commit bd5ea6b1b3745d807d6ded9261ba49df1c292e55
Author: Dominic Mayers <dominic.mayers@meditationstudies.org>
Date:   Wed Jan 3 15:43:41 2024 -0500

    parser: Find the closing tag with a balanced inner text
    
    The current code for the parser fails when the code contains
    two levels of nesting with an empty ref tag. For example,
    it fails on <ref>A<ref>B<ref name="N" /></ref></ref>. You
    will not see it on a real page, because it will be hidden
    behind other problems. But with the series of patches that
    I am offering for the Cite extension, these other problems
    disappear and then this limitation can be seen. The faulted
    code is this pregmatch in Preprocessor_Hash.php :
    preg_match( "/<\/" . preg_quote( $name, '/' ) . "\s*>/i",
              $text, $matches, PREG_OFFSET_CAPTURE, $offset ) )
    It expects the $offset to be the start of the inner text.
    In the above example, the $offset will be 5. The code
    computes the inner text with the expression
    $inner = substr( $text, $offset, $matches[0][1] - $offset );
    In the above example, the returned $inner is
    A<ref>B<ref name="N" />. The correct value is
    A<ref>B<ref name="N" /></ref>. The commit defines
    a function pregFindClosingTag() to replace the
    faulty preg_match.
    
    Bug: T22707
    Change-Id: I5e6c17f553623d4f80a8f73b527fb4b9a7dbb421