adamtornhill/code-maat

Parsing error

Opened this issue · 2 comments

I am analyzing wikimedia/mediawiki with code-maat.

I seem to be getting parsing errors for the commit messages:

java -Xmx512M -Xms64M -jar \winmaat0.8.5\code-maat-0.8.5-standalone.jar -l ../mediawiki/maat_evo.log -c git -a summary
Invalid argument: input: [ebbf38d2ab] Antoine Musso 2010-10-29 Follow up r75682 : fix private function naming. Fix cruise control build 2010-10-29T22:18:27
3 3 maintenance/tests/phpunit/includes/UserIsValidEmailAddrTest.php
, reason: Parse error at line 1, column 122:
[ebbf38d2ab] Antoine Musso 2010-10-29 Follow up r75682 : fix private function naming. Fix cruise control build 2010-10-29T22:18:27

When I manually fix this comment, I run into the next parsing error:

java -Xmx512M -Xms64M -jar \winmaat0.8.5\code-maat-0.8.5-standalone.jar -l ../mediawiki/maat_evo.log -c git -a summary
Invalid argument: input: [c7723263fc] Niklas Laxstr▒m 2010-08-08 Removed $wgUseZhdaemon, $wgZhdaemonHost and $wgZhdaemonPort. Apparently unused since r6878 at 2004-12-29.

It would be nice if code-maat would be more robust in parsing comments.

@Meffi42 , thanks for reporting this. I suspect that it's due to the dates included in the commit messages.

Code Maat supports a second log format that is more robust: git log --all --numstat --date=short --pretty=format:'--%h--%ad--%aN' --no-renames

Logs on this format have to be parsed using the -c git2 parse option in Code Maat.

Hi,
I'm trying to run the tool as in Readme:
image

But receive this error:
image

Tried various combinations, but got the same error

  • legacy git format for the logs and -c git for run
  • log format from here:
    ;;; git log --all -M -C --numstat --date=short --pretty=format:'--%h--%cd--%cn'
  • different paths to code-maat: <path_to_projects>/code-maat/src/code_maat:/data and <path_to_projects>/code-maat:/data

Not sure what I'm doing wrong, could someone please help?