short the url

goaman

Hello guys,
I wonder if there are any nice function to short very long urls such as: http://www.google.com/search?q=php+tutorials&sourceid=mozilla-search&start=0&start=0&ie=utf-8&oe=utf-8&client=firefox&rls=org.mozilla:en-US:official

http://www.google.com

TIA

jkurrle

I suppose you could cheeseball it by registering the long link at http://www.tinyurl.com, then calling the smaller supplied link...

jkurrle

Or if you want to just rip off the trailing dross, you could use a pair of explode() statements on the url. The first one uses the // as a delimiter and the second uses / as a delimiter. Then you just parse the pieces you want back together...

goaman

thanks for the reply jkurrle,
using explode will bring the wrong result, if the url is http://www.google.com/?blabla it may work, but if the url is http://www.google.com.ca/aaskd?asnd&asasd, will it work?

laserlight

hmm... perhaps you want to use [man]parse_url/man?

goaman

thanks for the reply laserlight..

I've checked php.net/parse_url, that example is little hard to understand, there are few examples ppl posted under the function on that site.. but they doesn't work well.. is there any other ways to do this trick?

jkurrle

Here are some examples using explode() and parse_url()

First, explode()

//creates a two item array, the header (http, https,etc.)
$start_array=explode("//",$url);

//now get the first item off the array and use it as the starting point.  If there isn't a
//second array item, don't do this.  Just use http: as your starter...
if ($start_array[1]=="")
   {
   $starter="http:";
   }
else
   {
   $starter=array_shift($start_array);
   }

// now that we have stripped off the first starting part, we want to get the url base
$base=explode("/",$start_array);
//popping off first item of the array...
$item=array_shift($start_array);
//Just in case the "?" immediately follows the url...
$test=explode("?";$item);
$item=array_shift($test);
//checking for prefaced login info to the site...
$test=explode(":",$item);
if ($test[1]<>"") $item=array_pop($test);

//now, we parse it back together
$returned_url=$starter."//".$item

Now let's look at parse_url():

//split the url into it's elements
$start_array=parse_url($url);
//now parse the pieces back together
$return_url=$start_array[scheme]."://".$start_array[host];

All in all, laser's solution is much better than mine. Wish I'd known that feature before...

goaman

SweeeeT.. jkurrle,
works without any problem.. clean.. thanks, really appreciate it.. I owe you a beer 😉

jkurrle

Remember that parse_url() will return a value of false if it cannot properly parse out a url. You probably want to code for that possibility...

goaman

oh.. I've got 2 questions.. 1) how it could fail, any idea?
2) how can I check if the url is wrongly parse out?

jkurrle

If the url is incorrectly formed, it may not parse out correctly. Examples might be "http:/bob.com" or perhaps even "www.bob.com" or "http://bob". If parse_url fails it returns a value of false. Therefore you need to run it through a statement like this:

//initialize $return_val to an empty value
$return_val="";
//set $start_array and run this loop if the parse_url gets a good url passed into it.
while ($start_array=parse_url($url))
   {
   //set the $return_url variable to a new, truncated path
   $return_url=$start_array[scheme]."://".$start_array[host]; 
   }
//If the parse_url got a bad url passed to it, $return_url is still empty so...
if ($return_url=="")
   {
   //set the $return_url to an error page...
   $return_url="http://www.bob.com/bogus_url.html";
   }

It's an ugly little piece of code, but I've never been that good at making code pretty. Laser could probably do it better...

Weedpacket

parse_url() will handle "http:/bob.com", and "www.bob.com" and "http://bob", and does so in compliance with the URI standard; of course, if the input isn't well-formed, there's no reason to expect the result to be what you want:

"http:/bob.com" parses into a scheme of "http" and a path of "/bob.com"
"www.bob.com" parses into a path of "www.bob.com"
"http://bob" parses into a scheme of "http" and a host of "bob".

All three have been parsed correctly.

One thing the return value never is is an empty string (or even an empty array).

So, assuming from the original post that what is wanted are the scheme and host:

$parsed_url = parse_url($url);
if(isset($parsed_url['scheme']) && isset($parsed_url['host']))
{
    $returned_url = $parsed_url['scheme'].'://'.$parsed_url['host'].'/';
}
else
{
    $returned_url = false;
}

$returned_url will be false if the string passed in didn't start with "scheme://host".

laserlight

The URL might be parsed by parse_url() and still lack the scheme and host components. If parse_url() fails, it not only returns false but emits a warning, so you may wish to suppress that.

if (($url_parts = @parse_url($url))
	&& array_key_exists('scheme', $url_parts)
	&& array_key_exists('host', $url_parts))
{
	$return_url = $url_parts['scheme'] . '://' . $url_parts['host'];
}
else
{
	$return_url = "http://www.example.com/bogus_url/";
}

Weedpacket

How'd you get it to emit a warning? The only way I found was to call it without an argument, or something stupid like a resource (and the code shouldn't be so badly written to allow that in the first place).

goaman

Thanks for the replies guys.. Now I've fixed the code as you guys suggested.. 🙂
Appreciate it very much.

laserlight

How'd you get it to emit a warning?

Something like this should do the trick:

parse_url('bogus://');

Weedpacket

Oh, right. No host when the URI said there was one.