@naturalcycles/github-db

GitHub branch as a Database

Intro

Store your data for free in GitHub repo branch. Similar to gh-pages, but for data.

Implements CommonDB interface.

Store as *.ndjson (newline-delimited json), or *.ndjson.gz (gzipped), or *.sqlite (binary).

Push to preserve git commit history or force-push to reset history and only store the recent state.

Example

const db = new FileDB({
  plugin: new GithubPersistencePlugin({
    token: 'YOUR_GITHUB_TOKEN',
    repo: 'NaturalCycles/github-db',
  }),
})

const items = [
  { id: 'id1', a: 'a1' },
  { id: 'id2', a: 'a2' },
]

await db.saveBatch('TEST_TABLE', items)

It will create, commit and push data/TEST_TABLE.ndjson file to the repo/branch that you've specified (branch defaults to gh-data, similarly to gh-pages).

Next saves will be either new commits, or force-push of new commit over previous (with forcePush: true option).

Currently, it needs some preparation steps to work (create empty branch), described further down.

Prepare an empty branch to store your data

Replace git@github.com:NaturalCycles/github-db.git with your repo.

Replace gh-data with the desired branch name (but gh-data is the default).

mkdir data && cd data
git clone git@github.com:NaturalCycles/github-db.git ./
git checkout --orphan gh-data
git reset --hard
git commit --allow-empty -m "chore: init empty data branch"
git push --set-upstream origin gh-data

TODO

• .ndjson.gzip option
• .sqlite option
• command to vacuum/force-push all commits in gh-data into one