PEARSax3.php

Go to the documentation of this file.

<?php

class HTMLPurifier_Lexer_PEARSax3 extends HTMLPurifier_Lexer
{

    protected $tokens = array();

    public function tokenizeHTML($string, $config, $context) {

        $this->tokens = array();

        $string = $this->normalize($string, $config, $context);

        $parser = new XML_HTMLSax3();
        $parser->set_object($this);
        $parser->set_element_handler('openHandler','closeHandler');
        $parser->set_data_handler('dataHandler');
        $parser->set_escape_handler('escapeHandler');

        // doesn't seem to work correctly for attributes
        $parser->set_option('XML_OPTION_ENTITIES_PARSED', 1);

        $parser->parse($string);

        return $this->tokens;

    }

    public function openHandler(&$parser, $name, $attrs, $closed) {
        // entities are not resolved in attrs
        foreach ($attrs as $key => $attr) {
            $attrs[$key] = $this->parseData($attr);
        }
        if ($closed) {
            $this->tokens[] = new HTMLPurifier_Token_Empty($name, $attrs);
        } else {
            $this->tokens[] = new HTMLPurifier_Token_Start($name, $attrs);
        }
        return true;
    }

    public function closeHandler(&$parser, $name) {
        // HTMLSax3 seems to always send empty tags an extra close tag
        // check and ignore if you see it:
        // [TESTME] to make sure it doesn't overreach
        if ($this->tokens[count($this->tokens)-1] instanceof HTMLPurifier_Token_Empty) {
            return true;
        }
        $this->tokens[] = new HTMLPurifier_Token_End($name);
        return true;
    }

    public function dataHandler(&$parser, $data) {
        $this->tokens[] = new HTMLPurifier_Token_Text($data);
        return true;
    }

    public function escapeHandler(&$parser, $data) {
        if (strpos($data, '--') === 0) {
            $this->tokens[] = new HTMLPurifier_Token_Comment($data);
        }
        // CDATA is handled elsewhere, but if it was handled here:
        //if (strpos($data, '[CDATA[') === 0) {
        //    $this->tokens[] = new HTMLPurifier_Token_Text(
        //        substr($data, 7, strlen($data) - 9) );
        //}
        return true;
    }

}

// vim: et sw=4 sts=4