<?php
/**
* Markup Handler: Dereferer
*
* $Id$
*
* @author gERD Schaufelberger <gerd@php-tools.net>
* @license PHP License
* @package wb
* @subpackage Markup
*/
WBClass::load('WBMarkup_Handler');
/**
* Markup Scanner Handler: Dereferer
*
* Find references in XML
*
* @version 0.3.0
* @package wb
* @subpackage Markup
*/
class WBMarkup_Handler_Dereferer implements WBMarkup_Handler
{
/**
* current depth
* @var int
*/
private $depth = 0;
/**
* complete content
* @var string
*/
private $content = '';
/**
* document nodes
* @var array
*/
private $doc = array();
/**
* list of running dererferer
* @var array
*/
private $dref = array();
/**
* list of ids for namespaces and tags
* @var array
*/
private $ids = array();
/**
* URL
* @var WBDictionary_URL
*/
private $url;
/**
* path
* @var string
*/
private $urlPath;
/**
* table
* @var WBDatasource_Table
*/
private $table;
/**
* set url
*
* Set refering URL
*
* @param WBDictionary_URL $url
* @param string $path
*/
public function setUrl($url, $path = '')
{
$this->urlPath = $path;
if ($url instanceof WBDictionary_URL) {
$this->url = $url;
return;
}
$this->url = WBClass::create('WBDictionary_URL');
if (preg_match('/^\\d+$/', $url)) {
$this->url->load($url);
return;
}
$this->url->addWord($url);
}
/**
* handler on start beginning of scan
*
*
*
* @param string $content
* @return bool usually true, false to stop the scanner
*/
public function onScanStart(&$content)
{
$this->url = null;
$this->dref = array();
$this->depth = 0;
$this->content = $content;
$this->doc = array();
$this->doc[0] = array(
'ns' => '',
'tag' => '',
'attributes' => array(),
'isEmpty' => false,
'cData' => array(),
);
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()
);
return true;
}
/**
* handler fo cData
*
* Simply add cDate to document
*
* Copied regular expression to find and replace URL from
* {@link http://snipplr.com/view/6889/regular-expressions-for-uri-validationparsing/}
* and removed parentheses
*
* @uses replaceUrlString()
* @param string $cData character data
* @return bool usually true, false to stop the scanner
*/
public function onCharacterData($cData)
{
$this->doc[$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
* @param boolean $isUnicode true = the element is a unicode string
*/
public function onEntityElement($entity, $isUnicode)
{
$cData = '&';
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);
if ($node['ns']) {
$this->extractReferences($node['ns'], $node['tag'], $node);
}
--$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()
{
// remove empty lists and make them unique
foreach ($this->ids as $ns => &$nIds) {
foreach ($nIds as $tag => &$ids) {
if (empty($ids)) {
unset($nIds[$tag]);
continue;
}
$ids = array_unique($ids);
}
if (empty($nIds)) {
unset($this->ids[$ns]);
}
}
return true;
}
/**
* get list of refering ids
*
* @return array
*/
public function getIds()
{
return $this->ids;
}
/**
* update reference list
*
* Remove old references and insert new ones.
*/
public function save()
{
$this->table = WBClass::create('WBDatasource_Table');
$clause = array();
$clause[] = array(
'field' => $this->table->getIdentifier('url'),
'value' => $this->url->getId()
);
$clause[] = array(
'field' => 'path',
'value' => $this->urlPath
);
$this->table->delete('reference', null, null, $clause);
$proto = array(
$this->table->getIdentifier('url') => $this->url->getId(),
'path' => $this->urlPath,
'xnamespace' => '',
'xtag' => '',
'xid' => ''
);
$save = array();
foreach ($this->ids as $ns => $ids) {
$proto['xnamespace'] = $ns;
foreach ($ids as $tag => $id) {
$proto['xtag'] = $tag;
for ($i = 0; $i < count($id); ++$i) {
$proto['xid'] = $id[$i];
$save[] = $proto;
}
}
}
// nothing to save
if (empty($save)) {
return;
}
$this->table->save('reference', '__new', $save);
}
/**
* receive translated content string
*
* @return string
*/
public function getParsedContent()
{
return $this->content;
}
/**
* convert node
*
* transform namespaced tags to HTML
*
* @todo convert empty namespace (standard HTML) to known WB namespace, if possible
* @param string $ns
* @param string $tag
* @param array $node
*/
protected function extractReferences($ns, $tag, &$node)
{
$dref = $this->getDereferencer($ns, $tag);
$refs = $dref->getReferences($node);
// backward compatibility
if (isset($refs[0])) {
$this->ids[$ns][$tag] = array_merge( $this->ids[$ns][$tag], $refs);
return;
}
foreach ($refs as $ns => $ref) {
foreach ($ref as $tag => $r) {
if (!is_array($this->ids[$ns][$tag])) {
$this->ids[$ns][$tag] = array();
}
$this->ids[$ns][$tag] = array_merge($this->ids[$ns][$tag], $r);
}
}
}
/**
* cet instance of matching converter
*
* See whether convert is already running, start it otherwise
*
* @param $ns
* @param $tag
* @return WBMarkup_Dereferencer
*/
protected function getDereferencer($ns, $tag)
{
$name = ucfirst(strtolower($ns))
. '_'
. ucfirst(strtolower($tag));
if (isset($this->dref[$name])) {
return $this->dref[$name];
}
// try full quallified class: ns+ tag
try{
$this->dref[$name] = WBClass::create('WBMarkup_Dereferencer_' . $name);
/** @var $this->con[$con] WBMarkup_Dereferencer */
} catch(WBException_Class $e) {
// try just the ns
try{
$this->dref[$name] = WBClass::create('WBMarkup_Dereferencer_' . $ns);
/** @var $this->con[$con] WBMarkup_Dereferencer */
} catch(WBException_Class $e) {
// very generic dereferencer
$this->dref[$name] = WBClass::create('WBMarkup_Dereferencer');
/** @var $this->con[$con] WBMarkup_Dereferencer */
}
}
if (!isset($this->ids[$ns])) {
$this->ids[$ns] = array();
}
$this->ids[$ns][$tag] = array();
return $this->dref[$name];
}
/**
* 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;
}
}
?>