bbcode-parser.php

<?php

$bbcode = '[p][url=http://www.gmail.com][u]gmail[/u][/url]';
$bbcode .= '[b][size=3]foo[/size][/b]';
$bbcode .= '[color=red][u][i]bar[/i][/u][/color][/p]';

$root = new BBCodeNode('root');
$root = build_tree($bbcode, $root);
echo 'BBCode: ', $bbcode, "\n\n";
traverse_tree($root, 'printer', 0);

function build_tree($bbcode, $node)
{
    if ($bbcode == '') {
        return $node;
    }
    $i = 0;
    if ($bbcode{0} != '[') {
        // text node
        $text = '';
        while ($bbcode{$i} != '[') {
            $text .= $bbcode{$i};
            ++$i;
        }

        $child = new BBCodeNode($text);
        $child->type = 'text';
        $child->value = $text;
        $child->parentNode = $node;

        $node->children[$node->childCount] = $child;
        ++$node->childCount;
        
        $node = build_tree(substr($bbcode, $i), $node);
        return $node;
    } else {
        ++$i;
        if ($bbcode{$i} != '/') {
            // a open tag
            $state = 'in_open_tag';
            $open_tag = '';
            $param = '';
            while ($bbcode{$i} != ']') {
                if ($bbcode{$i} != '=') {
                    $open_tag .= $bbcode{$i};
                    ++$i;
                } else {
                    $state = 'in_param';
                    break;
                }
            }    
            if ($state == 'in_param') {
                while($bbcode{++$i} != ']') {
                    $param .= $bbcode{$i};
                }
                $state = 'in_open_tag';
            }    
            if ($state == 'in_open_tag' && $bbcode{$i} == ']') {
                // open_tag finished
                $child = new BBCodeNode($open_tag);
                if ($param) {
                    $child->param = $param;
                }
                $child->parentNode = $node;

                $node->children[$node->childCount] = $child;
                ++$node->childCount;

                $node = build_tree(substr($bbcode, ++$i), $child);
                return $node;
            } else {
                throw new Exception('invalid bbcode, index[' . $i . ']');
            }
        } else {
            // a close tag
            // skip close tag
            while ($bbcode{++$i} != ']') {
                ;
            }
            if ($node->parentNode) {
                ++$i;
                $node->parentNode = build_tree(substr($bbcode, $i), $node->parentNode);
                return $node->parentNode;
            } else {
                throw new Exception('bbcode not match, index[' . $i . ']');
            }
        }
    }
}

function traverse_tree($node, $handler, $depth)
{
    $handler($node, $depth);
    if ($node->childCount) {
        foreach ($node->children as $v) {
            traverse_tree($v, $handler, $depth + 1);
        }
    }
}

function printer($node, $depth)
{
    for ($i = 0; $i < $depth; $i++) {
        echo '  ';
    }
    echo 'node(', $node->name;
    if ($node->param) {
        echo '=', $node->param;
    }
    echo ")\n";
}

class BBCodeNode
{
    public $name = null;
    public $parentNode = null;
    public $children = array();
    public $childCount = 0;
    public $type = 'node';
    public $value = null;

    function __construct($name)
    {
        $this->name = $name;
    }
}

?>