Question

Being a designer with limited coding experience, I always considered regex to be some kind of black magic. Recently, I’ve been reading up a bit – and I’m getting pretty intrigued by its possibilities. So I decided to give it a first try in my current php project.

I want to find all URLs of the following structure:

http://[any subdomain, only a-z].domain.com/[any subfolder, can contain a-z,A-Z,0-9,- and _]/

Examples:

• http://sub.domain.com/A4_3/
• http://othersub.domain.com/c_44/

My regex:

http://[a-z]*\.domain\.com/[A-Za-z0-9\_\-]*/

My questions:

• The regex is working, but I’m just wondering whether it could be improved. For instance, I tried adding case insencitive with (i?) but couldn’t get it working.
• I could only get it working in php if I added double quotes at start and end of the expression, why is that? $ref = preg_replace('"http://[a-z]*\.domain\.com/[A-Za-z0-9\_\-]*/"','',$ref);

Answer 1

In php regex must be delimited, usualy by / but it can be almost any character.

The reason why your second attempt works is because you’re using " as delimiter.

To be case insensitive you have to put the flag i after the second delimiter:

$ref = preg_replace('"http://[a-z]*\.domain\.com/[A-Za-z0-9\_\-]*/"i','',$ref);
                                                           here ___^

With the i flag there’re no needs for [a-zA-Z] and [a-z] would suffice. Moreover you don’t need to escape the underscore _ in the character class and not the dash - if it’s placed at the first or the last position within the character class

$ref = preg_replace('"http://[a-z]*\.domain\.com/[a-z0-9_-]*/"i','',$ref);

Note that [a-zA-Z0-9_] can be abbreviated as \w, then your code could look like:

$ref = preg_replace('"http://[a-z]*\.domain\.com/[\w-]*/"i','',$ref);

Take into account that * stands for 0 or more times, so your regex will match something like:

http://.domain.com//

Change * by + that means 1 or more time to be sure you have at least one char for the subdomain and one char for subfolder:

$ref = preg_replace('"http://[a-z]+\.domain\.com/[\w-]+/"i','',$ref);

And then " is unusual for delimiter, use for example #, ~ or !:

$ref = preg_replace('#http://[a-z]+\.domain\.com/[\w-]+/#i','',$ref);