Test if links are external with jQuery / javascript?

Question

How do I test to see if links are external or internal? Please note:

1. I cannot hard code the local domain.
2. I cannot test for "http". I could just as easily be linking to my own site with an http absolute link.
3. I want to use jQuery / javascript, not css.

I suspect the answer lies somewhere in location.href, but the solution evades me.

Thanks!

Solution 1

var comp = new RegExp(location.host);

$('a').each(function(){
   if(comp.test($(this).attr('href'))){
       // a link that contains the current host           
       $(this).addClass('local');
   }
   else{
       // a link that does not contain the current host
       $(this).addClass('external');
   }
});

Note: this is just a quick & dirty example. It would match all href="#anchor" links as external too. It might be improved by doing some extra RegExp checking.

---

Update 2016-11-17

This question still got a lot of traffic and I was told by a ton of people that this accepted solution will fail on several occasions. As I stated, this was a very quick and dirty answer to show the principal way how to solve this problem. A more sophisticated solution is to use the properties which are accessible on a <a> (anchor) element. Like @Daved already pointed out in this answer, the key is to compare the hostname with the current window.location.hostname. I would prefer to compare the hostname properties, because they never include the port which is included to the host property if it differs from 80.

So here we go:

$( 'a' ).each(function() {
  if( location.hostname === this.hostname || !this.hostname.length ) {
      $(this).addClass('local');
  } else {
      $(this).addClass('external');
  }
});

State of the art:

Array.from( document.querySelectorAll( 'a' ) ).forEach( a => {
    a.classList.add( location.hostname === a.hostname || !a.hostname.length ? 'local' : 'external' );
});

Solution 2

I know this post is old but it still shows at the top of results so I wanted to offer another approach. I see all the regex checks on an anchor element, but why not just use window.location.host and check against the element's host property?

function link_is_external(link_element) {
    return (link_element.host !== window.location.host);
}

With jQuery:

$('a').each(function() {
    if (link_is_external(this)) {
        // External
    }
});

and with plain javascript:

var links = document.getElementsByTagName('a');
for (var i = 0; i < links.length; i++) {
    if (link_is_external(links[i])) {
        // External
    }
}

Solution 3

And the no-jQuery way

var nodes = document.getElementsByTagName("a"), i = nodes.length;
var regExp = new RegExp("//" + location.host + "($|/)");
while(i--){
    var href = nodes[i].href;
    var isLocal = (href.substring(0,4) === "http") ? regExp.test(href) : true;
    alert(href + " is " + (isLocal ? "local" : "not local"));
}

All hrefs not beginning with http (http://, https://) are automatically treated as local

Solution 4

var external = RegExp('^((f|ht)tps?:)?//(?!' + location.host + ')');

Usage:

external.test('some url'); // => true or false

Solution 5

Here's a jQuery selector for only external links:

$('a[href^="(http:|https:)?//"])') 

A jQuery selector only for internal links (not including hash links within the same page) needs to be a bit more complicated:

$('a:not([href^="(http:|https:)?//"],[href^="#"],[href^="mailto:"])')

Additional filters can be placed inside the :not() condition and separated by additional commas as needed.

http://jsfiddle.net/mblase75/Pavg2/

---

Alternatively, we can filter internal links using the vanilla JavaScript href property, which is always an absolute URL:

$('a').filter( function(i,el) {
    return el.href.indexOf(location.protocol+'//'+location.hostname)===0;
})

http://jsfiddle.net/mblase75/7z6EV/

View more solutions

Comments

[Matrym]

This works fairly well, although I'm going to hold off on an answer in case someone else has a more elegant solution. Out of curiosity, why do anchor links register as external?

[Sean Kinsey]

I'm pretty sure that this will not work with relative urls. attr is supposed to return the attribute, not the property (the property might be resolved, not the attribute).

[Sean Kinsey]

jsfiddle.net/zuSeh It is verified that this method does not work for relative urls.

[Savageman]

This answer is more accurate. Relatives URL are also important.

[Grimace of Despair]

If I'm not mistaken, "($|/" should actually be "($|/)" with a closing brace

[Grimace of Despair]

Just ran into this and want to confirm that this code indeed has a problem with relative urls. So +1 on Sean and Damian.

[feeela]

+1 for the regexp, though it does not completely solve the question

[framp]

This is close to the solution but you should also check if the href property does begin with location.protocol+'//'+location.host. Check this fiddle: jsfiddle.net/framp/Ag3BT/1

[1nfiniti]

Why are you doing this with a while loop? Seems to me like it would make more sense to use event delegation via $(document).on('click', 'a', function({}); and test the specific link that was clicked (at the point of being clicked). That way, you don't needlessly loop through all the links on the page, and it will allow for any elements added to the page via ajax after the initial DOM ready... There's actually a point to using jQuery sometimes (beyond being a "fanboy").

[rossipedia]

This won't work if you use protocol agnostic urls, ie: href="//somedomain.com/some-path"

[Adam Plocher]

Not all relative URLs start with /. You can reference something like images/logo.png which is one folder down from your current location. In that case you're referencing a relative path in your relative URL, it will be a different meaning in different directories on your site. /images/logo.png is an absolute path of whatever site it's running on (hence the relativity). Your code will not include relative paths like images/logo.png.

[Kevin B]

It will work for relative too if you use the href property instead of the href attribute.

[Gone Coding]

+1: OK there was a way to up-vote you again for this answer :)

[tremby]

This is the only answer which makes sense to me -- the others are way overengineered. Are there any arguments against this method?

[Daved]

I've seen people still viewing this question and solution so I wanted to link to my proposed solution below which should handle all checks without RegEx or relative issues: stackoverflow.com/a/18660968/2754848

[mikesir87]

First one doesn't work as "//code.jquery.com/jquery.min.js" is a completely legit URL, but not internal. The protocol and colon are not required, as the browser will use whatever the current site is using (a semi-sorta protocol relative URL).

[Mark Ribau]

The jQuery above works for me in Safari, and handles all of the issues that the others handle -- local anchors, relative URLs, cross-protocol URLs, etc. Note: example.com and example.com will be marked as internal; that may be important to you.

[dana]

This is the best answer. This makes use of the a.host property which is probably unknown to the average JavaScript developer (including myself before reading this).

[kthornbloom]

FYI, your external selector gives an error: Uncaught Error: Syntax error, unrecognized expression: a[href^="(http:|https:)?//"])

[Nicholas]

Perfect, just what I was looking for! Clean and simple without all the regex hassle which isn't needed.

[Laogeodritt]

jAndy's solution will also fail (false detection of internal link) in the case that an external URL contains the hostname, e.g. external-site.com/domaintools/original.hostname.com/stats or something like that. Overall I favour @Daved's solution as the most elegant and most direct check—far less error prone when the browser has already resolved the URL and you're checking its resolved hostname!

[jAndy]

Updated answer.

[Danny Coulombe]

Works flawlessly. Best answer.

[Andrew]

this will fail when port number is specified, also if "www" is left out

[Daved]

I'm a little late to the follow up here, but I would argue to @Andrew that "www" is an identifier that would indicate a different match. And though the domain might be the same, the host would be different with WWW and I think that's a valid condition. The port can be checked by using a ".port" comparison as well if they both != ''.

[Mo Ali]

@jAndy I believe that, correct me if I'm wrong, there is no need for using Array.from in last part of your answer, instead just wrap the expression in parens to become (document.querySelectorAll( 'a' )).forEach

[jAndy]

@MoAli At the time I wrote that, most browsers didn't provide the forEach method on the HTMLElement prototype. I still wouldn't dare to directly call that even today to be honest. Of course there are still some versions and mobile browsers which also don't support Array.from, but chances are way higher here, than .forEach being on the prototype.

[Rahul J. Rane]

@jAndy - Thanks for this solution. one more point is that if the href has relative URL having # in it, then it is not working. may you please give a solution for the same?

[Vincente]

@jAndy, why do you check "!this.hostname.length". In chrome it's always not empty even if href does not have hostname. Can it be empty in some browser?

[jAndy]

@Vincente I just did a quick check here on the SO site... If you filter all results from document.querySelectorAll( 'a' ).forEach( anchor => {}); for if( !anchor.hostname.length) you get lots of results. For instance anchors which only have a name or class but no href.

[Vincente]

@jAndy, thank you. I also tested it in IE11, link with href="/page1" does not have hostname property, unlike in Chrome.

[giovannipds]

@tremby Yes, I have just one argument against it, it depends on a DOM node. Imagine that I can have only the href string, I wouldn't be able to test it without creating a ghost node for it.

[tremby]

@Daved Looking back, maybe testing .origin of each would make most sense. That'll include the scheme and port.

[tremby]

@giovannipds, you can use new URL(myHrefString) instead of a new DOM node.

[Cezary Tomczyk]

Note to new URL(href, window.location): Argument of type 'Location' is not assignable to parameter of type 'string | URL | undefined'.

[Tyler2P]

A good answer will always include an explanation why this would solve the issue, so that the OP and any future readers can learn from it.

[Mahdi Bashirpour]

The description was great. Thankful

[Vah Run]

This is a worthy answer which could use its own space...