从字符串中的链接中删除部件

Question

How can I change all links in a string from:

...<p><a href="https://www.somesite.com/url?q=http://www.someothersite.se/&amp;q1=xxx&q2=xxx">Some text</a>...

Into:

...<p><a href="http://www.someothersite.se/">Some text</a>...

"..." means that there are lots of other code. Also there are multiple links like this in the string. All links looks like this.

Answer 1

Working solution:

$regex = <<<EOF
%(<[aA]\s[^>]*href=['"])([^"']+url\?q=([A-z]+:\/{2}[^"'&]+)[^"']*)(["'][^>]*>)%im
EOF;

$replacement = '$1$3$4';

$html = <<<EOF
...<p><a href="https://www.somesite.com/url?q=http://www.secondsite.se/&amp;q1=xxx&q2=xxx">Some text</a>...
...<p><a class="lnk" href="https://www.somesite.com/url?q=http://www.thirdsite.se" id="lnk">Some text</a>...
...<p><a class="lnk2" href="https://www.somesite.com/">Some text</a>...
EOF;

$new_html = preg_replace($regex, $replacement, $html);

Regex explained:

(                     - Group 1 (tag A from beginning to href parameter)
  <[aA]\s             - Match <a or <A followed by white character
  [^>]*               - Match anything after it except > because we want to match all parameters (like class, id etc.)
  href=['"]           - match href parameter with equal sign and ' or " after it
)                     - End group 1
(                     - Group 2 (content of href parameter)
    [^"']+            - everything that is not ' or "
    url\?q=           - url?q=
    (                 - Group 3 (URL we are really interested in)
        [A-z]+:\/{2}  - match protocol of the url http:// https:// ftp:// etc.
        [^"'&]+       - match anything except ' " or &. those characters represents end of the url we are interested in.
    )                 - End group 3
    [^"']*            - Anything except " or ' - this represents end of href parameter
)                     - End group 2
(                     - Group 4 - rest of the tag
    ["']              - " or ' closing href parameter
    [^>]*             - anything except > so we match rest of the tag
    >                 - finally we match closing character >
)                     - End group 4

Then we just replace whole thing with just groups 1, 3 and 4.