[mypyc] Add bytes and bytearray initialization ops (#10900)

Use `PyBytes_FromObject` for `bytes(o)`.
Use `PyByteArray_FromObject` for `bytearray(o)`.

Current approach doesn't support empty initialization.

Related to mypyc/mypyc#880.

mypyc/common.py

@@ -55,9 +55,10 @@
     'getargs.c',
     'getargsfast.c',
     'int_ops.c',
+    'str_ops.c',
+    'bytes_ops.c',
     'list_ops.c',
     'dict_ops.c',
-    'str_ops.c',
     'set_ops.c',
     'tuple_ops.c',
     'exc_ops.c',

mypyc/lib-rt/CPy.h

@@ -397,6 +397,10 @@ bool CPyStr_IsTrue(PyObject *obj);
 Py_ssize_t CPyStr_Size_size_t(PyObject *str);
 
 
+// Bytes operations
+
+
+
 // Set operations
 
 

mypyc/lib-rt/bytes_ops.c

@@ -0,0 +1,6 @@
+// Bytes primitive operations
+//
+// These are registered in mypyc.primitives.bytes_ops.
+
+#include <Python.h>
+#include "CPy.h"

mypyc/primitives/bytes_ops.py

@@ -1,11 +1,31 @@
 """Primitive bytes ops."""
 
-from mypyc.ir.rtypes import object_rprimitive
-from mypyc.primitives.registry import load_address_op
+from mypyc.ir.ops import ERR_MAGIC
+from mypyc.ir.rtypes import (
+    object_rprimitive, bytes_rprimitive, list_rprimitive, dict_rprimitive,
+    str_rprimitive, RUnion
+)
+from mypyc.primitives.registry import load_address_op, function_op
 
 
 # Get the 'bytes' type object.
 load_address_op(
     name='builtins.bytes',
     type=object_rprimitive,
     src='PyBytes_Type')
+
+# bytes(obj)
+function_op(
+    name='builtins.bytes',
+    arg_types=[RUnion([list_rprimitive, dict_rprimitive, str_rprimitive])],
+    return_type=bytes_rprimitive,
+    c_function_name='PyBytes_FromObject',
+    error_kind=ERR_MAGIC)
+
+# bytearray(obj)
+function_op(
+    name='builtins.bytearray',
+    arg_types=[object_rprimitive],
+    return_type=bytes_rprimitive,
+    c_function_name='PyByteArray_FromObject',
+    error_kind=ERR_MAGIC)

mypyc/primitives/int_ops.py

@@ -28,7 +28,7 @@
     type=object_rprimitive,
     src='PyLong_Type')
 
-# Convert from a float to int. We could do a bit better directly.
+# int(float). We could do a bit better directly.
 function_op(
     name='builtins.int',
     arg_types=[float_rprimitive],
@@ -52,7 +52,7 @@
     c_function_name='CPyLong_FromStrWithBase',
     error_kind=ERR_MAGIC)
 
-# str(n) on ints
+# str(int)
 int_to_str_op = function_op(
     name='builtins.str',
     arg_types=[int_rprimitive],

mypyc/test-data/irbuild-bytes.test

@@ -0,0 +1,64 @@
+[case testBytesBasics]
+def f(num: int, l: list, d: dict, s: str) -> None:
+    b1 = bytes()
+    b2 = bytes(num)
+    b3 = bytes(l)
+    b4 = bytes(d)
+    b5 = bytes(s)
+[out]
+def f(num, l, d, s):
+    num :: int
+    l :: list
+    d :: dict
+    s :: str
+    r0, r1 :: object
+    r2, b1 :: bytes
+    r3, r4, r5 :: object
+    r6, b2, r7, b3, r8, b4, r9, b5 :: bytes
+L0:
+    r0 = load_address PyBytes_Type
+    r1 = PyObject_CallFunctionObjArgs(r0, 0)
+    r2 = cast(bytes, r1)
+    b1 = r2
+    r3 = load_address PyBytes_Type
+    r4 = box(int, num)
+    r5 = PyObject_CallFunctionObjArgs(r3, r4, 0)
+    r6 = cast(bytes, r5)
+    b2 = r6
+    r7 = PyBytes_FromObject(l)
+    b3 = r7
+    r8 = PyBytes_FromObject(d)
+    b4 = r8
+    r9 = PyBytes_FromObject(s)
+    b5 = r9
+    return 1
+
+[case testBytearrayBasics]
+def f(s: str, num: int) -> None:
+    a = bytearray()
+    b = bytearray(s)
+    c = bytearray(num)
+[out]
+def f(s, num):
+    s :: str
+    num :: int
+    r0 :: object
+    r1 :: str
+    r2, r3, a :: object
+    r4 :: bytes
+    b, r5 :: object
+    r6 :: bytes
+    c :: object
+L0:
+    r0 = builtins :: module
+    r1 = 'bytearray'
+    r2 = CPyObject_GetAttr(r0, r1)
+    r3 = PyObject_CallFunctionObjArgs(r2, 0)
+    a = r3
+    r4 = PyByteArray_FromObject(s)
+    b = r4
+    r5 = box(int, num)
+    r6 = PyByteArray_FromObject(r5)
+    c = r6
+    return 1
+

mypyc/test-data/run-bytes.test

@@ -24,6 +24,24 @@ try:
 except TypeError:
     pass
 
+[case testBytesInit]
+def test_bytes_init() -> None:
+    b1 = bytes([5])
+    assert b1 == b'\x05'
+    b2 = bytes([5, 10, 12])
+    assert b2 == b'\x05\n\x0c'
+    b3 = bytes(bytearray(b'foo'))
+    assert b3 == b'foo'
+    b4 = bytes(b'aaa')
+    assert b4 == b'aaa'
+    b5 = bytes(5)
+    assert b5 == b'\x00\x00\x00\x00\x00'
+    try:
+        bytes('x')
+        assert False
+    except TypeError:
+        pass
+
 [case testBytesOps]
 def test_indexing() -> None:
     # Use bytes() to avoid constant folding

mypyc/test/test_irbuild.py

@@ -16,20 +16,21 @@
 
 files = [
     'irbuild-basic.test',
+    'irbuild-int.test',
     'irbuild-lists.test',
+    'irbuild-tuple.test',
     'irbuild-dict.test',
+    'irbuild-set.test',
+    'irbuild-str.test',
+    'irbuild-bytes.test',
     'irbuild-statements.test',
     'irbuild-nested.test',
     'irbuild-classes.test',
     'irbuild-optional.test',
-    'irbuild-tuple.test',
     'irbuild-any.test',
     'irbuild-generics.test',
     'irbuild-try.test',
-    'irbuild-set.test',
-    'irbuild-str.test',
     'irbuild-strip-asserts.test',
-    'irbuild-int.test',
     'irbuild-vectorcall.test',
     'irbuild-unreachable.test',
     'irbuild-isinstance.test',