Replace all relative URLs with absolute URLS

Question

I've seen a few answers (like this one), but I have some more complex scenarios I'm not sure how to account for.

I essentially have full HTML documents. I need to replace every single relative URL with absolute URLs.

Elements from potential HTML look as follows, may be other cases as well:

<img src="/relative/url/img.jpg" />
<form action="/">
<form action="/contact-us/">
<a href='/relative/url/'>Note the Single Quote</a>
<img src="//example.com/protocol-relative-img.jpg" />

Desired Output would be:

// "//example.com/" is ideal, but "http(s)://example.com/" are acceptable

<img src="//example.com/relative/url/img.jpg" />
<form action="//example.com/">
<form action="//example.com/contact-us/">
<a href='//example.com/relative/url/'>Note the Single Quote</a>
<img src="//example.com/protocol-relative-img.jpg" /> <!-- Unmodified -->

I DON'T want to replace protocol relative URLs, since they already function as absolute URLs. I've come up with some code that works, but I'm wondering if I can clean it up a little, as it's extremely repetitive.

But I have to account for single and double quoted attribute values for src, href, and action (am I missing any attributes that can have relative URLs?) while simultaneously avoiding protocol relative URLs.

Here's what I have so far:

// Make URL replacement protocol relative to not break insecure/secure links
$url = str_replace( array( 'http://', 'https://' ), '//', $url );

// Temporarily Modify Protocol-Relative URLS
$str = str_replace( 'src="//', 'src="::TEMP_REPLACE::', $str );
$str = str_replace( "src='//", "src='::TEMP_REPLACE::", $str );
$str = str_replace( 'href="//', 'href="::TEMP_REPLACE::', $str );
$str = str_replace( "href='//", "href='::TEMP_REPLACE::", $str );
$str = str_replace( 'action="//', 'action="::TEMP_REPLACE::', $str );
$str = str_replace( "action='//", "action='::TEMP_REPLACE::", $str );

// Replace all other Relative URLS
$str = str_replace( 'src="/', 'src="'. $url .'/', $str );
$str = str_replace( "src='/", "src='". $url ."/", $str );
$str = str_replace( 'href="/', 'href="'. $url .'/', $str );
$str = str_replace( "href='/", "href='". $url ."/", $str );
$str = str_replace( 'action="/', 'action="'. $url .'/', $str );
$str = str_replace( "action='/", "action='". $url ."/", $str );

// Change Protocol Relative URLs back
$str = str_replace( 'src="::TEMP_REPLACE::', 'src="//', $str );
$str = str_replace( "src='::TEMP_REPLACE::", "src='//", $str );
$str = str_replace( 'href="::TEMP_REPLACE::', 'href="//', $str );
$str = str_replace( "href='::TEMP_REPLACE::", "href='//", $str );
$str = str_replace( 'action="::TEMP_REPLACE::', 'action="//', $str );
$str = str_replace( "action='::TEMP_REPLACE::", "action='//", $str );

I mean, it works, but it's uuugly, and I was thinking there's probably a better way to do it.

Answer 1

New Answer

If your real html document is valid (and has a parent/containing tag), then the most appropriate and reliable technique will be to use a proper DOM parser.

Here is how DOMDocument and Xpath can be used to elegantly target and replace your designated tag attributes:

Code1 - Nested Xpath Queries: (Demo)

$domain = '//example.com';
$tagsAndAttributes = [
    'img' => 'src',
    'form' => 'action',
    'a' => 'href'
];

$dom = new DOMDocument; 
$dom->loadHTML($html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
$xpath = new DOMXPath($dom);
foreach ($tagsAndAttributes as $tag => $attr) {
    foreach ($xpath->query("//{$tag}[not(starts-with(@{$attr}, '//'))]") as $node) {
        $node->setAttribute($attr, $domain . $node->getAttribute($attr));
    }
}
echo $dom->saveHTML();

Code2 - Single Xpath Query w/ Condition Block: (Demo)

$domain = '//example.com';
$targets = [
    "//img[not(starts-with(@src, '//'))]",
    "//form[not(starts-with(@action, '//'))]",
    "//a[not(starts-with(@href, '//'))]"
];

$dom = new DOMDocument; 
$dom->loadHTML($html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
$xpath = new DOMXPath($dom);
foreach ($xpath->query(implode('|', $targets)) as $node) {
    if ($src = $node->getAttribute('src')) {
        $node->setAttribute('src', $domain . $src);
    } elseif ($action = $node->getAttribute('action')) {
        $node->setAttribute('action', $domain . $action);
    } else {
        $node->setAttribute('href', $domain . $node->getAttribute('href'));
    }
}
echo $dom->saveHTML();

---

Old Answer: (...regex is not "DOM-aware" and is vulnerable to unexpected breakage)

If I understand you properly, you have a base value in mind, and you only want to apply it to relative paths.

Pattern Demo

Code: (Demo)

$html=<<<HTML
<img src="/relative/url/img.jpg" />
<form action="/">
<a href='/relative/url/'>Note the Single Quote</a>
<img src="//site.com/protocol-relative-img.jpg" />
HTML;

$base='https://example.com';

echo preg_replace('~(?:src|action|href)=[\'"]\K/(?!/)[^\'"]*~',"$base$0",$html);

Output:

<img src="https://example.com/relative/url/img.jpg" />
<form action="https://example.com/">
<a href='https://example.com/relative/url/'>Note the Single Quote</a>
<img src="//site.com/protocol-relative-img.jpg" />

Pattern Breakdown:

~                      #Pattern delimiter
(?:src|action|href)    #Match: src or action or href
=                      #Match equal sign
[\'"]                  #Match single or double quote
\K                     #Restart fullstring match (discard previously matched characters
/                      #Match slash
(?!/)                  #Negative lookahead (zero-length assertion): must not be a slash immediately after first matched slash
[^\'"]*                #Match zero or more non-single/double quote characters
~                      #Pattern delimiter