Python hypothesis.strategies.recursive() Examples

def test_json_dumps(value):
    """Checks that value is serialisable as JSON."""
    # We expect this test to always pass - the point of this exercise is
    # to define a recursive strategy, and then investigate the values it
    # generates for a *passing* test.
    hypothesis.note("type: {}".format(type(value)))
    hypothesis.event("type: {}".format(type(value)))
    json.dumps(value)


# Takeaway: you've seen and played with a few ways to see what a
# passing test is doing, without having to inject a failure.


##############################################################################
# `@st.composite` exercise

# This goal of this exercise is to play with a contrived data dependency,
# using a composite strategy to generate inputs.  You can use the same tricks
# as above to check what's being generated, so try to keep the test passing! 

def action_structures(draw):
    """
    A Hypothesis strategy that creates a tree of L{ActionStructure} and
    L{unicode}.
    """
    tree = draw(st.recursive(labels, st.lists, max_leaves=20))

    def to_structure(tree_or_message):
        if isinstance(tree_or_message, list):
            return ActionStructure(
                type=draw(labels),
                failed=draw(st.booleans()),
                children=[to_structure(o) for o in tree_or_message],
            )
        else:
            return tree_or_message

    return to_structure(tree) 

def assert_roundtrip(obj):
    """Assert that we can successfully round-trip the given object."""
    buf = pencode(obj)
    assert isinstance(buf, bytes)
    obj2 = pdecode(buf)

    assert type(obj) == type(obj2)

    # By definition NaN does not equal anything, even itself
    if isinstance(obj2, float) and math.isnan(obj2):
        return obj2

    try:
        assert obj == obj2
    except RuntimeError as e:
        if 'maximum recursion depth exceeded' not in e.args[0]:
            raise
        # If we hit a RecursionError, we correctly decoded a recursive
        # structure, so test passes :)
    except RecursionError:
        pass

    return obj2 

def _create_hyp_nested_strategy(simple_class_strategy):
    """
    Create a recursive attrs class.
    Given a strategy for building (simpler) classes, create and return
    a strategy for building classes that have as an attribute:
        * just the simpler class
        * a list of simpler classes
        * a dict mapping the string "cls" to a simpler class.
    """
    # A strategy producing tuples of the form ([list of attributes], <given
    # class strategy>).
    attrs_and_classes = st.tuples(
        lists_of_attrs(defaults=True), simple_class_strategy
    )

    return (
        attrs_and_classes.flatmap(just_class)
        | attrs_and_classes.flatmap(just_class_with_type)
        | attrs_and_classes.flatmap(list_of_class)
        | attrs_and_classes.flatmap(list_of_class_with_type)
        | attrs_and_classes.flatmap(dict_of_class)
    ) 

def regex_patterns(draw: Any) -> str:
    """Return a recursive strategy for simple regular expression patterns."""
    fragments = st.one_of(
        st.just("."),
        st.from_regex(r"\[\^?[A-Za-z0-9]+\]"),
        REGEX_PATTERNS.map("{}+".format),
        REGEX_PATTERNS.map("{}?".format),
        REGEX_PATTERNS.map("{}*".format),
    )
    result = draw(st.lists(fragments, min_size=1, max_size=3).map("".join))
    assert isinstance(result, str)
    try:
        re.compile(result)
    except re.error:
        assume(False)
    return result 

def add_bools(list_of_lists):
    """
    Given recursive list that can contain other lists, return tuple of that plus
    a booleans strategy for each list.
    """
    l = []
    def count(recursive):
        l.append(1)
        for child in recursive:
            if isinstance(child, list):
                count(child)
    count(list_of_lists)
    return st.tuples(st.just(list_of_lists), st.tuples(*[st.sampled_from([True, False]) for i in l])) 

def from_json(cls, string):
        """Decodes a JSON-string back into a `Record` instance

        This is a *bad* method. This needs to be fixed
        """
        value = string
        return cls(value)


# We can define recursive strategies like so: 

def test_map_odd_numbers(x):
    assert x[-1] in "13579"


# Takeaway
# --------
# `.map` permits us to extend Hypothesis' core strategies in powerful
# ways. See that it can be used to affect the individual values being
# produced by a strategy (e.g. mapping integers to even-valued
# integers), as well as to cast the values to a different type (e.g.
# mapping an integer to a string.
#
# If it seems like a data-type is missing from Hypothesis'
# strategies, then it is likely that a simple application of `.map`
# will suffice. E.g. suppose you want a strategy that generate deques,
# then
#     `deque_strat = st.lists(...).map(deque)`
# will serve nicely - we don't even need a lambda!


##############################################################################
# Defining recursive data.

# There are a couple of ways to define recursive data with Hypothesis,
# leaning on the fact that strategies are lazily instantiated.
#
# `st.recursive` takes a base strategy, and a function that takes a strategy
# and returns an extended strategy.  All good if we want that structure!
# If you want mutual recursion though, or have a complicated kind of data
# (or just limited time in a tutorial), `st.deferred` is the way to go.
#
# The `Record` exercise in pbt-101.py defined JSON using `st.recursive`,
# if you want to compare them, and has some extension problems that you
# could write as tests here instead.


# JSON values are defined as one of null, false, true, a finite number,
# a string, an array of json values, or a dict of string to json values. 

def hashable_containers(primitives):
    def extend(base):
        return st.one_of(
            st.frozensets(base, max_size=50),
            st.lists(base, max_size=50).map(tuple),
        )
    return st.recursive(primitives, extend) 

def containers(primitives):
    def extend(base):
        return st.one_of(
            st.lists(base, max_size=50),
            st.lists(base, max_size=50).map(tuple),
            st.dictionaries(
                keys=hashable_containers(primitives),
                values=base,
                max_size=10
            ),
        )

    return st.recursive(primitives, extend, max_leaves=50) 

def nested_dictionaries():
    simple_strings_alphabet = 'abcdefghijklmnopqrstuvwxyz\'"\r\n '
    simple_text = st.text(alphabet=simple_strings_alphabet, min_size=5)

    def extend(base):
        return st.one_of(
            st.lists(base, min_size=5),
            st.dictionaries(keys=simple_text, values=base, min_size=1)
        )

    return st.recursive(simple_text, extend, max_leaves=50) 

def json(value_limit=5):
    """Hypothesis strategy for generating values that can be passed to
    `json.dumps` to produce valid JSON data.

    :param value_limit: A limit on the number of values in the JSON data -
                        setting this too high can cause value generation to
                        time out.
    :type value_limit: int
    """
    return hy_st.recursive(
        hy_st.floats() | hy_st.booleans() | hy_st.text() | hy_st.none(),
        lambda children: hy_st.dictionaries(hy_st.text(), children),
        max_leaves=value_limit) 

def simple_classes(defaults=None, min_attrs=0):
    """
    Return a strategy that yields tuples of simple classes and values to
    instantiate them.
    """
    return lists_of_attrs(defaults, min_size=min_attrs).flatmap(
        _create_hyp_class
    )


# Ok, so st.recursive works by taking a base strategy (in this case,
# simple_classes) and a special function. This function receives a strategy,
# and returns another strategy (building on top of the base strategy). 

def test_nestedInteractions(self, values):
        """
        Nested interactions operate independently of parent interactions.

        :param values: a two-tuple composed of:
           - a recursive list of unicode and other recursive lists - list start
             means begin interaction, string means node resolve, list end means
             finish interaction.
           - list of False/True; True means failed interaction
        """
        requested_interactions, failures = values
        failures = iter(failures)
        assume(not isinstance(requested_interactions, unicode))
        self.init()
        ws_actor = self.connector.expectSocket()
        self.connector.connect(ws_actor)

        failures = iter(failures)
        created_services = {}
        expected_success_nodes = Counter()
        expected_failed_nodes = Counter()

        def run_interaction(children):
            should_fail = next(failures)
            failed = []
            succeeded = []
            self.session.start_interaction()
            for child in children:
                if isinstance(child, unicode):
                    # Make sure disco knows about the node:
                    if child in created_services:
                        node = created_services[child]
                    else:
                        node = create_node(child, child)
                        created_services[child] = node
                    self.disco.onMessage(None, NodeActive(node))
                    # Make sure the child Node is resolved in the interaction
                    self.session.resolve(node.service, "1.0")
                    if should_fail:
                        expected_failed_nodes[node] += 1
                        failed.append(node)
                    else:
                        expected_success_nodes[node] += 1
                        succeeded.append(node)
                else:
                    run_interaction(child)
            if should_fail:
                self.session.fail_interaction("OHNO")
            self.session.finish_interaction()
            self.connector.advance_time(5.0) # Make sure interaction is sent
            ws_actor.swallowLogMessages()
            self.connector.expectInteraction(
                self, ws_actor, self.session, failed, succeeded)

        run_interaction(requested_interactions)
        for node in set(expected_failed_nodes) | set(expected_success_nodes):
            policy = self.disco.failurePolicy(node)
            self.assertEqual((policy.successes, policy.failures),
                             (expected_success_nodes[node],
                              expected_failed_nodes[node])) 

def _create_hyp_nested_strategy(simple_class_strategy):
    """
    Create a recursive attrs class.

    Given a strategy for building (simpler) classes, create and return
    a strategy for building classes that have as an attribute: either just
    the simpler class, a list of simpler classes, a tuple of simpler classes,
    an ordered dict or a dict mapping the string "cls" to a simpler class.
    """
    # Use a tuple strategy to combine simple attributes and an attr class.
    def just_class(tup):
        combined_attrs = list(tup[0])
        combined_attrs.append(attr.ib(default=attr.Factory(tup[1])))
        return _create_hyp_class(combined_attrs)

    def list_of_class(tup):
        default = attr.Factory(lambda: [tup[1]()])
        combined_attrs = list(tup[0])
        combined_attrs.append(attr.ib(default=default))
        return _create_hyp_class(combined_attrs)

    def tuple_of_class(tup):
        default = attr.Factory(lambda: (tup[1](),))
        combined_attrs = list(tup[0])
        combined_attrs.append(attr.ib(default=default))
        return _create_hyp_class(combined_attrs)

    def dict_of_class(tup):
        default = attr.Factory(lambda: {"cls": tup[1]()})
        combined_attrs = list(tup[0])
        combined_attrs.append(attr.ib(default=default))
        return _create_hyp_class(combined_attrs)

    def ordereddict_of_class(tup):
        default = attr.Factory(lambda: OrderedDict([("cls", tup[1]())]))
        combined_attrs = list(tup[0])
        combined_attrs.append(attr.ib(default=default))
        return _create_hyp_class(combined_attrs)

    # A strategy producing tuples of the form ([list of attributes], <given
    # class strategy>).
    attrs_and_classes = st.tuples(list_of_attrs, simple_class_strategy)

    return st.one_of(
        attrs_and_classes.map(just_class),
        attrs_and_classes.map(list_of_class),
        attrs_and_classes.map(tuple_of_class),
        attrs_and_classes.map(dict_of_class),
        attrs_and_classes.map(ordereddict_of_class),
    ) 

def simple_classes(
    draw, slots=None, frozen=None, weakref_slot=None, private_attrs=None
):
    """
    A strategy that generates classes with default non-attr attributes.

    For example, this strategy might generate a class such as:

    @attr.s(slots=True, frozen=True, weakref_slot=True)
    class HypClass:
        a = attr.ib(default=1)
        _b = attr.ib(default=None)
        c = attr.ib(default='text')
        _d = attr.ib(default=1.0)
        c = attr.ib(default={'t': 1})

    By default, all combinations of slots, frozen, and weakref_slot classes
    will be generated.  If `slots=True` is passed in, only slotted classes will
    be generated, and if `slots=False` is passed in, no slotted classes will be
    generated. The same applies to `frozen` and `weakref_slot`.

    By default, some attributes will be private (i.e. prefixed with an
    underscore). If `private_attrs=True` is passed in, all attributes will be
    private, and if `private_attrs=False`, no attributes will be private.
    """
    attrs = draw(list_of_attrs)
    frozen_flag = draw(st.booleans()) if frozen is None else frozen
    slots_flag = draw(st.booleans()) if slots is None else slots
    weakref_slot_flag = (
        draw(st.booleans()) if weakref_slot is None else weakref_slot
    )

    if private_attrs is None:
        attr_names = maybe_underscore_prefix(gen_attr_names())
    elif private_attrs is True:
        attr_names = ("_" + n for n in gen_attr_names())
    elif private_attrs is False:
        attr_names = gen_attr_names()

    cls_dict = dict(zip(attr_names, attrs))
    post_init_flag = draw(st.booleans())
    if post_init_flag:

        def post_init(self):
            pass

        cls_dict["__attrs_post_init__"] = post_init

    return make_class(
        "HypClass",
        cls_dict,
        slots=slots_flag,
        frozen=frozen_flag,
        weakref_slot=weakref_slot_flag,
    )


# st.recursive works by taking a base strategy (in this case, simple_classes)
# and a special function.  This function receives a strategy, and returns
# another strategy (building on top of the base strategy).