Add !R conversion to f-strings and str.format()

Add _PyObject_AltRepr()

Author: vstinner

Include/ceval.h

@@ -130,8 +130,9 @@ PyAPI_FUNC(void) PyEval_ReleaseThread(PyThreadState *tstate);
 #define FVC_STR       0x1
 #define FVC_REPR      0x2
 #define FVC_ASCII     0x3
-#define FVS_MASK      0x4
-#define FVS_HAVE_SPEC 0x4
+#define FVC_ALT_REPR  0x4
+#define FVS_MASK      0x8
+#define FVS_HAVE_SPEC 0x8
 
 #ifndef Py_LIMITED_API
 #  define Py_CPYTHON_CEVAL_H

Include/internal/pycore_object.h

@@ -1037,6 +1037,8 @@ static inline Py_ALWAYS_INLINE void _Py_INCREF_MORTAL(PyObject *op)
  * references. */
 PyAPI_FUNC(int) _PyObject_VisitType(PyObject *op, visitproc visit, void *arg);
 
+extern PyObject* _PyObject_AltRepr(PyObject *);
+
 #ifdef __cplusplus
 }
 #endif

Lib/test/test_fstring.py

@@ -1365,6 +1365,14 @@ def test_conversions(self):
         self.assertEqual(f'{"a"!r}', "'a'")
         self.assertEqual(f'{"a"!a}', "'a'")
 
+        # Test !R conversion
+        class OverrideRepr(str):
+            def __repr__(self):
+                return 'CUSTOM REPR'
+        abc = OverrideRepr('abc')
+        self.assertEqual(f'{abc!r}', 'CUSTOM REPR')
+        self.assertEqual(f'{abc!R}', "'abc'")
+
         # Conversions can have trailing whitespace after them since it
         # does not provide any significance
         self.assertEqual(f"{3!s  }", "3")
@@ -1390,7 +1398,7 @@ def test_conversions(self):
         for conv_identifier in 'g', 'A', 'G', 'ä', 'ɐ':
             self.assertAllRaise(SyntaxError,
                                 "f-string: invalid conversion character %r: "
-                                "expected 's', 'r', or 'a'" % conv_identifier,
+                                "expected 's', 'r', 'R', or 'a'" % conv_identifier,
                                 ["f'{3!" + conv_identifier + "}'"])
 
         for conv_non_identifier in '3', '!':
@@ -1406,7 +1414,7 @@ def test_conversions(self):
 
         self.assertAllRaise(SyntaxError,
                             "f-string: invalid conversion character 'ss': "
-                            "expected 's', 'r', or 'a'",
+                            "expected 's', 'r', 'R', or 'a'",
                             ["f'{3!ss}'",
                              "f'{3!ss:}'",
                              "f'{3!ss:s}'",

Lib/test/test_str.py

@@ -1279,6 +1279,14 @@ def __repr__(self):
         self.assertEqual('{0!a}'.format(F('Hello')), 'F(Hello)')
         self.assertEqual('{0!a}'.format(F('\u0374')), 'F(\\u0374)')
 
+        # !R conversion
+        class OverrideRepr(str):
+            def __repr__(self):
+                return 'CUSTOM REPR'
+        abc = OverrideRepr('abc')
+        self.assertEqual('{0!r}'.format(abc), 'CUSTOM REPR')
+        self.assertEqual('{0!R}'.format(abc), "'abc'")
+
         # test fallback to object.__format__
         self.assertEqual('{0}'.format({}), '{}')
         self.assertEqual('{0}'.format([]), '[]')

Modules/_testinternalcapi/test_cases.c.h

@@ -209,6 +209,9 @@ _PyInterpolation_Build(PyObject *value, PyObject *str, int conversion, PyObject
             case FVC_REPR:
                 interpolation->conversion = _Py_LATIN1_CHR('r');
                 break;
+            case FVC_ALT_REPR:
+                interpolation->conversion = _Py_LATIN1_CHR('R');
+                break;
             case FVC_STR:
                 interpolation->conversion = _Py_LATIN1_CHR('s');
                 break;

Objects/interpolationobject.c

@@ -797,6 +797,19 @@ PyObject_Repr(PyObject *v)
     return res;
 }
 
+PyObject *
+_PyObject_AltRepr(PyObject *v)
+{
+    if (v != NULL && PyUnicode_Check(v)) {
+        if (PyErr_CheckSignals()) {
+            return NULL;
+        }
+        reprfunc repr_func = PyUnicode_Type.tp_repr;
+        return repr_func(v);
+    }
+    return PyObject_Repr(v);
+}
+
 PyObject *
 PyObject_Str(PyObject *v)
 {

Objects/object.c

@@ -770,6 +770,8 @@ do_conversion(PyObject *obj, Py_UCS4 conversion)
     switch (conversion) {
     case 'r':
         return PyObject_Repr(obj);
+    case 'R':
+        return _PyObject_AltRepr(obj);
     case 's':
         return PyObject_Str(obj);
     case 'a':

Objects/stringlib/unicode_format.h

@@ -2974,8 +2974,7 @@ unicode_fromformat_arg(_PyUnicodeWriter *writer,
         assert(obj);
         PyObject *repr;
         if (flags & F_ALT) {
-            reprfunc repr_func = PyUnicode_Type.tp_repr;
-            repr = repr_func(obj);
+            repr = _PyObject_AltRepr(obj);
         }
         else {
             repr = PyObject_Repr(obj);

Objects/unicodeobject.c

@@ -1047,9 +1047,9 @@ _PyPegen_check_fstring_conversion(Parser *p, Token* conv_token, expr_ty conv)
 
     Py_UCS4 first = PyUnicode_READ_CHAR(conv->v.Name.id, 0);
     if (PyUnicode_GET_LENGTH(conv->v.Name.id) > 1 ||
-            !(first == 's' || first == 'r' || first == 'a')) {
+            !(first == 's' || first == 'r' || first == 'R' || first == 'a')) {
         RAISE_SYNTAX_ERROR_KNOWN_LOCATION(conv,
-                                            "%c-string: invalid conversion character %R: expected 's', 'r', or 'a'",
+                                            "%c-string: invalid conversion character %R: expected 's', 'r', 'R', or 'a'",
                                             formatted_string_prefix(p),
                                             conv->v.Name.id);
         return NULL;