Bug 223686 – git-webkit find doesn't work well with unicode characters in author name

• Status: RESOLVED
• Resolution: FIXED
• Component: Tools / Tests

[Aakash Jain]

git-webkit find doesn't work well with unicode characters in author name

https://build.webkit.org/#/builders/28/builds/566
python Tools/Scripts/git-webkit find r274645

Title: Nullptr crash in RenderStyle::shapeOutside()

Traceback (most recent call last):
  File "Tools/Scripts/git-webkit", line 49, in <module>
    subversion='https://svn.webkit.org/repository/webkit',
  File "/Volumes/Data/worker/ios-simulator-14-release-tests-wk2/build/Tools/Scripts/libraries/webkitscmpy/webkitscmpy/program/__init__.py", line 84, in main
    subversion=subversion,
  File "/Volumes/Data/worker/ios-simulator-14-release-tests-wk2/build/Tools/Scripts/libraries/webkitscmpy/webkitscmpy/program/find.py", line 118, in main
    return Info.main(args, repository=repository, reference=args.argument[0], **kwargs)
  File "/Volumes/Data/worker/ios-simulator-14-release-tests-wk2/build/Tools/Scripts/libraries/webkitscmpy/webkitscmpy/program/find.py", line 87, in main
    print(u'Author: {}'.format(commit.author))
UnicodeEncodeError: 'ascii' codec can't encode character u'\xe9' in position 10: ordinal not in range(128)

[Aakash Jain]

Author name which git-webkit couldn't handle properly was: Frédéric Wang

[Aakash Jain]

Steps to reproduce (on python 2):

>>> u'Author: {}'.format('Fréd')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 2: ordinal not in range(128)

[Jonathan Bedard]

I can't reproduce this problem on trunk:

python Tools/Scripts/git-webkit find r274645
Title: Nullptr crash in RenderStyle::shapeOutside()
Author: Frédéric Wang <[e-mail]>
Date: Thu Mar 18 07:44:01 2021
Revision: 274645
Identifier: 235470@trunk

The issue is that there aren't any commits in the last week that could have fixed this problem

[Radar WebKit Bug Importer]

<rdar://problem/76047116>

[Aakash Jain]

This issue still happening, probably for every commit by authors containing non-ascii characters in their name. e.g.:

https://ews-build.webkit.org/#/builders/8/builds/49976
https://ews-build.webkit.org/#/builders/9/builds/44389
https://ews-build.webkit.org/#/builders/36/builds/32660
https://ews-build.webkit.org/#/builders/47/builds/13126
https://ews-build.webkit.org/#/builders/48/builds/13220
https://ews-build.webkit.org/#/builders/49/builds/12750
https://ews-build.webkit.org/#/builders/50/builds/12986
https://ews-build.webkit.org/#/builders/51/builds/10740
https://ews-build.webkit.org/#/builders/52/builds/12394
https://ews-build.webkit.org/#/builders/53/builds/12048
https://ews-build.webkit.org/#/builders/55/builds/6174
https://ews-build.webkit.org/#/builders/58/builds/6215
https://ews-build.webkit.org/#/builders/59/builds/5290
https://ews-build.webkit.org/#/builders/61/builds/1718
https://ews-build.webkit.org/#/builders/61/builds/1717

UnicodeEncodeError: 'ascii' codec can't encode character u'\xf1' in position 19: ordinal not in range(128)

These builds were for r275596 by 'Enrique Ocaña González'

[Aakash Jain]

Created attachment 425391 [details]
Patch

[Jonathan Bedard]

Confused by my inability to reproduce, but a ? Seems better than blowing up.

[Alexey Proskuryakov]

Is there a plan for how to fix it for good, so that names are visible?

[Aakash Jain]

(In reply to Jonathan Bedard from comment #7)
> Confused by my inability to reproduce, but a ? Seems better than blowing up.
I am also unable to reproduce this. This works fine with the problematic commit on my local machine. I even tried on an ews machine which indicated this failure, and I wasn't able to reproduce.

[Aakash Jain]

(In reply to Alexey Proskuryakov from comment #8)
> Is there a plan for how to fix it for good, so that names are visible?
This issue doesn't happen on Python 3. Also, we are not able to reproduce, so seems like some corner case which fails only in buildbot (and in buildbot we anyways hide this build-step, and only show the commit-identifier). 

So, people aren't likely to hit this problem on their machines. Automation doesn't show the author name anyways. We can prioritize the real fix if we notice this problem anywhere else, where it has any visible impact.

[EWS]

Committed r275612 (236257@main): <https://commits.webkit.org/236257@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 425391 [details].

[Alexey Proskuryakov]

If no one is going to see this, it's better to say something like "cannot print the name, please file a bug if seeing this locally", not just "?".

[Aakash Jain]

Reopening to attach another patch.

[Aakash Jain]

Created attachment 425420 [details]
[fast-cq] Follow-up Patch

[EWS]

Committed r275778 (236351@main): <https://commits.webkit.org/236351@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 425420 [details].