使用PHP查找字符串中的链接。 与普通链接和youtube链接不同

Question

I have a string that contain links. I want my php to do different things with my links, depending on the url.

Answer:

function fixLinks($text)
{       
        $links = array();
        $text = strip_tags($text);    
        $pattern = '!(https?://[^\s]+)!'; 
        if (preg_match_all($pattern, $text, $matches)) {
            list(, $links) = ($matches);
        }

        $i = 0;
        $links2 = array();
        foreach($links AS $link) {
            if(strpos($link,'youtube.com') !== false) {
                $search = "!(http://.*youtube\.com.*v=)?([a-zA-Z0-9_-]{11})(&.*)?!";
                $youtube = '<a href="youtube.php?id=\\2" class="fancy">http://www.youtube.com/watch?v=\\2</a>';
                $link2 = preg_replace($search, $youtube, $link);
            } else {
                $link2 = preg_replace('@(https?://([-\w\.]+)+(:\d+)?(/([\-\w/_\.]*(\?\S+)?)?)?)@', '<a href="$1" target="_blank"><u>$1</u></a>', $link);
            }
            $links2[$i] = $link2;
            $i++;     
        }
        $text = str_replace($links, $links2, $text);
        $text = nl2br($text);

    return $text;
}

Answer 1

First of all, ditch eregi. It's deprecated and will disappear soon.

Then, doing this in just one pass is maybe a stretch too far. I think you'll be better off splitting this into three phases.

Phase 1 runs a regex search over your input, finding everything that looks like a link, and storing it in a list.

Phase 2 iterates over the list, checking whether a link goes to youtube (parse_url is tremendously useful for this), and putting a suitable replacement into a second list.

Phase 3: you now have two lists, one containing the original matches, one containing the desired replacements. Run str_replace over your original text, providing the match list for the search parameter and the replacement list for the replacements.

There are several advantages to this approach:

1. The regular expression for extracting links can be kept relatively simple, since it doesn't have to take special hostnames into account
2. It is easier to debug; you can dump the search and replace arrays prior to phase 3, and see if they contain what you expect
3. Because you perform all replacements in one go, you avoid problems with overlapping matches or replacing a piece of already-replaced text (after all, the replaced text still contains a URL, and you don't want to replace that again)