Implement C API function PyUnicode_Decode.

Author: fangerer

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

@@ -476,3 +476,11 @@ UPCALL_ID(PyUnicode_AsUnicodeEscapeString);
 PyObject * PyUnicode_AsUnicodeEscapeString(PyObject *unicode) {
 	return UPCALL_CEXT_O(_jls_PyUnicode_AsUnicodeEscapeString, native_to_java(unicode));
 }
+
+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 UPCALL_CEXT_O(_jls_PyUnicode_Decode, s, size, polyglot_from_string(encoding, SRC_CS), polyglot_from_string(errors, SRC_CS));
+}

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

@@ -57,6 +57,7 @@
 import java.nio.charset.CoderResult;
 import java.nio.charset.CodingErrorAction;
 import java.nio.charset.StandardCharsets;
+import java.nio.charset.UnsupportedCharsetException;
 import java.text.DecimalFormat;
 import java.text.ParseException;
 import java.text.ParsePosition;
@@ -945,6 +946,22 @@ protected static String getUTF32Name(int byteorder) {
             }
             return csName;
         }
+
+        @TruffleBoundary
+        protected static CharBuffer allocateCharBuffer(int cap) {
+            return CharBuffer.allocate(cap);
+        }
+
+        @TruffleBoundary
+        protected static String toString(CharBuffer cb) {
+            return cb.toString();
+        }
+
+        @TruffleBoundary
+        protected static int remaining(ByteBuffer cb) {
+            return cb.remaining();
+        }
+
     }
 
     @Builtin(name = "TrufflePInt_AsPrimitive", minNumOfPositionalArgs = 3)
@@ -2555,21 +2572,6 @@ Object doUtf8Decode(VirtualFrame frame, Object module, Object cByteArray, String
             }
         }
 
-        @TruffleBoundary
-        private static CharBuffer allocateCharBuffer(int cap) {
-            return CharBuffer.allocate(cap);
-        }
-
-        @TruffleBoundary
-        private static String toString(CharBuffer cb) {
-            return cb.toString();
-        }
-
-        @TruffleBoundary
-        private static int remaining(ByteBuffer cb) {
-            return cb.remaining();
-        }
-
         @TruffleBoundary
         private CoderResult decodeUTF8(CharBuffer resultBuffer, ByteBuffer inputBuffer, String errors) {
             CharsetDecoder decoder = StandardCharsets.UTF_8.newDecoder();
@@ -2626,4 +2628,35 @@ private Number parse(String source) throws ParseException {
             return DecimalFormat.getInstance().parse(source);
         }
     }
+
+    @Builtin(name = "PyUnicode_Decode", minNumOfPositionalArgs = 5, declaresExplicitSelf = true)
+    @GenerateNodeFactory
+    abstract static class PyUnicode_Decode extends NativeUnicodeBuiltin {
+
+        @Specialization
+        Object doDecode(VirtualFrame frame, Object module, Object cByteArray, long size, String encoding, String errors,
+                            @Cached CExtNodes.ToSulongNode toSulongNode,
+                            @Cached GetByteArrayNode getByteArrayNode,
+                            @Cached GetNativeNullNode getNativeNullNode) {
+
+            try {
+                ByteBuffer inputBuffer = wrap(getByteArrayNode.execute(frame, cByteArray, size));
+                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, "unknown encoding: " + encoding);
+            } catch (InteropException e) {
+                return raiseNative(frame, getNativeNullNode.execute(module), PythonErrorType.TypeError, "%m", e);
+            }
+        }
+
+        @TruffleBoundary
+        private CoderResult decode(CharBuffer resultBuffer, ByteBuffer inputBuffer, String encoding, String errors) {
+            CharsetDecoder decoder = Charset.forName(encoding).newDecoder();
+            CodingErrorAction action = BytesBuiltins.toCodingErrorAction(errors, this);
+            return decoder.onMalformedInput(CodingErrorAction.REPORT).onUnmappableCharacter(action).decode(inputBuffer, resultBuffer, true);
+        }
+    }
 }