Introduction

Cloud Firestore is a NoSQL document database built for automatic scaling, high performance, and ease of application development. While the Cloud Firestore interface has many of the same features as traditional databases, as a NoSQL database it differs from them in the way it describes relationships between data objects. It is a part of the Google Cloud platform, and a spiritual successor to the Firebase Real-Time Database.

Firestore makes it easy to securely store and retrieve data, and already has a powerful API for querying data. Firegraph builds on that awesome foundation by making it even easier to retrieve data across collections, subcollections, and document references.

Primary Goals

  • Wrap the Firestore SDK in its entirety. This means that, in time, we hope to support features like real-time updates, caching, index management, and other APIs available through Firestore's SDK.
  • Leverage features of GraphQL query syntax. When creating this library, I initially planned to create a "GraphQL-esque" query language specifically for Firestore. I have since decided that is the wrong way to go, and opted to ensure that all Firegraph queries are valid GraphQL. This should make it easier if you decide to roll your own GraphQL backend at some point.
  • Operate as a lightweight wrapper. As we move toward supporting all Firestore APIs, the goal is to also introduce support for some common (but not directly supported) Firestore use cases. That said, Firegraph should retain a small footprint and avoid depending on other NPM modules as much as possible.

Getting Started

Getting started with Firegraph is very easy! You do not need to host a GraphQL server to use Firegraph. However, your project does require some GraphQL-related dependencies.

Installing

# npm
npm install --save graphql graphql-tag firegraph

# Yarn
yarn add graphql graphql-tag firegraph

Queries

You can either write queries inside your JavaScript files with gql, or if you use webpack, you can use graphql-tag/loader to import GraphQL query files (*.graphql) directly.

Collections

Every top-level name in a query is considered a Firestore collection. For example, in the query below, we are querying every document in the posts collection and retrieving the id, title, and body values from each document in the response. Note: id is a special field that actually retrieves the document key.

const { posts } = await firegraph.resolve(firestore, gql`
    query {
        posts {
            id
            title
            body
        }
    }
`)

Subcollections

When you have nested values (e.g. in the query below), they are processed as child collections. To clarify, for each doc in the posts collection, we also retrieve the posts/${doc.id}/comments collection. This result is stored in the comments key for each document that is returned.

const { posts: postsWithComments } = await firegraph.resolve(firestore, gql`
    query {
        posts {
            id
            title
            body
            comments {
                id
                body
            }
        }
    }
`)

Document References

Right now, we are assuming that post.author is a string that matches the ID of some document in the users collection. In the future we will leverage Firestore's DocumentReference value type to handle both use cases.

const { posts: postsWithAuthorAndComments } = await firegraph.resolve(firestore, gql`
    query {
        posts {
            id
            title
            body
            author(matchesKeyFromCollection: "users") {
                id
                displayName
            }
            comments {
                id
                body
                author(matchesKeyFromCollection: "users") {
                    id
                    displayName
                }
            }
        }
    }
`)

Filtering Results

One of our primary goals is to wrap the Firestore API in its entirety. That said, the where clause in Firegraph maps directly to the expected behavior in Firestore:

  • someKey: someValue maps to .where(someKey, '==', someValue)
  • someKey_gt: someValue maps to .where(someKey, '>', someValue)
  • someKey_gte: someValue maps to .where(someKey, '>=', someValue)
  • someKey_lt: someValue maps to .where(someKey, '<', someValue)
  • someKey_lte: someValue maps to .where(someKey, '>=', someValue)
  • someKey_contains: someValue maps to .where(someKey, 'array-contains', someValue)

For the last one, of course, someKey would have to use Firestore's array type. All of the restrictions related to compound queries with Firestore (no logical OR or inequality testing) still apply but those are some of the first things we are hoping to add support for.

const authorId = 'sZOgUC33ijsGSzX17ybT';
const { posts: postsBySomeAuthor } = await firegraph.resolve(firestore, gql`
    query {
        posts(where: {
            author: ${authorId},
        }) {
            id
            message
            author(matchesKeyFromCollection: "users") {
                id
            }
        }
    }
`);

Roadmap

  • [x] Querying values from collections
  • [x] Querying nested collections
  • [ ] GraphQL mutations allowing updates to multiple documents at once
  • [ ] Basic search functionality (on par with current Firestore API)
  • [ ] More advanced search functionality (GraphQL params, fragments, etc)

Contributing

Thank you for your interest! You are welcome (and encouraged) to submit Issues and Pull Requests. If you want to add features, check out the roadmap above (which will have more information as time passes). You are welcome to ping me on Twitter as well: @sjroot

New Features

To submit a new feature, you should follow these steps:

  1. Clone the repository and write tests that describe how your new feature is used and the results you would expect.
  2. Implement the appropriate changes to our code base. The test directory includes a Firestore instance that is ready to go; just provide your Firebase app config as environment variables.
  3. Submit a PR once you've implemented changes and ensured that your new tests pass without causing problems with other tests.