Unify PyUnicode_Decode with codecs module

Author: msimacek

graalpython/com.oracle.graal.python.cext/src/unicodeobject.c

@@ -568,13 +568,16 @@ PyObject * PyUnicode_AsUnicodeEscapeString(PyObject *unicode) {
 	return UPCALL_CEXT_O(_jls_PyUnicode_AsUnicodeEscapeString, native_to_java(unicode));
 }
 
-typedef PyObject* (*unicode_PyUnicode_Decode_fun_t)(void *data, const char *encoding, const char *errors);
-UPCALL_TYPED_ID(PyUnicode_Decode, unicode_PyUnicode_Decode_fun_t);
+UPCALL_ID(PyUnicode_Decode);
 PyObject * PyUnicode_Decode(const char *s, Py_ssize_t size, const char *encoding, const char *errors) {
     if (encoding == NULL) {
         return PyUnicode_DecodeUTF8Stateful(s, size, errors, NULL);
     }
-	return _jls_PyUnicode_Decode(polyglot_from_i8_array(s, size), polyglot_from_string(encoding, SRC_CS), convert_errors(errors));
+    PyObject *mv = PyMemoryView_FromMemory(s, size, PyBUF_READ);
+    if (!mv) {
+        return NULL;
+    }
+	return UPCALL_CEXT_O(_jls_PyUnicode_Decode, mv, polyglot_from_string(encoding, SRC_CS), convert_errors(errors));
 }
 
 PyObject * PyUnicode_DecodeASCII(const char *s, Py_ssize_t size, const char *errors) {

graalpython/com.oracle.graal.python/src/com/oracle/graal/python/builtins/modules/PythonCextBuiltins.java

@@ -2595,35 +2595,21 @@ protected static boolean isReprFormatCode(int formatCode) {
 
     @Builtin(name = "PyUnicode_Decode", minNumOfPositionalArgs = 4, declaresExplicitSelf = true)
     @GenerateNodeFactory
-    abstract static class PyUnicode_Decode extends NativeUnicodeBuiltin {
+    abstract static class PyUnicode_Decode extends NativeBuiltin {
 
         @Specialization
-        Object doDecode(VirtualFrame frame, Object module, Object cByteArray, String encoding, String errors,
-                        @Cached CExtNodes.ToSulongNode toSulongNode,
-                        @Cached GetByteArrayNode getByteArrayNode,
+        Object doDecode(VirtualFrame frame, Object module, PMemoryView mv, String encoding, String errors,
+                        @Cached CodecsModuleBuiltins.DecodeNode decodeNode,
+                        @Cached CExtNodes.ToNewRefNode toSulongNode,
+                        @Cached TransformExceptionToNativeNode transformExceptionToNativeNode,
                         @Cached GetNativeNullNode getNativeNullNode) {
-
             try {
-                ByteBuffer inputBuffer = wrap(getByteArrayNode.execute(cByteArray, -1));
-                int n = remaining(inputBuffer);
-                CharBuffer resultBuffer = allocateCharBuffer(n * 4);
-                decode(resultBuffer, inputBuffer, encoding, errors);
-                return toSulongNode.execute(factory().createTuple(new Object[]{toString(resultBuffer), n - remaining(inputBuffer)}));
-            } catch (IllegalArgumentException e) {
-                return raiseNative(frame, getNativeNullNode.execute(module), PythonErrorType.LookupError, ErrorMessages.UNKNOWN_ENCODING, encoding);
-            } catch (InteropException e) {
-                return raiseNative(frame, getNativeNullNode.execute(module), PythonErrorType.TypeError, "%m", e);
-            } catch (OverflowException e) {
-                return raiseNative(frame, getNativeNullNode.execute(module), PythonErrorType.SystemError, ErrorMessages.INPUT_TOO_LONG);
+                return toSulongNode.execute(decodeNode.executeWithStrings(frame, mv, encoding, errors));
+            } catch (PException e) {
+                transformExceptionToNativeNode.execute(frame, e);
+                return toSulongNode.execute(getNativeNullNode.execute(module));
             }
         }
-
-        @TruffleBoundary
-        private void decode(CharBuffer resultBuffer, ByteBuffer inputBuffer, String encoding, String errors) {
-            CharsetDecoder decoder = Charset.forName(encoding).newDecoder();
-            CodingErrorAction action = BytesBuiltins.toCodingErrorAction(errors, this);
-            decoder.onMalformedInput(CodingErrorAction.REPORT).onUnmappableCharacter(action).decode(inputBuffer, resultBuffer, true);
-        }
     }
 
     @Builtin(name = "PyObject_Size", minNumOfPositionalArgs = 1)