GitHub Action
Add & Commit
You can use this GitHub Action to commit changes made in your workflow run directly to your repo: for example, you use it to lint your code, update documentation, commit updated builds, etc...
Add a step like this to your workflow:
- uses: EndBug/add-and-commit@v8 # You can change this to use a specific version.
with:
# The arguments for the `git add` command (see the paragraph below for more info)
# Default: '.'
add: 'src'
# The name of the user that will be displayed as the author of the commit.
# Default: depends on the default_author input
author_name: Author Name
# The email of the user that will be displayed as the author of the commit.
# Default: depends on the default_author input
author_email: mail@example.com
# Additional arguments for the git commit command. The --message argument is already set by the message input.
# Default: ''
commit: --signoff
# The name of the custom committer you want to use, if different from the author of the commit.
# Default: the name of the author (set with either author_name or default_author)
committer_name: Committer Name
# The email of the custom committer you want to use, if different from the author of the commit.
# Default: the email of the author (set with either author_email or default_author)
committer_email: mail@example.com
# The local path to the directory where your repository is located. You should use actions/checkout first to set it up.
# Default: '.'
cwd: './path/to/the/repo'
# Determines the way the action fills missing author name and email. Three options are available:
# - github_actor -> UserName <UserName@users.noreply.github.com>
# - user_info -> Your Display Name <your-actual@email.com>
# - github_actions -> github-actions <email associated with the github logo>
# Default: github_actor
default_author: github_actor
# The message for the commit.
# Default: 'Commit from GitHub Actions (name of the workflow)'
message: 'Your commit message'
# If this input is set, the action will push the commit to a new branch with this name.
# Default: ''
new_branch: custom-new-branch
# The way the action should handle pathspec errors from the add and remove commands. Three options are available:
# - ignore -> errors will be logged but the step won't fail
# - exitImmediately -> the action will stop right away, and the step will fail
# - exitAtEnd -> the action will go on, every pathspec error will be logged at the end, the step will fail.
# Default: ignore
pathspec_error_handling: ignore
# Arguments for the git pull command. By default, the action does not pull.
# Default: ''
pull: '--rebase --autostash ...'
# Whether to push the commit and, if any, its tags to the repo. It can also be used to set the git push arguments (see the paragraph below for more info)
# Default: true
push: false
# The arguments for the `git rm` command (see the paragraph below for more info)
# Default: ''
remove: './dir/old_file.js'
# Arguments for the git tag command (the tag name always needs to be the first word not preceded by an hyphen)
# Default: ''
tag: 'v1.0.0 --force'
Multiple options let you provide the git arguments that you want the action to use. It's important to note that these arguments are not actually used with a CLI command, but they are parsed by a package called string-argv
, and then used with simple-git
.
What does this mean for you? It means that string that contain a lot of nested quotes may be parsed incorrectly, and that specific ways of declaring arguments may not be supported by this libraries. If you're having issues with your argument strings you can check whether they're being parsed correctly either by enabling debug logging for your workflow runs or by testing it directly with string-argv
(RunKit demo): if each argument and option is aprsed correctly you'll see an array where every string is an option or value.
The action adds files using a regular git add
command, so you can put every kind of argument in the add
option. For example, if you want to force-add a file: ./path/to/file.txt --force
.
The script will not stop if one of the git commands doesn't match any file. E.g.: if your command shows a "fatal: pathspec 'yourFile' did not match any files" error the action will go on.
You can also use JSON or YAML arrays (e.g. '["first", "second"]'
, "['first', 'second']"
) to make the action run multiple git add
commands: the action will log how your input has been parsed. Please mind that your input still needs to be a string because of how GitHub Actions works with inputs: just write your array inside the string, the action will parse it later.
You can delete files with the remove
option: that runs a git rm
command that will stage the files in the given path for removal. As with the add
argument, you can use every option git rm
allows (e.g. add --force
to ignore .gitignore
rules).
The script will not stop if one of the git commands doesn't match any file. E.g.: if your command shows a "fatal: pathspec 'yourFile' did not match any files" error the action will go on.
You can also use JSON or YAML arrays (e.g. '["first", "second"]'
, "['first', 'second']"
) to make the action run multiple git rm
commands: the action will log how your input has been parsed. Please mind that your input still needs to be a string because of how GitHub Actions works with inputs: just write your array inside the string, the action will parse it later.
By default the action runs the following command: git push origin ${new_branch input} --set-upstream
. You can use the push
input to modify this behavior, here's what you can set it to:
true
: this is the default value, it will behave as usual.false
: this prevents the action from pushing at all, nogit push
command is run.- any other string:
The action will use your string as the arguments for thegit push
command. Please note that nothing is used other than your arguments, and the command will result ingit push ${push input}
(no remote, no branch, no--set-upstream
, you have to include them yourself).
One way to use this is if you want to force push to a branch of your repo: you'll need to set the push
input to, for example, origin yourBranch --force
.
If you want the action to commit in a new branch, you can use the new_branch
input.
Please note that if the branch exists, the action will still try push to it, but it's possible that the push will be rejected by the remote as non-straightforward.
If that's the case, you need to make sure that the branch you want to commit to is already checked out before you run the action.
If you're really sure that you want to commit to that branch, you can also force-push by setting the push
input to something like origin yourBranchName --set-upstream --force
.
If you want to commit files "across different branches", here are two ways to do it:
- You can check them out in two different directories, generate your files, move them to your destination and then run
add-and-commit
in the destination directory using thecwd
input. - You can manually commit those files with
git
commands as you would on your machine. There are several ways to do this depending on the scenario. One of them if to stash your changes, checkout the destination branch, and popping the stash. You can then use theadd-and-commit
action as usual. Please note that this is just an example and may not work for you, since your use case may be different.
You can use the tag
option to enter the arguments for a git add
command. In order for the action to isolate the tag name from the rest of the arguments, it should be the first word not preceded by an hyphen (e.g. -a tag-name -m "some other stuff"
is ok).
The action provides these outputs:
committed
: whether the action has created a commit ('true'
or'false'
)commit_long_sha
: the full SHA of the commit that has just been createdcommit_sha
: the short 7-character SHA of the commit that has just been createdpushed
: whether the action has pushed to the remote ('true'
or'false'
)tagged
: whether the action has created a tag ('true'
or'false'
)
For more info on how to use outputs, see "Context and expression syntax".
By default, when you use actions/checkout
on a PR, it will checkout the head commit in a detached head state.
If you want to make some changes, you have to checkout the branch the PR is coming from in the head repo.
You can set it up like this:
- uses: actions/checkout@v2
with:
repository: ${{ github.event.pull_request.head.repo.full_name }}
ref: ${{ github.event.pull_request.head.ref }}
You can find the full docs for payloads of pull_request
events here.
If you're planning on running this only on "internal" PRs (where head and base are in the same repo) then you can omit the repository
input.
If you're planning to use this with PRs coming from other forks, please keep in mind that you might not have write access to those repos.
You can try setting up the repo with your PAT, as explained in the "About tokens" paragraph of this section.
Keep in mind that this "custom checkout" is meant only for PRs: if your workflow runs on multiple events (like push
or workflow_dispatch
), you could try having this step run only for pull_request
events, while other ones will trigger the usual checkout.
If you wish to do so, you can use the step.if
property, here's the docs.
When pushing, the action uses the token that the local git repository has been configured with: that means that if you want to change it you'll need to do it in the steps that run before this action. For example: if you set up your repo with actions/checkout
then you have to add the token there.
Changing the token with which the repo is configured can be useful if you want to run CI checks on the commit pushed by this action; anyway, it has to be set up outside of this action.
The action automatically gets the GitHub token from a github_token
input: this input should not be modified by the user, since it doesn't affect the commits as it's only used to access the GitHub API to get user info, in case they selected that option for the commit author.
That's because you're checking out the repo using the built-in GITHUB_TOKEN
secret: GitHub sees that the push event has been triggered by a commit generated by CI, and doesn't run any further checks to avoid unintentional check loops.
If you're sure that you want the commits generated during CI to trigger other workflow runs, you can checkout the repo using a Personal Access Token (PAT): this will make the resulting commit the same as if you made it yourself.
If you're using actions/checkout
, check out their docs to see how to set your repo token.
The token you use when setting up the repo with this action will determine what token add-and-commit
will use.
Some users reported that they were getting an error:
> fatal: could not read Username for 'https://github.com': No such device or address
If you're getting this error and you're using actions/checkout@v1
, try upgrading to actions/checkout@v2
. If you're still having problems after upgrading, feel free to open an issue. Issue ref: #146
If you don't want to use your GitHub username for the CI commits, you can use the default_author
option to make it appear as if it was made by "github-actions"
on: push
jobs:
build:
runs-on: ubuntu-latest
steps:
- uses: EndBug/add-and-commit@v8
with:
default_author: github_actions
You can also use the committer_name
and committer_email
inputs to make it appear as if GitHub Actions is the committer, here are a couple of example steps:
- uses: EndBug/add-and-commit@v8
with:
message: Show GitHub Actions logo
committer_name: GitHub Actions
committer_email: actions@github.com
- uses: EndBug/add-and-commit@v8
with:
message: Show GitHub logo
committer_name: GitHub Actions
committer_email: 41898282+github-actions[bot]@users.noreply.github.com
Do you want to lint your JavaScript files, located in the src
folder, with ESLint, so that fixable changes are done without your intervention? You can use a workflow like this:
name: Lint source code
on: push
jobs:
run:
name: Lint with ESLint
runs-on: ubuntu-latest
steps:
- name: Checkout repo
uses: actions/checkout@v2
- name: Set up Node.js
uses: actions/setup-node@v1
with:
node-version: 12.x
- name: Install dependencies
run: npm install
- name: Update source code
run: eslint "src/**" --fix
- name: Commit changes
uses: EndBug/add-and-commit@v8
with:
author_name: Your Name
author_email: mail@example.com
message: 'Your commit message'
add: '*.js'
If you need to run the action on a repository that is not located in $GITHUB_WORKSPACE
, you can use the cwd
option: the action uses a cd
normal command, so the path should follow bash standards.
name: Use a different repository directory
on: push
jobs:
run:
name: Add a text file
runs-on: ubuntu-latest
steps:
# If you need to, you can check out your repo to a different location
- uses: actions/checkout@v2
with:
path: './pathToRepo/'
# You can make whatever type of change to the repo...
- run: echo "123" > ./pathToRepo/file.txt
# ...and then use the action as you would normally do, but providing the path to the repo
- uses: EndBug/add-and-commit@v8
with:
message: 'Add the very useful text file'
add: '*.txt --force'
cwd: './pathToRepo/'
Thanks goes to these wonderful people (emoji key):
This project follows the all-contributors specification. Contributions of any kind welcome!
This action is inspired by git-auto-commit-action
by Stefan Zweifel.
This action is distributed under the MIT license, check the license for more info.