.end] existing_as_int = int_from_bytes(existing) max_int_value = (256**len(existing)) - 1 if existing_as_int == max_int_value and upwards: return False if existing_as_int == 0 and not upwards: return False # We make a mix of small and large jumps. Neither are guaranteeed to # work, but each will work in circumstances the other might not. Small # jumps are especially important for the last steps towards a local # maximum - often large jumps will break some important property of the # test case while small, more careful, jumps will not. On the flip side # we sometimes end up in circumstances where small jumps don't work # but a random draw will produce a good result with reasonable # probability, and we don't want to be too timid in our optimisation as # it will take too long to complete. if self.random.randint(0, 1): if upwards: replacement_as_int = self.random.randint( existing_as_int + 1, max_int_value) else: replacement_as_int = self.random.randint( 0, existing_as_int - 1) elif upwards: replacement_as_int = existing_as_int + 1 else: replacement_as_int = existing_as_int - 1 replacement = int_to_bytes(replacement_as_int, len(existing)) attempt = self.engine.cached_test_function( prefix + replacement + suffix, extend=BUFFER_SIZE, ) if self.consider_new_test_data(attempt): return True if attempt.status < Status.VALID: return False ex_attempt = attempt.examples[example_index] replacement = attempt.buffer[ex_attempt.start:ex_attempt.end] return self.consider_new_buffer(prefix + replacement + suffix)
def parse_buf(b): return flt.lex_to_float(int_from_bytes(b))
def do_draw(self, data): b = int_from_bytes(data.draw_bytes(2)) assume(b == 3) print('ohai')
def calc_label_from_name(name): hashed = hashlib.sha384(str_to_bytes(name)).digest() return int_from_bytes(hashed[:8])
def hill_climb(self): """The main hill climbing loop where we actually do the work: Take data, and attempt to improve its score for target. select_example takes a data object and returns an index to an example where we should focus our efforts.""" blocks_examined = set() prev = None i = len(self.current_data.blocks) - 1 while i >= 0 and self.improvements <= self.max_improvements: if prev is not self.current_data: i = len(self.current_data.blocks) - 1 prev = self.current_data if i in blocks_examined: i -= 1 continue blocks_examined.add(i) data = self.current_data block = data.blocks[i] prefix = data.buffer[:block.start] existing = data.buffer[block.start:block.end] existing_as_int = int_from_bytes(existing) max_int_value = (256**len(existing)) - 1 if existing_as_int == max_int_value: continue def attempt_replace(v): """Try replacing the current block in the current best test case with an integer of value i. Note that we use the *current* best and not the one we started with. This helps ensure that if we luck into a good draw when making random choices we get to keep the good bits.""" if v < 0 or v > max_int_value: return False v_as_bytes = int_to_bytes(v, len(existing)) # We make a couple attempts at replacement. This only matters # if we end up growing the buffer - otherwise we exit the loop # early - but in the event that there *is* some randomized # component we want to give it a couple of tries to succeed. for _ in range(3): attempt = self.engine.cached_test_function( prefix + v_as_bytes + self.current_data.buffer[block.end:] + bytes(BUFFER_SIZE), ) if self.consider_new_test_data(attempt): return True if attempt.status < Status.INVALID or len( attempt.buffer) == len(self.current_data.buffer): return False for i, ex in enumerate(self.current_data.examples): if ex.start >= block.end: break if ex.end <= block.start: continue ex_attempt = attempt.examples[i] if ex.length == ex_attempt.length: continue replacement = attempt.buffer[ex_attempt. start:ex_attempt.end] if self.consider_new_test_data( self.engine.cached_test_function( prefix + replacement + self.current_data.buffer[ex.end:])): return True return False # We unconditionally scan both upwards and downwards. The reason # for this is that we allow "lateral" moves that don't increase the # score but instead leave it constant. All else being equal we'd # like to leave the test case closer to shrunk, so afterwards we # try lowering the value towards zero even if we've just raised it. if not attempt_replace(max_int_value): find_integer(lambda k: attempt_replace(k + existing_as_int)) existing = self.current_data.buffer[block.start:block.end] existing_as_int = int_from_bytes(existing) if not attempt_replace(0): find_integer(lambda k: attempt_replace(existing_as_int - k))