Merge "Truncate encoded instance message to 255 or fewer"

nova/compute/utils.py

@@ -19,6 +19,7 @@ import string
 import traceback
 
 from oslo.config import cfg
+from oslo.utils import encodeutils
 
 from nova import block_device
 from nova.compute import flavors
@@ -64,7 +65,19 @@ def exception_to_dict(fault):
     # NOTE(dripton) The message field in the database is limited to 255 chars.
     # MySQL silently truncates overly long messages, but PostgreSQL throws an
     # error if we don't truncate it.
-    u_message = unicode(message)[:255]
+    b_message = encodeutils.safe_encode(message)[:255]
+
+    # NOTE(chaochin) UTF-8 character byte size varies from 1 to 6. If
+    # truncating a long byte string to 255, the last character may be
+    # cut in the middle, so that UnicodeDecodeError will occur when
+    # converting it back to unicode.
+    decode_ok = False
+    while not decode_ok:
+        try:
+            u_message = encodeutils.safe_decode(b_message)
+            decode_ok = True
+        except UnicodeDecodeError:
+            b_message = b_message[:-1]
 
     fault_dict = dict(exception=fault)
     fault_dict["message"] = u_message

nova/tests/unit/compute/test_compute_utils.py

@@ -22,6 +22,7 @@ import string
 import mock
 from oslo.config import cfg
 from oslo.serialization import jsonutils
+from oslo.utils import encodeutils
 from oslo.utils import importutils
 import six
 import testtools
@@ -825,3 +826,25 @@ class ComputeUtilsGetRebootTypes(test.NoDBTestCase):
     def test_get_reboot_not_running_hard(self):
         reboot_type = compute_utils.get_reboot_type('foo', 'bar')
         self.assertEqual(reboot_type, 'HARD')
+
+
+class ComputeUtilsTestCase(test.NoDBTestCase):
+    def test_exception_to_dict_with_long_message_3_bytes(self):
+        # Generate Chinese byte string whose length is 300. This Chinese UTF-8
+        # character occupies 3 bytes. After truncating, the byte string length
+        # should be 255.
+        msg = encodeutils.safe_decode('\xe8\xb5\xb5' * 100)
+        exc = exception.NovaException(message=msg)
+        fault_dict = compute_utils.exception_to_dict(exc)
+        byte_message = encodeutils.safe_encode(fault_dict["message"])
+        self.assertEqual(255, len(byte_message))
+
+    def test_exception_to_dict_with_long_message_2_bytes(self):
+        # Generate Russian byte string whose length is 300. This Russian UTF-8
+        # character occupies 2 bytes. After truncating, the byte string length
+        # should be 254.
+        msg = encodeutils.safe_decode('\xd0\x92' * 150)
+        exc = exception.NovaException(message=msg)
+        fault_dict = compute_utils.exception_to_dict(exc)
+        byte_message = encodeutils.safe_encode(fault_dict["message"])
+        self.assertEqual(254, len(byte_message))