Twitter样式URL正则表达匹配

Question

I'm trying to achieve very lax Regex match for a chat client using PHP.

The chat client must be able to pick up both complete and incomplete URLS.

For example:

http://www.example.com or www.example.com or example.com

I have set up a preg_replace that tries to achieve this:

$find = array("/([\w]+:\/\/[\w-?&;#~=\.\/\@]+[\w\/])/is","/(^(?!http:\/\/)[a-zA-Z0-9\-\.]+\.[a-zA-Z]{2,4}(\/?\S*)?)/is");
$replace = array( "<a target=\"_blank\" href=\"http://\\1\">\\1</a>","<a target=\"_blank\" href=\"\\1\">\\1</a>");
$output = preg_replace($find, $replace, $input); 

So, the aim is to 1st find "complete" URLs with the protocol, then try to find "lazy" URLs that do not have the protocol.

Currently it works great for the "complete" URLs, but the "lazy" URLs do not get picked up.

Any help will be greatly appreciated.

Thanks.

Answer 1

I set up something similar a while ago. My thinking was ... anything that starts with a protocol identifier or a "www" is a URL, plus anything that matches a domain that ends in a valid TLD (two letters, or a known gTLD) if it's followed by a path. Domains by themselves are just domains.

$gtlds="com|net|org|biz|edu|gov|int|pro|xxx";
$gtlds+="|aero|arpa|asia|coop|museum|name|travel";
#$gtlds+="|xn-[a-z0-9]+";
$a = array(
  '/(f|ht)tps?:\/\/[^ ]+/',
  '/(ftp|www)\.[a-z0-9.-]+(/[^ ]*)/',
  "/([a-z0-9][a-z0-9-]*\.)+([a-z]{2}|$gtlds)\/[^ ]*/"
);

Note that I'm an old-school regexp user, so this is ERE, not that fancy PREG stuff all the kids are using these days.

The absurdly long list of gTLDs is from IANA . I've updated it so it's valid as of the time of this answer except for .XN--* . You can include the list of TLDs that start with .XN-- if you like, either with a pattern or by matching them directly and growing the $gtlds variable. I have never encountered any problems caused by simply ignoring their existence, so that's my strategy.

The above REs worked for my specific use. I make no claim that they'll work for every case that is not mine. (For example, they will include trailing quotes, if a domain or URL is quoted. That was never something I had to deal with, so I didn't deal with it.)

Note that when you're doing your replacement, while you want your generator anchor's HREF attribute to be the URL that you match or generate, you probably want to leave the original text as-is for purposes of layout and display.

And depending how you use these, word boundaries may be helpful ... but you already know how to do those.

Answer 2

So after hours of toiling with it.. I managed to find a relatively easy way to match both http:// URLs and www. URLs in order to turn them into anchors.

This is the final solution:

$output = preg_replace("/\b((http(s)?:\/\/)?(www\.[a-zA-Z0-9\/\\\:\?\%\.\&\;=#\-\_\!\+\~\,]*))/is","<a target=\"_blank\" href=\"http$3://$4\">$0</a>",$output);

Thanks to tamouse for the regex.