make sure we fail fast on when we get a string that's not utf-8

pymongo/_cbsonmodule.c

@@ -104,7 +104,14 @@ static PyObject* _cbson_element_to_bson(PyObject* self, PyObject* args) {
      * here we check for type equivalence, not isinstance in some
      * places. */
     if (PyString_CheckExact(value)) {
-        return build_string(0x02, value, name);
+        // we have to do the encoding so we can fail fast if they give us non utf-8
+        PyObject* encoded = PyString_AsEncodedObject(value, "utf-8", "strict");
+        if (!encoded) {
+            return NULL;
+        }
+        PyObject* result = build_string(0x02, encoded, name);
+        Py_DECREF(encoded);
+        return result;
     } else if (PyUnicode_CheckExact(value)) {
         PyObject* encoded = PyUnicode_AsUTF8String(value);
         if (!encoded) {

pymongo/bson.py

@@ -52,10 +52,8 @@ def _get_c_string(data):
 
     return (unicode(data[:end], "utf-8"), data[end + 1:])
 
-def _make_c_string(string, encode=True):
-    if encode and isinstance(string, unicode):
-        return string.encode("utf-8") + "\x00"
-    return string + "\x00"
+def _make_c_string(string):
+    return string.encode("utf-8") + "\x00"
 
 def _validate_number(data):
     assert len(data) >= 8
@@ -298,11 +296,11 @@ def _element_to_bson(key, value):
             value = struct.pack("<i", len(value)) + value
         return "\x05" + name + struct.pack("<i", len(value)) + chr(subtype) + value
     if isinstance(value, Code):
-        cstring = _make_c_string(value, False)
+        cstring = _make_c_string(value)
         length = struct.pack("<i", len(cstring))
         return "\x0D" + name + length + cstring
     if isinstance(value, str):
-        cstring = _make_c_string(value, False)
+        cstring = _make_c_string(value)
         length = struct.pack("<i", len(cstring))
         return "\x02" + name + length + cstring
     if isinstance(value, unicode):
@@ -342,7 +340,7 @@ def _element_to_bson(key, value):
             flags += "u"
         if value.flags & re.VERBOSE:
             flags += "x"
-        return "\x0B" + name + _make_c_string(pattern) + _make_c_string(flags, False)
+        return "\x0B" + name + _make_c_string(pattern) + _make_c_string(flags)
     if isinstance(value, DBRef):
         ns = _make_c_string(value.collection())
         return "\x0C" + name + struct.pack("<i", len(ns)) + ns + _shuffle_oid(str(value.id()))

test/test_bson.py

@@ -156,5 +156,9 @@ class TestBSON(unittest.TestCase):
                     f.write(bson)
                     f.close()
 
+    def test_bad_encode(self):
+        self.assertRaises(UnicodeDecodeError, BSON.from_dict,
+                          {"lalala": '\xf4\xe0\xf0\xe1\xc0 Color Touch'})
+
 if __name__ == "__main__":
     unittest.main()