protected function compare()
{
$bench = new Ubench();
$url = 'tests/templated-retrospect/index.html';
$file = 'test.html';
if (!file_exists($file)) {
$htmlstr = file_get_contents($url);
file_put_contents($file, $htmlstr);
}
$htmlstr = file_get_contents($file);
$this->log('', true);
$this->log('Measuring Simple HTML DOM Parser...');
$resultsSimpleHtmlDomParser = $bench->run(function ($htmlstr) {
$results = [];
$html = HtmlDomParser::str_get_html($htmlstr);
$html->find('title', 0)->innertext('New Title');
$results[1] = $html->__toString();
$tpl = HtmlDomParser::str_get_html(file_get_contents('tests/templated-retrospect/index.html'));
foreach ($tpl->find('link') as $elem) {
$elem->href = '//localhost/xparser/tests/templated-retrospect/' . $elem->href;
}
foreach ($tpl->find('img, script') as $elem) {
$elem->src = '//localhost/xparser/tests/templated-retrospect/' . $elem->src;
}
$results[2] = $tpl->__toString();
return $results;
}, $htmlstr);
//$this->log('distance: ' . similar_text($htmlstr, $result));
$this->logBench($bench);
$this->log('', true);
$this->log('Measuring XParser...');
$resultsXParser = $bench->run(function ($htmlstr) {
$results = [];
$html = new XNode($htmlstr);
$html->find('title')->inner('New Title');
$results[1] = $html->__toString();
$tpl = new XNode(file_get_contents('tests/templated-retrospect/index.html'));
foreach ($tpl('link') as $elem) {
$elem->href = '//localhost/xparser/tests/templated-retrospect/' . $elem->href;
}
foreach ($tpl('img, script') as $elem) {
$elem->src = '//localhost/xparser/tests/templated-retrospect/' . $elem->src;
}
$results[2] = $tpl->__toString();
return $results;
}, $htmlstr);
//$this->log('distance: ' . similar_text($htmlstr, $result));
$this->logBench($bench);
$this->log('', true);
$this->log('Measuring Ganon...');
$resultsGanon = $bench->run(function ($htmlstr) {
$html = str_get_dom($htmlstr);
foreach ($html('title') as $title) {
$title->setInnerText('New Title');
}
$results[1] = $html->__toString();
$tpl = new XNode(file_get_contents('tests/templated-retrospect/index.html'));
foreach ($tpl('link') as $elem) {
$elem->href = '//localhost/xparser/tests/templated-retrospect/' . $elem->href;
}
foreach ($tpl('img, script') as $elem) {
$elem->src = '//localhost/xparser/tests/templated-retrospect/' . $elem->src;
}
$results[2] = $tpl->__toString();
return $results;
}, $htmlstr);
//$this->log('distance: ' . similar_text($htmlstr, $result));
$this->logBench($bench);
$this->log('', true);
$this->log('Symfony CSS Selector combined with DOMDocument and DOMXPath...');
$resultsXParser = $bench->run(function ($htmlstr) {
$results = [];
$html = new DOMDocument();
libxml_use_internal_errors(true);
$html->loadHTML($htmlstr);
$converter = new CssSelectorConverter();
$xpath = new DOMXPath($html);
$elements = $xpath->query($converter->toXPath('title'));
foreach ($elements as $element) {
$element->innserHTML = 'New Title';
}
$results[1] = $html->saveHTML();
$tpl = new DOMDocument();
$tpl->load('tests/templated-retrospect/index.html');
foreach ($xpath->query($converter->toXPath('link')) as $elem) {
$elem->setAttribute('href', '//localhost/xparser/tests/templated-retrospect/' . $elem->getAttribute('href'));
}
foreach ($xpath->query($converter->toXPath('img, script')) as $elem) {
$elem->setAttribute('src', '//localhost/xparser/tests/templated-retrospect/' . $elem->getAttribute('src'));
}
$results[2] = $tpl->saveHTML();
return $results;
}, $htmlstr);
//$this->log('distance: ' . similar_text($htmlstr, $result));
$this->logBench($bench);
$this->log('', true);
$this->log('Simple HTML DOM Parser vs Ganon distance: ' . similar_text($resultsSimpleHtmlDomParser[2], $resultsGanon[2]));
$this->log('Simple HTML DOM Parser vs XParser distance: ' . similar_text($resultsSimpleHtmlDomParser[2], $resultsXParser[2]));
$this->log('Ganon vs XParser distance: ' . similar_text($resultsGanon[2], $resultsXParser[2]));
$this->log('', true);
$this->log('', true);
}
public function find($select, $index = null)
{
$ret = new XNodeList([], $this);
if (!preg_match('/^[\\.\\#\\w\\s\\,]+$/is', $select)) {
return $this->findViaSymfony($select, $index);
}
$selects = preg_split('/\\s*,\\s*/', $select);
foreach ($selects as $select) {
$words = preg_split('/\\s+/', trim($select));
$founds = [];
foreach ($words as $wkey => $word) {
self::parseSelectorWord($word, $tag, $id, $classes);
if (!$id && !$classes) {
$founds = $this->getElementsArray($tag);
} else {
if ($id && !$classes) {
//foreach($ids as $id) {
$founds = array_merge($founds, $this->getElementsArray($tag, 'id', $id));
//}
} else {
if (!$id && $classes) {
foreach ($classes as $class) {
$foundsByClass[$class] = $this->getElementsByTagAndClassArray($tag, $class);
}
if (count($foundsByClass) > 1) {
$founds = array_merge($founds, call_user_func_array('array_intersect', $foundsByClass));
} else {
$founds = array_merge($founds, $foundsByClass[$class]);
}
} else {
if ($id && $classes) {
$foundsById = [];
//foreach($ids as $id) {
$foundsById = array_merge($foundsById, $this->getElementsArray($tag, 'id', $id));
//}
$foundsByClass = [];
foreach ($classes as $class) {
$foundsByClass = array_merge($foundsByClass, $this->getElementsByClassArray($class));
}
$founds = array_intersect($foundsById, $foundsByClass);
} else {
// hmmm.. interesting.
throw new XParserException('?');
}
}
}
}
if (!$founds) {
break;
}
if (isset($words[$wkey + 1])) {
$rest = implode(' ', array_slice($words, $wkey + 1));
foreach ($founds as $found) {
$inner = self::getInner($found);
$innerElement = new XNode($inner, $this);
$restElements = $innerElement->find($rest);
foreach ($restElements as $restElement) {
$ret->addElement($restElement);
}
}
return $ret;
}
}
$ret->addElementsArray($founds, $this);
}
if (!is_null($index)) {
return $ret->getElement($index);
}
return $ret;
}
public function mainTest()
{
$tpl = new XNode('<html>
<head>
<title>Test page</title>
</head>
<body>
<h1>Lorem ipsum</h1>
<div />
<div>asd</div>
<div id= "hello01" asdasdw />
<!--
<div id= "hello02" asdasdw class="message" asdasd />
-->
<hr>
<div id="hello1" class="message"> Hello World! </div>
<hr>
<div id="hello2" class="message selected"> before <span>Hello World!</span> after </div>
<hr>
<div id="hello3" class="message"> before <div>Hello <span>here</span> World!</div> after </div>
<hr>
<input type="text" id="myinput1" value="my value here..">
</body>
</html>');
$before = $tpl->find('div#hello2.selected.message, div#hello1')->inner();
$tpl->find('div#hello2.selected.message, div#hello1')->inner('yupeeee!');
$after = $tpl->find('div#hello2.selected.message, div#hello1')->inner();
$this->equ($before, ' Hello World! ');
$this->equ($after, 'yupeeee!');
$before = $tpl->find('html body input')->attr('value');
$tpl->find('input')->attr('value', 'elembe!');
$after = $tpl->find('html body input')->attr('value');
$this->equ($before, 'my value here..');
$this->equ($after, 'elembe!');
$before = $tpl->outer();
$this->equ(count($tpl('#hello02')->getElements()), 0);
$after = $tpl->outer();
$this->equ($before, $after);
$this->equ(count($tpl('#hello02')->getElements()), $tpl->getCount('#hello02'));
}
