intrinsified python_cext PyUnicode_XXX

Author: tomasstupka

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

@@ -98,6 +98,56 @@ def _reference_contains(args):
         raise TypeError
     return args[1] in args[0]
 
+def _reference_compare(args):
+    if not isinstance(args[0], str) or not isinstance(args[1], str):
+        if sys.version_info.minor >= 6:
+            raise SystemError
+        else:
+            raise TypeError  
+
+    if args[0] == args[1]:
+        return 0
+    elif args[0] < args[1]:
+        return -1
+    else:
+        return 1
+
+def _reference_as_encoded_string(args):
+    if not isinstance(args[0], str):
+        if sys.version_info.minor >= 6:
+            raise SystemError
+        else:
+            raise TypeError  
+
+    s = args[0]
+    encoding = args[1]
+    errors = args[2]
+    return s.encode(encoding, errors)
+
+_codecs_module = None
+def _reference_as_unicode_escape_string(args):
+    if not isinstance(args[0], str):
+        if sys.version_info.minor >= 6:
+            raise SystemError
+        else:
+            raise TypeError  
+    global _codecs_module
+    if not _codecs_module:
+        import _codecs as _codecs_module
+    return _codecs_module.unicode_escape_encode(args[0])[0]
+
+def _reference_tailmatch(args):
+    if not isinstance(args[0], str) or not isinstance(args[1], str):
+        return -1;
+
+    s = args[0]
+    substr = args[1]
+    start = args[2]
+    end = args[3]
+    direction = args[4]
+    if direction > 0:
+        return 1 if s[start:end].endswith(substr) else 0
+    return 1 if s[start:end].startswith(substr) else 0
 
 class CustomString(str):
     pass
@@ -509,6 +559,37 @@ def compile_module(self, name):
         arguments=["PyObject* str", "PyObject* seq"],
         cmpfunc=unhandled_error_compare
     )
+    
+    test_PyUnicode_Compare = CPyExtFunction(
+        _reference_compare,
+        lambda: (
+            ("a", "a"),
+            ("a", "b"),
+            ("a", None),
+            ("a", 1),
+        ),
+        resultspec="i",
+        argspec='OO',
+        arguments=["PyObject* left", "PyObject* right"],
+        cmpfunc=unhandled_error_compare
+    )
+
+    test_PyUnicode_Tailmatch = CPyExtFunction(
+        _reference_tailmatch,
+        lambda: (
+            ("abc", "a", 0, 1, 0),
+            ("abc", "a", 0, 1, 1),
+            ("abc", "a", 0, 0, 1),
+            ("abc", "c", 0, 1, 0),
+            ("abc", "c", 0, 1, 1),
+            ("abc", None, 0, 1, 1),
+            ("abc", 1, 1, 0, 1),
+        ),
+        resultspec="i",
+        argspec='OOiii',
+        arguments=["PyObject* left", "PyObject* right", "Py_ssize_t start", "Py_ssize_t end", "int direction"],
+        cmpfunc=unhandled_error_compare
+    )
 
     test_PyUnicode_FromOrdinal = CPyExtFunction(
         lambda args: chr(args[0]),
@@ -545,21 +626,34 @@ def compile_module(self, name):
 
 
     test_PyUnicode_AsEncodedString = CPyExtFunction(
-        lambda args: args[0].encode(args[1], args[2]),
+        _reference_as_encoded_string,
         lambda: (
             ("abcd", "ascii", "report"),
             ("abcd", "utf8", "report"),
             ("öüä", "ascii", "report"),
             ("öüä", "utf8", "report"),
             ("öüä", "ascii", "ignore"),
             ("öüä", "ascii", "replace"),
+            (1, "ascii", "replace"),
         ),
         resultspec="O",
         argspec='Oss',
         arguments=["PyObject* str", "const char* encoding", "const char* errors"],
         cmpfunc=unhandled_error_compare
     )
 
+    test_PyUnicode_AsUnicodeEscapeString = CPyExtFunction(
+        _reference_as_unicode_escape_string,
+        lambda: (
+            ("abcd",),
+            ("öüä",),            
+            (1,),
+        ),
+        resultspec="O",
+        argspec='O',
+        arguments=["PyObject* s"],
+        cmpfunc=unhandled_error_compare
+    )
 
     # NOTE: this test assumes that Python uses UTF-8 encoding for source files
     test_PyUnicode_FromWideChar = CPyExtFunction(

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

@@ -1335,7 +1335,7 @@ Object encode(VirtualFrame frame, Object obj, Object errors,
 
     @Builtin(name = "unicode_escape_encode", minNumOfPositionalArgs = 1, parameterNames = {"obj", "errors"})
     @GenerateNodeFactory
-    abstract static class UnicodeEscapeEncodeNode extends PythonBinaryBuiltinNode {
+    public abstract static class UnicodeEscapeEncodeNode extends PythonBinaryBuiltinNode {
         @Specialization
         Object encode(VirtualFrame frame, Object obj, Object errors,
                         @Cached CodecsEncodeNode encode) {