Switch to hasher based hashing.

Signed-off-by: Maxim Zaks <[email protected]>

stdlib/src/builtin/dtype.mojo

@@ -16,7 +16,6 @@ These are Mojo built-ins, so you don't need to import them.
 """
 
 from collections import KeyElement
-from hashlib._hasher import _HashableWithHasher, _Hasher
 from sys import sizeof, bitwidthof, os_is_windows
 
 alias _mIsSigned = UInt8(1)
@@ -33,7 +32,6 @@ struct DType(
     Representable,
     KeyElement,
     CollectionElementNew,
-    _HashableWithHasher,
 ):
     """Represents DType and provides methods for working with it."""
 
@@ -299,15 +297,7 @@ struct DType(
             self._as_i8(), rhs._as_i8()
         )
 
-    fn __hash__(self) -> UInt:
-        """Return a 64-bit hash for this `DType` value.
-
-        Returns:
-            A 64-bit integer hash of this `DType` value.
-        """
-        return hash(UInt8(self._as_i8()))
-
-    fn __hash__[H: _Hasher](self, inout hasher: H):
+    fn __hash__[H: Hasher](self, inout hasher: H):
         """Updates hasher with this `DType` value.
 
         Parameters:

stdlib/src/builtin/int.mojo

@@ -19,8 +19,6 @@ from collections import KeyElement
 
 from builtin._documentation import doc_private
 from builtin._math import Ceilable, CeilDivable, Floorable, Truncable
-from hashlib.hash import _hash_simd
-from hashlib._hasher import _HashableWithHasher, _Hasher
 from builtin.io import _snprintf
 from collections.string import (
     _calc_initial_buffer_size_int32,
@@ -287,7 +285,6 @@ struct Int(
     KeyElement,
     Roundable,
     IntLike,
-    _HashableWithHasher,
 ):
     """This type represents an integer value."""
 
@@ -1103,18 +1100,7 @@ struct Int(
         """
         return str(self)
 
-    fn __hash__(self) -> UInt:
-        """Hash the int using builtin hash.
-
-        Returns:
-            A 64-bit hash value. This value is _not_ suitable for cryptographic
-            uses. Its intended usage is for data structures. See the `hash`
-            builtin documentation for more details.
-        """
-        # TODO(MOCO-636): switch to DType.index
-        return _hash_simd(Scalar[DType.int64](self))
-
-    fn __hash__[H: _Hasher](self, inout hasher: H):
+    fn __hash__[H: Hasher](self, inout hasher: H):
         """Updates hasher with this int value.
 
         Parameters:

stdlib/src/builtin/simd.mojo

@@ -37,8 +37,6 @@ from bit import pop_count
 from builtin._documentation import doc_private
 from builtin._math import Ceilable, CeilDivable, Floorable, Truncable
 from builtin.dtype import _uint_type_of_width
-from hashlib.hash import _hash_simd
-from hashlib._hasher import _HashableWithHasher, _Hasher
 from builtin.format_int import _try_write_int
 from collections import InlineArray
 from memory import bitcast, UnsafePointer
@@ -177,7 +175,6 @@ struct SIMD[type: DType, size: Int](
     Floorable,
     Writable,
     Hashable,
-    _HashableWithHasher,
     Intable,
     Powable,
     Representable,
@@ -1536,17 +1533,7 @@ struct SIMD[type: DType, size: Int](
         # TODO: see how can we implement this.
         return llvm_intrinsic["llvm.round", Self, has_side_effect=False](self)
 
-    fn __hash__(self) -> UInt:
-        """Hash the value using builtin hash.
-
-        Returns:
-            A 64-bit hash value. This value is _not_ suitable for cryptographic
-            uses. Its intended usage is for data structures. See the `hash`
-            builtin documentation for more details.
-        """
-        return _hash_simd(self)
-
-    fn __hash__[H: _Hasher](self, inout hasher: H):
+    fn __hash__[H: Hasher](self, inout hasher: H):
         """Updates hasher with this SIMD value.
 
         Parameters:

stdlib/src/builtin/string_literal.mojo

@@ -19,7 +19,7 @@ from sys.ffi import c_char
 
 from memory import memcpy, UnsafePointer
 from collections import List
-from hashlib._hasher import _HashableWithHasher, _Hasher
+from hashlib.hasher import Hashable, Hasher
 from utils import StringRef, Span, StringSlice, StaticString
 from utils import Writable, Writer
 from utils._visualizers import lldb_formatter_wrapping_type
@@ -49,7 +49,6 @@ struct StringLiteral(
     Stringable,
     FloatableRaising,
     BytesCollectionElement,
-    _HashableWithHasher,
 ):
     """This type represents a string literal.
 
@@ -264,17 +263,7 @@ struct StringLiteral(
         """
         return self.__str__().__repr__()
 
-    fn __hash__(self) -> UInt:
-        """Hash the underlying buffer using builtin hash.
-
-        Returns:
-            A 64-bit hash value. This value is _not_ suitable for cryptographic
-            uses. Its intended usage is for data structures. See the `hash`
-            builtin documentation for more details.
-        """
-        return hash(self.unsafe_ptr(), len(self))
-
-    fn __hash__[H: _Hasher](self, inout hasher: H):
+    fn __hash__[H: Hasher](self, inout hasher: H):
         """Updates hasher with the underlying bytes.
 
         Parameters:

stdlib/src/builtin/uint.mojo

@@ -18,14 +18,12 @@ These are Mojo built-ins, so you don't need to import them.
 from sys import bitwidthof
 from utils._visualizers import lldb_formatter_wrapping_type
 from builtin._documentation import doc_private
-from hashlib.hash import _hash_simd
-from hashlib._hasher import _HashableWithHasher, _Hasher
 
 
 @lldb_formatter_wrapping_type
 @value
 @register_passable("trivial")
-struct UInt(IntLike, _HashableWithHasher):
+struct UInt(IntLike, Hashable):
     """This type represents an unsigned integer.
 
     An unsigned integer is represents a positive integral number.
@@ -151,18 +149,7 @@ struct UInt(IntLike, _HashableWithHasher):
         """
         return "UInt(" + str(self) + ")"
 
-    fn __hash__(self) -> UInt:
-        """Hash the UInt using builtin hash.
-
-        Returns:
-            A 64-bit hash value. This value is _not_ suitable for cryptographic
-            uses. Its intended usage is for data structures. See the `hash`
-            builtin documentation for more details.
-        """
-        # TODO(MOCO-636): switch to DType.index
-        return _hash_simd(Scalar[DType.uint64](self))
-
-    fn __hash__[H: _Hasher](self, inout hasher: H):
+    fn __hash__[H: Hasher](self, inout hasher: H):
         """Updates hasher with this uint value.
 
         Parameters:

stdlib/src/collections/dict.mojo

@@ -213,7 +213,7 @@ struct DictEntry[K: KeyElement, V: CollectionElement](
         V: The value type of the dict.
     """
 
-    var hash: Int
+    var hash: UInt64
     """`key.__hash__()`, stored so hashing isn't re-computed during dict lookup."""
     var key: K
     """The unique key for the entry."""
@@ -315,7 +315,7 @@ struct _DictIndex:
     fn __moveinit__(inout self, owned existing: Self):
         self.data = existing.data
 
-    fn get_index(self, reserved: Int, slot: Int) -> Int:
+    fn get_index(self, reserved: Int, slot: UInt64) -> Int:
         if reserved <= 128:
             var data = self.data.bitcast[Int8]()
             return int(data.load(slot & (reserved - 1)))
@@ -329,7 +329,7 @@ struct _DictIndex:
             var data = self.data.bitcast[Int64]()
             return int(data.load(slot & (reserved - 1)))
 
-    fn set_index(inout self, reserved: Int, slot: Int, value: Int):
+    fn set_index(inout self, reserved: Int, slot: UInt64, value: Int):
         if reserved <= 128:
             var data = self.data.bitcast[Int8]()
             return data.store(slot & (reserved - 1), value)
@@ -770,7 +770,7 @@ struct Dict[K: KeyElement, V: CollectionElement](
         """
         var hash = hash(key)
         var found: Bool
-        var slot: Int
+        var slot: UInt64
         var index: Int
         found, slot, index = self._find_index(hash, key)
         if found:
@@ -835,7 +835,7 @@ struct Dict[K: KeyElement, V: CollectionElement](
         """
         var hash = hash(key)
         var found: Bool
-        var slot: Int
+        var slot: UInt64
         var index: Int
         found, slot, index = self._find_index(hash, key)
         if found:
@@ -962,7 +962,7 @@ struct Dict[K: KeyElement, V: CollectionElement](
         if not safe_context:
             self._maybe_resize()
         var found: Bool
-        var slot: Int
+        var slot: UInt64
         var index: Int
         found, slot, index = self._find_index(entry.hash, entry.key)
 
@@ -972,30 +972,30 @@ struct Dict[K: KeyElement, V: CollectionElement](
             self.size += 1
             self._n_entries += 1
 
-    fn _get_index(self, slot: Int) -> Int:
+    fn _get_index(self, slot: UInt64) -> Int:
         return self._index.get_index(self._reserved(), slot)
 
-    fn _set_index(inout self, slot: Int, index: Int):
+    fn _set_index(inout self, slot: UInt64, index: Int):
         return self._index.set_index(self._reserved(), slot, index)
 
-    fn _next_index_slot(self, inout slot: Int, inout perturb: UInt64):
+    fn _next_index_slot(self, inout slot: UInt64, inout perturb: UInt64):
         alias PERTURB_SHIFT = 5
         perturb >>= PERTURB_SHIFT
         slot = ((5 * slot) + int(perturb + 1)) & (self._reserved() - 1)
 
-    fn _find_empty_index(self, hash: Int) -> Int:
+    fn _find_empty_index(self, hash: UInt64) -> UInt64:
         var slot = hash & (self._reserved() - 1)
-        var perturb = bitcast[DType.uint64](Int64(hash))
+        var perturb = hash
         while True:
             var index = self._get_index(slot)
             if index == Self.EMPTY:
                 return slot
             self._next_index_slot(slot, perturb)
 
-    fn _find_index(self, hash: Int, key: K) -> (Bool, Int, Int):
+    fn _find_index(self, hash: UInt64, key: K) -> (Bool, UInt64, Int):
         # Return (found, slot, index)
         var slot = hash & (self._reserved() - 1)
-        var perturb = bitcast[DType.uint64](Int64(hash))
+        var perturb = hash
         while True:
             var index = self._get_index(slot)
             if index == Self.EMPTY:

stdlib/src/collections/set.mojo

@@ -21,7 +21,7 @@ from .dict import (
 )
 
 
-struct Set[T: KeyElement](Sized, Comparable, Hashable, Boolable):
+struct Set[T: KeyElement](Sized, Comparable, Boolable):
     """A set data type.
 
     O(1) average-case amortized add, remove, and membership check.
@@ -287,21 +287,6 @@ struct Set[T: KeyElement](Sized, Comparable, Hashable, Boolable):
         """
         return len(self._data)
 
-    fn __hash__(self) -> UInt:
-        """A hash value of the elements in the set.
-
-        The hash value is order independent, so s1 == s2 -> hash(s1) == hash(s2).
-
-        Returns:
-            A hash value of the set suitable for non-cryptographic purposes.
-        """
-        var hash_value = 0
-        # Hash combination needs to be commutative so iteration order
-        # doesn't impact the hash value.
-        for e in self:
-            hash_value ^= hash(e[])
-        return hash_value
-
     @no_inline
     fn __str__[U: RepresentableKeyElement](self: Set[U]) -> String:
         """Returns the string representation of the set.

stdlib/src/collections/string.mojo

@@ -26,7 +26,6 @@ from memory import UnsafePointer, memcmp, memcpy
 from python import PythonObject
 
 from sys.intrinsics import _type_is_eq
-from hashlib._hasher import _HashableWithHasher, _Hasher
 
 from utils import (
     Span,
@@ -704,7 +703,6 @@ struct String(
     Writer,
     CollectionElementNew,
     FloatableRaising,
-    _HashableWithHasher,
 ):
     """Represents a mutable string."""
 
@@ -1949,17 +1947,7 @@ struct String(
             l_idx += 1
         return self[l_idx:]
 
-    fn __hash__(self) -> UInt:
-        """Hash the underlying buffer using builtin hash.
-
-        Returns:
-            A 64-bit hash value. This value is _not_ suitable for cryptographic
-            uses. Its intended usage is for data structures. See the `hash`
-            builtin documentation for more details.
-        """
-        return hash(self.as_string_slice())
-
-    fn __hash__[H: _Hasher](self, inout hasher: H):
+    fn __hash__[H: Hasher](self, inout hasher: H):
         """Updates hasher with the underlying bytes.
 
         Parameters:

stdlib/src/hashlib/__init__.mojo

@@ -11,4 +11,4 @@
 # limitations under the License.
 # ===----------------------------------------------------------------------=== #
 """Implements the hashlib package that provides various hash algorithms."""
-from .hash import hash, Hashable
+from .hash import hash