PYTHON-1682 UTF-8 encode unicode error messages on Python 2

pymongo/errors.py

@@ -14,6 +14,8 @@
 
 """Exceptions raised by PyMongo."""
 
+import sys
+
 from bson.errors import *
 
 try:
@@ -26,6 +28,7 @@ class PyMongoError(Exception):
     """Base class for all PyMongo exceptions."""
     def __init__(self, message='', error_labels=None):
         super(PyMongoError, self).__init__(message)
+        self._message = message
         self._error_labels = set(error_labels or [])
 
     def has_error_label(self, label):
@@ -43,6 +46,11 @@ def _remove_error_label(self, label):
         """Remove the given label from this error."""
         self._error_labels.remove(label)
 
+    def __str__(self):
+        if sys.version_info[0] == 2 and isinstance(self._message, unicode):
+            return self._message.encode('utf-8', errors='replace')
+        return str(self._message)
+
 
 class ProtocolError(PyMongoError):
     """Raised for failures related to the wire protocol."""

test/test_collection.py

@@ -1316,6 +1316,21 @@ def test_write_error_text_handling(self):
                           db.test.insert_many,
                           [{"text": text}])
 
+    def test_write_error_unicode(self):
+        coll = self.db.test
+        self.addCleanup(coll.drop)
+
+        coll.create_index('a', unique=True)
+        coll.insert_one({'a': u'unicode \U0001f40d'})
+        with self.assertRaisesRegex(
+                DuplicateKeyError,
+                'E11000 duplicate key error') as ctx:
+            coll.insert_one({'a': u'unicode \U0001f40d'})
+
+        # Once more for good measure.
+        self.assertIn('E11000 duplicate key error',
+                      str(ctx.exception))
+
     def test_wtimeout(self):
         # Ensure setting wtimeout doesn't disable write concern altogether.
         # See SERVER-12596.