r7yh6g
Last Updated: July 14, 2017
·
15.24K
· drumsrgr8forn8
1450297 10151756018017117 1934558545 n

Create a new Gitlab repo from the command line

Git does some funny magic so you can create script named git-foo and run the like they're arguments to git like git foo.

Create a script and name it something like git-init-remote

Note: I have used GitLab's free instance they now provide. Change the url to your own if you're running a private instance of GitLab

#!/bin/sh

repo=$1
token=put_your_api_token_here

test -z $repo_name && echo "Repo name required." 1>&2 && exit 1

curl -H "Content-Type:application/json" https://gitlab.com/api/v3/projects?private_token=$token -d "{ \"name\": \"$repo\" }"

Run chmod +x git-init-remote (or whatever you called your script). And move it somewhere into you PATH.

Now you can run git init-remote my-cool-repo to create a new repo. You'll get a bunch of json back... Someone could probably write a fancier script that actually checks for success or failure.

Say Thanks
Respond

4 Responses
Add your response

19856
None

You have a small typo in this script that prevents it from working. Replace the $repo_name variable with $repo. Thanks for writing this... I was doing something similar on GitHub, but wasn't using this as a git script. Perfect for my project creation automation.

over 1 year ago ·
21208
None

Last paragraph, small typo. Should be:

Now you can run git-init-remote my-cool-repo to create a new repo.

over 1 year ago ·
25141
0 hc1ylgy1yabrqnoqmq dl0m1y2rvq1uqg6f3l0shtdxiibunukitwxjakckxbqs43ta8oz07v u9

There are multiple errors in the script ("repo" and "repo_name" variables are used interchangeably) for those who are getting errors, use this rectified version:

!/bin/sh

reponame=$1
token=your
apitokenhere

test -z $repo_name && echo "Repo name required." 1>&2 && exit 1

curl -H "Content-Type:application/json" http://your_domain_or_ip_here/api/v3/projects?private_token=$token -d "{ \"name\": \"$repo_name\" }"

over 1 year ago ·
29103
Squidward
#!/bin/bash

# Script to automate the deployment of new private gitlab repos
# TODO: simplify and enhance error handling (already existing repos etc.)
#       expand script to handle: public repos, deletion, re-naming

token="your_token_here"

function usage {
  echo "Usage: $0 -n name"
  exit 1
}

number_args=$#
if [[ !("$number_args" -eq 2) ]]; then
  echo "Incorrect number of args" >&2
  echo "If two strings in name, quote them: e.g. \"str1 str2\""
  usage
fi

while getopts "n:" name; do
  case "${name}" in
    n)
      repo=${OPTARG}
      ;;
    \?)
      usage
      ;;
    *)
      usage
      ;;
  esac
done

if [[ -z ${repo} ]]; then
  echo "Option -n requires an argument." >&2
  usage
else
  echo "#######\nCreating new repo: ${repo}\n#######\n"
fi

# In case space separate words are used for the repo name, the first word will be used
# when naming the json output file.
repo_short=$(echo ${repo} | cut -d " " -f1)

response=$(curl -s -o ./temp.json -w '%{http_code}' \
-H "Content-Type:application/json" https://gitlab.com/api/v3/projects?private_token=$token \
-d "{ \"name\": \"${repo}\" }")

# Format JSON log
cat ./temp.json | python -m json.tool > ./${repo_short}_repo.json
rm -f ./temp.json

echo "Any JSON output is logged in ./${repo_short}_repo.json"
if [ $response != 201 ]; then
  echo "Error\nResponse code: $response"
  exit 1
else
  echo "Repo successfully created\nResponse code: $response"
  exit 0
fi
7 months ago ·
Filed Under