<?php
/**
 * Markup Handler: XML2Html
 *
 * $Id$
 *
 * @author gERD Schaufelberger <gerd@php-tools.net>
 * @license PHP License
 * @package wb
 * @subpackage Markup
 */

WBClass::load('WBMarkup_Handler');

/**
 * Markup Scanner Handler: Xml2Xml
 *
 * Convert Wombat XML data to well formed XML
 *
 * @version 0.2.0
 * @package wb
 * @subpackage Markup
 */
class WBMarkup_Handler_Xml2xml implements WBMarkup_Handler
{
   /**
    * current depth
    * @var int
    */
    protected $depth = 0;

   /**
    * content
    * @var string
    */
    protected $content = '';

    /**
     * document's md5 sum
     * @var string
     */
    protected $md5;

    /**
     * document nodes
     * @var array
     */
    protected $doc  =   array();

    /**
     * list of running converters (WBMarkup_Converter)
     * @var array
     */
    protected $con   =   array();

   /**
    * handler on start beginning of scan
    *
    *
    *
    * @param string $content
    * @return bool usually true, false to stop the scanner
    */
    public function onScanStart(&$content)
    {
        $this->con      =   array();
        $this->depth    =   0;
        $this->content  =&  $content;
        $this->md5      =   md5($content);

        $this->doc     =   array();
        $this->doc[0]  =   array(
                        'ns'            =>  '',
                        'tag'           =>  '',
                        'attributes'    =>  array(),
                        'isEmpty'       =>  false,
                        'cData'         =>  array(),
                        'depth'         =>  $this->depth
                        );

        // inform converter about document's md5 and content
        foreach (array_keys($this->con) as $con) {
            $this->con[$con]->preset($this, $content);
        }

        return true;
    }

   /**
    * handler on start element
    *
    * @param string $ns The used namespace, if set. otherwise null
    * @param string $tag the tag itself
    * @param array  $attributes the attributes of the found element
    * @param bool $isEmpty true if it is a start- and closing-Element (e.g. <br />)
    * @return bool usually true, false to stop the scanner
    */
    public function onStartElement($ns, $tag, $attributes, $isEmpty)
    {
        ++$this->depth;
        $this->doc[$this->depth]   =   array(
                        'ns'            =>  $ns,
                        'tag'           =>  $tag,
                        'attributes'    =>  $attributes,
                        'isEmpty'       =>  $isEmpty,
                        'cData'         =>  array(),
                        'depth'         =>  $this->depth
                        );

        return true;
    }

   /**
    * handler fo cData
    *
    * Simply add cDate to document
    *
    * @uses replaceUrlString()
    * @param string $cData character data
    * @return bool usually true, false to stop the scanner
    */
    public function onCharacterData($cData)
    {
        // replace vertical tabs
        $cData  =   str_replace("\v", ' ', $cData);
        $cData  =   htmlentities($cData, ENT_NOQUOTES | ENT_XML1, "UTF-8");
        $this->doc[$this->depth]['cData'][] =   $cData;
        return true;
    }

   /**
    * handler for entities
    *
    * Add entities to document's cData
    *
    * @param string $entity the entity element, e.g. nbsp for &nbsp;
    * @param boolean $isUnicode true = the element is a unicode string
    */
    public function onEntityElement($entity, $isUnicode)
    {
        $cData  =   '&';
        if ('nbsp' == $entity) {
            $entity =   '160';
            $isUnicode  =   true;
        }
        if( $isUnicode ) {
            $cData  .=  '#';
        }

        $cData  .=  $entity . ';';
        $this->doc[$this->depth]['cData'][] =   $cData;
        return true;
    }

   /**
    * test handler on end element
    *
    * @param string $ns The used namespace, if set. otherwise null
    * @param string $tag the tag itself
    * @param bool $empty defines if the tag is empty
    * @return bool usually true, false to stop the scanner
    */
    public function onEndElement($ns, $tag, $empty)
    {
        // serialize node
        if( !$this->depth ) {
            return true;
        }
        $node   =   array_pop($this->doc);

        --$this->depth;
        if( $this->depth < 0 ) {
            $this->depth    =   0;
        }

        $this->doc[$this->depth]['cData'][]    =   $this->node2String($node);
        return true;
    }

   /**
    * called right after scan is complete
    *
    * @return bool usually true, false to stop the scanner
    */
    public function onScanComplete()
    {
        $this->content  =   $this->node2String($this->doc[0]);
        return true;
    }

   /**
    * receive translated content string
    *
    * @return string
    */
    public function getParsedContent()
    {
        return $this->content;
    }

   /**
    * serialize document node
    *
    * @param array $node
    * @return string serialized attributes
    */
    protected function node2String($node)
    {
        $cData  =   implode('', $node['cData']);
        if (!$node['tag']) {
            return $cData;
        }

        $tag    =   '';
        if ($node['ns']) {
            $tag    .=  $node['ns'] . ':';
        }
        $tag    .=  $node['tag'];

        $element    =   '<' . $tag . $this->attributes2String($node['attributes']);

        if ($node['isEmpty']) {
            $element    .=  ' />';
            return $element;
        }

        $element    .=  '>' . $cData . '</' . $tag . '>';
        return $element;
    }

   /**
    * serialize attribute list to HTML attribute string
    *
    * @param array $att attribute list
    * @return string serialized attributes
    */
    protected function attributes2String($att)
    {
        if (empty($att)) {
            return '';
        }
        $str    =   '';
        foreach ($att as $k => $v) {
            $str    .=  ' ' . $k . '="' .  $v . '"';
        }
        return $str;
    }
}
?>