Back to All Posts

Building a Shell Function to Copy the Latest Commit SHA

Frequently enough, I find myself needing to copy the SHA of the latest commit in project, and doing it manually was becoming a chore. I’d been getting more comfortable and excited about improving efficiency by rolling custom shell functions (like I wrote about here), and so this felt like a good candidate for another one. After a bit of searching, this was confirmed – very little was out there on someone else building a simple terminal command to do such a thing. So, I went for it.

Problem: Copy the SHA of my last commit is clunky and takes time.

Solution: Make a shell function that’ll allow me to do it with a simple terminal command.

Building the Function

Most of the experimentation I did for this was done in a regular shell script on my machine. But to quickly test the script as a terminal command, I copied my iterations inside my ~/.zshrc and sourced it with source ~/.zshrc. Over time, this is how those iterations progressed.

Iteration #1 :: Stupid Simple

Setting up the basics of the function was pretty straightforward:

# `clc` stands for `copy last commit`.function clc {    LAST_COMMIT_SHA=$(git rev-parse HEAD)    echo "$LAST_COMMIT_SHA" | tr -d '\n' | pbcopy    echo "Copied ${LAST_COMMIT_SHA}."}

At this point, I can run clc in my terminal, and by doing so, grab the last commit SHA, remove the line breaks, copy it to the clipboard, and spit out a nice confirmation message. Gets the job done! But.

Iteration #2 :: Copy From Different Branch

When I’m cherry-picking, I often want the latest commit SHA from different branch – not my current one. So, I upgraded the function to accept a branch name as a parameter. If the parameter is set, check out that branch, grab the latest commit SHA, and return to the original branch.

function clc {+    # The original script wrapped in a nested function:+    function copy_last_commit() {        LAST_COMMIT_SHA=$(git rev-parse HEAD)        echo "$LAST_COMMIT_SHA" | tr -d '\n' | pbcopy        echo "Copied ${LAST_COMMIT_SHA}."+    }++    # Added to check out branch, if parameter is set.+    if [ ! -z "$1" ]; then+        if git checkout $1 >/dev/null; then+            copy_last_commit+            git checkout - >/dev/null+        else+            echo "Checkout wasn't successful. Didn't copy anything."+        fi+    else+        copy_last_commit+    fi}

Now, we can run something like clc some-branch. That’ll cause the branch to be checked out (git checkout $1 >/dev/null), and if that’s successful, the copying will commence, spitting out something like this:

Switched to branch 'my-branch'
Copied 3ccbd742f916659c50cbff6c2f63e2ba28a168b5.
Switched to branch 'master'

If, by chance, I pass a non-existent branch name and it fails, an error message is output. And, of course, if no branch parameter is passed, the just check go straight to copying the commit SHA of the current branch. But!

Iteration #3 :: Respect Unstaged Changes

Often times, my active branch has a bunch of unstaged changes, and running this command to grab a SHA from a different branch produces this message:

Please commit your changes or stash them before you switch branches.
Aborting
Checkout wasn't successful. Didn't copy anything.

So, I went for upgrading this again – this time, instead of completely checking out the branch, using git stash to stash away our changes and restore them. It’s lot more flexible, not requiring that you have a clean working directory.

In doing this, it’s important that we only stash when it’s actually needed. Running git stash when there’s nothing to stash does nothing – no new stash is created. And if we automatically run git stash pop when a new stash wasn’t actually created, we might end up restoring a previous stash we don’t want (you can view all the stashes in your repo using git stash list).

I performed this check using git status -s and saving it to the variable IS_DIRTY. If the working tree is “dirty,” stash the changes, and restore them when we’re all done.

function clc {    # The original script wrapped in a nested function:    function copy_last_commit() {        LAST_COMMIT_SHA=$(git rev-parse HEAD)        echo "$LAST_COMMIT_SHA" | tr -d '\n' | pbcopy        echo "Copied ${LAST_COMMIT_SHA}."    }    # Added to check out branch, if parameter is set.    if [ ! -z "$1" ]; then+        IS_DIRTY=$(git status -s)++        if [[ ! -z $IS_DIRTY ]]; then+            git stash push -u >/dev/null+            echo "Stashed unstaged stages."+        fi        if git checkout $1 >/dev/null; then            copy_last_commit            git checkout - >/dev/null        else            echo "Checkout wasn't successful. Didn't copy anything."        fi+        if [[ ! -z $IS_DIRTY ]]; then+            git stash pop >/dev/null+            echo "Restored unstaged changes."+        fi    else        copy_last_commit    fi}

Because it tripped me up, take extra notice that I’m not using a simple git stash command to stash my changes. Instead I’m using git stash push -u. This is because I want to stash away all my current changes, including files I might have just created but not yet committed. The more verbose git stash push followed by the -u (which stands for --include-untracked) flag will do just that.

Iteration #4 :: Tear Down & Rebuild Everything 🤦

At this point, I was feeling pretty good about myself. I had worked through all the weird shell issues I had hit along the way, and actually published this very blog post on the whole process.

And then, a couple of Reddit users (thanks, nunull and austin-schaffer!) pointed out that I don’t actually need to perform a checkout just to get at a commit SHA. This should have been obvious since was already using git rev-parse HEAD to pull the SHA. Swapping out HEAD for whatever branch I need would have done the trick, completely removing the need for any of that complicated checkout and stash logic 🤦.

With that revelation, the function goes from all of those lines of code down to just a few:

function clc {    [[ -z $1 ]] && BRANCH=$(git rev-parse --abbrev-ref HEAD) || BRANCH=$1    LAST_COMMIT_SHA=$(git rev-parse $BRANCH | tail -n 1)    echo "$LAST_COMMIT_SHA" | tr -d '\n' | pbcopy    echo "Copied ${LAST_COMMIT_SHA}."}

See, Reddit ain’t so bad! Thanks again to the two who called this out!

See the Touched Up Final Product

I’ve added a couple of pretty terminal colors in the version that lives in this Gist.

Using the Function

To use this in your local shell, you could throw it in your ~/.bashrc or ~/.zshrc file, but it’s probably better to store it somewhere else on your system. I can’t speak for bash users, but if you’re using zsh, that’s just a matter of putting the file in your $ZSH/custom directory and sourcing it.

Install as Custom ZSH Function

To make it super easy, run the following command, which will retrieve the function from my GitHub Gist and put it into the appropriate location:

curl https://gist.githubusercontent.com/alexmacarthur/933a50c3e072baaf7b6ed18b94e0e873/raw/59f22ae740d83f39a88b70f4aebb0c27b2f9805f/copy-last-commit.zsh -o $ZSH/custom/clc.zsh

Running the Command

After doing that, source it up with source ~/.zshrc, and you should be able to run the command.

Running clc will return something like this:

Copied 3ccbd742f916659c50cbff6c2f63e2ba28a168b5 from master.

Running clc new-branch will return something like this:

Copied 3ccbd742f916659c50cbff6c2f63e2ba28a168b5 from new-branch.

Did I Miss Something?

It wouldn’t surprise me. If you have any suggestions or improvements, let me know!


Alex MacArthur is a software engineer working for Dave Ramsey in Nashville-ish, TN.
Soli Deo gloria.

Get irregular emails about new posts or projects.

No spam. Unsubscribe whenever.
Leave a Free Comment

1 comment
  • Austin Schaffer

    Hey, mentioned this on the reddit thread, but for everyone reading this from somewhere other than reddit, I'd recommend using LAST_COMMIT_SHA=$(git log -1 --format="%H" some_branch), instead of checkout and rev-parse. Git checkout can take a lot of time in some situations, and git log allows you to target any branch without doing checkout, even remote branches. It also simplifies the logic of this command a lot.


    4 replies
    • Austin Schaffer

      No reason, I tried rev-parse but didn't have immediate success. I didn't know it existed and honestly I don't know what it does. I am familiar with git log though, so I opted to go with immediate feedback before any novices start recklessly switching branches back and forth while at work.


    • Alex MacArthur

      Thanks for that feedback, @austin_schaffer! Already made an update moving away from the entire checkout approach. I did notice git rev-parse BRANCH-NAME appears to work too -- even against with remote branches. Any reason why you'd prefer the more verbose git log -1 --format="%H" over that?


    • Alex MacArthur

      Good call. There are few things in development more dangerous than a junior who... checks out a branch.