On Wed, Sep 08, 2021 at 11:40:20PM +0300, Aleksey Midenkov wrote:
> Is that possible/how to print hunk numbers with git diff?
>
> F.ex. instead of:
>
> @@ -106,7 +110,6 @@ while ($r < $statement_count)
> ...
>
> To print something like:
>
> @@ -106,7 +110,6 @@ 4 @@ while ($r < $statement_count)
> ...
>
> filterdiff uses hunk numbers intensively. Work with line-number ranges
> is not so effective.
No, Git doesn't know how to do any annotations on hunk lines (aside from
finding and reporting the funcname lines from the source). So you'd have
to post-process it, like:
git diff ... |
perl -pe 's/^@@.*?@@/join(" ", $&, ++$i, "@@")/e'
but I'm not sure if that's quite what you're after. If you're using
filterdiff to pick out hunks, then piping through "filterdiff
--annotate" does something similar.
If you want to post-process your diffs all the time, you can do
something like:
git config pager.color false
git config pager.diff 'filterdiff --annotate | less'
to show the annotations anytime the output is going to a terminal.
Though sadly filterdiff does not handle the colors; other
post-processors like diff-highlight parse around them.
And finally, if your ultimate goal is to use filterdiff to pick out
hunks, you might find using Git's picking tools like "checkout -p"
easier. Even if you are starting with an actual patch, you can apply it
and then pick out bits, like:
git checkout --detach ;# temporary head for applying patch
git apply </path/to/patch
git commit -m "temporary commit for patch"
git checkout - ;# back to the original branch
git checkout -p HEAD@{1} ;# now selectively grab parts
Of course that only helps if the patch actually applies. If your goal is
to filter out hunks that don't apply, it won't help. :)
-Peff
