Friday, 3 November, 2017 UTC


Summary

Gatsby is an incredible static site generator that allows for React to be used as the underlying rendering engine to scaffold out a static site that truly has all the benefits expected in a modern web application. It does this by rendering dynamic React components into static HTML content via server side rendering at build time. This means that your users get all the benefits of a static site such as the ability to work without JavaScript, search engine friendliness, speedy load times, etc. without losing the dynamism and interactivity that is expected of the modern web. Once rendered to static HTML, client-site React/JavaScript can take over (if creating stateful components or logic in componentDidMount) and add dynamism to the statically generated content.
Gatsby recently released a v1.0.0 with a bunch of new features, including (but not limited to) the ability to create content queries with GraphQL, integration with various CMSs—including WordPress, Contentful, Drupal, etc., and route based code splitting to keep the end-user experience as snappy as possible. In this post, we’ll take a deep dive into Gatsby and some of these new features by creating a static blog. Let’s get on it!
Getting started

Installing the CLI

npm install -g gatsby-cli
Gatsby ships with a great CLI (command line interface) that contains the functionality of scaffolding out a working site, as well as commands to help develop the site once created.
gatsby new personal-blog && cd $_
This command will create the folder personal-blog and then change into that directory. A working gatsby statically generated application can now be developed upon. The Gatsby CLI includes many common development features such as gatsby build (build a production, statically generated version of the project), gatsby develop (launch a hot-reload enabled web development server), etc.
We can now begin the exciting task of actually developing on the site, and creating a functional, modern blog. You’ll generally want to use gatsby develop to launch the local development server to validate functionality as we progress through the steps.
Adding necessary plugins
Gatsby supports a rich plugin interface, and many incredibly useful plugins have been authored to make accomplishing common tasks a breeze. Plugins can be broken up into three main categories: functional plugins, source plugins, and transformer plugins.

Functional plugins

Functional plugins either implement some functionality (e.g. offline support, generating a sitemap, etc.) or they extend Gatsby’s webpack configuration adding support for typescript, sass, etc.
For this particular blog post, we want a single page app-like feel (without page reloads), as well as the ability to dynamically change the title tag within the head tags. As noted, the Gatsby plugin ecosystem is rich, vibrant, and growing, so oftentimes a plugin already exists that solves the particular problem you’re trying to solve. To address the functionality we want for this blog, we’ll use the following plugins:
  • gatsby-plugin-catch-links
    • implements the history pushState API, and does not require a page reload on navigating to a different page in the blog
  • gatsby-plugin-react-helmet
    • react-helmet is a tool that allows for modification of the head tags; Gatsby statically renders any of these head tag changes
with the following command:
yarn add gatsby-plugin-catch-links gatsby-plugin-react-helmet 
We’re using yarn, but npm can just as easily be used with npm i --save [deps].
After installing each of these functional plugins, we’ll edit gatsby-config.js, which Gatsby loads at build-time to implement the exposed functionality of the specified plugins.
module.exports = { siteMetadata: { title: `Your Name - Blog`, author: `Your Name`, },  plugins: [  'gatsby-plugin-catch-links',  'gatsby-plugin-react-helmet',  ], } 
Without any additional work besides a yarn install and editing a config file, we now have the ability to edit our site’s head tags, as well as implement a single page app feel without reloads. Now let’s enhance the base functionality by implementing a source plugin which can load blog posts from our local file system.

Source plugins

Source plugins create nodes which can then be transformed into a usable format (if not already usable) by a transformer plugin. For instance, a typical workflow often involves using gatsby-source-filesystem, which loads files off of disk—e.g. Markdown files—and then specifying a Markdown transformer to transform the Markdown into HTML.
Since the bulk of the blog’s content, and each article, will be authored in Markdown, let’s add that gatsby-source-filesystem plugin. Similarly to our previous step, we’ll install the plugin and then inject into our gatsby-config.js, like so:
yarn add gatsby-source-filesystem 
module.exports = { // previous configuration plugins: [ 'gatsby-plugin-catch-links', 'gatsby-plugin-react-helmet',  {  resolve: `gatsby-source-filesystem`,  options: {  path: `${__dirname}/src/pages`,  name: 'pages',  },  }  ] } 
Some explanation will be helpful here! An options object can be passed to a plugin, and we’re passing the filesystem path (i.e. where our Markdown files will be located), and then a name for the source files. Now that Gatsby knows about our source files, we can begin applying some useful transformers to convert those files into usable data!

Transformer plugins

As mentioned, a transformer plugin takes some underlying data format that is not inherently usable in its current form (e.g. Markdown, json, yaml, etc.), and transforms it into a format that Gatsby can understand, and that we can query against with GraphQL. Jointly, the filesystem source plugin will load file nodes (as Markdown) off of our filesystem, and then the Markdown transformer will take over and convert to usable HTML.
We’ll only be using one transformer plugin (for Markdown), so let’s get that installed.
  • gatsby-transformer-remark
    • Uses the remark Markdown parser to transform .md files on disk into HTML; additionally this transformer can optionally take plugins to further extend functionality—e.g. add syntax highlighting with gatsby-remark-prismjs, gatsby-remark-copy-linked-files to copy relative files specified in markdown, gatsby-remark-images to compress images and add responsive images with srcset, etc.
The process should be familiar by now, install and then add to config.
yarn add gatsby-transformer-remark 
and editing gatsby-config.js
module.exports = { // previous setup plugins: [ 'gatsby-plugin-catch-links', 'gatsby-plugin-react-helmet', { resolve: `gatsby-source-filesystem`, options: { path: `${__dirname}/src/pages`, name: 'pages', }, },  {  resolve: 'gatsby-transformer-remark',  options: {  plugins: [] // just in case those previously mentioned remark plugins sound cool :)  }  },  ] }; 
Whew! Seems like a lot of set up, but collectively these plugins are going to super charge Gatsby, and give us an incredibly powerful (yet relatively simple!) development environment. We have one more set up step and it’s an easy one. We’re simply going to create a Markdown file that will contain the content of our first blog post. Let’s get to it.
Writing our first Markdown blog post
The gatsby-source-filesystem plugin we configured earlier expects our content to be in src/pages, so that’s exactly where we’ll put it!
Gatsby is not at all prescriptive in naming conventions, but a typical practice for blog posts is to name the folder something like MM-DD-YYYY-title, e.g. 07-12-2017-hello-world. Let’s do just that, and create the folder src/pages/07-12-2017-getting-started, and place an index.md inside!
The content of this Markdown file will be our blog post, authored in Markdown (of course!). Here’s what it’ll look like:
--- path: "/hello-world" date: "2017-07-12T17:12:33.962Z" title: "My First Gatsby Post" --- Oooooh-weeee, my first blog post! 
Fairly typical stuff, except for the block surrounded in dashes. What is that? That is what is referred to as frontmatter, and the contents of the block can be used to inject React components with the specified data, e.g. path, date, title, etc. Any piece of data can be injected here (e.g. tags, sub-title,draft, etc.), so feel free to experiment and find what necessary pieces of frontmatter are required to achieve an ideal blogging system for your usage. One important note is that path will be used when we dynamically create our pages to specify the URL/path to render the file (in a later step!). In this instance, http://localhost:8000/hello-world will be the path to this file.
Now that we have created a blog post with frontmatter and some content, we can begin actually writing some React components that will display this data!
Creating the (React) template
As Gatsby supports server side rendering (to string) of React components, we can write our template in… you guessed it, React! (Or Preact, if that’s more your style)
We’ll want to create the file src/templates/blog-post.js (please create the src/templates folder if it does not yet exist!).
import React from 'react'; import Helmet from 'react-helmet'; // import '../css/blog-post.css'; // make it pretty! export default function Template({ data // this prop will be injected by the GraphQL query we'll write in a bit }) { const { markdownRemark: post } = data; // data.markdownRemark holds our post data return ( <div className="blog-post-container"> <Helmet title={`Your Blog Name - ${post.frontmatter.title}`} /> <div className="blog-post"> <h1>{post.frontmatter.title}</h1> <div className="blog-post-content" dangerouslySetInnerHTML={{ __html: post.html }} /> </div> </div> ); } 
Whoa, neat! This React component will be rendered to a static HTML string (for each route/blog post we define), which will serve as the basis of our routing/navigation for our blog.
At this point, there is a reasonable level of confusion and “magic” occuring, particularly with the props injection. What is markdownRemark? Where is this data prop injected from? All good questions, so let’s answer them by writing a GraphQL query to seed our <Template /> component with content!

Writing the GraphQL query

Below the Template declaration, we’ll want to add a GraphQL query. This is an incredibly powerful utility provided by Gatsby which lets us pick and choose very simply the pieces of data that we want to display for our blog post. Each piece of data our query selects will be injected via the data property we specified earlier.
import React from 'react'; import Helmet from 'react-helmet'; // import '../css/blog-post.css'; export default function Template({ data }) { const { markdownRemark: post } = data; return ( <div className="blog-post-container"> <Helmet title={`Your Blog Name - ${post.frontmatter.title}`} /> <div className="blog-post"> <h1>{post.frontmatter.title}</h1> <div className="blog-post-content" dangerouslySetInnerHTML={{ __html: post.html }} /> </div> </div> ); } export const pageQuery = graphql`  query BlogPostByPath($path: String!) {  markdownRemark(frontmatter: { path: { eq: $path } }) {  html  frontmatter {  date(formatString: "MMMM DD, YYYY")  path  title  }  }  } `; 
If you’re not familar with GraphQL, this may seem slightly confusing, but we can break down what’s going down here piece by piece.
Note: To learn more about GraphQL, consider this excellent resource
The underlying query name BlogPostByPath (note: these query names need to be unique!) will be injected with the current path, e.g. the specific blog post we are viewing. This path will be available as $path in our query. For instance, if we were viewing our previously created blog post, the path of the file that data will be pulled from will be /hello-world.
markdownRemark will be the injected property available via the prop data, as named in the GraphQL query. Each property we pull via the GraphQL query will be available under this markdownRemark property. For example, to access the transformed HTML we would access the data prop via data.markdownRemark.html.
frontmatter, is of course our data structure we provided at the beginning of our Markdown file. Each key we define there will be available to be injected into the query.
At this point, we have a bunch of plugins installed to load files off of disk, transform Markdown to HTML, and other utilities. We have a single, lonely Markdown file that will be rendered as a blog post. Finally, we have a React template for blog posts, as well as a wired up GraphQL query to query for a blog post and inject the React template with the queried data. Next up: programmatically creating the necessary static pages (and injecting the templates) with Gatsby’s Node API. Let’s get down to it.
An important note to make at this point is that the GraphQL query takes place at build time. The component is injected with the data prop that is seeded by the GraphQL query. Unless anything dynamic (e.g. logic in componentDidMount, state changes, etc.) occurs, this component will be pure, rendered HTML generated via the React rendering engine, GraphQL, and Gatsby!
Creating the static pages
Gatsby exposes a powerful Node API, which allows for functionality such as creating dynamic pages (blog posts!), extending the babel or webpack configs, modifying the created nodes or pages, etc. This API is exposed in the gatsby-node.js file in the root directory of your project—e.g. at the same level as gatsby-config.js. Each export found in this file will be parsed by Gatsby, as detailed in its Node API specification. However, we only care about one particular API in this instance, createPages.
const path = require('path'); exports.createPages = ({ boundActionCreators, graphql }) => { const { createPage } = boundActionCreators; const blogPostTemplate = path.resolve(`src/templates/blog-post.js`); } 
Nothing super complex yet! We’re using the createPages API (which Gatsby will call at build time with injected parameters). We’re also grabbing the path to our blogPostTemplate we created earlier. Finally, we’re using the createPage action creator/function made available in boundActionCreators. Gatsby uses Redux internally to manage its state, and boundActionCreators are simply the exposed action creators of Gatsby, of which createPage is one of the action creators! For the full list of exposed action creators, check out Gatsby’s documentation. We can now construct the GraphQL query, which will fetch all of our Markdown posts.

Querying for posts

const path = require('path'); exports.createPages = ({ boundActionCreators, graphql }) => { const { createPage } = boundActionCreators; const blogPostTemplate = path.resolve(`src/templates/blog-post.js`);  return graphql(`{  allMarkdownRemark(  sort: { order: DESC, fields: [frontmatter___date] }  limit: 1000  ) {  edges {  node {  excerpt(pruneLength: 250)  html  id  frontmatter {  date  path  title  }  }  }  }  }`)  .then(result => {  if (result.errors) {  return Promise.reject(result.errors);  }  }); } 
We’re using GraphQL to get all Markdown nodes and making them available under the allMarkdownRemark GraphQL property. Each exposed property (on node) is made available for querying against. We’re effectively seeding a GraphQL “database” that we can then query against via page-level GraphQL queries. One note here is that the exports.createPages API expects a Promise to be returned, so it works seamlessly with the graphql function, which returns a Promise (although note a callback API is also available if that’s more your thing).
One cool note here is that the gatsby-plugin-remark plugin exposes some useful data for us to query with GraphQL, e.g. excerpt (a short snippet to display as a preview), id (a unique identifier for each post), etc.
We now have our query written, but we haven’t yet programmatically created the pages (with the createPage action creator). Let’s do that!

Creating the pages

const path = require('path'); exports.createPages = ({ boundActionCreators, graphql }) => { const { createPage } = boundActionCreators; const blogPostTemplate = path.resolve(`src/templates/blog-post.js`); return graphql(`{ allMarkdownRemark( sort: { order: DESC, fields: [frontmatter___date] } limit: 1000 ) { edges { node { excerpt(pruneLength: 250) html id frontmatter { date path title } } } } }`) .then(result => { if (result.errors) { return Promise.reject(result.errors); }  result.data.allMarkdownRemark.edges  .forEach(({ node }) => {  createPage({  path: node.frontmatter.path,  component: blogPostTemplate,  context: {} // additional data can be passed via context  });  });  }); } 
We’ve now tied into the Promise chain exposed by the graphql query. The actual posts are available via the path result.data.allMarkdownRemark.edges. Each edge contains an internal node, and this node holds the useful data that we will use to construct a page with Gatsby. Our GraphQL “shape” is directly reflected in this data object, so each property we pulled from that query will be available when we are querying in our GraphQL blog post template.
The createPage API accepts an object which requires path and component properties to be defined, which we have done above. Additionally, an optional property context can be used to inject data and make it available to the blog post template component via injected props (log out props to see each available prop!). Each time we build with Gatsby, createPage will be called, and Gatsby will create a static HTML file of the path we specified in the post’s frontmatter—the result of which will be our stringified and parsed React template injected with the data from our GraphQL query. Whoa, it’s actually starting to come together!
We can run yarn develop at this point, and then navigate to http://localhost:8000/hello-world to see our first blog post, which should look something like below:
At this point, we’ve created a single static blog post as an HTML file, which was created by a React component and several GraphQL queries. However, this isn’t a blog! We can’t expect our users to guess the path of each post, we need to have an index or listing page, where we display each blog post, a short snippet, and a link to the full blog post. Wouldn’t you know it, we can do this incredibly easily with Gatsby, using a similar strategy as we used in our blog template, e.g. a React component and a GraphQL query.
Creating the Blog Listing
I won’t go into quite as much detail for this section, because we’ve already done something very similar for our blog template! Look at us, we’re pro Gatsby-ers at this point!
Gatsby has a standard for “listing pages,” and they’re placed in the root of our filesystem we specified in gatsby-source-filesystem, e.g. src/pages/index.js. So create that file if it does not exist, and let’s get it working! Additionally note that any static JavaScript files (that export a React component!) will get a corresponding static HTML file. For instance, if we create src/pages/tags.js, the path http://localhost:8000/tags/ will be available within the browser and the statically generated site.
import React from 'react'; import Link from 'gatsby-link'; import Helmet from 'react-helmet'; // import '../css/index.css'; // add some style if you want! export default function Index({ data }) { const { edges: posts } = data.allMarkdownRemark; return ( <div className="blog-posts"> {posts .filter(post => post.node.frontmatter.title.length > 0) .map(({ node: post }) => { return ( <div className="blog-post-preview" key={post.id}> <h1> <Link to={post.frontmatter.path}>{post.frontmatter.title}</Link> </h1> <h2>{post.frontmatter.date}</h2> <p>{post.excerpt}</p> </div> ); })} </div> ); } export const pageQuery = graphql` query IndexQuery { allMarkdownRemark(sort: { order: DESC, fields: [frontmatter___date] }) { edges { node { excerpt(pruneLength: 250) id frontmatter { title date(formatString: "MMMM DD, YYYY") path } } } } } `; 
OK! So we’ve followed a similar approach to our blog post template, so this should hopefully seem pretty familiar. Once more we’re exporting pageQuery which contains a GraphQL query. Note that we’re pulling a slightly different data set, specifically we are pulling an excerpt of 250 characters rather than the full HTML, as well as we are formatting the pulled date with a format string! GraphQL is awesome.
The actual React component is fairly trivial, but one important note should be made. It’s important that when linking to internal content, i.e. other blog links, that you should always use gatsby-link. Gatsby does not work if pages are not routed via this utility. Additionally, this utility also works with pathPrefix, which allows for a Gatsby site to be deployed a non-root domain. This is useful if this blog will be hosted on something like Github Pages, or perhaps hosted at /blog.
Now this is getting exciting and it feels like we’re finally getting somewhere! At this point, we have a fully functional blog generated by Gatsby, with real content authored in Markdown, a blog listing, and the ability to navigate around in the blog. If you run yarn develop, http://localhost:8000 should display a preview of each blog post, and each post title links to the content of the blog post. A real blog!
It’s now on you to make something incredible with the knowledge you’ve gained in following along with this tutorial! You can not only make it pretty and style with CSS (or styled-components!), but you could improve it functionally by implementing some of the following:
  • Add a tag listing and tag search page
    • hint: the createPages API in gatsby-node.js file is useful here, as is frontmatter
  • adding navigation between a specific blog post and past/present blog posts (the context API of createPages is useful here), etc.
With our new found knowledge of Gatsby and its API, you should feel empowered to begin to utilize Gatsby to its fullest potential. A blog is just the starting point; Gatsby’s rich ecosystem, extensible API, and advanced querying capabilities provide a powerful toolset for building truly incredible, performant sites.
Now go build something great.
Links
  • @dschau/gatsby-blog-starter-kit
    • A working repo demonstrating all of the aforementioned functionality of Gatsby
  • @dschau/create-gatsby-blog-post
    • A utility and CLI I created to scaffold out a blog post following the predefined Gatsby structure with frontmatter, date, path, etc.
  • Source code for my blog
    • The source code for my blog, which takes the gatsby-starter-blog-post (previous link), and expands upon it with a bunch of features and some more advanced functionality