Extend PHP regex to cover "srcset" and "style" attributes

Question

I've created a plugin that turn all links into protocol-relative URLs (removing http: and https:) based off the tags and attributes that I list in the $tag and $attribute variables. This is part of the function. To save space, the rest of the code can be found here.

$tag = 'a|base|div|form|iframe|img|link|meta|script|svg';
$attribute = 'action|content|data-project-file|href|src|srcset|style';
# If 'Protocol Relative URL' option is checked, only apply change to internal links
if ( $this->option == 1 ) {
    # Remove protocol from home URL
    $website = preg_replace( '/https?:\/\//', '', home_url() );
    # Remove protocol from internal links
    $links = preg_replace( '/(<(' . $tag . ')([^>]*)(' . $attribute . ')=["\'])https?:\/\/' . $website . '/i', '$1//' . $website, $links );
}
# Else, remove protocols from all links
else {
    $links = preg_replace( '/(<(' . $tag . ')([^>]*)(' . $attribute . ')=["\'])https?:\/\//i', '$1//', $links );
}

This works as intended, but it doesn't work on these examples:

<!-- Within the 'style' attribute -->
<div class="some-class" style='background-color:rgba(255,255,255,0);background-image:url("http://placehold.it/300x200");background-position:center center;background-repeat:no-repeat'>
<!-- Within the 'srcset' attribute -->
<img src="http://placehold.it/600x300" srcset="http://placehold.it/500 500x, http://placehold.it/100 100w">

However, the code partially works for these examples.

<div class="some-class" style='background-color:rgba(255,255,255,0);background-image:url("http://placehold.it/300x200");background-position:center center;background-repeat:no-repeat'>
<img src="http://placehold.it/600x300" srcset="//placehold.it/500 500x, http://placehold.it/100 100w">

I've played around with adding additional values to the $tag and $attribute variables, but that didn't help. I'd assume I need to update the rest of my regex to cover these two additional tags? Or is there is a different way to approach it, such as DOMDocument?

Answer 1

I was able to use the following regex to cover all tags & attributes with protocols:

/<(?:input\b[^<]*\bvalue=[\"\']https?:\/\/|link\b[^<]*?\brel=[\'\"]canonical[\'\"][^<]*?>)(*SKIP)(*F)|https?:\/\//

This is the updated section:

# If 'Relative' option is selected, remove domain from all internal links
$exceptions = '<(?:input\b[^<]*\bvalue=[\"\']https?:\/\/|link\b[^<]*?\brel=[\'\"]canonical[\'\"][^<]*?>)(*SKIP)(*F)';
if ( $this->option == 2 ) {
    $website = preg_replace( '/https?:\/\//', '', home_url() );
    $links = preg_replace( '/' . $exceptions . '|https?:\/\/' . $website . '/', '', $links );
}
# For all external links, remove protocols
$links = preg_replace( '/<(?:input\b[^<]*\bvalue=[\"\']https?:\/\/|link\b[^<]*?\brel=[\'\"]canonical[\'\"][^<]*?>)(*SKIP)(*F)|https?:\/\//', '//', $links );