regex to trim down subdomain in the url - php

Question

I have a regexp that match to something like : wiseman.google.com.jp, me.co.uk, paradise.museum, abcd-abc.net, www.google.jp, 12345-daswe-23dswe-dswedsswe-54eddss.info, del.icio.us, jo.ggi.ng, all of this is from a textarea value.
used regexp (in preg_match_all($regex1, $str, $match)) to get the above values: /(?:[a-zA-Z0-9]{2,}\.)?[-a-zA-Z0-9]{2,}\.[a-zA-Z0-9]{2,7}(?:\.[-a-zA-Z0-9]{2,3})?/
Now, my question is : how can I make the regexp to trim down the "wiseman.google.com.jp" into "google.com.jp" and "www.google.jp" into "google.jp"?
I am thingking to make a second preg_match($regex2, $str, $match) function with each value coming from the preg_match_all function.
I have tried this regexp in $regex2 : ([-a-zA-Z0-9\x{0080}-\x{00FF}]{2,}+)\.[a-zA-Z0-9\x{0080}-\x{00FF}]{2,7}(?:\.[-a-zA-Z0-9\x{0080}-\x{00FF}]{2,3})? but it doesn't work.
Any inputs? TIA
here is my little solution :
preg_match_all($regex, $str, $matches, PREG_PATTERN_ORDER);
$arrlength=count($matches[0]);
for($x=0;$x<$arrlength;$x++){
$dom = $matches[0][$x];
$newstringcount = substr_count($dom, '.'); // this line is to count how many "." present in the string.
if($newstringcount == 3){ // if there are 3 '.' present in the string = true
$pos = strpos($dom, '.', 0); // this line is to find the first occurence of the '.' in the string
$find = substr($dom, $pos+1); //this line is to get the value after the first occurence of the '.' in the string
echo $find;
}else if($newstringcount == 2){
if ($pos = strpos($dom,'www.') !== false) {
$find = substr($dom, $pos+3);
echo $find;
}else{
echo $dom;
}
}else if($newstringcount == 1){
echo $dom;
}
echo "<br>";
}

Answer 1

(Caution: this answer will only fit your needs if you HAVE to use regex or you're somewhat... desperate...)
What you want to achieve isn't possible with general rules due to domains like .com.jp or .co.uk.
The only general rule one can find is:
When read from right to left there are one or two TLDs followed by one second level domain
Thus, we have to whitelist all available TLDs. I think i'll call the following the "domain-kraken".
Release the kraken!
([a-z0-9\-]{2,63}(?:\.(?:a(?:cademy|ero|rpa|sia|[cdefgilmnoqrstuwxz])|b(?:ike
|iz|uilders|uzz|[abdefghijlmnoqrstvwyz])|c(?:ab|amera|amp|areers|at|enter|eo
|lothing|odes|offee|om(?:pany|puter)?|onstruction|ontractors|oop|
[acdfghiklmnoruvwxyz])|d(?:iamonds|irectory|omains|[ejkmoz])|e(?:du(?:cation)?
|mail|nterprises|quipment|state|[ceghrstu])|f(?:arm|lorist|[ijkmor])|g(?:allery|
lass|raphics|uru|[abdefghlmnpqrstuwy])|h(?:ol(?:dings|iday)|ouse|[kmnrtu])|
i(?:mmobilien|n(?:fo|stitute|ternational)|[delmnoqrst])|j(?:obs|[emop])|
k(?:aufen|i(?:tchen|wi)|[eghimnprwxyz])|l(?:and|i(?:ghting|mo)|[abcikrstuvy])|
m(?:anagement|enu|il|obi|useum|[acdefghklmnopqrstuvwxyz])|n(?:ame|et|inja|
[acefgilopruz])|o(?:m|nl|rg)|p(?:hoto(?:graphy|s)|lumbing|ost|ro|[aefghklmnrstwy])|
r(?:e(?:cipes|pair)|uhr|[eosuw])|s(?:exy|hoes|ingles|ol(?:ar|utions)|upport|
ystems|[abcdeghijklmnorstuvxyz])|t(?:attoo|echnology|el|ips|oday|
[cdfghjklmnoprtvwz])|u(?:no|[agkmsyz])|v(?:entures|iajes|oyage|[aceginu])|
w(?:ang|ien|[fs])|xxx|y(?:[et])|z(?:[amw]))){1,2})$
Use it together with the i and m flags.
This supposes your data is on mutiple lines.
In case your data is seperated by a ,, change the last character in the regex ($) to ,? and use the g and i flags.
Demos are available on regex101 and debuggex.
(Both of the demos have an explanation: regex101 describes it with text while debuggex visualizes the beast)
A list of available TLDs can be found at iana.org, the used TLDs in the regex are as of January 2014.