matix.io

Using Asynchronous Requests to Mimic Single Page Apps in Django

November 10, 2019

I've created an example Django project for the tutorial here on Github to demonstrate what's discussed in this tutorial.

Single Page Apps (SPAs) built on frameworks like Vue, React, and Angular, create excellent user-experiences. The apps are built fully on the client side, so for any action the user performs, there will be immediate feedback. Instead of loading a new page when a link is clicked or a form is submitted, the SPA will perform HTTP requests asynchronously and update the view.

The resulting experience is much more pleasant than a traditional web application, where a link click or a form submit will cause the page to fully reload. On a slow connection, this often means a blank white browser while the next page finishes loading.

So why not build an SPA?

SPAs add complexity to a project. Instead of working on a single web application, you'll now have a separate backend API and frontend SPA. These two components interact with each-other and with two components instead of one, your project complexity increases. Bugs become harder to track down.

How can we make a traditional web application feel like a Single Page App?

Consider how a traditional web application works:

  1. An HTTP request is made to the server (GET if we're loading a page, POST if we're submitting a form)
  2. The server processes the request, and returns HTML, to be displayed in the web browser

The HTML that is returned to the browser is complete; that is, it starts with <html>, ends with </html>, contains the <head>, and the <body> and everything in between.

Let's say we have a simple website with two pages:

Page 1:



  
  
    
    

This is Page 1

and Page 2:



  
  
    
    

This is Page 2

(please don't comment on HTML semantics, this is just an example)

As you can see, the majority of the HTML structure is very similar here. The only thing that is unique to Page 1 and to Page 2 is the innerHTML of #content.

Why is that important? If we want to change the page asynchronously, all we have to do is fetch the innerHTML of #content from the server.

How does that work in general, for web applications?

The web-server will need to be able to return an HTML partial (instead of the complete HTML document) if requested. Using the example above, here's how the webapp might do this:

For a full-page load:


GET /page-1


  
  
    
    

This is Page 1

For a partial-page load:


GET /page-1?async

This is Page 1

This would be similar for form-submits, where a POST /page-3?async would also return a partial-HTML response. The particulars of the implementation are up to you, of course, but this is how the server should function in-general.

How would you hook your browser up to make the async requests?


function hook(scope) {
  scope.querySelectorAll('a').forEach(function (anchor) {

    // check if the anchor is internal
    if (anchor.href.indexOf(window.location.origin) === 0) {
      anchor.addEventListener('click', function (e) {

        // stop the anchor click from actually loading the next page
        e.preventDefault();

        // add our "async" GET parameter to the URL of the next page
        var url = anchor.href

        if (url.indexOf('?') === -1) {
          url += '?async';
        } else {
          url += '&async';
        }

        // asynchronously load the next page
        // if you want to display a loading spinner, now would be the time
        fetch(url)
          .then(function (res) {
            return res.text();
          })
          .then(function (res) {

            // inject the contents of the next page into our #content div
            document.querySelector('#content').innerHTML = res;

            // make sure that any links in the updated scope are also "hooked"
            hook(document.querySelector('#content'));

            // if you used a loading spinner, you could hide it now
          });
      });
    }
  });
}

document.addEventListener('DOMContentLoaded', function () {
  hook(document);
});

That's an example of how the Javascript would work. Of course in practice, it's likely to be more complicated for your project.

How does this work in Django?

We'll use a context processor to make sure that all of our views know whether to return a partial or full HTML response.

My main module is named "core"

In core/context_processors.py:


def check_async(request):
  ctx = {}
  if 'async' in request.GET:
    ctx['async'] = True
  return ctx

Now, make sure this context processor will run by making sure that your app name (core in my case) is added to INSTALLED_APPS in core/settings.py, and also by making sure that TEMPLATES['OPTIONS']['context_processors'] includes core.context_processors.check_async in core/settings.py.

Then, the variable should be getting passed down to your templates. Let's look at the templates folder.

In templates/base.html:


{% if not async %}

  
  
    
    
{% endif %} {% block content %}{% endblock %} {% if not async %}
{% endif %}

In templates/page-1.html:


{% extends 'base.html' %}

{% block content %}

Page 1

{% endblock %}

And in templates/page-2.html:


{% extends 'base.html' %}

{% block content %}

Page 2

{% endblock %}

Finally, make sure static/hook.js contains the Javascript defined above.

That's it! Now, your webapp will function like a traditional webapp, but if the user has Javascript enabled, those page transitions will be asynchronous.

Feel free to check out the sample Django project I've created that implements what we've discussed in this tutorial!

Final comments

Of course, if you want this to feel like a real Single Page App, you're going to need to modify that Javascript, but this should be enough to get you started to a smoother webapp.

Good luck!