[mypyc] Add primitive for bytes decode() method

Author: EvgenyVashkevich

mypyc/lib-rt/CPy.h

@@ -662,7 +662,6 @@ int CPyList_Insert(PyObject *list, CPyTagged index, PyObject *value);
 PyObject *CPyList_Extend(PyObject *o1, PyObject *o2);
 int CPyList_Remove(PyObject *list, PyObject *obj);
 CPyTagged CPyList_Index(PyObject *list, PyObject *obj);
-PyObject *CPySequence_Sort(PyObject *seq);
 PyObject *CPySequence_Multiply(PyObject *seq, CPyTagged t_size);
 PyObject *CPySequence_RMultiply(CPyTagged t_size, PyObject *seq);
 PyObject *CPySequence_InPlaceMultiply(PyObject *seq, CPyTagged t_size);
@@ -764,6 +763,7 @@ CPyTagged CPyBytes_GetItem(PyObject *o, CPyTagged index);
 PyObject *CPyBytes_Concat(PyObject *a, PyObject *b);
 PyObject *CPyBytes_Join(PyObject *sep, PyObject *iter);
 CPyTagged CPyBytes_Ord(PyObject *obj);
+PyObject *CPy_DecodeUtf8(PyObject *bytes_obj, const char *errors);
 
 
 int CPyBytes_Compare(PyObject *left, PyObject *right);

mypyc/lib-rt/bytes_ops.c

@@ -162,3 +162,16 @@ CPyTagged CPyBytes_Ord(PyObject *obj) {
     PyErr_SetString(PyExc_TypeError, "ord() expects a character");
     return CPY_INT_TAG;
 }
+
+
+PyObject *CPy_DecodeUtf8(PyObject *bytes_obj, const char *errors) {
+    if (!PyBytes_Check(bytes_obj)) {
+        PyErr_SetString(PyExc_TypeError, "expected bytes object");
+        return NULL;
+    }
+
+    char *data = PyBytes_AS_STRING(bytes_obj);
+    Py_ssize_t size = PyBytes_GET_SIZE(bytes_obj);
+
+    return PyUnicode_DecodeUTF8(data, size, errors);
+}

mypyc/primitives/bytes_ops.py

@@ -107,3 +107,11 @@
     c_function_name="CPyBytes_Ord",
     error_kind=ERR_MAGIC,
 )
+
+method_op(
+    name="decode",
+    arg_types=[bytes_rprimitive, bytes_rprimitive],
+    return_type=str_rprimitive,
+    c_function_name="CPy_DecodeUtf8",
+    error_kind=ERR_MAGIC,
+)

mypyc/test-data/irbuild-bytes.test

@@ -185,3 +185,15 @@ L0:
     r10 = CPyBytes_Build(2, var, r9)
     b4 = r10
     return 1
+
+[case testDecodeUtf8]
+def f(b: bytes) -> str:
+    return b.decode("utf-8")
+[out]
+def f(b):
+    b :: bytes
+    r0, r1 :: str
+L0:
+    r0 = 'utf-8'
+    r1 = CPy_Decode(b, r0, 0)
+    return r1

mypyc/test-data/run-bytes.test

@@ -323,3 +323,29 @@ class A:
 def test_bytes_dunder() -> None:
     assert b'%b' % A() == b'aaa'
     assert b'%s' % A() == b'aaa'
+
+[case testDecodeUtf8]
+from typing import Any
+from testutil import assertRaises
+from a import bytes_subclass
+
+def test_decode_utf8() -> None:
+    assert b'hello'.decode('utf-8') == 'hello'
+    assert b''.decode('utf-8') == ''
+
+    x: bytes = bytearray(b'hello')
+    assert x.decode('utf-8') == 'hello'
+    assert type(x.decode('utf-8')) == str
+
+    y: Any = bytes_subclass()
+    assert y.decode('utf-8') == 'spook'
+
+    n: Any = 123
+    with assertRaises(AttributeError):
+        n.decode('utf-8')
+
+
+[file a.py]
+class bytes_subclass(bytes):
+    def decode(self, encoding='utf-8'):
+        return 'spook'