summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorSam Thursfield <sam.thursfield@codethink.co.uk>2015-10-16 21:49:02 (GMT)
committerBaserock Gerrit <gerrit@baserock.org>2016-01-26 09:47:47 (GMT)
commitb3ecd02236e58386ac4d7566ef70e751ff0d7e26 (patch)
treef2d8ac09db161d7e6c91a6db7c4a49afa6619353
parentfa431c3ce38204f5237337fd2045ad6269413258 (diff)
downloadmorph-b3ecd02236e58386ac4d7566ef70e751ff0d7e26.tar.gz
distbuild: Avoid UnicodeEncodeError when writing build output to log files
Text encoding in Python 2 is a total mess so I can only pretend to understand what's going on. The 'stdout' and 'stderr' messages are Python 'unicode' instances, which isn't really important, but I know that because when we try to write them the log file, and they contain non-ASCII data, we see this error: File "/usr/lib/python2.7/site-packages/distbuild/initiator.py", line 231, in _handle_step_output_message f.write(msg['stdout']) UnicodeEncodeError: 'ascii' codec can't encode character u'\u2022' in position 29: ordinal not in range(128) Who said anything about encoding 'unicode' to 'ascii'? It turns out that when you write to a file, Python implicitly tries to encode the data to the 'default encoding', which happens to be 'ascii'. You lose! The only way to fix this is to tell Python that the file has a different encoding, a nice one like UTF-8. (I tried opening the file with 'b' mode, that doesn't seem to help). UTF-8 can only encode valid Unicode data, of course, so we need to make sure the data we write is valid UTF-8. You can to this by calling decode('unicode-escape'), which converts *from* Unicode *to* Unicode, but replacing any invalid characters with escape codes so that we don't get any UnicodeDecodeErrors during the conversion, or when we try to write it to the UTF-8 file. See this presentation for more info: http://farmdev.com/talks/unicode/ Or just use Python 3. Change-Id: I6316d346f5cca2c75f198b48ec9878ac647ae7e5
-rw-r--r--distbuild/initiator.py6
1 files changed, 4 insertions, 2 deletions
diff --git a/distbuild/initiator.py b/distbuild/initiator.py
index f2cb77e..e02f640 100644
--- a/distbuild/initiator.py
+++ b/distbuild/initiator.py
@@ -16,11 +16,13 @@
import cliapp
+
+import codecs
import itertools
import logging
import os
-import uuid
import time
+import uuid
import distbuild
@@ -194,7 +196,7 @@ class Initiator(distbuild.StateMachine):
path = self._get_step_output_dir()
filename = os.path.join(path,
'build-step-%s.log' % msg['step_name'])
- f = open(filename, 'a')
+ f = codecs.open(filename, 'a', encoding='utf-8')
self._step_outputs[msg['step_name']] = f
def _close_output(self, msg):