This seems to be working but also is quite horrendously slow and seems a little hacky to me.
IFS=$'\n'
for currentFile in $(git status | grep "modified:" | cut -c 14-)
do
gitDiff=$(git diff "$currentFile")
gitDiffStr="[$gitDiff]"
if [ "$gitDiffStr" == "[]" ]
then
echo match
echo git checkout "$currentFile"
git checkout "$currentFile"
fi
done
1 Answer 1
The slowness
The slowness most probably comes from executing multiple git diff
commands.
A faster way might be using a programming language with a Git library that would let you run a single git diff
command,
and iterate over the entries of the diff to find the empty ones.
The hacky part
This is hacky:
git status | grep "modified:" | cut -c 14-
The output of the git status
is not API,
and may also be subject to aliases and user settings.
It's not safe to use.
When parsing the output of Git commands,
look for a --porcelain
option to make it safe.
Another hacky part is changing the value of IFS
.
It's not a huge problem,
but it's good to minimize the scope of the change,
typically by using it in the form of IFS=... cmd
,
which would limit the effect of the change to the execution of the cmd
command.
Lastly, some minor style issues:
- The
$gitDiffStr
variable is pointless, you can simply inline it. - The
==
operator within a[ ... ]
is undocumented. Use=
instead.
Alternative implementation
Applying the above suggestions, this is less hacky, and probably just as slow:
git status --porcelain -z | grep -z '^ M' | while IFS= read -r -d '' path0; do
path=${path0:3}
gitDiff=$(git diff "$path")
if [ "$gitDiff" = "" ]; then
echo git checkout "$currentFile"
git checkout "$currentFile"
fi
done
git checkout -- .
? \$\endgroup\$git diff
. I'd like to maintain the ones that have genuine changes. This is definitely not explained well enough in my post, though. \$\endgroup\$git status
output; doesgit diff --name-only
list such files? \$\endgroup\$