php regex problem with url

[temex]

Anybody familiar with regex, please keep reading:

On one of my databases the external links are saved in this format:

[url=www.google.com/index/]Link[/url]

It works fine if one of the following are used:

• www.link.com/dir/
• www.link.com/dir/dir/
• www.link.com/dir/filename.gif

BUT, it won't recognize a bare domain in this format:

• www.link.com

Can anybody see the problem right out? Here's the regex:

$table_field = preg_replace(
"/\[url=www\.([a-z0-9\-]+\.(com|org|net|us|eu|fi))(\/[a-zA-Z0-9\-\_\/]+)(\.(jpg|gif))?\](.+?)\[\/url\]/",
"<a href='http://www.$1$3$4'>$6</a>",
$table_field);

Thanks in advance!

Have you tried removing the "www" parts? Most sites don't need it anymore, and some sites might look something like blogs.domain.com. It may be causing some weird conflict....

It looks fine from what I can tell.

[shayward]

I'm not much good with php but I like regular expressions a lot. I analyzed your regex and I think I see the problem. I'm assuming * means "zero or more" in php regex.

\[url=www\.([a-z0-9\-]+\.(com|org|net|us|eu|fi))(\/[a-zA-Z0-9\-\_\/]+)*(\.(jpg|gif))?\](.+?)\[\/url\]

What you were doing is saying that you always require at least one / followed by some various characters. In the case of the last URL you posted, there was no / followed by extra characters. So by adding the * after (\/[a-zA-Z0-9\-\_\/]+) it made it optional.

Now with all of that said, I don't believe the pattern you are using will properly validate all URLs. Even with the * to solve one problem, this URL won't validate:

java.sun.com

Have a look at the specification at http://www.w3.org/Addressing/rfc1738.txt and you'll be surprised how much simpler you could make your regex.

[shayward]

Here's perhaps something to have a look at:

PHP Code:

((?<Protocol>[a-z]+)://)?
(?<HostName>
    (?<Label>[a-z0-9][[a-z0-9-]*)
    ([.](?<Label>[a-z0-9][[a-z0-9-]*))*
)
([:](?<Port>d{1,5}))?
(?<Path>([/]([a-z0-9-.]|(%[0-9a-f]{2}))*)*|[/])? 


I haven't tested a huge variety of host names and protocols but it supports things like:

http://sun.java.com:90/dir1/dir2/dir3/pic.jpg
http://sun.java.com:90/dir1/dir2/dir3
http://sun.java.com:90/
http://sun.java.com
ftp://downloads.somesite.org
http://sun.java.com/special%0Fcaracters

So it supports:
- capturing groups... feel free to strip them out if you don't need them
- protocol (http, ftp) optionally
- any number of labels in the host name (but not imposing length limits)
- port numbers optionally
- path including special characters (%xx) where "x" is a hex digit.

I didn't do query strings. Also, this is compliant with .NET (C#, VB, ASP.NET) so you might need to add some escape characters for php.

[temex]

Thanks Shayward!

Honestly I was not expecting much of an answer from this forum, but I'm lazy to sign up for new forums, so I gave it a shot...fortunately!

I'm intentionally avoiding any sun.java.com or :90 -type urls and especially any ftp urls. All I'm allowing is the basic www.java.com or with some subdirs and with/without jpg|gif file. Nothing else.

But thanks a thousands for your input! I hope that * would solve the problem. I'll give it a shot in the morning.

[shayward]

I'm glad it helped

You never know what someone on CMN is going to come up with

[temex]

Worked out perfect! Thanks shayward! The * was the fix.

It's always easier to invent something if you have sample to look at