Regex match and replace advanced

Question

I'm trying to write a little WordPress plugin to support some migrated content.

The syntax highlighter expects (for proper highlighting):

<pre lang='something'>
  <code>
    The code...
  </code>
</pre>

However, my markdown code has the following:

<pre>
  <code>
    :::something
    The code...
  </code>
</pre>

I think you can see where this is going. What I want to achieve is this:

1. :::something should be removed, and the <pre> tag should be updated to <pre lang="something">.
2. If :::something does not exist, the <pre> tag should be <pre lang="plain">
3. There may be multiple occurrences per page that need to be updated.

How would a PHP function achieving the above look like?

function set_syntax_lang($content) {
  // Do stuff here
  return $new_content;
}

What I gathered so far is this regex:

/<pre.*>\s*<code>\s*:::(\w)/

This even yields me, using preg_match, the actual syntax indicator (something), but I don't know how to update the pre-tag correctly.

It's been a very long time since I coded PHP and regexes are not really my strong suit. So all help is appreciated.

Answer 1

Finding :::something

preg_replace( '/<pre(.*>\s*<code>\s*):::(\w+)/U', '<pre lang="$2"$1' , $html );

This is an edge-case. But normally I should advise you to NOT use regex for html (bobince someone?).

Also next time try be less verbouse on your question. I took more time to read you than to write this answer.

Finding code without :::something

preg_replace( '/<pre(.*>\s*<code>\s*)(?!:::\w+)/U', '<pre lang="plain"$1' , $html );

Fixing <code>

preg_replace( array( '/(<pre.*>)\s*<code>/U' , '/<\/code>\s*(<\/pre>)/U' ),
              '$1' , $html );
//> Completly untested

Answer 2

You answered most of your question in the steps you gave. Break it down into those chunks -- FIRST see if you have :::something, THEN update your <pre> tag and REPEAT.

You'll have a much easier time of it if you use the DOM instead of regex. It will make the job of navigating through the <pre> and <code> tags very simple. As has been said many, many times here, html is not a regular language, so a regular expression cannot parse it correctly. Even for a limited subset of HTML, it's really not the right tool. The regex for :::something is trivial once you use the DOM to get the text between <code> and </code>: /:::(\w+)/

Answer 3

First of all some points I ran over:

/<pre.*>\s*<code>\s*:::(\w)/
     ^ 

According to your question, there never is a space in there if you make use of :::something. But you add it into your regex. I wonder why.

/<pre.*>\s*<code>\s*:::(\w)/
                         ^ 

If the language specifier is larger than one character (which I assume) you must write that into the regex, like \w+ for one or more letters.

The rest looks quite like you have already everything. Probably not the replacement:

$result = preg_replace( '((<pre)(>\s*<code>\s*):::(\w+))', '$1 lang="$3"$2' , $subject );

Hopefully this helps.