Transform plain text into dynamic blogs and websites using the latest web technologies. A React.js static site generator.
Supports Markdown, HTML, and React.js pages out of the box. Easy to add support for additional file types.
Leverages React's component model and React Router's "nested component hierarchy" to make templating incredibly powerful and intuitive.
All templates, css, and content are hot reloadable — powered by webpack.
Build sites like it's 1995. Files are translated into HTML pages at the
same position within the file system. Add a markdown file at /docs/index.md
and
it'll be converted to /docs/index.html
.
We've started work on a 1.0 release of Gatsby!
- No-reload page transitions
- Hot reload editing. Tweak your pages, templates, and styles and see changes in real time.
- Make React.js component model and ecosystem available for building static sites
- Intuitive directory-based URLs. The URL of a page is derived from its spot on the file system.
- Support "Starters" — install starter sites directly from Github. Use open sourced starters or build your own.
- No-refresh page transitions
- The awesome React.js component model
- Live editing on every part of your site. Dramatically speed development.
- bricolage.io (source)
- relaterocket.co
- reindex.io
- syncano.io (source)
- graphene-python.org (source)
- back-to-the-basics.io
- ashk.io (source)
- nordnet.se/brand
- likescoffee.com (source)
- vii.campjs.com (source)
- michaeljdeeb.com (source)
- brittcrawford.com (source)
- React-MDL (source)
- Snapgit (source)
- iamdustan.com (source)
- React Hardware (source)
- peterp.me (source)
- k-create.com (source)
- kylegach.com (source)
- jazlal.li (source)
- React Headroom (source)
- Völlig Ohne (source)
- michaelcereda.com (source)
- openFDA (source)
- emilyaviva.com (source)
- dynamicext.com
- React Gravatar (source)
- johnm.io (source)
- twnsnd.co
- Beau Han Xu
- HalfAtheist (source)
- joshuahorwitz.net
- DOSH
- Outcomes.com
- Codefellows.org
- ollieglass.com
- waigojs.com (source)
- The State Of JavaScript
- ZBT MIT Website (source)
- ethereumclassic.org (source)
- Husam Machlovi, Portfolio & Blog
- numenta.com (source)
- chiedolabs.com
- Yuppies (source)
- Go7hic
- Video streaming devops blog (source)
- blog.rphl.io
- oliverbenns.com (source)
- Kapadiya.net (source)
- anvilabs.co (source)
- Cardiogram
- Edit this file to add yours!
Note, for the sites that have made their source available, you can
install them locally by running gatsby new SITE_NAME SOURCE_URL
.
- Scott Nonnenberg presents on Gatsby to the Seattle React.js Meetup https://blog.scottnonnenberg.com/static-site-generation-with-gatsby-js/
- Kyle Mathews gave a lightning talk @ the 2016 React.js Conference on Gatsby https://www.youtube.com/watch?v=RFkNRKL6ZoE
- Kyle Mathews spoke on Gatsby to the San Francisco Static Web Tech meetup http://www.staticwebtech.com/presentations/developing-with-react-gatsbyjs/
- ReactJS E-Commerce With No Backend Using Snipcart & Gatsby
- A step-by-step guide: Gatsby on Netlify
- Hosting GatsbyJS – A React Static Site Generator
- Michael Cereda's multi-part series of blog posts on building his site with Gatsby
- Talk on GatsbyJS @ ViennaJS
If your site falls closer to the site end of the app<---->site spectrum then yes.
Gatsby is an excellent fit for blogs, marketing sites, docs sites, etc. Proper web apps should probably remain as normal web apps (though I'd love to be proved wrong!).
Gatsby is not yet stable. APIs will break. Functionality is missing. It's usable but if you plan on building with it, expect a rocky road for some time.
Contributions welcome!
npm install -g gatsby
- Create new Gatsby site
gatsby new my-test-gatsby-site
This creates the directory for your Gatsby project and adds the minimal files needed. cd my-test-gatsby-site
gatsby develop
— Gatsby will start a hot-reloading development server accessible at localhost:8000- See the tutorial below for more.
The Gatsby CLI tool lets you install "starters". These are partially built sites preconfigured to help you get moving faster on creating a certain type of site.
When creating a new site, you can optionally specify a starter to
base your new site on e.g. gatsby new [SITE_DIRECTORY] [URL_OF_STARTER]
For example, to quickly create a blog using Gatsby, you could install the Gatsby Starter Blog by running:
gatsby new blog https://github.com/gatsbyjs/gatsby-starter-blog
This downloads the files and initializes the site by running npm install
If you don't specify a custom starter, your site will be created from the default starter.
There are several starters that have been created. Create a PR to include yours!
- Kitchen sink default starter (Demo)
- Simple blog (Demo)
- Simple documentation site (Demo)
- Lumen (Demo)
- DrunkenBlog (Demo)
- Clean start (Demo)
- Project AMP blog
- Install gatsby
npm install -g gatsby
- Install documentation site starter
gatsby new docs-site gh:gatsbyjs/gatsby-starter-documentation
- type
cd docs-site
- type
gatsby develop
- Open site in browser at localhost:8000. Verify clicking on links works.
- Try editing the site's config file
config.toml
. Change thesiteTitle
key. The site's title should change shortly after saving. - Next try editing a doc page. Open
/pages/docs/getting-started/index.md
and edit it. Again any saved changes should load without refreshing in the browser. - Add a new markdown page to the documentation. Copy the
getting-started
directory tosome-additional-steps
. Then edit the markdown file within the new directory. If you're familiar with other static site generation software, you'll be familiar with the "frontmatter" at the top of the file. Edit the title there + change the order to "5". Save this. Ideally this new file would be hot reloaded like other changes but I haven't figured out how to make this happen yet (help appreciated here). So to see your new page, restartgatsby develop
and then refresh your browser. - Build your site
gatsby build
. The site is built to the/public
directory. Test that the build worked by runninggatsby serve-build
which serves the contents of the/public
directory.
The process is file --> Webpack loader --> React.js wrapper component --> static HTML page.
Gatsby leverages Webpack extensively. Webpack is a sophisticated module bundler that can turn any sort of file into a commonjs module. Webpack uses "Loaders" to convert a file into a module. These loaded modules are then wrapped inside a React.js component that's specific to a given file type. Gatsby then generates a static HTML page from this component.
Gatsby ships with default loaders and wrappers for HTML, Markdown, and JSX/CJSX but for most projects you'll want to write your own loaders and wrappers (very easy to do).
As an example of how this process works, let's walk quickly through converting a markdown file into an HTML page.
The default Gatsby markdown loader parses the markdown into HTML and uses Highlight.js to syntax highlight code blocks.
Our markdown file.
---
title: This is a title
---
# Hi friends.
This is a markdown file.
When loaded and required, the resulting javascript object looks like the following:
{
file: {
// Information about file on disk e.g. extension, directory path, etc.
},
data: {
title: "This is a title",
body: "<h1>Hi friends.</h1><p>This is a markdown file</p>"
}
}
Now Gatsby wraps the markdown file in this very simple React.js component.
module.exports = React.createClass({
displayName: "MarkdownWrapper",
render: function() {
var post = this.props.route.page.data
return <div className="markdown">
<h1>{post.title}</h1>
<div dangerouslySetInnerHTML={{__html: post.body}}/>
</div>
}
})
config.toml
- Core application configuration is stored here. Available via arequire
orimport
of 'config'. Values:noProductionJavascript
- set to a truthy value to prevent generation of bundle.js (containing your client-side Single Page App) during agatbsy build
. You'll need to update your top-levelhtml.js
file so that it doesn't pull inbundle.js
in production, but you'll want to keep it forgatsby develop
mode.
/pages
- All pages go here. Everything is turned into a page except files which start with an underscore:_template
files under/pages
are treated as parent templates for other pages in the same directory tree.- (optional)
pages/404.js
orpages/404.html
- automatically picked up as your 'not found' page. If you<Link>
to an unknown URL, this page will be shown. Note: in production, you'll need to set up your server host to show this page when it can't find the requested file.
- (optional)
gatsby-browser.js
- a way to hook into key application events. ExportonRouteUpdate
of typefunction()
to be notified whenever React-Router navigates. - (optional)
gatsby-node.js
- a way to hook into events during build and development.
Gatsby uses webpack-configurator to make changing the webpack loaders easy. The default set of loaders is organized by key.
To modify the Webpack configuration, create a gatsby-node.js
in the root of your project
and export there a modifyWebpackConfig
function.
exports.modifyWebpackConfig = function(config, stage) {
// edit loaders here
return config
}
Gatsby calls this function with the webpack-configurator object and "stage" string when it creates a Webpack config. It first loads the defaults and then allows you to modify it.
The stage
can be
- develop: for
gatsby develop
command, hot reload and CSS injection into page - develop-html: same as develop without react-hmre in the babel config for html renderer
- build-css: build styles.css file
- build-html: build all HTML files
- build-javascript: Build bundle.js for Single Page App in production
Consider the following example which removes the default css loader and replaces it with a loader that uses css-modules.
exports.modifyWebpackConfig = function(config, stage) {
config.removeLoader('css')
config.loader('css', function(cfg) {
cfg.test = /\.css$/
cfg.loader = 'style!css?modules'
return cfg
})
return config
}
Each loader (cfg
in the above example) can be a valid
webpack loader
and there are a host of
preexisting loaders
which you can use to enhance Gatsby.
It is also possible to write your own loaders.
Gatsby includes some default loaders that you can also override.
To write your own loader or override a Gatsby loader, make a loaders
directory at the root of your site that contains directories for custom loaders.
e.g. loaders/markdown-loader/index.js
will take precedence
over the markdown-loader that Gatsby includes.
See an example of a custom loader in the default starter.
Similar to the loaders, plugins are handled via
webpack-configurator
and gatsby-node.js
.
Note: the following example is now redundant since ExtractTextWebpackPlugin is now setup by default but you can still use it as an example of how to modify the Webpack plugins.
If we wanted to extract all of the css in our project into a since
styles.css
file for production, we could add the
ExtractTextWebpackPlugin
. To do this, we need to modify the loader
and add the plugin when generating the static HTML for our site.
var ExtractTextPlugin = require("extract-text-webpack-plugin")
exports.modifyWebpackConfig = function(config, stage) {
if(stage === 'build-html') {
config.removeLoader('css')
config.loader('css', function(cfg) {
cfg.test = /\.css$/
cfg.loader = ExtractTextPlugin.extract('css?minimize')
return cfg
})
config.plugin('extract-css',
ExtractTextPlugin,
["styles.css", { allChunks: true }])
}
return config
}
Each plugin (extract-css
in the above example) can be a valid
webpack plugin
and there are a host of
preexisting plugins
which you can use to enhance Gatsby.
It is also possible to write your own plugins.
Gatsby also uses gatsby-node.js
to pass control of the final build step over
to the user when running gatsby build
. The post build function takes two arguments, the pages and the callback for completing the build:
exports.postBuild = function(pages, callback) {
// perform actions on pages here
callback()
}
- Coming...
Python v2 is required to install Gatsby. Please ensure python is in your path before running npm install -g gatsby
.
Webpack doesn't currently support hot-reloading new files added to a context. When you add a new file, restart the gatsby develop
process and your new page will show up.
Make sure you also including the trailing slash in your URLs:
- Bad:
http://localhost:8000/foo
- Good:
http://localhost:8000/foo/
A neat performance feature supported by Gatsby is inlining your CSS in
the <head>
of each HTML page. Not referencing external style sheets
significantly speeds up the initial render of your site by avoiding
another round trip to your server as the initial render of a page is
blocked by external CSS files. This is a best practice suggested by
many groups including Google's AMP
project.
Each of the official starters supports this pattern. The code to make it happen is in brief:
// In your html.js
let css
// In development, css is injected by Javascript by the Webpack style-loader.
if (process.env.NODE_ENV === 'production') {
css = <style dangerouslySetInnerHTML={{ __html: require('!raw!./public/styles.css') }} />
}
// Then in your <head>
{css}
CSS modules are support by default for all files with .module.(css|less|scss|sass)
extension.
// Uses CSS Modules
import './my-component.module.css'
// Doesn't use CSS Modules
import './main.css'
You can modify Babel's behavior as normal by either providing a .babelrc
in
your site's root directory or by adding a "babel" section in your site's
package.json
. You can find out more about how to configure babel
here.
Gatsby by default will use your Babel configuration over the default if it can find it. Gatsby will automatically add react-hmre to your Babel config during development.
Note that if you want to use babel-plugin that is not provided by Gatsby, you will have to also add it to your package.json. You can use any babel-plugin that Gatsby packs as a dependency without having to add it to your own package.json:
- babel-plugin-add-module-exports
- babel-plugin-transform-object-assign
- babel-preset-es2015
- babel-preset-react
- babel-preset-stage-0
If you need to change the loader to be something completely custom. You will have to define your own webpack loader by following the steps described above.
Gatsby uses markdown-it to parse markdown files into HTML. By default Gatsy ships with only basic markdown support. You can extend the syntax (e.g. for mathematical equations) by installing markdown-it plugins.
If you want to do this you will need to use a custom markdown loader. You can
copy the one provided in the default starter here.
Add the relevant packages to your dependencies, including the markdown-it
plugins that you want to use and enable them with md.use(require('markdown-it-plugin-name'))
within the markdown loader file.
Gatsby supports automatically prefixing links with its prefixLink
helper function.
First set the prefix in your config file e.g. linkPrefix = '/your-project'
Then simply import the function and run all links in your site thorough it e.g.
import { prefixLink } from 'gatsby-helpers'
prefixLink('/')
// During development this will return "/"
// When deployed to example.github.io/your-project/ this will return "/your-project/"
Then finally, when building your site, run gatsby build --prefix-links
The built site is now in /public
. These files need copied to your
gh-pages
branch and committed and pushed. You can do this manually or
use the handy gh-pages
CLI tool.
Both the sample sites are deployed to github pages and use link prefixing. Read their source for more help: documentation/blog.
Jekyll has a comprehensive import tool for these and many other website tools.
Once your pages are converted to markdown, change the file extensions to
.md
from the .markdown
the tool outputs and then use them in your
site.