Convert unicode objects to utf-8 encoded G_STRINGs

gst/pygstvalue.c

@@ -76,15 +76,13 @@ pygst_value_as_pyobject(const GValue *value, gboolean copy_boxed)
           (gstintrange_class,
           Py_BuildValue ("ii",
               gst_value_get_int_range_min (value),
-                        gst_value_get_int_range_max (value)),
+              gst_value_get_int_range_max (value)), NULL);
-         NULL);
     } else if (GST_VALUE_HOLDS_DOUBLE_RANGE (value)) {
       ret = PyObject_Call
           (gstdoublerange_class,
           Py_BuildValue ("dd",
               gst_value_get_double_range_min (value),
-                        gst_value_get_double_range_max (value)),
+              gst_value_get_double_range_max (value)), NULL);
-         NULL);
     } else if (GST_VALUE_HOLDS_LIST (value)) {
       int i, len;
       len = gst_value_list_get_size (value);
@@ -108,8 +106,7 @@ pygst_value_as_pyobject(const GValue *value, gboolean copy_boxed)
           (gstfraction_class,
           Py_BuildValue ("ii",
               gst_value_get_fraction_numerator (value),
-                        gst_value_get_fraction_denominator (value)),
+              gst_value_get_fraction_denominator (value)), NULL);
-         NULL);
     } else if (GST_VALUE_HOLDS_FRACTION_RANGE (value)) {
       const GValue *min, *max;
       min = gst_value_get_fraction_range_min (value);
@@ -118,16 +115,27 @@ pygst_value_as_pyobject(const GValue *value, gboolean copy_boxed)
           (gstfractionrange_class,
           Py_BuildValue ("OO",
               pygst_value_as_pyobject (min, copy_boxed),
-			pygst_value_as_pyobject (max, copy_boxed)),
+              pygst_value_as_pyobject (max, copy_boxed)), NULL);
-	 NULL);
     } else if (GST_VALUE_HOLDS_BUFFER (value)) {
       return pygstminiobject_new (gst_value_get_mini_object (value));
     } else {
       gchar buf[256];
-      g_snprintf (buf, 256, "unknown type: %s", g_type_name (G_VALUE_TYPE (value)));
+      g_snprintf (buf, 256, "unknown type: %s",
+          g_type_name (G_VALUE_TYPE (value)));
       PyErr_SetString (PyExc_TypeError, buf);
     }
   }
+
+  if (G_VALUE_TYPE (value) == G_TYPE_STRING) {
+    PyObject *u = NULL;
+
+    /* FIXME: docs are not clear on whether this sets a python
+       exception when it fails */
+    u = PyUnicode_FromEncodedObject (ret, "utf-8", NULL);
+    Py_DECREF (ret);
+    ret = u;
+  }
+
   return ret;
 }
 
@@ -172,6 +180,10 @@ pygst_value_init_for_pyobject (GValue *value, PyObject *obj)
     } else if (PyList_Check (obj)) {
       PyErr_Clear ();
       t = GST_TYPE_LIST;
+    } else if (PyUnicode_Check (obj)) {
+      /* unicode strings should be converted to UTF-8 */
+      PyErr_Clear ();
+      t = G_TYPE_STRING;
     } else {
       /* pyg_type_from_object already set the error */
       return FALSE;
@@ -198,6 +210,15 @@ pygst_value_from_pyobject (GValue *value, PyObject *obj)
 {
   GType f = g_type_fundamental (G_VALUE_TYPE (value));
 
+  /* Unicode objects should be converted to utf-8 strings */
+  if (PyObject_TypeCheck (obj, &PyUnicode_Type)) {
+    PyObject *v;
+
+    v = PyUnicode_AsUTF8String (obj);
+    Py_DECREF (obj);
+    obj = v;
+  }
+
   /* work around a bug in pygtk whereby pyg_value_from_pyobject claims success
      for unknown fundamental types without actually doing anything */
   if (f < G_TYPE_MAKE_FUNDAMENTAL (G_TYPE_RESERVED_USER_FIRST)
@@ -357,11 +378,13 @@ pygst_value_init(void)
   NULL_CHECK (gstfourcc_class);
   gstintrange_class = (PyObject *) PyDict_GetItemString (dict, "IntRange");
   NULL_CHECK (gstintrange_class);
-  gstdoublerange_class = (PyObject*)PyDict_GetItemString (dict, "DoubleRange");
+  gstdoublerange_class =
+      (PyObject *) PyDict_GetItemString (dict, "DoubleRange");
   NULL_CHECK (gstdoublerange_class);
   gstfraction_class = (PyObject *) PyDict_GetItemString (dict, "Fraction");
   NULL_CHECK (gstfraction_class);
-  gstfractionrange_class = (PyObject*)PyDict_GetItemString (dict, "FractionRange");
+  gstfractionrange_class =
+      (PyObject *) PyDict_GetItemString (dict, "FractionRange");
   NULL_CHECK (gstfractionrange_class);
   return TRUE;
 

testsuite/test_struct.py

@@ -44,11 +44,11 @@ class StructureTest(TestCase):
 
     def testString(self):
         assert self.struct.has_key('foo')
-        assert isinstance(self.struct['foo'], str)
+        assert isinstance(self.struct['foo'], unicode)
         assert self.struct['foo'] == 'bar', self.struct['foo']
         self.struct['foo'] = 'baz'
         assert self.struct.has_key('foo')
-        assert isinstance(self.struct['foo'], str)
+        assert isinstance(self.struct['foo'], unicode)
         assert self.struct['foo'] == 'baz', self.struct['foo']
 
     def testBoolean(self):

testsuite/test_taglist.py

@@ -43,3 +43,25 @@ class TestTagList(TestCase):
         keys = taglist.keys()
         keys.sort()
         self.assertEqual(keys, ['key1', 'key2'])
+
+    def testUnicode(self):
+        taglist = gst.TagList()
+
+        # normal ASCII text
+        taglist[gst.TAG_ARTIST] = 'Artist'
+        self.failUnless(isinstance(taglist[gst.TAG_ARTIST], unicode))
+        self.assertEquals(taglist[gst.TAG_ARTIST], u'Artist')
+        self.assertEquals(taglist[gst.TAG_ARTIST], 'Artist')
+
+        # normal ASCII text as unicode
+        taglist[gst.TAG_ARTIST] = u'Artist'
+        self.failUnless(isinstance(taglist[gst.TAG_ARTIST], unicode))
+        self.assertEquals(taglist[gst.TAG_ARTIST], u'Artist')
+        self.assertEquals(taglist[gst.TAG_ARTIST], 'Artist')
+        
+        # real unicode
+        taglist[gst.TAG_ARTIST] = u'S\xc3\xadgur R\xc3\xb3s'
+        self.failUnless(isinstance(taglist[gst.TAG_ARTIST], unicode))
+        self.assertEquals(taglist[gst.TAG_ARTIST], u'S\xc3\xadgur R\xc3\xb3s')
+        
+