vrid

Single Page Apps for GitHub Pages

Live example

This is a lightweight solution for deploying single page apps with GitHub Pages. You can easily deploy a React single page app with React Router <BrowserRouter />, like the one in the live example, or a single page app built with any frontend library or framework.

Why it's necessary

GitHub Pages doesn't natively support single page apps. When there is a fresh page load for a url like example.tld/foo, where /foo is a frontend route, the GitHub Pages server returns 404 because it knows nothing of /foo.

How it works

When the GitHub Pages server gets a request for a path defined with frontend routes, e.g. example.tld/foo, it returns a custom 404.html page. The custom 404.html page contains a script that takes the current url and converts the path and query string into just a query string, and then redirects the browser to the new url with only a query string and hash fragment. For example, example.tld/one/two?a=b&c=d#qwe, becomes example.tld/?p=/one/two&q=a=b~and~c=d#qwe.

The GitHub Pages server receives the new request, e.g. example.tld?p=/..., ignores the query string and hash fragment and returns the index.html file, which has a script that checks for a redirect in the query string before the single page app is loaded. If a redirect is present it is converted back into the correct url and added to the browser's history with window.history.replaceState(...), but the browser won't attempt to load the new url. When the single page app is loaded further down in the index.html file, the correct url will be waiting in the browser's history for the single page app to route accordingly. (Note that these redirects are only needed with fresh page loads, and not when navigating within the single page app once it's loaded).

A quick SEO note - while it's never good to have a 404 response, it appears based on Search Engine Land's testing that Google's crawler will treat the JavaScript window.location redirect in the 404.html file the same as a 301 redirect for its indexing. From my testing I can confirm that Google will index all pages without issue, the only caveat is that the redirect query is what Google indexes as the url. For example, the url example.tld/about will get indexed as example.tld/?p=/about. When the user clicks on the search result, the url will change back to example.tld/about once the site loads.

Usage instructions

For general information on using GitHub Pages please see GitHub Pages Basics, note that pages can be User, Organization or Project Pages
 

Basic instructions - there are two things you need from this repo for your single page app to run on GitHub Pages

1. Copy over the 404.html file to your repo as is
   • Note that if you are setting up a Project Pages site and not using a custom domain (i.e. your site's address is username.github.io/repo-name), then you need to set segmentCount to 1 in the 404.html file in order to keep /repo-name in the path after the redirect.
2. Copy the redirect script in the index.html file and add it to your index.html file
   • Note that the redirect script must be placed before your single page app script in your index.html file  

Detailed instructions - using this repo as a boilerplate for a React single page app hosted with GitHub Pages

1. Clone this repo ($ git clone https://github.com/rafrex/spa-github-pages.git)
2. Delete the .git directory (cd into the spa-github-pages directory and run $ rm -rf .git)
3. Instantiate the repository
   • If you're using this boilerplate as a new repository
     • $ git init in the spa-github-pages directory, and then $ git add . and $ git commit -m "Add SPA for GitHub Pages boilerplate" to initialize a fresh repository
     • If this will be a Project Pages site, then change the branch name from master to gh-pages ($ git branch -m gh-pages), if this will be a User or Organization Pages site, then leave the branch name as master
     • Create an empty repo on GitHub.com (don't add a readme, gitignore or license), and add it as a remote to the local repo ($ git remote add origin <your-new-github-repo-url>)
     • Feel free to rename the local spa-github-pages directory to anything you want (e.g. your-project-name)
   • If you're adding this boilerplate as the gh-pages branch of an existing repository
     • Create and checkout a new orphaned branch named gh-pages for your existing repo ($ git checkout --orphan gh-pages), note that the gh-pages branch won't appear in the list of branches until you make your first commit
     • Delete all of the files and directories (except the .git directory) from the directory of your existing repo ($ git rm -rf .)
     • Copy all of the files and directories (including hidden dot files) from the cloned spa-github-pages directory into your project's now empty directory ($ mv path/to/spa-github-pages/{.[!.],}* path/to/your-projects-directory)
     • $ git add . and $ git commit -m "Add SPA for GitHub Pages boilerplate" to instantiate the gh-pages branch
4. Set up a custom domain (optional) - see GitHub Pages instructions for setting up a custom domain
   • Update the CNAME file with your custom domain, don't include http://, but do include a subdomain if desired, e.g. www or your-subdomain
   • Update your CNAME and/or A record with your DNS provider
   • Run $ dig your-subdomain.your-domain.tld to make sure it's set up properly with your DNS (don't include http://)
5. Set up without using a custom domain (optional)
   • Delete the CNAME file
   • If you are creating a User or Organization Pages site, then that's all you need to do
   • If you are creating a Project Pages site, (i.e. your site's address is username.github.io/repo-name):
     • Set segmentCount to 1 in the 404.html file in order to keep /repo-name in the path after the redirect
     • Add your repo-name to the absolute path of assets in index.html
       • Change the bundle.js src to "/repo-name/build/bundle.js"
     • If you are using React Router you'll need to tell it to use the repo-name as the basename, for example <BrowserRouter basename="/repo-name" />
6. Run $ npm install to install React and other dependencies, and then run $ npm run build to update the build
7. $ git add . and $ git commit -m "Update boilerplate for use with my domain" and then push to GitHub ($ git push origin gh-pages for Project Pages or $ git push origin master for User or Organization Pages) - the example site should now be live on your domain
8. Create your own site
   • Write your own React components, create your own routes, and add your own style
     • Note that the example site is created with all inline styles and uses React Interactive for the links and other interactive components (there is no CSS except for a reset in index.html)
   • Change the title in index.html and the title in 404.html to your site's title
   • Remove the favicon links from the header of index.html
   • Change the readme, license and package.json as you see fit
   • For testing changes locally see development environment info below
   • To publish your changes to GitHub Pages run $ npm run build (this runs webpack -p for production) to update the build, then $ git commit and $ git push to make your changes live

Development environment

I have included webpack-dev-server for testing changes locally. It can be accessed by running $ npm start (details below), or you can use your own dev setup by running $ webpack and serving the index.html file and the 404.html file for 404s. Note that webpack-dev-server automatically creates a new bundle whenever the source files change and serves the bundle from memory, so you'll never see the bundle as a file saved to disk.

• $ npm start runs the start script in package.json, which runs the command $ webpack-dev-server --devtool eval-source-map --history-api-fallback --open
  • -devtool eval-source-map is for generating source maps in while in development
  • --history-api-fallback allows for frontend routing and will serve index.html when the requested file can't be found
  • --open will open automatically open the site in your browser
• webpack-dev-server will serve index.html at http://localhost:8080 (port 8080 is the default). Note that you must load the index.html from the server and not just open it directly in the browser or the scripts won't load.

Miscellaneous

• The .nojekyll file in this repo turns off Jekyll for GitHub Pages
• Need form submission on your static site? Use Formspree
• One of the great things about the GitHub Pages CDN is that all files are automatically compressed with gzip, so no need to worry about compressing your JavaScript, HTML or CSS files for production