Strange behavior of preg_replace - php

Question

I want to switch the protocol of a link. If it is http, it should become https, and https should become http. I'm using pre_replace but something is going wrong.
Could someone look at my code and tell me what I am missing in my thinking process?
Here is the code:
$pattern = array(
0 => '/^(http\:)/',
1 => '/^(https\:)/'
);
$replace = array(
0 => 'https:',
1 => 'http:'
);
ksort($pattern);
ksort($replace);
$url = 'http://someurl.com';
echo $url."<br />";
$url = preg_replace($pattern, $replace, trim($url),1);
die($url);

Answer 1

You do not need to escape :, it is not a special character.
You don't need a capture group ().
You don't need to call ksort(), your arrays are already sorted by key when you declare them.
You appear to have your code replacing 'http' with 'https' AND replacing 'https' with 'http'. Why?
$url = preg_replace('/^http:/', 'https', trim($url)); will work just fine if you're simply looking to force to https.
edit
I still don't know why anyone would want to switch both http/https concurrently, but here you go:
function protocol_switcheroo($url) {
if( preg_match('/^http:/', $url) ) {
return preg_replace('/^http:/', 'https:', $url); // http to https
} else if( preg_match('/^https:/', $url) ) {
return preg_replace('/^https:/', 'http:', $url); // https to http
} else {
return $url; // for URIs with protocols other than http/https
}
}
You need to separate out the calls to replace so that you do not accidentally chain them like in the original code in the question.

Answer 2

The reason this isn't working for http -> https (but does work for https -> http) is that preg_replace() first changes the http to https with the first set of key/variable (0), but then immediately back to https -> http, because then the second set of variables (1) in each array is another valid match.

Answer 3

//$url = 'http://example.com/https://www';
$url = 'https://example.com/http://www';
$url = (0 === strpos($url, 'http:'))
? substr_replace($url, 's:', 4, 1)
: substr_replace($url, '', 4, 1);
echo $url;
This will convert HTTP -> HTTPS and HTTPS -> HTTP
It does not use a regex which would be slower, and does not use str_replace() which can inadvertently replace other portions of the URL. It will only replace the first prefix.
Breakdown : it looks to see if the URL begins with http: is it does it will replace the 5th character : with s: making it HTTPS. Otherwise it will replace the 5th character s with nothing making it HTTP.

Answer 4

Your URL is getting replaced twice. First, first expression matches and http://someurl.com becomes https://someurl.com. Then, second expression matches and https://someurl.com becomes http://someurl.com.
It's easier to see with this other example:
echo preg_replace(
array('/fox/', '/turtle/'),
array('turtle', 'sparrow'),
'fox', 1);
... that prints sparrow.

Answer 5

The problem you have is that preg_replace() does the two replacements one after the other, so that after the first one has run, the second one reverses the effect of the first one.
You need to specify both patterns in a single expression in order to have them run together.
I suggest using preg_replace_callback() instead of preg_replace(). With this, you can write a more complex output expression, which makes it easier to combine them into a single pattern. Something like this will do the trick:
$outputString = preg_replace_callback(
'/^(http|ftp)(s)?(:)/',
function($matches) {return $matches[1].($matches[2]=='s'?'':'s').':';},
$inputString
);
Hope that helps.
[EDIT] Edited the code so it works for ftp/ftps as well as http/https, after comment by OP.