DEV Community

Cover image for A small Tutorial for GraphQL Pagination with MongoDB
Karolinkas
Karolinkas

Posted on • Updated on

A small Tutorial for GraphQL Pagination with MongoDB

Why is GraphQL so great?

Developers that are regularily writing API endpoints are pretty excited about GraphQL since it came out in 2015. But what's so special about it? Basically it gives you more freedom as a developer to request only exactly as much data you really need, it's aiding type safety and allows your API to evolve without versioning. GraphQL is actually it's own language for querying databases that is very much worth it to look into because it might make API development more smooth like this guy, almost like this guy:

Alt Text

Is this for me?

If you have never tried out GraphQL with MongoDB, but you are super curious about it, don't despair, but try this tutorial before, that explains you how to bring together GraphQL with MongoDB: https://medium.com/the-ideal-system/graphql-and-mongodb-a-quick-example-34643e637e49

What pagination means here

First of all let's confirm we agree on what pagination means for us. Pagination according to our understanding could for example give the user of a web application the chance to customize how much of the search results of list he wants to see. If there's a lot of details he would prefer to see 10 results in a list rather than 100.

Now let's jump into the actual tutorial:

Tutorial

๐Ÿ‘‰ Tip:
This assumes you have mongoDB installed locally, if you realise you don't, this is a good guide: https://docs.mongodb.com/manual/administration/install-community/

Clone my repository with the example project.

git clone git@github.com:karolinkas/graphQL-pagination.git
Enter fullscreen mode Exit fullscreen mode

Run yarn to install all dependencies.

Make sure you local mongo instance is running, if not start it with the mongo daemon

mongod
Enter fullscreen mode Exit fullscreen mode

To illustrate you how and where we might be needing pagination I created a graphql server that connects to a local mongo db instance.

The mineral collection as an example

Imagine you are really obsessed with minerals and would like to have digital representation of your collection. Using graphQL queries you will be creating minerals that have a name and a content
( with important containing chemical elements ) property each time you add a new one to your collection.

To see the project in action get things ready by running

yarn start
Enter fullscreen mode Exit fullscreen mode

which builds and starts an express server.

Open http://localhost:3001/graphiql
GraphiQL is an online IDE (editor) that allows you to run queries and mutations against your database.

Making minerals

To create a mineral ( a database entry ) in our mineral collection run this in GraphiQL. :

mutation {
  createMineral(name: "Quartz", content: "Silicon"){
    _id
        name
        content
    }
}
Enter fullscreen mode Exit fullscreen mode

Create a bunch of them with different mineral names to have a good list of example minerals.

Exploring minerals

To see and explore what minerals we already created run in GraphiQL:

query {
  minerals {
    _id
    name
    content
  }
}
Enter fullscreen mode Exit fullscreen mode

You'll see that this query returns the list of all the minerals that you ever created, but what if you have a huge collection of 250 minerals and you don't want to show the all in a giant list.
That's where pagination comes into play.

An example for this would be to limit the query's return value to 4 minerals instead of showing all 12 at once. If we are showing only the minerals in the second row there is an offset of 4 minerals. We could leave the decision how many results should be shown at a time to the user, by presenting some UI that lets the user pick from some options that then set the limit parameter.

Alt Text

Adjusting Query

Let's try to implement this with a graphql query.
In src/start.js you will be able to find the definition of query and it's types.

type Query {
    minerals: [Mineral]
}
Enter fullscreen mode Exit fullscreen mode

If we want to give the user new functionality we could just add new parameters to the query.

type Query {
    minerals(limit: Int, offset: Int): [Mineral]
}
Enter fullscreen mode Exit fullscreen mode

Then in the resolver, which is what brings together the mongodDB Query and the GraphQL query,
this resolver:

Query: {
    minerals: async () => {
        return (await Minerals.find({})
            .toArray()).map(prepare)
    },
},
Enter fullscreen mode Exit fullscreen mode

turns into

Query: {
    minerals: async (root, {limit, offset}) => {
        return (await Minerals.find({})
            .limit(parseInt(limit))
            .skip(parseInt(offset))
            .toArray()).map(prepare)
    },
},
Enter fullscreen mode Exit fullscreen mode

Here we are using the parameters that we defined in the Query type to limit/offset the results the query will return.

Now when trying out the new pagination functionality we set the new parameters in GraphiQL, to implement the same pagination like in the infographic we would do:

query {
    minerals(limit: 4, offset: 4) {
        _id
        name
        content
    }
}
Enter fullscreen mode Exit fullscreen mode

On the other hand the limit/offset pagination comes with a couple of disadvantages, i.e. if one mineral gets removed the whole mineral collection gets shifted forward, which might change the query's response in unexpected ways. That's where cursor-based pagination might bring more specific results. The idea of cursor pased pagination is that we create references to each object, in this case mineral, where the user might have left off while exploring all minerals. From this reference point we can then fetch more data.

Alt Text

Let's implement this with GraphQL, we'll still need the limit of results to show but will know pass in a cursor, which to simplify things for the sake of this tutorial will be the index of the mineral that's currenlty being looked at by the user. This is the type:

type Query {
    minerals(limit: Int, index: Int): [Mineral]
}
Enter fullscreen mode Exit fullscreen mode

And the query will now be:

Query: {
minerals: async (root, {limit, index) => {
return (await Minerals.find({})
.limit(parseInt(limit))
.skip(parseInt(index))
.toArray()).map(prepare)
},
},
Enter fullscreen mode Exit fullscreen mode




Next Steps

This implementation could be improved in the next steps, i.e. we could use the id of the mineral to access the right mineral instead of the index to make sure the selection of one mineral is more specific. Also we could add a pageInfo field to each mineral to store information about the startCursor and endCursor assuming the limit hasn't changed and the nextPage boolean that will allow you to be aware if there's more results to follow.

Top comments (0)