handle bool/float cases correctly in choice_key

Author: tybug

hypothesis-python/src/hypothesis/internal/conjecture/choice.py

@@ -67,6 +67,9 @@ class BooleanKWargs(TypedDict):
     IntegerKWargs, FloatKWargs, StringKWargs, BytesKWargs, BooleanKWargs
 ]
 ChoiceNameT: "TypeAlias" = Literal["integer", "string", "boolean", "float", "bytes"]
+ChoiceKeyT: "TypeAlias" = Union[
+    int, str, bytes, tuple[Literal["bool"], bool], tuple[Literal["float"], int]
+]
 
 
 def _size_to_index(size: int, *, alphabet_size: int) -> int:
@@ -456,13 +459,17 @@ def choice_permitted(choice: ChoiceT, kwargs: ChoiceKwargsT) -> bool:
         raise NotImplementedError(f"unhandled type {type(choice)} with value {choice}")
 
 
-def choices_key(choices: Sequence[ChoiceT]) -> tuple[ChoiceT, ...]:
+def choices_key(choices: Sequence[ChoiceT]) -> tuple[ChoiceKeyT, ...]:
     return tuple(choice_key(choice) for choice in choices)
 
 
-def choice_key(choice: ChoiceT) -> ChoiceT:
-    if type(choice) is float:
-        return float_to_int(choice)
+def choice_key(choice: ChoiceT) -> ChoiceKeyT:
+    if isinstance(choice, float):
+        # distinguish -0.0/0.0, signaling/nonsignaling nans, etc.
+        return ("float", float_to_int(choice))
+    if isinstance(choice, bool):
+        # avoid choice_key(0) == choice_key(False)
+        return ("bool", choice)
     return choice
 
 

hypothesis-python/src/hypothesis/internal/conjecture/engine.py

@@ -345,7 +345,7 @@ def __stoppable_test_function(self, data: ConjectureData) -> None:
                     # correct engine.
                     raise
 
-    def _cache_key(self, choices: Sequence[ChoiceT]) -> tuple[ChoiceT, ...]:
+    def _cache_key(self, choices: Sequence[ChoiceT]) -> tuple[ChoiceKeyT, ...]:
         return choices_key(choices)
 
     def _cache(self, data: ConjectureData) -> None:

hypothesis-python/tests/conjecture/test_ir.py

@@ -29,6 +29,7 @@
     choice_from_index,
     choice_permitted,
     choice_to_index,
+    choices_key,
 )
 from hypothesis.internal.conjecture.data import (
     COLLECTION_DEFAULT_MAX_SIZE,
@@ -884,3 +885,26 @@ def test_draw_directly_explicit():
         )
         == 10
     )
+
+
+@pytest.mark.parametrize(
+    "choices1, choices2",
+    [
+        [(True,), (1,)],
+        [(False,), (0,)],
+        [(0.0,), (-0.0,)],
+    ],
+)
+def test_choices_key_distinguishes_weird_cases(choices1, choices2):
+    assert choices_key(choices1) != choices_key(choices2)
+
+
+@given(st.lists(ir_nodes()), st.lists(ir_nodes()))
+def test_choices_key_respects_inequality(nodes1, nodes2):
+    choices1 = [n.value for n in nodes1]
+    choices2 = [n.value for n in nodes2]
+    if choices_key(choices1) != choices_key(choices2):
+        assert set(choices1) != set(choices2)
+
+    # note that the other direction is not necessarily true: {False} == {0},
+    # but choices_key([False]) != choices_key([0]).

hypothesis-python/tests/cover/test_database_backend.py

@@ -478,7 +478,7 @@ def test_background_write_database():
 @example(ir("a"))
 @example(ir(b"a"))
 @example(ir(b"a" * 50))
-def test_ir_nodes_rountrips(nodes1):
+def test_ir_nodes_roundtrips(nodes1):
     s1 = ir_to_bytes([n.value for n in nodes1])
     assert isinstance(s1, bytes)
     ir2 = ir_from_bytes(s1)