Fix handling of parentheses in URLs
| | |
| | | // Support unicode/punycode in top-level domain part |
| | | $utf_domain = '[^?&@"\'\\/()\s\r\t\n]+\\.?([^\\x00-\\x2f\\x3b-\\x40\\x5b-\\x60\\x7b-\\x7f]{2,}|xn--[a-zA-Z0-9]{2,})'; |
| | | $url1 = '.:;,'; |
| | | $url2 = 'a-zA-Z0-9%=#$@+?!&\\/_~\\[\\]{}\*-'; |
| | | $url2 = 'a-zA-Z0-9%=#$@+?!&\\/_~\\[\\]\\(\\){}\*-'; |
| | | |
| | | $this->link_pattern = "/([\w]+:\/\/|\W[Ww][Ww][Ww]\.|^[Ww][Ww][Ww]\.)($utf_domain([$url1]?[$url2]+)*)/"; |
| | | $this->mailto_pattern = "/(" |
| | |
| | | // "http://example.com/?a[b]=c". However we need to handle |
| | | // properly situation when a bracket is placed at the end |
| | | // of the link e.g. "[http://example.com]" |
| | | // Yes, this is not perfect handles correctly only paired characters |
| | | // but it should work for common cases |
| | | |
| | | if (preg_match('/(\\[|\\])/', $url)) { |
| | | $in = false; |
| | | for ($i=0, $len=strlen($url); $i<$len; $i++) { |
| | |
| | | } |
| | | } |
| | | |
| | | // Do the same for parentheses |
| | | if (preg_match('/(\\(|\\))/', $url)) { |
| | | $in = false; |
| | | for ($i=0, $len=strlen($url); $i<$len; $i++) { |
| | | if ($url[$i] == '(') { |
| | | if ($in) |
| | | break; |
| | | $in = true; |
| | | } |
| | | else if ($url[$i] == ')') { |
| | | if (!$in) |
| | | break; |
| | | $in = false; |
| | | } |
| | | } |
| | | |
| | | if ($i < $len) { |
| | | $suffix = substr($url, $i); |
| | | $url = substr($url, 0, $i); |
| | | } |
| | | } |
| | | |
| | | return $suffix; |
| | | } |
| | | } |
| | |
| | | array('Start http://localhost/?foo End', 'Start <a href="http://localhost/?foo" target="_blank">http://localhost/?foo</a> End'), |
| | | array('www.domain.tld', '<a href="http://www.domain.tld" target="_blank">www.domain.tld</a>'), |
| | | array('WWW.DOMAIN.TLD', '<a href="http://WWW.DOMAIN.TLD" target="_blank">WWW.DOMAIN.TLD</a>'), |
| | | array('[http://link.com]', '[<a href="http://link.com" target="_blank">http://link.com</a>]'), |
| | | array('http://link.com?a[]=1', '<a href="http://link.com?a[]=1" target="_blank">http://link.com?a[]=1</a>'), |
| | | array('http://link.com?a[]', '<a href="http://link.com?a[]" target="_blank">http://link.com?a[]</a>'), |
| | | array('(http://link.com)', '(<a href="http://link.com" target="_blank">http://link.com</a>)'), |
| | | array('http://link.com?a(b)c', '<a href="http://link.com?a(b)c" target="_blank">http://link.com?a(b)c</a>'), |
| | | array('http://link.com?(link)', '<a href="http://link.com?(link)" target="_blank">http://link.com?(link)</a>'), |
| | | ); |
| | | } |
| | | |