Add modified str to IR generator.

src/codegen/irgen/irgenerator.cpp

@@ -1705,7 +1705,7 @@ private:
             // end code for handling of softspace
 
 
-            llvm::Value* v = emitter.createCall(unw_info, g.funcs.str, converted->getValue());
+            llvm::Value* v = emitter.createCall(unw_info, g.funcs.strOrUnicode, converted->getValue());
             v = emitter.getBuilder()->CreateBitCast(v, g.llvm_value_type_ptr);
             auto s = new ConcreteCompilerVariable(STR, v, true);
             r = dest->callattr(emitter, getOpInfoForNode(node, unw_info), &write_str, flags, ArgPassSpec(1), { s },

src/codegen/runtime_hooks.cpp

@@ -206,6 +206,7 @@ void initGlobalFuncs(GlobalState& g) {
     GET(importStar);
     GET(repr);
     GET(str);
+    GET(strOrUnicode);
     GET(exceptionMatches);
     GET(yield);
     GET(getiterHelper);

src/codegen/runtime_hooks.h

@@ -38,7 +38,7 @@ struct GlobalFuncs {
         *decodeUTF8StringPtr;
     llvm::Value* getattr, *setattr, *delattr, *delitem, *delGlobal, *nonzero, *binop, *compare, *augbinop, *unboxedLen,
         *getitem, *getclsattr, *getGlobal, *setitem, *unaryop, *import, *importFrom, *importStar, *repr, *str,
-        *exceptionMatches, *yield, *getiterHelper, *hasnext;
+        *strOrUnicode, *exceptionMatches, *yield, *getiterHelper, *hasnext;
 
     llvm::Value* unpackIntoArray, *raiseAttributeError, *raiseAttributeErrorStr, *raiseNotIterableError,
         *assertNameDefined, *assertFail;

src/runtime/inline/link_forcer.cpp

@@ -100,6 +100,7 @@ void force() {
     FORCE(assertNameDefined);
     FORCE(assertFail);
 
+    FORCE(strOrUnicode);
     FORCE(printFloat);
     FORCE(listAppendInternal);
     FORCE(getSysStdout);

src/runtime/objmodel.cpp

@@ -2028,6 +2028,14 @@ extern "C" BoxedString* str(Box* obj) {
     return static_cast<BoxedString*>(obj);
 }
 
+extern "C" BoxedString* strOrUnicode(Box* obj) {
+    // Like str, but returns unicode objects unchanged.
+    if (obj->cls == unicode_cls) {
+        return static_cast<BoxedString*>(obj);
+    }
+    return str(obj);
+}
+
 extern "C" BoxedString* repr(Box* obj) {
     static StatCounter slowpath_repr("slowpath_repr");
     slowpath_repr.log();

src/runtime/objmodel.h

@@ -61,6 +61,7 @@ extern "C" BoxedString* str(Box* obj);
 extern "C" BoxedString* repr(Box* obj);
 extern "C" BoxedString* reprOrNull(Box* obj); // similar to repr, but returns NULL on exception
 extern "C" BoxedString* strOrNull(Box* obj);  // similar to str, but returns NULL on exception
+extern "C" BoxedString* strOrUnicode(Box* obj);
 extern "C" bool exceptionMatches(Box* obj, Box* cls);
 extern "C" BoxedInt* hash(Box* obj);
 extern "C" Box* abs_(Box* obj);

test/tests/unicode_test.py

@@ -49,11 +49,9 @@ import gc
 for i in xrange(100):
     print repr(BaseException().__unicode__())
     gc.collect()
-    # TODO / FIXME: For some reason, uncommenting the loop below will cause printing unicode
-    # to fail!!!
     # do some allocations:
-    # for j in xrange(100):
-    #     [None] * j
+    for j in xrange(100):
+        [None] * j
 
 print u'' in ''
 print '' in u''