Backup All Your Github Repos Script
I had a nightmare that someone had access to my Github account and was deleting every repo I had before my eyes. I couldn’t figure out how to stop them and invalidate the proper token. I woke up and decided that I am not ok with this one point-of-failure of my historical work. So I wrote a script to back it up.
You can find the script on my Github repo (the irony is not escaping me). But this is basically what I thought:
- I want to backup my Github repos that I own - not the ones that I belong to - once every 6 months. There might be some duplicate backups but that’s ok. This is a nice mix for the size of my repos compared to the amount / frequency of new ones.
- I will put that in my iCloud drive so it will back up to icloud. I also use Backblaze so it will backup there as well
- I will run it every 6 months with a reminder in Things3 using an Alfred workflow
- I will zip the file myself (even though I could build that in).
So, to do that, I created the script, the reminders, the Alfred workflow, etc.
To call it, I simply run FOLDER=/home/aaron/ghbackups GHTOKEN=myAccessTokenHere node run-backup.js
and then it runs for me.
There is an output with the progress bar. It updates each repo I’m checking out. Pretty simple.
Let’s take a quick look at the Node code:
const { Octokit } = require("@octokit/rest");
const cliProgress = require('cli-progress');
const { mkdirSync } = require('fs');
const path = require('path');
const { execSync } = require('child_process');
process.env.GHTOKEN || (() => {throw new Error('The GHTOKEN env var is missing.')})();
process.env.FOLDER || (() => {throw new Error('The FOLDER env var is missing.')})();
(async () => {
const octokit = new Octokit({
auth: process.env.GHTOKEN
});
console.info('Loading up on repos...');
const repos = await octokit.paginate("GET /user/repos", {
affiliation: "owner",
sort: "full_name",
per_page: 100
});
const total = repos.length;
console.info(`Retrieved all repos - there are ${total} of them.`);
const rootFolder = path.join(
process.env.FOLDER,
(new Date()).toISOString().substring(0, 10)
);
console.info(`Beginning to write them to folder ${rootFolder}`);
mkdirSync(rootFolder);
const progress = new cliProgress.SingleBar({
format: '[{bar}] {percentage}% | ETA: {eta}s | {value}/{total} | {repo}'
}, cliProgress.Presets.shades_classic);
progress.start(total, 0, {
repo: ""
});
let count = 0;
repos.forEach(repo => {
execSync(`git clone ${repo.ssh_url} --quiet`, {
cwd: rootFolder
});
progress.update(++count, { repo: repo.full_name });
});
progress.stop();
console.info('We are done!');
})();
Let’s go through it real quick.
First, bring in the @octokit/rest
and cli-progress
libs. Grab the path
, child_process.execSync
and fs.mkdir
modules from Node.
Then, make sure we have the GHTOKEN
and FOLDER
environment variables. These could be set in your environment in general, or passed on the CLI like I showed above.
Then, it’s a self-executing closure so that I can use await
at the ’top level’ of the script.
We create an Octokit instance with the token, and then paginate the results of all of the repos owned by the token’s user into a variable. This is sorted by
full_name
just for ease of following the progress output. I create a directory based on the current date. It’s important to note that this will crash if
the directory already exists. That’s unintentionally by design and should stop duplicating backups. I create a progress bar with the total amount and a custom formatter. The custom formatter is
because I want to show the repository’s full name on every update. That way if something breaks, or if I’m just curious, I know what I’m cloning at the moment.
Then, looping over each repository, spawning a sync child exec process that git clones via SSH quietly into the date-named folder. And progressing the progress bar.
There you go - simple as that. Feel free to use this to back up your own repos - or - to build off of!