Better error message when C extension raises InvalidStringData, PYTHON-533

bson/_cbsonmodule.c

@@ -655,8 +655,24 @@ static int _write_element_to_buffer(PyObject* self, buffer_t buffer, int type_by
         if (status == NOT_UTF_8) {
             PyObject* InvalidStringData = _error("InvalidStringData");
             if (InvalidStringData) {
-                PyErr_SetString(InvalidStringData,
-                                "strings in documents must be valid UTF-8");
+                PyObject* repr = PyObject_Repr(value);
+                char* repr_as_cstr = repr ? PyString_AsString(repr) : NULL;
+                if (repr_as_cstr) {
+                    PyObject *message = PyString_FromFormat(
+                        "strings in documents must be valid UTF-8: %s",
+                        repr_as_cstr);
+
+                    if (message) {
+                        PyErr_SetObject(InvalidStringData, message);
+                        Py_DECREF(message);
+                    }
+                } else {
+                    /* repr(value) failed, use a generic message. */
+                    PyErr_SetString(
+                        InvalidStringData,
+                        "strings in documents must be valid UTF-8");
+                }
+                Py_XDECREF(repr);
                 Py_DECREF(InvalidStringData);
             }
             return 0;

test/test_bson.py

@@ -343,18 +343,26 @@ class TestBSON(unittest.TestCase):
         w = {u"aéあ": u"aéあ"}
         self.assertEqual(w, BSON.encode(w).decode())
 
-        if not PY3:
-            # In python3 this would be stored as BSON binary
-            # subtype 0 and wouldn't raise an exception.
-            y = {"hello": u"aé".encode("iso-8859-1")}
-            self.assertRaises(InvalidStringData, BSON.encode, y)
+        iso8859_bytes = u"aé".encode("iso-8859-1")
+        y = {"hello": iso8859_bytes}
+        if PY3:
+            # Stored as BSON binary subtype 0.
+            out = BSON.encode(y).decode()
+            self.assertTrue(isinstance(out['hello'], bytes))
+            self.assertEqual(out['hello'], iso8859_bytes)
+        else:
+            # Python 2.
+            try:
+                BSON.encode(y)
+            except InvalidStringData, e:
+                self.assertTrue(repr(iso8859_bytes) in str(e))
 
             # The next two tests only make sense in python 2.x since
             # you can't use `bytes` type as document keys in python 3.x.
             x = {u"aéあ".encode("utf-8"): u"aéあ".encode("utf-8")}
             self.assertEqual(w, BSON.encode(x).decode())
 
-            z = {u"aé".encode("iso-8859-1"): "hello"}
+            z = {iso8859_bytes: "hello"}
             self.assertRaises(InvalidStringData, BSON.encode, z)
 
     def test_null_character(self):