Fix cell detection broken by DokuWiki Mort's lexer anchor change

[gh189]

Problem

Since DokuWiki release 2026-07-14 "Mort", cellbg stops setting cell background colors — @color: markers in table cells are rendered as literal text instead of being converted to a bgcolor attribute.

Root cause

connectTo() registers the trigger pattern:

'^@#?[0-9a-zA-Z]*:(?=[^\n]*\|[[:space:]]*\n)'

The leading ^ was relying on an old lexer implementation detail: DokuWiki's ParallelRegex::split() used to match patterns against only the remaining unparsed substring, so ^ effectively matched "wherever the parser cursor currently is" — which happened to be right after the | that opens a table cell, even though that's mid-line, not a real line start.

Mort's parser/lexer rewrite (needed to correctly support lookbehind assertions for the new GFM emphasis/closer engine) changed ParallelRegex::split() to match against the full document with a real byte offset via PREG_OFFSET_CAPTURE. Under normal PCRE multiline semantics, ^ now only matches right after a literal \n (or at the very start of the document). Since @color: inside a table cell always follows a literal |/^ delimiter on the same source line, ^ no longer matches there, the special pattern is never triggered, and handle()/render() are never called for it.

I reproduced this against the actual dokuwiki-2026-07-14 release tarball and confirmed:

• @red:test| as the very first line of a document → still matches (real line start).
• The same marker inside a table cell (| @lightgreen:**Span** | ... |) → never matches anymore.

Fix

Replace the ^ anchor with a bounded-width lookbehind for the actual preceding table delimiter (| or ^, optionally followed by spaces/tabs). This matches at the same effective position under both the old cursor-relative lexer and Mort's full-document/offset-based lexer, so it works across DokuWiki versions:

'(?<=[|^][\t ]{0,20})@#?[0-9a-zA-Z]*:(?=[^\n]*\|[[:space:]]*\n)'

Testing

Verified against the real dokuwiki-2026-07-14 "Mort" release (via bin/render.php) using all the example tables from the plugin's documentation page (solid color name, hex color, default @: color, and colors on non-first cells) — all now render with the correct bgcolor attribute again. Also confirmed no regression for the plain-first-line-of-document case.

[michieltimmers]

The fix doesn't work for me. I now get "A PCRE internal error occured. This might be caused by a faulty plugin" when I try to load a page. And the markup of the page has been converted to text only. Don't see any errors in the nginx log.

$ lsb_release -a
No LSB modules are available.
Distributor ID:	Ubuntu
Description:	Ubuntu 24.04.4 LTS
Release:	24.04
Codename:	noble

$ php -v
PHP 8.3.6 (cli) (built: Jul 16 2026 18:30:41) (NTS)
Copyright (c) The PHP Group
Zend Engine v4.3.6, Copyright (c) Zend Technologies
    with Zend OPcache v8.3.6, Copyright (c), by Zend Technologies

This seems to be a bit of an old plugin.. don't think it will be fixed properly? are there any alternatives to change the colors in a table?

[michieltimmers]

It might be an issue with the PCRE version on ubuntu 24.04

$ php -r 'echo PCRE_VERSION, PHP_EOL;'
10.42 2022-12-11

This should probably be fine in ubuntu 26.04. But I wait to upgrade to that version until 26.04.1 is released (scheduled August 27, 2026)

[ProjectsKoryHasWorkedOn]

@michieltimmers

I also had same PCRE error

This is working for me

After editing page and saving it again

Replacement syntax.php

<?php
/**
 * cellbg Plugin: Allows user-defined colored cells in tables
 *
 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
 * @author     dr4Ke <dr4ke@dr4ke.net>
 * @link       http://git.dr4ke.net/?p=dr4Ke/forks/dokuwiki-cells-bg.git
 *
 * Derived from the highlight plugin from : http://www.dokuwiki.org/plugin:highlight
 * and : http://www.staddle.net/wiki/plugins/highlight
 */

/**
 * All DokuWiki plugins to extend the parser/rendering mechanism
 * need to inherit from this class
 */
class syntax_plugin_cellbg extends \dokuwiki\Extension\SyntaxPlugin {

    // What kind of syntax are we?
    function getType(){ return 'substition'; }

    // What kind of syntax do we allow (optional)
    function getAllowedTypes() {
        return array('formatting', 'substition', 'disabled');
    }

    // What about paragraphs? (optional)
    function getPType(){ return 'normal'; }

    // Where to sort in?
    function getSort(){ return 200; }

    // Connect pattern to lexer
    function connectTo($mode) {
        $this->Lexer->addSpecialPattern('(?<=[\t |^])@#?[0-9a-zA-Z]+:', $mode, 'plugin_cellbg');
    }

    // Handle the match
    function handle($match, $state, $pos, \Doku_Handler $handler){
        switch ($state) {
            case DOKU_LEXER_ENTER :
                break;
            case DOKU_LEXER_MATCHED :
                break;
            case DOKU_LEXER_UNMATCHED :
                break;
            case DOKU_LEXER_EXIT :
                break;
            case DOKU_LEXER_SPECIAL :
                preg_match("/@([^:]*)/", $match, $color); // get the color
                if ($this->_isValid($color[1])) return array($state, $color[1], $match);
                break;
        }
        return array($state, "yellow", $match);
    }

    // Create output
    function render($mode, \Doku_Renderer $renderer, $data) {
        if ($mode == 'xhtml') {
            list($state, $color, $text) = $data;
            switch ($state) {
                case DOKU_LEXER_SPECIAL :
                    // Escape the color value to prevent XSS
                    $color = hsc($color);

                    // Match the last opened <td or <th tag. Newer DokuWiki versions
                    // (Greebo+) may insert additional markup (wrappers, classes, etc.)
                    // between the opening tag and our position, so we use a broader
                    // match that looks back further in the document.
                    if (preg_match('/(<t[dh]\b[^>]*)>((?:(?!<t[dh]\b).)*?)$/s', $renderer->doc, $m)) {
                        $renderer->doc = preg_replace(
                            '/(<t[dh]\b[^>]*)>((?:(?!<t[dh]\b).)*?)$/s',
                            '$1 style="background-color:' . $color . ';">$2',
                            $renderer->doc
                        );
                    } else {
                        $renderer->doc .= $text;
                    }
                    break;
            }
            return true;
        }
        return false;
    }

    // validate color value $c
    // this is cut price validation - only to ensure the basic format is
    // correct and there is nothing harmful
    // three basic formats  "colorname", "#fff[fff]", "rgb(255[%],255[%],255[%])"
    function _isValid($c) {

        $c = trim($c);

        $pattern = "/
            (^[a-zA-Z]+$)|                                #colorname - not verified
            (^\#([0-9a-fA-F]{3}|[0-9a-fA-F]{6})$)|       #colorvalue
            (^rgb\(([0-9]{1,3}%?,){2}[0-9]{1,3}%?\)$)    #rgb triplet
            /x";

        return (preg_match($pattern, $c));

    }
}