141

How can I iterate through all the local branches in my repository using bash script. I need to iterate and check is there any difference between the branch and some remote branches. Ex

for branch in $(git branch); do git log --oneline $branch ^remotes/origin/master; done 

I need to do something like given above, but the issue I'm facing is $(git branch) gives me the folders inside the repository folder along with the branches present in the repository.

Is this the correct way to solve this issue? Or is there another way to do it?

Thank you

6
  • Possible duplicate of for loop over all git branches with certain name Commented Aug 24, 2016 at 12:19
  • 4
    @pihentagy This was written before the linked question. Commented Jul 1, 2019 at 0:21
  • Eg: - for b in "$(git branch)"; do git branch -D $b; done Commented Jan 2, 2020 at 7:39
  • Not a duplicate. This question goes for all branches while the other goes for ones with specific name Commented Feb 24, 2021 at 8:28
  • @Abhi YOU GOT ME! lol I didn't even notice the -D in there Commented Jan 12, 2023 at 19:06

18 Answers 18

198

You should not use git branch when writing scripts. Git provides a “plumbing” interface that is explicitly designed for use in scripting (many current and historical implementations of normal Git commands (add, checkout, merge, etc.) use this same interface).

The plumbing command you want is git for-each-ref:

git for-each-ref --shell \ --format='git log --oneline %(refname) ^origin/master' \ refs/heads/ 

Note: You do not need the remotes/ prefix on the remote ref unless you have other refs that cause origin/master to match multiple places in the ref name search path (see “A symbolic ref name. …” in the Specifying Revisions section of git-rev-parse(1)). If you are trying to explictly avoid ambiguity, then go with the full ref name: refs/remotes/origin/master.

You will get output like this:

git log --oneline 'refs/heads/master' ^origin/master git log --oneline 'refs/heads/other' ^origin/master git log --oneline 'refs/heads/pu' ^origin/master 

You can pipe this output into sh.

If you do not like the idea of generating the shell code, you could give up a bit of robustness* and do this:

for branch in $(git for-each-ref --format='%(refname)' refs/heads/); do git log --oneline "$branch" ^origin/master done 

* Ref names should be safe from the shell’s word splitting (see git-check-ref-format(1)). Personally I would stick with the former version (generated shell code); I am more confident that nothing inappropriate can happen with it.

Since you specified bash and it supports arrays, you could maintain safety and still avoid generating the guts of your loop:

branches=() eval "$(git for-each-ref --shell --format='branches+=(%(refname))' refs/heads/)" for branch in "${branches[@]}"; do # … done 

You could do something similar with $@ if you are not using a shell that supports arrays (set -- to initialize and set -- "$@" %(refname) to add elements).

Sign up to request clarification or add additional context in comments.

8 Comments

Seriously. There isn't a simpler way to do this?
But what if I want to use one of the filtering options of git branch, like --merged, would I have to duplicate the logic in git branch? There has to be a better way to do this.
Simplier version: git for-each-ref refs/heads | cut -d/ -f3-
@wid: Or, simply, git for-each-ref refs/heads --format='%(refname)'
@Thayne: this question is old, but the Git folks have finally addressed the problem: for-each-ref now supports all the branch selectors like --merged and git branch and git tag are now actually implemented in terms of git for-each-ref itself, at least for the list-existing cases. (Creating new branches and tags is not, and should not be, part of for-each-ref.)
|
57

This is because git branch marks the current branch with an asterisk, e.g.:

$ git branch * master mybranch $ 

so $(git branch) expands to e.g. * master mybranch, and then the * expands to the list of files in the current directory.

I don't see an obvious option for not printing the asterisk in the first place; but you could chop it off:

$(git branch | cut -c 3-) 

5 Comments

If you surround in double-quotes you can stop bash from expanding the asterisk - though you'll still want to remove it from the output. A more robust way of removing an asterisk from any point would be $(git branch | sed -e s/\\*//g).
nice, i really like your 3- solution.
Slightly simpler sed version: $(git branch | sed 's/^..//')
slightly simpler tr version: $(git branch | tr -d " *")
This should be the marked answer. The currently marked answer is just preachy without answering the question about why it's not working
23

The bash builtin, mapfile, is built for this

all git branches: git branch --all --format='%(refname:short)'

all local git branches: git branch --format='%(refname:short)'

all remote git branches: git branch --remotes --format='%(refname:short)'

iterate through all git branches: mapfile -t -C my_callback -c 1 < <( get_branches )

example:

my_callback () { INDEX=${1} BRANCH=${2} echo "${INDEX} ${BRANCH}" } get_branches () { git branch --all --format='%(refname:short)' } # mapfile -t -C my_callback -c 1 BRANCHES < <( get_branches ) # if you want the branches that were sent to mapfile in a new array as well # echo "${BRANCHES[@]}" mapfile -t -C my_callback -c 1 < <( get_branches ) 

for the OP's specific situation:

#!/usr/bin/env bash _map () { ARRAY=${1?} CALLBACK=${2?} mapfile -t -C "${CALLBACK}" -c 1 <<< "${ARRAY[@]}" } get_history_differences () { REF1=${1?} REF2=${2?} shift shift git log --oneline "${REF1}" ^"${REF2}" "${@}" } has_different_history () { REF1=${1?} REF2=${2?} HIST_DIFF=$( get_history_differences "${REF1}" "${REF2}" ) return $( test -n "${HIST_DIFF}" ) } print_different_branches () { read -r -a ARGS <<< "${@}" LOCAL=${ARGS[-1]?} for REMOTE in "${SOME_REMOTE_BRANCHES[@]}"; do if has_different_history "${LOCAL}" "${REMOTE}"; then # { echo; echo; get_history_differences "${LOCAL}" "${REMOTE}" --color=always; } # show differences echo local branch "${LOCAL}" is different than remote branch "${REMOTE}"; fi done } get_local_branches () { git branch --format='%(refname:short)' } get_different_branches () { _map "$( get_local_branches )" print_different_branches } # read -r -a SOME_REMOTE_BRANCHES <<< "${@}" # use this instead for command line input declare -a SOME_REMOTE_BRANCHES SOME_REMOTE_BRANCHES=( origin/master remotes/origin/another-branch another-remote/another-interesting-branch ) DIFFERENT_BRANCHES=$( get_different_branches ) echo "${DIFFERENT_BRANCHES}" 

source: List all local git branches without an asterisk

Comments

13
for branch in $(git for-each-ref --format='%(refname:short)' refs/heads); do ... done 

This uses git plumbing commands, which are designed for scripting. It's also simple and standard.

Reference: Git's Bash completion

Comments

7

I iterate as it for example :

for BRANCH in `git branch --list|sed 's/\*//g'`; do git checkout $BRANCH git fetch git branch --set-upstream-to=origin/$BRANCH $BRANCH done git checkout master; 

Comments

7

Finally nailed how to make an output for git branch without an asterisk and without magic in git for-each-ref arguments:

$ git branch --format="%(refname:short)" 

Why this is valuable? git branch command has some additional filters like --merged, which is not easy to implement using git for-each-ref (at least as far as I see).

Comments

4

The accepted answer is correct and really should be the approach used, but solving the problem in bash is a great exercise in understanding how shells work. The trick to doing this using bash without performing additional text manipulation, is to ensure the output of git branch never gets expanded as part of a command to be executed by the shell. This prevents the asterisk from ever being expanding in the file name expansion (step 8) of shell expansion (see http://tldp.org/LDP/Bash-Beginners-Guide/html/sect_03_04.html)

Use the bash while construct with a read command to chop the git branch output into lines. The '*' will be read in as a literal character. Use a case statement to match it, paying special attention to the matching patterns.

git branch | while read line ; do case $line in \*\ *) branch=${line#\*\ } ;; # match the current branch *) branch=$line ;; # match all the other branches esac git log --oneline $branch ^remotes/origin/master done 

The asterisks in both the bash case construct and in the parameter substitution need to be escaped with backslashes to prevent the shell interpreting them as pattern matching characters. The spaces are also escaped (to prevent tokenization) because you are literally matching '* '.

Comments

4

Easiest option to remember in my opinion:

git branch | grep "[^* ]+" -Eo

Output:

bamboo develop master 

Grep's -o option (--only-matching) restricts the output to only the matching parts of the input.

Since neither space nor * are valid in Git branch names, this returns the list of branches without the extra characters.

Edit: If you're in 'detached head' state, you'll need to filter out the current entry:

git branch --list | grep -v "HEAD detached" | grep "[^* ]+" -oE

Comments

4

Keep it simple

The simple way of getting branch name in loop using bash script.

#!/bin/bash for branch in $(git for-each-ref --format='%(refname)' refs/heads/); do echo "${branch/'refs/heads/'/''}" done 

Output:

master other 

Comments

3

I would suggest $(git branch|grep -o "[0-9A-Za-z]\+") if your local branches are named by digits, a-z, and/or A-Z letters only

Comments

2

What I ended up doing, applied to your question (& inspired by ccpizza mentioning tr):

git branch | tr -d ' *' | while IFS='' read -r line; do git log --oneline "$line" ^remotes/origin/master; done

(I use while loops a lot. While for particular things you'd definitely want to use a pointed variable name ["branch", for example], most of the time I am only concerned with doing something with each line of input. Using 'line' here instead of 'branch' is a nod to reusability/muscle memory/efficiency.)

Comments

2

Googlian's answer, but without using for

git for-each-ref --format='%(refname:lstrip=-1)' refs/heads/ 

1 Comment

This doesn't work for namespaced branch names, as in branches that have a slash in them. This means that branches created by dependabot, which look something like "dependabot/npm_and_yarn/typescript-3.9.5", will appear instead as "typescript-3.9.5".
1

If you're at this state:

git branch -a * master remotes/origin/HEAD -> origin/master remotes/origin/branch1 remotes/origin/branch2 remotes/origin/branch3 remotes/origin/master 

And you run this code:

git branch -a | grep remotes/origin/* for BRANCH in `git branch -a | grep remotes/origin/*` ; do A="$(cut -d'/' -f3 <<<"$BRANCH")" echo $A done 

You'll get this result:

branch1 branch2 branch3 master 

3 Comments

This looks less complex solution to me +1
Eg: Delete all branches starting with abc: for b in $(git branch | grep abc); do git branch -D $b; done
Also, for i in $(git branch | grep feature/); do if [[ $i == "feature/"* ]]; then echo $i; fi; done. For all branches starting with feature/
1

The correct way to simply iterate over local branch names is to use for-each-ref over refs/heads/. For example:

for branch in $(git for-each-ref --format='%(refname:short)' refs/heads/); do echo branch="${branch}" done 

This works with standard/default values for IFS because spaces are illegal in branch names in git.

Comments

0

Extending on from @finn's answer (thank you!), the following will let you iterate over the branches without creating an intervening shell script. It's robust enough, as long as there's no newlines in the branch name :)

git for-each-ref --format='%(refname)' refs/heads | while read x ; do echo === $x === ; done 

The while loop runs in a subshell, which is usually fine unless you're setting shell variables that you want to access in the current shell. In that case you use process substitution to reverse the pipe:

while read x ; do echo === $x === ; done < <( git for-each-ref --format='%(refname)' refs/heads ) 

Comments

0

List heads (branches) in the local repository

git show-ref --heads 

This will list the heads something like

682e47c01dc8d0f4e4102f183190a48aaf34a3f0 refs/heads/main .... 

so if you're only interested in the name, you can use something like sed to obtain the output you want

git show-ref --heads | sed 's/.*refs\/heads\///' 

Iterate through the branches

With this output you can easily iterate through it, say using a bash loop, xargs, whatever floats your boat

for SHA in $(git show-ref --heads | awk '{ print $1 }'); do echo "magic! $SHA" done 
  • git show-ref --heads get the branches as per above
  • awk '{ print $1 }' obtain the SHA
  • echo "magic! $SHA" <- this is where you would do your magic

Comments

0

Of course in theory one should use a special interface that Git indeed has for use when scripting. But often you want something simpler — handy for a oneliner. Something that doesn't urge you remember stuff like git for-each-ref --format … refs … amen. And it's UNIX anyways finally. Then it goes like that:

  1. There's an utility widely known for its obscure but a terse way to print the last column.
  2. git branch puts asterisk before the branch name. Meaning we're seemingly always interested in the last column exactly.

Resulting:

git branch | awk '{print $NF}' 

This works because awk has an interval variable NF which is the number of the last field. Prefixing this number with $ will yield the contents of that field witch is exactly what is need here.

Comments

-1
#/bin/bash for branch in $(git branch -r); do echo $branch done 

1 Comment

This includes -> and both origin/HEAD and origin/main for me...

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.