Fix PyBytes_FromStringAndSize

Author: fangerer

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

@@ -50,6 +50,11 @@ PyTypeObject PyBytes_Type = PY_TRUFFLE_TYPE_WITH_ITEMSIZE("bytes", &PyType_Type,
 
 typedef PyObject* (*fromStringAndSize_fun_t)(int8_t* str, int64_t sz);
 
+UPCALL_ID(PyBytes_Size);
+Py_ssize_t PyBytes_Size(PyObject *bytes) {
+    return UPCALL_CEXT_L(_jls_PyBytes_Size, native_to_java(bytes));
+}
+
 UPCALL_ID(PyBytes_FromStringAndSize);
 UPCALL_ID(PyTruffle_Bytes_EmptyWithCapacity);
 PyObject* PyBytes_FromStringAndSize(const char* str, Py_ssize_t sz) {
@@ -71,22 +76,16 @@ char* PyBytes_AsString(PyObject *obj) {
     return (char*)(UPCALL_CEXT_NOCAST(_jls_PyTruffle_Bytes_AsString, native_to_java(obj), ERROR_MARKER));
 }
 
-UPCALL_ID(PyBytes_AsStringCheckEmbeddedNull);
+UPCALL_ID(PyTruffle_Bytes_CheckEmbeddedNull);
 int PyBytes_AsStringAndSize(PyObject *obj, char **s, Py_ssize_t *len) {
-    setlocale(LC_ALL, NULL);
-    const char* encoding = nl_langinfo(CODESET);
-    PyObject *result = UPCALL_CEXT_O(_jls_PyBytes_AsStringCheckEmbeddedNull, native_to_java(obj), polyglot_from_string(encoding, SRC_CS));
-    if(result == NULL) {
-        return -1;
-    }
-
-    *s = (char *)as_char_pointer(result);
-
+    PyObject* resolved = native_to_java(obj);
+    *s = (char*)(UPCALL_CEXT_NOCAST(_jls_PyTruffle_Bytes_AsString, resolved, ERROR_MARKER));
     if (len != NULL) {
-        *len = polyglot_as_i64(polyglot_invoke(PY_TRUFFLE_CEXT, "PyTruffle_Object_LEN", native_to_java(obj)));
+        *len = UPCALL_CEXT_L(_jls_PyBytes_Size, resolved);
+        return 0;
+    } else {
+    	return UPCALL_CEXT_I(_jls_PyTruffle_Bytes_CheckEmbeddedNull, resolved);
     }
-
-    return 0;
 }
 
 PyObject * PyBytes_FromFormat(const char *format, ...) {
@@ -261,11 +260,6 @@ void PyBytes_ConcatAndDel(PyObject **bytes, PyObject *newpart) {
     Py_DECREF(newpart);
 }
 
-UPCALL_ID(PyBytes_Size);
-Py_ssize_t PyBytes_Size(PyObject *bytes) {
-    return UPCALL_CEXT_L(_jls_PyBytes_Size, native_to_java(bytes));
-}
-
 int bytes_buffer_getbuffer(PyBytesObject *self, Py_buffer *view, int flags) {
     return PyBuffer_FillInfo(view, (PyObject*)self, (void *)self->ob_sval, Py_SIZE(self), 1, flags);
 }

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

@@ -171,6 +171,8 @@
 import com.oracle.graal.python.builtins.objects.common.PHashingCollection;
 import com.oracle.graal.python.builtins.objects.common.SequenceNodes;
 import com.oracle.graal.python.builtins.objects.common.SequenceStorageNodes;
+import com.oracle.graal.python.builtins.objects.common.SequenceStorageNodes.GetItemScalarNode;
+import com.oracle.graal.python.builtins.objects.common.SequenceStorageNodes.LenNode;
 import com.oracle.graal.python.builtins.objects.dict.PDict;
 import com.oracle.graal.python.builtins.objects.exception.PBaseException;
 import com.oracle.graal.python.builtins.objects.frame.PFrame;
@@ -3829,4 +3831,28 @@ static Object doGeneric(VirtualFrame frame, String format, Object vaList,
             }
         }
     }
+
+    @Builtin(name = "PyTruffle_Bytes_CheckEmbeddedNull", minNumOfPositionalArgs = 1)
+    @GenerateNodeFactory
+    abstract static class PyTruffleBytesCheckEmbeddedNull extends PythonUnaryBuiltinNode {
+
+        @Specialization
+        static int doBytes(PBytes bytes,
+                        @Cached SequenceStorageNodes.LenNode lenNode,
+                        @Cached GetItemScalarNode getItemScalarNode) {
+
+            SequenceStorage sequenceStorage = bytes.getSequenceStorage();
+            int len = lenNode.execute(sequenceStorage);
+            try {
+                for (int i = 0; i < len; i++) {
+                    if (getItemScalarNode.executeInt(sequenceStorage, i) == 0) {
+                        return -1;
+                    }
+                }
+            } catch (ClassCastException e) {
+                throw CompilerDirectives.shouldNotReachHere("bytes object contains non-int value");
+            }
+            return 0;
+        }
+    }
 }

graalpython/lib-graalpython/python_cext.py

@@ -283,23 +283,8 @@ def PyObject_LEN(obj):
     return len(obj)
 
 
-def PyTruffle_Object_LEN(obj):
-    return len(to_java(obj))
-
-
 ##################### BYTES
 
-@may_raise(-1)
-def PyBytes_AsStringCheckEmbeddedNull(obj, encoding):
-    if not PyBytes_Check(obj):
-        raise TypeError('expected bytes, {!s} found'.format(type(obj)))
-    result = obj.decode(encoding)
-    for ch in obj:
-        if ch == 0:
-            raise ValueError('embedded null byte')
-    return result
-
-
 def PyBytes_Size(obj):
     return PyObject_Size(obj)