exec and eval of unicode

src/codegen/irgen/hooks.cpp

@@ -38,6 +38,7 @@
 #include "core/stats.h"
 #include "core/types.h"
 #include "core/util.h"
+#include "runtime/capi.h"
 #include "runtime/objmodel.h"
 #include "runtime/types.h"
 
@@ -371,6 +372,13 @@ Box* eval(Box* boxedCode) {
     if (globals && globals->cls == attrwrapper_cls && unwrapAttrWrapper(globals) == module)
         globals = NULL;
 
+    if (boxedCode->cls == unicode_cls) {
+        boxedCode = PyUnicode_AsUTF8String(boxedCode);
+        if (!boxedCode)
+            throwCAPIException();
+        // cf.cf_flags |= PyCF_SOURCE_IS_UTF8
+    }
+
     // TODO error message if parse fails or if it isn't an expr
     // TODO should have a cleaner interface that can parse the Expression directly
     // TODO this memory leaks
@@ -428,6 +436,13 @@ Box* exec(Box* boxedCode, Box* globals, Box* locals) {
             PyDict_SetItemString(globals, "__builtins__", builtins_module);
     }
 
+    if (boxedCode->cls == unicode_cls) {
+        boxedCode = PyUnicode_AsUTF8String(boxedCode);
+        if (!boxedCode)
+            throwCAPIException();
+        // cf.cf_flags |= PyCF_SOURCE_IS_UTF8
+    }
+
     // TODO same issues as in `eval`
     RELEASE_ASSERT(boxedCode->cls == str_cls, "");
     const char* code = static_cast<BoxedString*>(boxedCode)->s.data();

test/integration/virtualenv_test.py

@@ -16,7 +16,7 @@ print "Running", args
 subprocess.check_call(args)
 
 sh_script = """
-set -e
+set -ex
 python -c 'import __future__'
 python -c 'import sys; print sys.executable'
 . test_env/bin/activate
@@ -24,6 +24,6 @@ pip install bcrypt==1.1.0
 python -c 'import bcrypt; assert bcrypt.__version__ == "1.1.0"; assert bcrypt.hashpw("password1", "$2a$12$0123456789012345678901").endswith("I1hdtg4K"); print "bcrypt seems to work"'
 """.strip()
 
-print sh_script
+# print sh_script
 subprocess.check_call(["sh", "-c", sh_script])
 

test/tests/exec_basic.py

@@ -3,3 +3,7 @@ a = 5
 print a"""
 
 exec ""
+
+# Exec of a unicode encodes as utf8:
+exec u"print repr('\u0180')"
+print repr(eval(u"'\u0180'"))