search developers
Edit

Custom queries

Learn how to write custom GraphQL queries, how to pass input values into GraphQL requests as variables, and how to retrieve and output the data from a scraper.

Sometimes, the queries found in the Network tab aren't good enough for your use case. Or, perhaps they're even returning more data than what you're after (which can slow down the queries depending on how much data they're giving back). In these situations, it's a good idea to dig a bit deeper into the API and start writing your own custom use-case specific queries.

In this lesson, we're building a scraper which expects a single number (in hours) and a query string as its input. As output, it should provide data about the first 1000 Cheddar posts published within the last n hours which match the provided query. Each post object should contain the title, the publishDate and the videoUrl of the post.

[
    {
        "title": "FDA Authorizes 1st Breath Test for COVID-19 Infection",
        "publishDate": "2022-04-15T11:58:44-04:00",
        "videoUrl": "https://vod.chdrstatic.com/source%3Dbackend%2Cexpire%3D1651782479%2Cpath%3D%2Ftranscode%2Fb68f8133-3aa9-4c96-ac26-047452bbc9ce%2Ctoken%3D581fd52bb7f634834edca5c201619c014cd21eb20448cf89525bf101ca8a6f64/transcode/b68f8133-3aa9-4c96-ac26-047452bbc9ce/b68f8133-3aa9-4c96-ac26-047452bbc9ce.mp4"
    }
    {
        "...": "..."
    }
]

Project setup

To make sure we're all on the same page, we're going to set up the project together by first creating a folder named graphql-scraper. Once navigated to the folder within your terminal, run the following command:

npm init -y && npm install graphql-tag puppeteer got-scraping

This command will first initialize the project with NPM, then will install the puppeteer, graphql-tag, and got-scraping packages, which we will need in this lesson.

Finally, create a file called index.js. This is the file we will be working in for the rest of the lesson.

Preparations

If we remember from the last lesson, we need to pass a valid "app token" within the X-App-Token header of every single request we make, or else we will be blocked. When testing queries, we just copied this value straight from the Network tab; however, since this is a dynamic value, we should farm it.

Since we know requests with this header are sent right when the front page is loaded, it can be farmed by simply visiting thee page and intercepting requests in Puppeteer like so:

// scrapeAppToken.js
import puppeteer from 'puppeteer';

const scrapeAppToken = async () => {
    const browser = await puppeteer.launch();
    const page = await browser.newPage();

    let appToken = null;

    page.on('response', async (res) => {
        // grab the token from the request headers
        const token = res.request().headers()?.['x-app-token'];

        // if there is a token, grab it and close the browser
        if (token) {
            appToken = token;
            await browser.close();
        }
    });

    await page.goto('https://cheddar.com/');

    await page.waitForNetworkIdle();

    // otherwise, just close the browser after networkidle
    // has been fired
    await browser.close();

    // return the apptoken (or null)
    return appToken;
};

export default scrapeAppToken;

With this code, we're doing the same exact thing as we did in the previous lesson to grab this header value, except programmatically.

To learn more about this method of scraping headers and tokens, refer to the Cookies, headers, and tokens lesson of the General API scraping section.

Now, we can import this function into our index.js and use it to create a token variable which will be passed as our X-App-Token header when scraping:

// index.js

// import the function
import scrapeAppToken from './scrapeAppToken.mjs';

const token = await scrapeAppToken();

Building the query

First, we'll write a skeleton query where we define which variables we're expecting (from the user of the scraper):

query SearchQuery($query: String!, $max_age: Int!) {
    # query will go here
}

Also in the previous lesson, we learned that the media type is dependent on the organization type. This means to get any media, it must be wrapped in the organization query:

query SearchQuery($query: String!, $max_age: Int!) {
  organization {
    media(query: $query, max_age: $max_age , first: 1000) {
        
    }
  }
}

Finally, since Cheddar is using cursor-based relay pagination for their API, we must access the data through the edges property, where each node is a result item:

query SearchQuery($query: String!) {
  organization {
    media(query: $query, max_age: $max_age , first: 1000) {
      edges {
        node {
            # here we will define the fields we want
        }
      }
    }
  }
}

The next step is to just fill out the fields we'd like back, and we've got our final query!

query SearchQuery($query: String!) {
  organization {
    media(query: $query, max_age: $max_age , first: 1000) {
      edges {
        node {
          title # title
          public_at # this will be publishDate
          hero_video {
            video_urls {
              url # the first URL from these results will be videoUrl
            }
          }
        }
      }
    }
  }
}

Making the request

Back in our code, we can import gql from graphql-tag and use it to store our query:

// index.js
import { gql, } from 'graphql-tag';
import scrapeAppToken from './scrapeAppToken.mjs';

const token = await scrapeAppToken();

const GET_LATEST = gql`
    query SearchQuery($query: String!, $max_age: Int!) {
        organization {
            media(query: $query, max_age: $max_age, first: 1000) {
                edges {
                    node {
                        title
                        public_at
                        hero_video {
                            video_urls {
                                url
                            }
                        }
                        thumbnail_url
                    }
                }
            }
        }
    }
`;

Alternatively, if you don't want to write your GraphQL queries right within your Javascript code, you can write them in files using the .graphql format, then read them from the filesystem or import them.

In order to receive nice GraphQL syntax highlighting in these template literals, download the GraphQL VSCode extension

Then, we'll take our input and use it to create a variables object which will be used for the request:

// find posts from the last 48 hours that include the keyword "stocks".
// since we don't have any real input, we'll simulate some input
const testInput = { hours: 48, query: 'stocks' };

// the API takes max_input in the format of minutes * 60
// to calculate this value, we do hours * 60^2
const variables = { query: testInput.query, max_age: Math.round(testInput.hours) * 60 ** 2 };

The final step is to take the query and variable and marry them within a gotScraping() call, which will return the API response:

const data = await gotScraping('https://api.cheddar.com/graphql', {
    // we are expecting a JSON response back
    responseType: 'json',
    // we must use a post request
    method: 'POST',
    // this is where we pass in our token
    headers: { 'X-App-Token': token, 'Content-Type': 'application/json' },
    // here is our query with our variables
    body: JSON.stringify({ query: GET_LATEST.loc.source.body, variables }),
});

The final step after making the query is to format the data to match the expected output schema.

Final code

Here's what our final project looks like:

// index.js
import { gql, } from 'graphql-tag';
import scrapeAppToken from './scrapeAppToken.mjs';
import { gotScraping } from 'got-scraping';

// Scrape the token
const token = await scrapeAppToken();

// Define our query
const GET_LATEST = gql`
    query SearchQuery($query: String!, $max_age: Int!) {
        organization {
            media(query: $query, max_age: $max_age, first: 1000) {
                edges {
                    node {
                        title
                        public_at
                        hero_video {
                            video_urls {
                                url
                            }
                        }
                        thumbnail_url
                    }
                }
            }
        }
    }
`;

// Grab our input
const testInput = { hours: 48, query: 'stocks' };

// Calculate and prepare our variables
const variables = { query: testInput.query, max_age: Math.round(testInput.hours) * 60 ** 2 };

// Make the request 
const { body: { data: { organization } } } = await gotScraping('https://api.cheddar.com/graphql', {
    responseType: 'json',
    method: 'POST',
    headers: { 'X-App-Token': token, 'Content-Type': 'application/json' },
    body: JSON.stringify({ query: GET_LATEST.loc.source.body, variables }),
});

// Format the data
const result = organization.media.edges.map(({ node }) => ({
    title: node?.title,
    publishDate: node?.public_at,
    videoUrl: node?.hero_video ? node.hero_video.video_urls[0].url : null,
}));

// Log the result
console.log(result);
// scrapeAppToken.js
import puppeteer from 'puppeteer';

const scrapeAppToken = async () => {
    const browser = await puppeteer.launch();
    const page = await browser.newPage();

    let appToken = null;

    page.on('response', async (res) => {
        const token = res.request().headers()?.['x-app-token'];

        if (token) {
            appToken = token;
            await browser.close();
        }
    });

    await page.goto('https://cheddar.com/');

    await page.waitForNetworkIdle();

    await browser.close();

    return appToken;
};

export default scrapeAppToken;

Wrap up

If you've made it this far, that means that you've conquered the king of API scraping - GraphQL, and that you're ready to take on writing scrapers for the majority of websites out there. Nice work!

Take a moment to review the skills you learned in this section:

  1. Modifying the variables of copied GraphQL queries
  2. Introspecting a GraphQL API
  3. Visualizing and understanding a GraphQL API introspection
  4. Writing custom queries
  5. Dealing with cursor-based relay pagination
  6. Writing a GraphQL scraper with custom queries