react-apollo-dynamic-query

This package provides a wrapper around react-apollo's graphql() function, allowing the query to be dynamically generated as a function of props.

Installation

Using npm:

npm install --save react-apollo-dynamic-query

Using yarn:

yarn add react-apollo-dynamic-query

Usage

Without graphql-apollo-dynamic-query

import { graphql } from "react-apollo"

const query = gql`
  # YOUR QUERY HERE
`

const config = {
  // YOUR CONFIG HERE
}

const MyApolloComponent = graphql(query, config)(MyDisplayComponent)

With graphql-apollo-dynamic-query

import { graphqlDynamic } from "react-apollo-dynamic-query"

const query = props => {
  return gql`
    # YOUR QUERY HERE
  `
}

const config = {
  // YOUR CONFIG HERE
}

const MyApolloComponent = graphqlDynamic(query, config)(MyDisplayComponent)

As you can see, graphqlDynamic() has the same interface as graphql(), except the query argument is now a function that takes in props.

Why?

When we use the graphql() function in react-apollo, we must specify an exact query. We can have variable arguments, but the fields we use have to be specified in our query. This is sufficient for most cases, because most components will need to display a specific type of data. But what if we want to build a component that can slightly modify the query based on the props we pass in? We would have to build a different component for each use case.

For example, let's say we want to build a component that can display a list of users or animals. If we're using the graphql() function, we'd have to build two separate components, like this:

import React from "react"
import gql from "graphql-tag"
import { graphql } from "react-apollo"

const ListEntities = props => {
  if (props.data.loading) return null

  return (
    <div>
      <h1>List of {props.field}:</h1>
      <ul>
        {props.data.entities.map(entity => (
          <li>{entity.name}</li>
        ))}
      </ul>
    </div>
  )
}

const usersQuery = gql`
  query ListUsers {
    entities: users {
      name
    }
  }
`

const animalsQuery = gql`
  query ListAnimals {
    entities: animals {
      name
    }
  }
`

const ListUsersWithData = graphql(usersQuery)(ListEntities)
const ListAnimalsWithData = graphql(animalsQuery)(ListEntities)

Through the use of aliasing (setting the field in both queries to "entities"), we're able to re-use our display component (ListEntities), but we have to create two different GraphQL-connected components (ListUsersWithData and ListAnimalsWithData) since we're dealing with two different queries.

It would be nice to be able to create a generic ListEntitiesWithData component that modifies the query based on the props we pass in.

Example

This package provides a function called graphqlDynamic(), which is a drop-in replacement for react-apollo's graphql() function. The only difference is that the query argument we pass in must now be a function instead of a GraphQL tag. This allows us to create a component that modifies the query based on the props we pass in.

import React from "react"
import gql from "graphql-tag"
import { graphqlDynamic } from "react-apollo-dynamic-query"

// This component hasn't changed from the example above.
const ListEntities = props => {
  if (props.data.loading) return null

  return (
    <div>
      <h1>List of {props.field}:</h1>
      <ul>
        {props.data.entities.map(entity => (
          <li>{entity.name}</li>
        ))}
      </ul>
    </div>
  )
}

// We can now pass in the name of the field as a prop,
// and generate the query using string interpolation.
const query = props => {
  return gql`
    query Foo {
      entities: ${props.field} {
        name
      }
    }
`
}

const ListEntitiesWithData = graphqlDynamic(query)(ListEntities)

Then, we can use our component like this:

<ListEntitiesWithData field="users" />

<ListEntitiesWithData field="animals" />

While not depicted in this example, we can optionally pass in a config object as the second argument, as we normally would with the standard graphql() function. See the basic usage example for details.

Do I need a whole package for this?

Absolutely not. Feel free to copy the source code into your application to avoid having yet another dependency.