Merge pull request #4254 from tybug/tcs-cleanup

Reorganize code for the typed choice sequence

hypothesis-python/RELEASE.rst

@@ -0,0 +1,3 @@
+RELEASE_TYPE: patch
+
+Registration of experimental :ref:`alternative-backends` is now done via ``hypothesis.internal.conjecture.providers.AVAILABLE_PROVIDERS`` instead of ``hypothesis.internal.conjecture.data.AVAILABLE_PROVIDERS``.

hypothesis-python/src/hypothesis/_settings.py

@@ -39,6 +39,7 @@
     InvalidArgument,
     InvalidState,
 )
+from hypothesis.internal.conjecture.providers import AVAILABLE_PROVIDERS
 from hypothesis.internal.reflection import get_pretty_function_description
 from hypothesis.internal.validation import check_type, try_convert
 from hypothesis.utils.conventions import not_set
@@ -296,8 +297,6 @@ def __setattr__(self, name: str, value: object) -> NoReturn:
         raise AttributeError("settings objects are immutable")
 
     def __repr__(self) -> str:
-        from hypothesis.internal.conjecture.data import AVAILABLE_PROVIDERS
-
         bits = sorted(
             f"{name}={getattr(self, name)!r}"
             for name in all_settings
@@ -731,8 +730,6 @@ def is_in_ci() -> bool:
 
 
 def _backend_validator(value: str) -> str:
-    from hypothesis.internal.conjecture.data import AVAILABLE_PROVIDERS
-
     if value not in AVAILABLE_PROVIDERS:
         if value == "crosshair":  # pragma: no cover
             install = '`pip install "hypothesis[crosshair]"` and try again.'

hypothesis-python/src/hypothesis/core.py

@@ -50,7 +50,7 @@
     settings as Settings,
 )
 from hypothesis.control import BuildContext
-from hypothesis.database import ir_from_bytes, ir_to_bytes
+from hypothesis.database import choices_from_bytes, choices_to_bytes
 from hypothesis.errors import (
     BackendCannotProceed,
     DeadlineExceeded,
@@ -77,17 +77,16 @@
     int_from_bytes,
 )
 from hypothesis.internal.conjecture.choice import ChoiceT
-from hypothesis.internal.conjecture.data import (
-    ConjectureData,
-    PrimitiveProvider,
-    Status,
-)
+from hypothesis.internal.conjecture.data import ConjectureData, Status
 from hypothesis.internal.conjecture.engine import BUFFER_SIZE, ConjectureRunner
 from hypothesis.internal.conjecture.junkdrawer import (
     ensure_free_stackframes,
     gc_cumulative_time,
 )
-from hypothesis.internal.conjecture.providers import BytestringProvider
+from hypothesis.internal.conjecture.providers import (
+    BytestringProvider,
+    PrimitiveProvider,
+)
 from hypothesis.internal.conjecture.shrinker import sort_key
 from hypothesis.internal.entropy import deterministic_PRNG
 from hypothesis.internal.escalation import (
@@ -323,7 +322,7 @@ def accept(test):
 
 
 def encode_failure(choices):
-    blob = ir_to_bytes(choices)
+    blob = choices_to_bytes(choices)
     compressed = zlib.compress(blob)
     if len(compressed) < len(blob):
         blob = b"\1" + compressed
@@ -353,7 +352,7 @@ def decode_failure(blob: bytes) -> Sequence[ChoiceT]:
             f"Could not decode blob {blob!r}: Invalid start byte {prefix!r}"
         )
 
-    choices = ir_from_bytes(decoded)
+    choices = choices_from_bytes(decoded)
     if choices is None:
         raise InvalidArgument(f"Invalid serialized choice sequence for blob {blob!r}")
 
@@ -1883,7 +1882,9 @@ def fuzz_one_input(
                     if settings.database is not None and (
                         known is None or sort_key(data.nodes) <= sort_key(known)
                     ):
-                        settings.database.save(database_key, ir_to_bytes(data.choices))
+                        settings.database.save(
+                            database_key, choices_to_bytes(data.choices)
+                        )
                         minimal_failures[data.interesting_origin] = data.nodes
                     raise
                 assert isinstance(data.provider, BytestringProvider)

hypothesis-python/src/hypothesis/database.py

@@ -775,8 +775,8 @@ def _unpack_uleb128(buffer: bytes) -> tuple[int, int]:
     return (i + 1, value)
 
 
-def ir_to_bytes(ir: Iterable[ChoiceT], /) -> bytes:
-    """Serialize a list of IR elements to a bytestring.  Inverts ir_from_bytes."""
+def choices_to_bytes(ir: Iterable[ChoiceT], /) -> bytes:
+    """Serialize a list of IR elements to a bytestring.  Inverts choices_from_bytes."""
     # We use a custom serialization format for this, which might seem crazy - but our
     # data is a flat sequence of elements, and standard tools like protobuf or msgpack
     # don't deal well with e.g. nonstandard bit-pattern-NaNs, or invalid-utf8 unicode.
@@ -815,7 +815,7 @@ def ir_to_bytes(ir: Iterable[ChoiceT], /) -> bytes:
     return b"".join(parts)
 
 
-def _ir_from_bytes(buffer: bytes, /) -> tuple[ChoiceT, ...]:
+def _choices_from_bytes(buffer: bytes, /) -> tuple[ChoiceT, ...]:
     # See above for an explanation of the format.
     parts: list[ChoiceT] = []
     idx = 0
@@ -846,15 +846,15 @@ def _ir_from_bytes(buffer: bytes, /) -> tuple[ChoiceT, ...]:
     return tuple(parts)
 
 
-def ir_from_bytes(buffer: bytes, /) -> Optional[tuple[ChoiceT, ...]]:
+def choices_from_bytes(buffer: bytes, /) -> Optional[tuple[ChoiceT, ...]]:
     """
-    Deserialize a bytestring to a tuple of choices. Inverts ir_to_bytes.
+    Deserialize a bytestring to a tuple of choices. Inverts choices_to_bytes.
 
     Returns None if the given bytestring is not a valid serialization of choice
     sequences.
     """
     try:
-        return _ir_from_bytes(buffer)
+        return _choices_from_bytes(buffer)
     except Exception:
         # deserialization error, eg because our format changed or someone put junk
         # data in the db.

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

@@ -9,7 +9,7 @@
 # obtain one at https://mozilla.org/MPL/2.0/.
 
 import math
-from collections.abc import Sequence
+from collections.abc import Iterable, Sequence
 from typing import (
     TYPE_CHECKING,
     Callable,
@@ -21,6 +21,8 @@
     cast,
 )
 
+import attr
+
 from hypothesis.errors import ChoiceTooLarge
 from hypothesis.internal.conjecture.floats import float_to_lex, lex_to_float
 from hypothesis.internal.conjecture.utils import identity
@@ -72,6 +74,16 @@ class BooleanKWargs(TypedDict):
 ]
 
 
+@attr.s(slots=True)
+class ChoiceTemplate:
+    type: Literal["simplest"] = attr.ib()
+    count: Optional[int] = attr.ib()
+
+    def __attrs_post_init__(self) -> None:
+        if self.count is not None:
+            assert self.count > 0
+
+
 def _size_to_index(size: int, *, alphabet_size: int) -> int:
     # this is the closed form of this geometric series:
     # for i in range(size):
@@ -494,3 +506,9 @@ def choice_kwargs_key(ir_type, kwargs):
             kwargs["shrink_towards"],
         )
     return tuple(kwargs[key] for key in sorted(kwargs))
+
+
+def choices_size(choices: Iterable[ChoiceT]) -> int:
+    from hypothesis.database import choices_to_bytes
+
+    return len(choices_to_bytes(choices))