Easily enable fast AJAX navigation on any website (using pushState() + XHR)
Pjax is a standalone JavaScript module that uses AJAX (XmlHttpRequest) and pushState() to deliver a fast browsing experience.
It allows you to completely transform the user experience of standard websites (server-side generated or static ones) to make them feel like they are browsing an app, especially for users with low bandwidth connection.
No more full page reloads. No more multiple HTTP requests.
Pjax does not rely on other libraries, like jQuery or similar. It is written entirely in vanilla JS.
-
You can install Pjax from npm:
npm install pjax
-
You can also link directly to the bundle:
<script src="https://cdn.jsdelivr.net/npm/pjax@VERSION/pjax.js"></script>
Or the minified bundle:
<script src="https://cdn.jsdelivr.net/npm/pjax@VERSION/pjax.min.js"></script>
Pjax loads pages using AJAX and updates the browser's current URL using pushState() without reloading your page's layout or any resources (JS, CSS), giving a fast page load.
But under the hood, it's just ONE HTTP request with a pushState() call.
Obviously, for browsers that don't support history.pushState() Pjax gracefully degrades and does not do anything at all.
It simply works with all permalinks and can update all parts of the page you want (including HTML metas, title, and navigation state).
- It's not limited to one container, like jQuery-Pjax is.
- It fully supports browser history (back and forward buttons).
- It supports keyboard browsing.
- Automatically falls back to standard navigation for external pages (thanks to Captain Obvious's help).
- Automatically falls back to standard navigation for internal pages that do not have an appropriate DOM tree.
- You can add pretty cool CSS transitions (animations) very easily.
- It's around 4kb (minified and gzipped).
- It listens to every click on links you want (by default all of them).
- When an internal link is clicked, Pjax grabs HTML from your server via AJAX.
- Pjax renders the page's DOM tree (without loading any resources - images, CSS, JS...).
- It checks that all defined parts can be replaced:
- If the page doesn't meet the requirements, standard navigation is used.
- If the page meets the requirements, Pjax does all defined DOM replacements.
- Then it updates the browser's current URL using
pushState().
Pjax is fully automatic. You don't need to setup anything in the existing HTML. You just need to designate some elements on your page that will be replaced when you navigate your site.
Consider the following page.
<!DOCTYPE html>
<html>
<head>
<!-- metas, title, styles, ... -->
</head>
<body>
<header class="my-Header"><nav><!-- a .is-active is in there --></nav></header>
<section class="my-Content">
Sha blah <a href="/blah ">blah</a>.
</section>
<aside class="my-Sidebar">Sidebar stuff</aside>
<footer class="my-Footer"></footer>
<script src="onDomReadystuff.js"></script>
<script><!-- analytics --></script>
</body>
</html>We want Pjax to intercept the URL /blah, and replace .my-Content with the results of the request.
Oh and the <nav> (that contains a status marker somewhere) can be updated too (or stay the same, as you wish).
And also the <aside> please.
So we want to update [".my-Header", ".my-Content", ".my-Sidebar"], without reloading styles nor scripts.
We do this by telling Pjax to listen on a tags and use CSS selectors defined above (without forgetting minimal meta):
var pjax = new Pjax({ selectors: ["title", ".my-Header", ".my-Content", ".my-Sidebar"] })Now, when someone in a Pjax-compatible browser clicks "blah", the content of all selectors will be replaced with the one found in the "blah" content.
Magic! For real! There is no need to do anything server-side!
Differences with jQuery-pjax
- No jQuery dependency
- Not limited to a container
- No server-side requirements
- Works for CommonJS environment (Webpack/Browserify), AMD (RequireJS) or even globally
- Allow page transition with CSS animations
- Can be easily tweaked, since every method is public (and as a result, overridable)
Pjax only works with browsers that support the history.pushState() API.
When the API isn't supported, Pjax goes into fallback mode (and it just does nothing).
To see if Pjax is actually supported by your browser, use Pjax.isSupported().
Let's talk more about the most basic way to get started.
When instantiating Pjax, you can pass options in to the constructor as an object:
var pjax = new Pjax({
elements: "a", // default is "a[href], form[action]"
selectors: ["title", ".my-Header", ".my-Content", ".my-Sidebar"]
})This will enable Pjax on all links, and designate the part to replace using CSS selectors "title", ".my-Header", ".my-Content", ".my-Sidebar".
For some reason, you might want to just target some elements to apply Pjax behavior. In that case, you can do two different things:
- Use a custom selector like "a.js-Pjax" or ".js-Pjax a" depending on what you want.
- Override
Pjax.prototype.getElementsthat just basicallyquerySelectorAlltheelementsoption. In this function you just need to return aNodeList.
// use case 1
var pjax = new Pjax({ elements: "a.js-Pjax" })
// use case 2
Pjax.prototype.getElements = function() {
return document.getElementsByClassName(".js-Pjax")
}
var pjax = new Pjax()With this method, you can manually trigger loading of a URL:
var pjax = new Pjax()
// use case 1 (without options override)
pjax.loadUrl("/your-url")
// use case 2 (with options override)
pjax.loadUrl("/your-other-url", {timeout: 10})This method takes the raw response, processes the URL, then calls pjax.loadContent() to actually load it into the DOM.
It is passed the following arguments:
- responseText (string): This is the raw response text. This is equivalent to
request.responseText. - request (XMLHttpRequest): This is the XHR object.
- href (string): This is the URL that was passed to
loadUrl().
You can override this if you want to process the data before, or instead of, it being loaded into the DOM.
For example, if you want to check for a JSON response, you could do the following:
var pjax = new Pjax();
pjax._handleResponse = pjax.handleResponse;
pjax.handleResponse = function(responseText, request, href) {
if (request.responseText.match("<html")) {
pjax._handleResponse(responseText, request, href);
} else {
// handle response here
}
}CSS selector to use to retrieve links to apply Pjax to.
CSS selectors to replace. If a query returns multiples items, it will just keep the index.
Example of what you can do:
<!doctype html>
<html>
<head>
<title>Page title</title>
</head>
<body>
<header class="js-Pjax"></header>
<section class="js-Pjax">...</section>
<footer class="my-Footer"></footer>
<script>...</script>
</body>
</html>This example is correct and should work "as expected". If the current page and new page do not have the same amount of DOM elements, Pjax will fall back to normal page load.
Objects containing callbacks that can be used to switch old elements with new elements. Keys should be one of the defined selectors.
Examples:
var pjax = new Pjax({
selectors: ["title", ".Navbar", ".js-Pjax"],
switches: {
// "title": Pjax.switches.outerHTML // default behavior
".Navbar": function(oldEl, newEl, options) {
// here it's a stupid example since it's the default behavior too
oldEl.outerHTML = newEl.outerHTML
this.onSwitch()
},
".js-Pjax": Pjax.switches.sideBySide
}
})Callbacks are bound to Pjax instance itself to allow you to reuse it (ex: this.onSwitch())
Pjax.switches.outerHTML: default behavior, replace elements using outerHTMLPjax.switches.innerHTML: replace elements using innerHTML and copy className tooPjax.switches.replaceNode: replace elements using replaceChildPjax.switches.sideBySide: smart replacement that allows you to have both elements in the same parent when you want to use CSS animations. Old elements are removed when all children have been fully animated (animationEnd event triggered)
Your function can do whatever you want, but you need to:
- replace
oldEl's content withnewEl's content in some fashion - call
this.onSwitch()to trigger the attached callback.
Here is the default behavior as an example:
function(oldEl, newEl, pjaxRequestOptions, switchesClasses) {
oldEl.outerHTML = newEl.outerHTML
this.onSwitch()
}These are options that can be used during switch by switchers (for now, only Pjax.switches.sideBySide uses it).
This is very convenient when you use something like Animate.css
with or without WOW.js.
var pjax = new Pjax({
selectors: ["title", ".js-Pjax"],
switches: {
".js-Pjax": Pjax.switches.sideBySide
},
switchesOptions: {
".js-Pjax": {
classNames: {
// class added on the element that will be removed
remove: "Animated Animated--reverse Animate--fast Animate--noDelay",
// class added on the element that will be added
add: "Animated",
// class added on the element when it go backward
backward: "Animate--slideInRight",
// class added on the element when it go forward (used for new page too)
forward: "Animate--slideInLeft"
},
callbacks: {
// to make a nice transition with 2 pages as the same time
// we are playing with absolute positioning for element we are removing
// & we need live metrics to have something great
// see associated CSS below
removeElement: function(el) {
el.style.marginLeft = "-" + (el.getBoundingClientRect().width/2) + "px"
}
}
}
}
})Note that remove class include Animated--reverse which is a simple way to not have
to create duplicate transition for (slideIn + reverse => slideOut).
The following CSS will be required to make something nice:
/*
if your content elements doesn't have a fixed width,
you can get issue when absolute positioning will be used
so you will need that rules
*/
.js-Pjax { position: relative } /* parent element where switch will be made */
.js-Pjax-child { width: 100% }
/* position for the elements that will be removed */
.js-Pjax-remove {
position: absolute;
left: 50%;
/* transform: translateX(-50%) */
/* transform can't be used since we already use generic translate for the remove effect (eg animate.css) */
/* margin-left: -width/2; // made with js */
/* you can totally drop the margin-left thing from switchesOptions if you use custom animations */
}
/* CSS animations */
.Animated {
animation-fill-mode: both;
animation-duration: 1s;
}
.Animated--reverse { animation-direction: reverse }
.Animate--fast { animation-duration: .5s }
.Animate--noDelay { animation-delay: 0s !important; }
.Animate--slideInRight { animation-name: Animation-slideInRight }
@keyframes Animation-slideInRight {
0% {
opacity: 0;
transform: translateX(100rem);
}
100% {
transform: translateX(0);
}
}
.Animate--slideInLeft { animation-name: Animation-slideInLeft }
@keyframes Animation-slideInLeft {
0% {
opacity: 0;
transform: translateX(-100rem);
}
100% {
transform: translateX(0);
}
}To give context to this CSS, here is an HTML snippet:
<!doctype html>
<html>
<head>
<title>Page title</title>
</head>
<body>
<section class="js-Pjax">
<div class="js-Pjax-child">
Your content here
</div>
<!--
when switching will be made you will have the following tree
<div class="js-Pjax-child js-Pjax-remove Animate...">
Your OLD content here
</div>
<div class="js-Pjax-child js-Pjax-add Animate...">
Your NEW content here
</div>
-->
</section>
<script>...</script>
</body>
</html>Enable the use of pushState(). Disabling this will prevent Pjax from updating browser history.
However, there is almost no use case where you would want to do that.
Internally, this option is used when a popstate event triggers Pjax (to not pushState() again).
analytics (Function | Boolean, default: a function that pushes _gaq _trackPageview or sends ga pageview
Function that allows you to add behavior for analytics. By default it tries to track a pageview with Google Analytics (if it exists on the page). It's called every time a page is switched, even for history navigation.
Set to false to disable this behavior.
When set to an integer, this is the value (in px from the top of the page) to scroll to when a page is switched.
When set to an array of 2 integers ([x, y]), this is the value to scroll both horizontally and vertically.
Set this to false to disable scrolling, which will mean the page will stay in that same position it was before loading the new elements.
When set to true, attempt to restore the scroll position when navigating backwards or forwards.
When set to true, append a timestamp query string segment to the requested URLs in order to skip browser cache.
Enables verbose mode. Useful to debug page layout differences.
When set to true, clicking on a link that points to the current URL will trigger a full page reload.
When set to false, clicking on such a link will cause Pjax to load the
current page like any page.
If you want to add some custom behavior, add a click listener to the link,
and set preventDefault to true, to prevent Pjax from receiving the event.
Note: this must be done before Pjax is instantiated. Otherwise, Pjax's event handler will be called first, and preventDefault() won't be called yet.
Here is some sample code:
var links = document.querySelectorAll(".js-Pjax");
for (var i = 0; i < links.length; i++) {
var el = links[i]
el.addEventListener("click", function(e) {
if (el.href === window.location.href.split("#")[0]) {
e.preventDefault();
console.log("Link to current page clicked");
// Custom code goes here.
}
})
}
var pjax = new Pjax()(Note that if cacheBust is set to true, the code that checks if the href
is the same as the current page's URL will not work, due to the query string
appended to force a cache bust).
The timeout in milliseconds for the XHR requests. Set to 0 to disable the timeout.
Pjax fires a number of events regardless of how it's invoked.
All events are fired from the document, not the link that was clicked.
pjax:send- Fired after the Pjax request begins.pjax:complete- Fired after the Pjax request finishes.pjax:success- Fired after the Pjax request succeeds.pjax:error- Fired after the Pjax request fails. The request object will be passed along asevent.options.request.
send and complete are a good pair of events to use if you are implementing a loading indicator (eg: topbar)
document.addEventListener('pjax:send', topbar.show)
document.addEventListener('pjax:complete', topbar.hide)Pjax uses several custom headers when it makes and receives HTTP requests. If the requests are going to your server, you can use those headers for some meta information about the response.
Pjax sends the following headers with every request:
X-Requested-With: "XMLHttpRequest"X-PJAX: "true"X-PJAX-Selectors: A serialized JSON array of selectors, taken fromoptions.selectors. You can use this to send back only the elements that Pjax will use to switch, instead of sending the whole page. UseJSON.parse()server-side to deserialize it back to an array.
Pjax looks for the following headers on the response:
-
X-PJAX-URLorX-XHR-Redirected-ToPjax first checks the
responseURLproperty on the XHR object to check if the request was redirected by the server. Most browsers support this, but not all. To ensure Pjax will be able to tell when the request is redirected, you can include one of these headers with the response, set to the final URL.
Most of the time, you will have code related to the current DOM that you only execute when the DOM is ready.
Since Pjax doesn't magically re-execute your previous code each time you load a page, you need to add some simple code to achieve this:
function whenDOMReady() {
// do your stuff
}
whenDOMReady()
var pjax = new Pjax()
document.addEventListener("pjax:success", whenDOMReady)Note: Don't create the Pjax instance in the whenDOMReady function.
For my concern and usage, I js-Pjax-ify all body children, including stuff like navigation and footer (to get navigation state easily updated).
The attached behavior is re-executed each time a page is loaded, like in the snippet above.
If you want to just update a specific part (it's totally a good idea), you can just add the DOM-related code in a function and re-execute this function when "pjax:success" event is fired.
// do your global stuff
//... DOM ready blah blah
function whenContainerReady() {
// do your container related stuff
}
whenContainerReady()
var pjax = new Pjax()
document.addEventListener("pjax:success", whenContainerReady)A: There is a few things you need to do:
- wrap your Disqus snippet into a DOM element that you will add to the
selectorproperty (or just wrap it withclass="js-Pjax") and be sure to have at least the empty wrapper on each page (to avoid differences of DOM between pages) - edit your Disqus snippet like the one below
<script>
var disqus_shortname = 'YOURSHORTNAME'
var disqus_identifier = 'PAGEID'
var disqus_url = 'PAGEURL'
var disqus_script = 'embed.js'
(function(d,s) {
s = d.createElement('script');s.async=1;s.src = '//' + disqus_shortname + '.disqus.com/'+disqus_script;
(d.getElementsByTagName('head')[0]).appendChild(s);
})(document)
</script><div class="js-Pjax"><!-- needs to be here on every Pjax-ified page, even if empty -->
<!-- if (blah blah) { // eventual server-side test to know whether or not you include this script -->
<script>
var disqus_shortname = 'YOURSHORTNAME'
var disqus_identifier = 'PAGEID'
var disqus_url = 'PAGEURL'
var disqus_script = 'embed.js'
// here we will only load the disqus <script> if not already loaded
if (!window.DISQUS) {
(function(d,s) {
s = d.createElement('script');s.async=1;s.src = '//' + disqus_shortname + '.disqus.com/'+disqus_script;
(d.getElementsByTagName('head')[0]).appendChild(s);
})(document)
}
// if disqus <script> already loaded, we just reset disqus the right way
// see http://help.disqus.com/customer/portal/articles/472107-using-disqus-on-ajax-sites
else {
DISQUS.reset({
reload: true,
config: function () {
this.page.identifier = disqus_identifier
this.page.url = disqus_url
}
})
}
</script>
<!-- } -->
</div>Note: Pjax only handles inline <script> blocks for the container you are switching.
Clone this repository and run npm run example, which will open the example app in your browser.
- ⇄ Pull requests and ★ Stars are always welcome.
- For bugs and feature requests, please create an issue.
- Pull requests must be accompanied by passing automated tests (
npm test). If the API is changed, please update the Typescript definitions as well (pjax.d.ts).