__str__ and __repr__ can return ascii-compatible unicode

src/runtime/objmodel.cpp

@@ -1862,8 +1862,13 @@ extern "C" BoxedString* str(Box* obj) {
         obj = callattrInternal(obj, &str_str, CLASS_ONLY, NULL, ArgPassSpec(0), NULL, NULL, NULL, NULL, NULL);
     }
 
-    if (obj->cls != str_cls) {
-        raiseExcHelper(TypeError, "__str__ did not return a string!");
+    if (isSubclass(obj->cls, unicode_cls)) {
+        obj = PyUnicode_AsASCIIString(obj);
+        checkAndThrowCAPIException();
+    }
+
+    if (!isSubclass(obj->cls, str_cls)) {
+        raiseExcHelper(TypeError, "__str__ returned non-string (type %s)", obj->cls->tp_name);
     }
     return static_cast<BoxedString*>(obj);
 }
@@ -1874,8 +1879,13 @@ extern "C" BoxedString* repr(Box* obj) {
 
     obj = callattrInternal(obj, &repr_str, CLASS_ONLY, NULL, ArgPassSpec(0), NULL, NULL, NULL, NULL, NULL);
 
-    if (obj->cls != str_cls) {
-        raiseExcHelper(TypeError, "__repr__ did not return a string!");
+    if (isSubclass(obj->cls, unicode_cls)) {
+        obj = PyUnicode_AsASCIIString(obj);
+        checkAndThrowCAPIException();
+    }
+
+    if (!isSubclass(obj->cls, str_cls)) {
+        raiseExcHelper(TypeError, "__repr__ returned non-string (type %s)", obj->cls->tp_name);
     }
     return static_cast<BoxedString*>(obj);
 }

src/runtime/str.cpp

@@ -1214,7 +1214,10 @@ extern "C" Box* strLen(BoxedString* self) {
 extern "C" Box* strStr(BoxedString* self) {
     assert(isSubclass(self->cls, str_cls));
 
-    return self;
+    if (self->cls == str_cls)
+        return self;
+
+    return new BoxedString(self->s);
 }
 
 static bool _needs_escaping[256]
@@ -1488,7 +1491,7 @@ extern "C" Box* strNew(BoxedClass* cls, Box* obj) {
     assert(isSubclass(cls, str_cls));
 
     Box* rtn = str(obj);
-    assert(rtn->cls == str_cls);
+    assert(isSubclass(rtn->cls, str_cls));
 
     if (cls == str_cls)
         return rtn;

test/tests/unicode_test.py

@@ -99,3 +99,41 @@ print "hello world".rsplit(u'l')
 
 with open(u"/dev/null", u"r") as f:
     print f.read()
+
+class CustomRepr(object):
+    def __init__(self, x):
+        self.x = x
+
+    def __str__(self):
+        return self.x
+
+    def __repr__(self):
+        return self.x
+
+print repr(str(CustomRepr(u'')))
+print repr(repr(CustomRepr(u'')))
+
+try:
+    str(CustomRepr(u'\u0180'))
+except Exception as e:
+    print type(e), e
+try:
+    repr(CustomRepr(u'\u0180'))
+except Exception as e:
+    print type(e), e
+try:
+    str(CustomRepr(1))
+except Exception as e:
+    print type(e), e
+try:
+    repr(CustomRepr(1))
+except Exception as e:
+    print type(e), e
+
+class MyStr(str):
+    pass
+
+print type(str(CustomRepr(MyStr("hi"))))
+
+print type(MyStr("hi").__str__())
+print type(str(MyStr("hi")))