I have a bunch of domain names coming in like this:
http://subdomain.example.com (example.com is always example.com, but the subdomain varies).
I need "subdomain".
Could some kind person who had the patience to learn regex help me out?
1
-
Yes, you can have string.string.domain.gtldDallas Clark– Dallas Clark2014-12-01 04:28:46 +00:00Commented Dec 1, 2014 at 4:28
Add a comment
|
7 Answers
The problem with the above regex is: if you do not know what the protocol is, or what the domain suffix is, you will get some unexpected results. Here is a little regex accounts for those situations. :D
/(?:http[s]*\:\/\/)*(.*?)\.(?=[^\/]*\..{2,5})/i //javascript
This should always return your subdomain (if present) in group 1. Here it is in a Javascript example, but it should also work for any other engine that supports positive look-ahead assertions:
// EXAMPLE of use
var regex = /(?:http[s]*\:\/\/)*(.*?)\.(?=[^\/]*\..{2,5})/i
, whoKnowsWhatItCouldBe = [
"www.mydomain.com/whatever/my-site" //matches: www
, "mydomain.com"// does not match
, "http://mydomain.com" // does not match
, "https://mydomain.com"// does not match
, "banana.com/somethingelse" // does not match
, "https://banana.com/somethingelse.org" // does not match
, "http://what-ever.mydomain.mu" //matches: what-ever
, "dev-www.thisdomain.com/whatever" // matches: dev-www
, "hot-MamaSitas.SomE_doma-in.au.xxx"//matches: hot-MamaSitas
, "http://hot-MamaSitas.SomE_doma-in.au.xxx" // matches: hot-MamaSitas
, "пуст.пустыня.ru" //even non english chars! Woohoo! matches: пуст
, "пустыня.ru" //does not match
];
// Run a loop and test it out.
for ( var i = 0, length = whoKnowsWhatItCouldBe.length; i < length; i++ ){
var result = whoKnowsWhatItCouldBe[i].match(regex);
if(result != null){
// YAY! We have a match!
} else {
// Boo... No subdomain was found
}
}
7 Comments
mastaBlasta
this is clearly the best answer because it accounts for protocol, none/multiple subdomains, and it is domain independent.
Pandem1c
I would wonder the desired output of multiple subdomains... Would you want it to return
one.two or just one? I suppose we could tweak the regex to pull all (.\.) groups prior to the domain... maybe later
Wesley Smith
Nice job, +1.
(file:\/\/|http:\/\/|https:\/\/|\/\/)*(.*?)\.(?=[^\/]*\..{2,5}) if you want to allow other prorocols
Ron
This worked in google analytics to filter by subdomain - had to drop the leading / and the trailing /i
(?:http[s]*\:\/\/)*(.*?)\.(?=[^\/]*\..{2,5})Pandem1c
@WebandFlow, The result
SomE_doma-in is the subdomain of your example, is it not? I am unclear what you had expected, vs. what you got. I personally expect SomE_doma-in as the match... |
/(http:\/\/)?(([^.]+)\.)?domain\.com/
Then $3 (or \3) will contain "subdomain" if one was supplied.
If you want to have the subdomain in the first group, and your regex engine supports non-capturing groups (shy groups), use this as suggested by palindrom:
/(?:http:\/\/)?(?:([^.]+)\.)?domain\.com/
3 Comments
Draemon
True. He didn't mention language/library so I wanted to make the regex as portable as possible - not sure if all implementations allow non-capturing groups.
Dallas Clark
What if you don't know what
domain is?Pandem1c
@DallasClark In that case, I would recommend my answer below
Purely the subdomain string (result is $1):
^http://([^.]+)\.domain\.com
Making http:// optional (result is $2):
^(http://)?([^.]+)\.domain\.com
Making the http:// and the subdomain optional (result is $3):
(http://)?(([^.]+)\.)?domain\.com
Comments
It should just be
\Qhttp://\E(\w+)\.domain\.com
The sub domain will be the first group.
Comments
#!/usr/bin/perl
use strict;
use warnings;
my $s = 'http://subdomain.example.com';
my $subdomain = (split qr{/{2}|\.}, $s)[1];
print "'$subdomain'\n";
Comments
To math sub domains with dot character in it, I used this one
https?:\/\/?(?:([^*]+)\.)?domain\.com
to get all matching characters after protocol until domain.
https://sub.domain.com (sub)
https://sub.sub.domain.com (sub.sub) ...
Comments
1st group of
http://(.*).example.com
1 Comment
Sinan Ünür
Forgetting, of course, that
.* will match an empty string and, more importantly, that the period stands for any character.