Merge branch 'master' into truedictprimitive

Author: BobTheBuidler

mypyc/ir/rtypes.py

@@ -192,7 +192,7 @@ def may_be_immortal(self) -> bool:
     def serialize(self) -> str:
         return "void"
 
-    def __eq__(self, other: object) -> bool:
+    def __eq__(self, other: object) -> TypeGuard[RVoid]:
         return isinstance(other, RVoid)
 
     def __hash__(self) -> int:
@@ -279,7 +279,7 @@ def serialize(self) -> str:
     def __repr__(self) -> str:
         return "<RPrimitive %s>" % self.name
 
-    def __eq__(self, other: object) -> bool:
+    def __eq__(self, other: object) -> TypeGuard[RPrimitive]:
         return isinstance(other, RPrimitive) and other.name == self.name
 
     def __hash__(self) -> int:
@@ -520,15 +520,15 @@ def __hash__(self) -> int:
 range_rprimitive: Final = RPrimitive("builtins.range", is_unboxed=False, is_refcounted=True)
 
 
-def is_tagged(rtype: RType) -> bool:
+def is_tagged(rtype: RType) -> TypeGuard[RPrimitive]:
     return rtype is int_rprimitive or rtype is short_int_rprimitive
 
 
-def is_int_rprimitive(rtype: RType) -> bool:
+def is_int_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return rtype is int_rprimitive
 
 
-def is_short_int_rprimitive(rtype: RType) -> bool:
+def is_short_int_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return rtype is short_int_rprimitive
 
 
@@ -542,7 +542,7 @@ def is_int32_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     )
 
 
-def is_int64_rprimitive(rtype: RType) -> bool:
+def is_int64_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return rtype is int64_rprimitive or (
         rtype is c_pyssize_t_rprimitive and rtype._ctype == "int64_t"
     )
@@ -561,88 +561,100 @@ def is_uint8_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return rtype is uint8_rprimitive
 
 
-def is_uint32_rprimitive(rtype: RType) -> bool:
+def is_uint32_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return rtype is uint32_rprimitive
 
 
-def is_uint64_rprimitive(rtype: RType) -> bool:
+def is_uint64_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return rtype is uint64_rprimitive
 
 
-def is_c_py_ssize_t_rprimitive(rtype: RType) -> bool:
+def is_c_py_ssize_t_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return rtype is c_pyssize_t_rprimitive
 
 
-def is_pointer_rprimitive(rtype: RType) -> bool:
+def is_pointer_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return rtype is pointer_rprimitive
 
 
-def is_float_rprimitive(rtype: RType) -> bool:
+def is_float_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.float"
 
 
-def is_bool_rprimitive(rtype: RType) -> bool:
+def is_bool_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.bool"
 
 
-def is_bit_rprimitive(rtype: RType) -> bool:
+def is_bit_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "bit"
 
 
-def is_bool_or_bit_rprimitive(rtype: RType) -> bool:
+def is_bool_or_bit_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return is_bool_rprimitive(rtype) or is_bit_rprimitive(rtype)
 
 
-def is_object_rprimitive(rtype: RType) -> bool:
+def is_object_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.object"
 
 
-def is_none_rprimitive(rtype: RType) -> bool:
+def is_none_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.None"
 
 
-def is_list_rprimitive(rtype: RType) -> bool:
+def is_list_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.list"
 
 
-def is_dict_rprimitive(rtype: RType) -> bool:
+def is_dict_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name in (
         "builtins.dict",
         "builtins.dict[exact]",
     )
 
 
-def is_exact_dict_rprimitive(rtype: RType) -> bool:
+def is_exact_dict_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.dict[exact]"
 
 
-def is_set_rprimitive(rtype: RType) -> bool:
+def is_set_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.set"
 
 
-def is_frozenset_rprimitive(rtype: RType) -> bool:
+def is_frozenset_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.frozenset"
 
 
-def is_str_rprimitive(rtype: RType) -> bool:
+def is_str_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.str"
 
 
-def is_bytes_rprimitive(rtype: RType) -> bool:
+def is_bytes_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.bytes"
 
 
-def is_tuple_rprimitive(rtype: RType) -> bool:
+def is_tuple_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.tuple"
 
 
-def is_range_rprimitive(rtype: RType) -> bool:
+def is_range_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and rtype.name == "builtins.range"
 
 
-def is_sequence_rprimitive(rtype: RType) -> bool:
+def is_sequence_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
     return isinstance(rtype, RPrimitive) and (
-        is_list_rprimitive(rtype) or is_tuple_rprimitive(rtype) or is_str_rprimitive(rtype)
+        is_list_rprimitive(rtype)
+        or is_tuple_rprimitive(rtype)
+        or is_str_rprimitive(rtype)
+        or is_bytes_rprimitive(rtype)
+    )
+
+
+def is_immutable_rprimitive(rtype: RType) -> TypeGuard[RPrimitive]:
+    return (
+        is_str_rprimitive(rtype)
+        or is_bytes_rprimitive(rtype)
+        or is_tuple_rprimitive(rtype)
+        or is_frozenset_rprimitive(rtype)
     )
 
 
@@ -731,7 +743,7 @@ def __str__(self) -> str:
     def __repr__(self) -> str:
         return "<RTuple %s>" % ", ".join(repr(typ) for typ in self.types)
 
-    def __eq__(self, other: object) -> bool:
+    def __eq__(self, other: object) -> TypeGuard[RTuple]:
         return isinstance(other, RTuple) and self.types == other.types
 
     def __hash__(self) -> int:
@@ -864,7 +876,7 @@ def __repr__(self) -> str:
             ", ".join(name + ":" + repr(typ) for name, typ in zip(self.names, self.types)),
         )
 
-    def __eq__(self, other: object) -> bool:
+    def __eq__(self, other: object) -> TypeGuard[RStruct]:
         return (
             isinstance(other, RStruct)
             and self.name == other.name
@@ -934,7 +946,7 @@ def attr_type(self, name: str) -> RType:
     def __repr__(self) -> str:
         return "<RInstance %s>" % self.name
 
-    def __eq__(self, other: object) -> bool:
+    def __eq__(self, other: object) -> TypeGuard[RInstance]:
         return isinstance(other, RInstance) and other.name == self.name
 
     def __hash__(self) -> int:
@@ -988,7 +1000,7 @@ def __str__(self) -> str:
         return "union[%s]" % ", ".join(str(item) for item in self.items)
 
     # We compare based on the set because order in a union doesn't matter
-    def __eq__(self, other: object) -> bool:
+    def __eq__(self, other: object) -> TypeGuard[RUnion]:
         return isinstance(other, RUnion) and self.items_set == other.items_set
 
     def __hash__(self) -> int:
@@ -1030,7 +1042,7 @@ def optional_value_type(rtype: RType) -> RType | None:
     return None
 
 
-def is_optional_type(rtype: RType) -> bool:
+def is_optional_type(rtype: RType) -> TypeGuard[RUnion]:
     """Is rtype an optional type with exactly two union items?"""
     return optional_value_type(rtype) is not None
 
@@ -1062,7 +1074,7 @@ def __str__(self) -> str:
     def __repr__(self) -> str:
         return f"<RArray {self.item_type!r}[{self.length}]>"
 
-    def __eq__(self, other: object) -> bool:
+    def __eq__(self, other: object) -> TypeGuard[RArray]:
         return (
             isinstance(other, RArray)
             and self.item_type == other.item_type

mypyc/irbuild/builder.py

@@ -91,6 +91,7 @@
     RType,
     RUnion,
     bitmap_rprimitive,
+    bytes_rprimitive,
     c_pyssize_t_rprimitive,
     exact_dict_rprimitive,
     int_rprimitive,
@@ -964,8 +965,12 @@ def get_sequence_type_from_type(self, target_type: Type) -> RType:
         elif isinstance(target_type, Instance):
             if target_type.type.fullname == "builtins.str":
                 return str_rprimitive
-            else:
+            elif target_type.type.fullname == "builtins.bytes":
+                return bytes_rprimitive
+            try:
                 return self.type_to_rtype(target_type.args[0])
+            except IndexError:
+                raise ValueError(f"{target_type!r} is not a valid sequence.") from None
         # This elif-blocks are needed for iterating over classes derived from NamedTuple.
         elif isinstance(target_type, TypeVarLikeType):
             return self.get_sequence_type_from_type(target_type.upper_bound)

mypyc/irbuild/for_helpers.py

@@ -49,6 +49,7 @@
     is_dict_rprimitive,
     is_exact_dict_rprimitive,
     is_fixed_width_rtype,
+    is_immutable_rprimitive,
     is_list_rprimitive,
     is_sequence_rprimitive,
     is_short_int_rprimitive,
@@ -211,9 +212,9 @@ def sequence_from_generator_preallocate_helper(
     there is no condition list in the generator and only one original sequence with
     one index is allowed.
 
-    e.g.  (1) tuple(f(x) for x in a_list/a_tuple)
-          (2) list(f(x) for x in a_list/a_tuple)
-          (3) [f(x) for x in a_list/a_tuple]
+    e.g.  (1) tuple(f(x) for x in a_list/a_tuple/a_str/a_bytes)
+          (2) list(f(x) for x in a_list/a_tuple/a_str/a_bytes)
+          (3) [f(x) for x in a_list/a_tuple/a_str/a_bytes]
     RTuple as an original sequence is not supported yet.
 
     Args:
@@ -230,7 +231,7 @@ def sequence_from_generator_preallocate_helper(
     """
     if len(gen.sequences) == 1 and len(gen.indices) == 1 and len(gen.condlists[0]) == 0:
         rtype = builder.node_type(gen.sequences[0])
-        if is_list_rprimitive(rtype) or is_tuple_rprimitive(rtype) or is_str_rprimitive(rtype):
+        if is_sequence_rprimitive(rtype):
             sequence = builder.accept(gen.sequences[0])
             length = builder.builder.builtin_len(sequence, gen.line, use_pyssize_t=True)
             target_op = empty_op_llbuilder(length, gen.line)
@@ -802,17 +803,31 @@ class ForSequence(ForGenerator):
     Supports iterating in both forward and reverse.
     """
 
+    length_reg: Value | AssignmentTarget | None
+
     def init(self, expr_reg: Value, target_type: RType, reverse: bool) -> None:
+        assert is_sequence_rprimitive(expr_reg.type), expr_reg
         builder = self.builder
         self.reverse = reverse
         # Define target to contain the expression, along with the index that will be used
         # for the for-loop. If we are inside of a generator function, spill these into the
         # environment class.
         self.expr_target = builder.maybe_spill(expr_reg)
+        if is_immutable_rprimitive(expr_reg.type):
+            # If the expression is an immutable type, we can load the length just once.
+            self.length_reg = builder.maybe_spill(self.load_len(self.expr_target))
+        else:
+            # Otherwise, even if the length is known, we must recalculate the length
+            # at every iteration for compatibility with python semantics.
+            self.length_reg = None
         if not reverse:
             index_reg: Value = Integer(0, c_pyssize_t_rprimitive)
         else:
-            index_reg = builder.builder.int_sub(self.load_len(self.expr_target), 1)
+            if self.length_reg is not None:
+                len_val = builder.read(self.length_reg)
+            else:
+                len_val = self.load_len(self.expr_target)
+            index_reg = builder.builder.int_sub(len_val, 1)
         self.index_target = builder.maybe_spill_assignable(index_reg)
         self.target_type = target_type
 
@@ -831,9 +846,13 @@ def gen_condition(self) -> None:
             second_check = BasicBlock()
             builder.add_bool_branch(comparison, second_check, self.loop_exit)
             builder.activate_block(second_check)
-        # For compatibility with python semantics we recalculate the length
-        # at every iteration.
-        len_reg = self.load_len(self.expr_target)
+        if self.length_reg is None:
+            # For compatibility with python semantics we recalculate the length
+            # at every iteration.
+            len_reg = self.load_len(self.expr_target)
+        else:
+            # (unless input is immutable type).
+            len_reg = builder.read(self.length_reg, line)
         comparison = builder.binary_op(builder.read(self.index_target, line), len_reg, "<", line)
         builder.add_bool_branch(comparison, self.body_block, self.loop_exit)
 

mypyc/irbuild/specialize.py

@@ -304,7 +304,7 @@ def translate_tuple_from_generator_call(
     """Special case for simplest tuple creation from a generator.
 
     For example:
-        tuple(f(x) for x in some_list/some_tuple/some_str)
+        tuple(f(x) for x in some_list/some_tuple/some_str/some_bytes)
     'translate_safe_generator_call()' would take care of other cases
     if this fails.
     """

mypyc/test-data/fixtures/ir.py

@@ -172,6 +172,7 @@ def __getitem__(self, i: int) -> int: ...
     def __getitem__(self, i: slice) -> bytes: ...
     def join(self, x: Iterable[object]) -> bytes: ...
     def decode(self, x: str=..., y: str=...) -> str: ...
+    def __iter__(self) -> Iterator[int]: ...
 
 class bytearray:
     @overload

mypyc/test-data/irbuild-generics.test

@@ -716,18 +716,18 @@ L0:
     r0 = __mypyc_self__.__mypyc_env__
     r1 = var_object_size args
     r2 = PyList_New(r1)
-    r3 = 0
+    r3 = var_object_size args
+    r4 = 0
 L1:
-    r4 = var_object_size args
-    r5 = r3 < r4 :: signed
+    r5 = r4 < r3 :: signed
     if r5 goto L2 else goto L4 :: bool
 L2:
-    r6 = CPySequenceTuple_GetItemUnsafe(args, r3)
+    r6 = CPySequenceTuple_GetItemUnsafe(args, r4)
     x = r6
-    CPyList_SetItemUnsafe(r2, r3, x)
+    CPyList_SetItemUnsafe(r2, r4, x)
 L3:
-    r7 = r3 + 1
-    r3 = r7
+    r7 = r4 + 1
+    r4 = r7
     goto L1
 L4:
     can_listcomp = r2