preg_match_all an array of items, then for each matches - php

Question

I am editing some Interspire Email code. Currently the program goes through the HTML of the email before sending, and looks for 'a href' code, to replace the links. I want it to also go through and get form action="" and replace the urls in them (it does not currently). I think I can use the regex from this stack post:
PHP - Extract form action url from mailchimp subscribe form code using regex
but I'm having some difficulty wrapping my head around how to handle the arrays. The current code that just does the 'a href=' is below:
preg_match_all('%<a.+(href\s*=\s*(["\']?[^>"\']+?))\s*.+>%isU', $this->body['h'], $matches);
$links_to_replace = $matches[2];
$link_locations = $matches[1];
arsort($link_locations);
reset($links_to_replace);
reset($link_locations);
foreach ($link_locations as $tlinkid => $url) {
// so we know whether we need to put quotes around the replaced url or not.
$singles = false;
$doubles = false;
// make sure the quotes are matched up.
// ie there is either 2 singles or 2 doubles.
$quote_check = substr_count($url, "'");
if (($quote_check % 2) != 0) {
...
I know (or I think I know), that I need to replace preg_match_all with:
preg_match_all(array('%<a.+(href\s*=\s*(["\']?[^>"\']+?))\s*.+>%isU', '|form action="([^"]*?)" method="post" id="formid"|i'), $this->body['h'], $matches);
but then how are the '$matches' handled?
$links_to_replace = $matches[2];
$link_locations = $matches[1];
does not still hold true does it? Is it possible to do what I'm thinking? Or would I need to write another function just to handle the 'forms action=' seperate from the 'a href'

Answer 1

A suggestion:
$pattern = <<<'LOD'
~
(?| # branch reset feature: allows to have the same named
# capturing group in an alternation. ("type" here)
<a\s # the link case
(?> # atomic group: possible content before the "href" attribute
[^h>]++ # all that is not a "h" or the end of the tag ">"
|
\Bh++ # all "h" not preceded by a word boundary
|
h(?!ref\s*+=) # all "h" not followed by "ref=" or "ref ="
)*+ # repeat the atomic group zero or more times.
(?<type> href )
| #### OR ####
<form\s # the form case
(?> # possible content before the "action" attribute. (same principle)
[^a>]++
|
\Ba++
|
a(?!ction\s*+=)
)*+
(?<type> action )
)
\s*+ = \s*+ # optional spaces before and after the "=" sign
\K # resets all on the left from match result
(?<quote> ["']?+ )
(?<url> [^\s"'>]*+ )
\g{quote} # backreference to the "quote" named capture (", ', empty)
~xi
LOD;
Note that this pattern will only match the url with possible quotes. However, the attribute name will be stored inside the named capture group "type" if you need it.
Then you can use all of this with:
$html = preg_replace_callback($pattern,
function ($m) {
$url = $m['url'];
$type = lowercase($m['type']);
$quote = $m['quote'];
// make what you want with the url, type and quotes
return $quote . $url . $quote;
}, $html);