Get commit hash for a commit in a branch - bash

Question

In a git branch I store the latest commit hash SHA using
latest_sha=$(git log --pretty=oneline | head -1 | cut -d ' ' -f 1)
After a bunch of commits in this branch, how do I get the next commit SHA after the latest_sha.
Say if there are 5 commits made to this branch after the $latest_sha.
I want to always get the SHA of the first commit after the latest_sha.
b8eead8ba4ff375911af6
c2452680eb7731e4d36ca
da2e113ca4768f5f34730
95b98d42a6e567ed56fc2
716c4f84a855f48bee55c
6a7223a74269f925cfd9e---I need this one
e945bcfabf3fbafc85084---latest_sha
159df375376ded565bec0
d725350982626f46a8b80
56a4b6ca91d93acc8d751
de584608616b1ed99a554
3cfc15339a98bb286d5baa
6ae834bf36c90fbd81854
fa9bdebd0f814f04ee05ba
cc44c4d9ff14314c1255da
5a6145586a8fdcaa2da659
bfea8cfe121d24a0ff1525
Thanks!

Answer 1

A Git branch name, in a sense, is the latest hash ID. That is, if git log branchX shows you commit b8eead8ba4ff375911af6 first, then branchX is a name representing b8eead8ba4ff375911af6, and:
git show branchX
will show the same commit as:
git show b8eead8ba4ff375911af6
If you need the hash ID for some reason—e.g., because you're going to change the hash ID to which the branch name points, by adding new commits—the simplest command to get it is git rev-parse:
hash=$(git rev-parse refs/heads/$branch)
For the rest, see alfunx's answer. Note that if the commits form a diamond-shaped graph, e.g.:
I--J
/ \
...--G--H M--N <-- branchX
\ /
K--L
then there are two commits that immediately follow H, but either I nor K are ancestors of each other, they're only related by both being descendants of H and ancestors (grandparents, in this case) of merge commit M. Using git rev-list --ancestry-path ^<anything-identifying-H> <anything-identifying-branchX> will list commits I, J, K, L, M, and N. The listing will start at N and move back to M as its second entry, but at this point, Git now has a choice of whether to list J or L. This is where the sorting options you choose go into effect. The default sort is chronological by committer date-and-time stamp.
Having listed either J or L, Git can now list the parent of whichever commit it listed, or the remaining commit on the other fork of history. Git will list one of them. If it chose to list J first, then I, it must now list L and then K in that order; if it chose to list L first, then K, it must now show I and then J in that order. But it might also list them in the order J, L, K, I, for instance; or J, L, I, K. Adding --topo-order constrains git rev-list to avoid interleaving commits from the two legs.
The linearization order in complex graphs is generally problematic: there's no single solution that handles all cases. That's why git rev-list offers multiple sorting options.

Answer 2

You could do that with rev-list:
git rev-list --ancestry-path HEAD ^${latest_sha} | tail -n1
rev-list lists all reachable commits in reverse chronological order for the given branches/commits. The caret (^) here means "not", which makes Git exclude all reachable commits starting at the given commit.
Concretely that means: Include all commits reachable from HEAD, exclude all commits reachable from ${latest_sha}, and then take the oldest one there using tail.
Edit: Add --ancestry-path to make sure only commits that are in the direct ancestry path between the specified commits are used (as mentioned by #jthill).

Answer 3

git logs have a parent commit field, so you could do something like
git log --pretty=format:"%P %H" | awk '$1 == "<YOUR_HASH>" {print $2}'