Implement C API function PyUnicode_ReadChar

Author: fangerer

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

@@ -659,3 +659,8 @@ void* native_unicode_as_string(PyObject *string) {
     assert(kind == PyUnicode_4BYTE_KIND);
     return polyglot_from_string_n((const char *)data, sizeof(Py_UCS4) * len, "UTF-32LE");
 }
+
+UPCALL_ID(PyUnicode_ReadChar);
+Py_UCS4 PyUnicode_ReadChar(PyObject *unicode, Py_ssize_t index) {
+	return UPCALL_CEXT_I(_jls_PyUnicode_ReadChar, native_to_java(unicode), index);
+}

graalpython/com.oracle.graal.python.test/src/tests/cpyext/test_unicode.py

@@ -84,6 +84,15 @@ def _reference_fromformat(args):
     return fmt % fmt_args
 
 
+def _reference_readchar(args):
+    s = args[0]
+    i = args[1]
+    if i < 0:
+        # just ensure that it is out of bounds
+        i = len(s) + 1
+    return ord(s[i])
+
+
 class CustomString(str):
     pass
 
@@ -602,4 +611,28 @@ def compile_module(self, name):
         cmpfunc=unhandled_error_compare
     )
 
+    test_PyUnicode_ReadChar = CPyExtFunction(
+        _reference_readchar,
+        lambda: (
+            ("hello", 0),
+            ("hello", 4),
+            ("hello", 100),
+            ("hello", -1),
+            ("höllö", 4),
+        ),
+        code='''PyObject* wrap_PyUnicode_ReadChar(PyObject* unicode, Py_ssize_t index) {
+            Py_UCS4 res = PyUnicode_ReadChar(unicode, index);
+            if (res == -1 && PyErr_Occurred()) {
+               return NULL;
+            }
+            return PyLong_FromLong((long) res);
+        }
+        ''',
+        resultspec="O",
+        argspec='On',
+        arguments=["PyObject* str", "Py_ssize_t index"],
+        callfunction="wrap_PyUnicode_ReadChar",
+        cmpfunc=unhandled_error_compare
+    )
+
 

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

@@ -4547,4 +4547,32 @@ static int doSet(PSet type,
             return lib.length(type.getDictStorage());
         }
     }
+
+    @Builtin(name = "PyUnicode_ReadChar", minNumOfPositionalArgs = 2)
+    @GenerateNodeFactory
+    abstract static class PyUnicodeReadChar extends PythonBinaryBuiltinNode {
+        @Specialization
+        int doGeneric(Object type, long lindex,
+                        @Cached CastToJavaStringNode castToJavaStringNode,
+                        @Cached TransformExceptionToNativeNode transformExceptionToNativeNode) {
+            try {
+                try {
+                    String s = castToJavaStringNode.execute(type);
+                    int index = PInt.intValueExact(lindex);
+                    // avoid StringIndexOutOfBoundsException
+                    if (index < 0 || index >= PString.length(s)) {
+                        throw raise(IndexError, ErrorMessages.STRING_INDEX_OUT_OF_RANGE);
+                    }
+                    return PString.charAt(s, index);
+                } catch (CannotCastException e) {
+                    throw raise(TypeError, ErrorMessages.BAD_ARG_TYPE_FOR_BUILTIN_OP);
+                } catch (OverflowException e) {
+                    throw raise(IndexError, ErrorMessages.STRING_INDEX_OUT_OF_RANGE);
+                }
+            } catch (PException e) {
+                transformExceptionToNativeNode.execute(e);
+                return -1;
+            }
+        }
+    }
 }