A small script to find stale branches

find-old-branches.sh

#!/bin/bash

# This is a very naive script, it doesn't do grouping and returns all branches
# I only really care about branches that have not seen commits in two months
#
# I am hoping to find some time to write a tool that can output these reports for me
# In the meantime, I am using this

echo "Merged branches"
for branch in `git branch -r --merged | grep -v HEAD`;do echo -e `git log --no-merges -n 1 --format="%ci, %cr, %an, %ae, "  $branch | head -n 1` \\t$branch; done | sort -r

echo ""
echo "Not merged branches"
for branch in `git branch -r --no-merged | grep -v HEAD`;do echo -e `git log --no-merges -n 1 --format="%ci, %cr, %an, %ae, " $branch | head -n 1` \\t$branch; done | sort -r

Quickly hacked together a powershell version in case it helps anyone
`#! /usr/bin/pwsh

echo ""
echo "Not merged branches"
$branches = git branch -r --no-merged | grep -v HEAD
$output = @()
foreach( $branch in $branches ) {
$branch = $branch.Trim()
$output += git log -n 1 --format="%ci, %cr, %an, %ae, $branch" --no-merges --first-parent $branch | Sort-Object
}
echo $output`

Could you please suggest me how to list all the branches under enitirr gitlab

It's a good idea to change the column separators because the dates in the output contain ,, so another column separator like | is more appropriate to avoid confusion.
Just change --format="%ci, %cr, %an, %ae, " to --format="%ci | %cr | %an | %ae | ".

A quick one liner to delete all merged branches: for branch in `git branch -r --merged | grep -v HEAD`;do git branch -D $branch; done

Here is my solution for powershell
https://gist.github.com/kondr1/b9252ed216bd327eb1063ad850b7c7f0
It will work faster, because it's using only one git command

Please help me understand this script. How does this script filter out branches that have not seen commits in two months?

Please help me understand this script. How does this script filter out branches that have not seen commits in two months?

From what I can tell, the command doesn't specifically point out branches that haven't had commits in 2 months but it puts the most recent commit for each branch in a readable, one-liner format. The author has split the commands for merged branches and non-merged branches. Also pointing out that they are filtering by non-merged commits so merged commits (ex. merging of branches) are excluded.

ok, someone please tell me how to run this script.
i mean how this script works , it doesn't require any login or anything

ok, someone please tell me how to run this script. i mean how this script works , it doesn't require any login or anything

It's a bash script.
You can save it into a folder where you have previously cloned a git repo.
Then run it: if you haven't prevously saved your git credentials, git will prompt for them.

In order to filter with awk keeping just branches that have not seen commits in two months
it would suffice prepending %ct (committer date, UNIX timestamp) to the git log fields
then filter and strip the added prefix

Since I want to produce a CSV I did it as follows (for unmerged branches)

for branch in `git branch -r --no-merged | grep -v HEAD`;do
  echo -e $(git log --no-merges -n 1 \
    --format="%ct,%cI,%an,%ae," $branch |                    `# prepend secs since EPOCH as 1st field` \
    head -n 1)$branch; \
done | \
  sort -r | \
  awk -F , '$1 < limit' limit=$(date -d "2 month ago" +%s) | `# split by comma and filter based on 1st field` \
  sed 's/^[0-9]*\,//g'                                       `# strip 1st field now`

It can clearly be optimized based on needs, moving sort after filter and so on

PS: please note here I added comments within ` (backticks) for readability, so in turn I had to replace subshell backticks with the $() syntax.
Also note that this approach leverages some additional comands, namely awk, date and sed

@trntsmn - The term 'grep' is not recognized as a name of a cmdlet, function, script file, or executable program.
=D