Script fails with exception when encountering UTF-8 character

[GoogleCodeExporter]

What steps will reproduce the problem?
1. run ./gitinspector.py script against repo with UTF-8 character in author's 
name


What is the expected output? What do you see instead?

raceback (most recent call last):
  File "./gitinspector.py", line 136, in <module>
    __run__.output()
  File "./gitinspector.py", line 57, in output
    outputable.output(changes.ChangesOutput(self.hard))
  File "/Users/tajima/Downloads/gitinspector/outputable.py", line 37, in output
    outputable.output_text()
  File "/Users/tajima/Downloads/gitinspector/changes.py", line 240, in output_text
    print(i.ljust(20)[0:20], end=" ")
UnicodeEncodeError: 'ascii' codec can't encode character u'\xfc' in position 8: 
ordinal not in range(128)

The person's name has a ü

What version of the product are you using? On what operating system?

Mac OS Snow Lion
 ./gitinspector.py  --version
gitinspector 0.2.2


Please provide any additional information below.

Original issue reported on code.google.com by [email protected] on 12 Jul 2013 at 10:26

[GoogleCodeExporter]

Thanks for the report. It "should" work just fine (TM) ;). It looks like it 
might be a terminal issue; the string itself is in UTF-8. Maybe something is 
causing it to fall back to ascii upon print?

Could you try running the following little script from your terminal?:

----

import locale
import sys

print locale.getpreferredencoding()
print sys.getdefaultencoding()
print sys.stdout.encoding
print sys.stdin.encoding

----

Tell me what output you get. My output is:

UTF-8
ascii
UTF-8
UTF-8

/Adam Waldenberg

Original comment by [email protected] on 12 Jul 2013 at 11:47

• Added labels: OpSys-OSX

[GoogleCodeExporter]

Hi,
I got:

$ python
Python 2.7.2 (default, Oct 11 2012, 20:14:37)
[GCC 4.2.1 Compatible Apple Clang 4.0 (tags/Apple/clang-418.0.60)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import locale
>>> import sys
>>>
>>> print locale.getpreferredencoding()
US-ASCII
>>> print sys.getdefaultencoding()
ascii
>>> print sys.stdout.encoding
US-ASCII
>>> print sys.stdin.encoding
US-ASCII

Original comment by [email protected] on 12 Jul 2013 at 11:53

[GoogleCodeExporter]

So, looking into a bit further, on my system I have:

$ locale
LANG="en_CA.US-ASCII"
LC_COLLATE="en_CA.US-ASCII"
LC_CTYPE="en_CA.US-ASCII"
LC_MESSAGES="en_CA.US-ASCII"
LC_MONETARY="en_CA.US-ASCII"
LC_NUMERIC="en_CA.US-ASCII"
LC_TIME="en_CA.US-ASCII"
LC_ALL=

So, then I did a bit of looking up and I'm supposed to add the following to my 
.bash_login:

export LC_ALL=en_US.UTF-8
export LANG=en_US.UTF-8

When I do that, running your little script I get this now:

$ python
Python 2.7.2 (default, Oct 11 2012, 20:14:37)
[GCC 4.2.1 Compatible Apple Clang 4.0 (tags/Apple/clang-418.0.60)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import locale
>>> import sys
>>>
>>> print locale.getpreferredencoding()
UTF-8
>>> print sys.getdefaultencoding()
ascii
>>> print sys.stdout.encoding
UTF-8
>>> print sys.stdin.encoding
UTF-8
>>>


Now it's working fine.

Thanks for your help and super fast response!! :)

Original comment by [email protected] on 12 Jul 2013 at 12:00

[GoogleCodeExporter]

Aha. So that is the problem then. The easiest fix is to switch terminal 
encoding to UTF-8.

Take a look at this post (seems to be a common issue):
http:https://yzisin.wordpress.com/2012/01/09/how-to-fix-locale-issues-in-mac-os-x-lion
-terminal/

You could also try to just run the following before starting gitinspector; if 
it behaves anything like a normal unix terminal it should also work:

LANG=en_US.UTF-8

/Adam Waldenberg

Original comment by [email protected] on 12 Jul 2013 at 12:13

[GoogleCodeExporter]

I see that you beat me to it :). Great that it is working.

/Adam Waldenberg

Original comment by [email protected] on 12 Jul 2013 at 12:14

[GoogleCodeExporter]

Marking as Semi-Invalid as it's not really an issue in gitinspector. However, 
some kind of fix or improvement could be added to help alleviate the problems 
caused by a non-unicode terminal encoding.

/Adam Waldenberg 

Original comment by [email protected] on 12 Jul 2013 at 12:26

• Changed state: Semi-Invalid

[GoogleCodeExporter]

Issue 54 has been merged into this issue.

Original comment by [email protected] on 27 Dec 2014 at 6:23

[dgruss]

Doesn't seem to work for me. Locales were already configured properly I think.

>>> import locale
>>> import sys
>>> 
>>> print locale.getpreferredencoding()
UTF-8
>>> print sys.getdefaultencoding()
ascii
>>> print sys.stdout.encoding
UTF-8
>>> print sys.stdin.encoding
UTF-8
>>> 

Still:
UnicodeEncodeError: 'ascii' codec can't encode character u'\u201c' in position 0: ordinal not in range(128)

[adam-waldenberg]

@dgruss It's not - because its trying to encode a UTF-8 character into ascii - which it can't do. This is not a locale issue, but rather a terminal configuration issue.

Your problem is;

>>> print sys.getdefaultencoding()
ascii

So it's doing exactly what it should. Either change the terminal encoding to whatever the repo uses (UTF-8 in this case), or use the environment variable PYTHONIOENCODING to force it into UTF-8 regardless of what the terminal says.

You can read more about it here;
https://docs.python.org/3/using/cmdline.html#envvar-PYTHONIOENCODING

Redirecting to a file should also do the trick, because that defaults to UTF-8 regardless.

[dgruss]

Sure, but if you run this on many thousand student repos you'll get various encodings...?

[adam-waldenberg]

No. The encoding for the terminal where you run gitinspector will always be the same. It doesn't matter what the source encoding is. Essentially, your problem here is that Python is trying to convert and show a character that is not available in the ascii charset. A UTF-8 destination, on the other hand, will support most characters and the conversion will work.

We can't display any data in the terminal if it's inherently impossible to do so. If the terminal doesnt support a certain character - it just doesn't. Python has ignoreor replace parameters that you can use when doing encoding. However, doing so would cause non-deterministic behavior where running on different terminals could create different results - something that's not desirable.

[dgruss]

Ok, then I'd add one more solution to the list here as PYTHONIOENCODING didn't change anything on my server:

Add to /usr/lib/python2.7/sitecustomize.py the code:

import sys
sys.setdefaultencoding('UTF-8')

Works then.