tt-rss/lib/readability/Readability.php

<?php
/** 
* Arc90's Readability ported to PHP for FiveFilters.org
* Based on readability.js version 1.7.1 (without multi-page support)
* Updated to allow HTML5 parsing with html5lib
* Updated with lightClean mode to preserve more images and youtube/vimeo/viddler embeds
* ------------------------------------------------------
* Original URL: http://lab.arc90.com/experiments/readability/js/readability.js
* Arc90's project URL: http://lab.arc90.com/experiments/readability/
* JS Source: http://code.google.com/p/arc90labs-readability
* Ported by: Keyvan Minoukadeh, http://www.keyvan.net
* More information: http://fivefilters.org/content-only/
* License: Apache License, Version 2.0
* Requires: PHP5
* Date: 2012-09-19
* 
* Differences between the PHP port and the original
* ------------------------------------------------------
* Arc90's Readability is designed to run in the browser. It works on the DOM 
* tree (the parsed HTML) after the page's CSS styles have been applied and 
* Javascript code executed. This PHP port does not run inside a browser. 
* We use PHP's ability to parse HTML to build our DOM tree, but we cannot 
* rely on CSS or Javascript support. As such, the results will not always 
* match Arc90's Readability. (For example, if a web page contains CSS style 
* rules or Javascript code which hide certain HTML elements from display, 
* Arc90's Readability will dismiss those from consideration but our PHP port, 
* unable to understand CSS or Javascript, will not know any better.)
* 
* Another significant difference is that the aim of Arc90's Readability is 
* to re-present the main content block of a given web page so users can 
* read it more easily in their browsers. Correct identification, clean up, 
* and separation of the content block is only a part of this process. 
* This PHP port is only concerned with this part, it does not include code 
* that relates to presentation in the browser - Arc90 already do 
* that extremely well, and for PDF output there's FiveFilters.org's 
* PDF Newspaper: http://fivefilters.org/pdf-newspaper/.
* 
* Finally, this class contains methods that might be useful for developers 
* working on HTML document fragments. So without deviating too much from 
* the original code (which I don't want to do because it makes debugging 
* and updating more difficult), I've tried to make it a little more 
* developer friendly. You should be able to use the methods here on 
* existing DOMElement objects without passing an entire HTML document to 
* be parsed.
*/

// This class allows us to do JavaScript like assignements to innerHTML
require_once(dirname(__FILE__).'/JSLikeHTMLElement.php');
libxml_use_internal_errors(true);

// Alternative usage (for testing only!)
// uncomment the lines below and call Readability.php in your browser 
// passing it the URL of the page you'd like content from, e.g.:
// Readability.php?url=http://medialens.org/alerts/09/090615_the_guardian_climate.php

/*
if (!isset($_GET['url']) || $_GET['url'] == '') {
	die('Please pass a URL to the script. E.g. Readability.php?url=bla.com/story.html');
}
$url = $_GET['url'];
if (!preg_match('!^https?://!i', $url)) $url = 'http://'.$url;
$html = file_get_contents($url);
$r = new Readability($html, $url);
$r->init();
echo $r->articleContent->innerHTML;
*/

class Readability
{
	public $version = '1.7.1-without-multi-page';
	public $convertLinksToFootnotes = false;
	public $revertForcedParagraphElements = true;
	public $articleTitle;
	public $articleContent;
	public $dom;
	public $url = null; // optional - URL where HTML was retrieved
	public $debug = false;
	public $lightClean = true; // preserves more content (experimental) added 2012-09-19
	protected $body = null; // 
	protected $bodyCache = null; // Cache the body HTML in case we need to re-use it later
	protected $flags = 7; // 1 | 2 | 4;   // Start with all flags set.
	protected $success = false; // indicates whether we were able to extract or not
	
	/**
	* All of the regular expressions in use within readability.
	* Defined up here so we don't instantiate them repeatedly in loops.
	**/
	public $regexps = array(
		'unlikelyCandidates' => '/combx|comment|community|disqus|extra|foot|header|menu|remark|rss|shoutbox|sidebar|sponsor|ad-break|agegate|pagination|pager|popup/i',
		'okMaybeItsACandidate' => '/and|article|body|column|main|shadow/i',
		'positive' => '/article|body|content|entry|hentry|main|page|attachment|pagination|post|text|blog|story/i',
		'negative' => '/combx|comment|com-|contact|foot|footer|_nav|footnote|masthead|media|meta|outbrain|promo|related|scroll|shoutbox|sidebar|sponsor|shopping|tags|tool|widget/i',
		'divToPElements' => '/<(a|blockquote|dl|div|img|ol|p|pre|table|ul)/i',
		'replaceBrs' => '/(<br[^>]*>[ \n\r\t]*){2,}/i',
		'replaceFonts' => '/<(\/?)font[^>]*>/i',
		// 'trimRe' => '/^\s+|\s+$/g', // PHP has trim()
		'normalize' => '/\s{2,}/',
		'killBreaks' => '/(<br\s*\/?>(\s|&nbsp;?)*){1,}/',
		'video' => '!//(player\.|www\.)?(youtube|vimeo|viddler)\.com!i',
		'skipFootnoteLink' => '/^\s*(\[?[a-z0-9]{1,2}\]?|^|edit|citation needed)\s*$/i'
	);	
	
	/* constants */
	const FLAG_STRIP_UNLIKELYS = 1;
	const FLAG_WEIGHT_CLASSES = 2;
	const FLAG_CLEAN_CONDITIONALLY = 4;
	
	/**
	* Create instance of Readability
	* @param string UTF-8 encoded string
	* @param string (optional) URL associated with HTML (used for footnotes)
	* @param string which parser to use for turning raw HTML into a DOMDocument (either 'libxml' or 'html5lib')
	*/	
	function __construct($html, $url=null, $parser='libxml')
	{
		$this->url = $url;
		/* Turn all double br's into p's */
		$html = preg_replace($this->regexps['replaceBrs'], '</p><p>', $html);
		$html = preg_replace($this->regexps['replaceFonts'], '<$1span>', $html);
		$html = mb_convert_encoding($html, 'HTML-ENTITIES', "UTF-8");
		if (trim($html) == '') $html = '<html></html>';
		if ($parser=='html5lib' && ($this->dom = HTML5_Parser::parse($html))) {
			// all good
		} else {
			$this->dom = new DOMDocument();
			$this->dom->preserveWhiteSpace = false;
			@$this->dom->loadHTML($html);
		}
		$this->dom->registerNodeClass('DOMElement', 'JSLikeHTMLElement');
	}

	/**
	* Get article title element
	* @return DOMElement
	*/
	public function getTitle() {
		return $this->articleTitle;
	}
	
	/**
	* Get article content element
	* @return DOMElement
	*/
	public function getContent() {
		return $this->articleContent;
	}	
	
	/**
	* Runs readability.
	* 
	* Workflow:
	*  1. Prep the document by removing script tags, css, etc.
	*  2. Build readability's DOM tree.
	*  3. Grab the article content from the current dom tree.
	*  4. Replace the current DOM tree with the new one.
	*  5. Read peacefully.
	*
	* @return boolean true if we found content, false otherwise
	**/
	public function init()
	{
		if (!isset($this->dom->documentElement)) return false;
		$this->removeScripts($this->dom);
		//die($this->getInnerHTML($this->dom->documentElement));
		
		// Assume successful outcome
		$this->success = true;

		$bodyElems = $this->dom->getElementsByTagName('body');
		if ($bodyElems->length > 0) {
			if ($this->bodyCache == null) {
				$this->bodyCache = $bodyElems->item(0)->innerHTML;
			}
			if ($this->body == null) {
				$this->body = $bodyElems->item(0);
			}
		}

		$this->prepDocument();
		
		//die($this->dom->documentElement->parentNode->nodeType);
		//$this->setInnerHTML($this->dom->documentElement, $this->getInnerHTML($this->dom->documentElement));
		//die($this->getInnerHTML($this->dom->documentElement));

		/* Build readability's DOM tree */
		$overlay        = $this->dom->createElement('div');
		$innerDiv       = $this->dom->createElement('div');
		$articleTitle   = $this->getArticleTitle();
		$articleContent = $this->grabArticle();

		if (!$articleContent) {
			$this->success = false;
			$articleContent = $this->dom->createElement('div');
			$articleContent->setAttribute('id', 'readability-content');
			$articleContent->innerHTML = '<p>Sorry, Readability was unable to parse this page for content.</p>';		
		}
		
		$overlay->setAttribute('id', 'readOverlay');
		$innerDiv->setAttribute('id', 'readInner');

		/* Glue the structure of our document together. */
		$innerDiv->appendChild($articleTitle);
		$innerDiv->appendChild($articleContent);
		$overlay->appendChild($innerDiv);
		
		/* Clear the old HTML, insert the new content. */
		$this->body->innerHTML = '';
		$this->body->appendChild($overlay);
		//document.body.insertBefore(overlay, document.body.firstChild);
		$this->body->removeAttribute('style');

		$this->postProcessContent($articleContent);
		
		// Set title and content instance variables
		$this->articleTitle = $articleTitle;
		$this->articleContent = $articleContent;
		
		return $this->success;
	}
	
	/**
	* Debug
	*/
	protected function dbg($msg) {
		if ($this->debug) echo '* ',$msg, "\n";
	}
	
	/**
	* Run any post-process modifications to article content as necessary.
	*
	* @param DOMElement
	* @return void
	*/
	public function postProcessContent($articleContent) {
		if ($this->convertLinksToFootnotes && !preg_match('/wikipedia\.org/', @$this->url)) { 
			$this->addFootnotes($articleContent);
		}
	}
	
	/**
	* Get the article title as an H1.
	*
	* @return DOMElement
	*/
	protected function getArticleTitle() {
		$curTitle = '';
		$origTitle = '';

		try {
			$curTitle = $origTitle = $this->getInnerText($this->dom->getElementsByTagName('title')->item(0));
		} catch(Exception $e) {}
		
		if (preg_match('/ [\|\-] /', $curTitle))
		{
			$curTitle = preg_replace('/(.*)[\|\-] .*/i', '$1', $origTitle);
			
			if (count(explode(' ', $curTitle)) < 3) {
				$curTitle = preg_replace('/[^\|\-]*[\|\-](.*)/i', '$1', $origTitle);
			}
		}
		else if (strpos($curTitle, ': ') !== false)
		{
			$curTitle = preg_replace('/.*:(.*)/i', '$1', $origTitle);

			if (count(explode(' ', $curTitle)) < 3) {
				$curTitle = preg_replace('/[^:]*[:](.*)/i','$1', $origTitle);
			}
		}
		else if(strlen($curTitle) > 150 || strlen($curTitle) < 15)
		{
			$hOnes = $this->dom->getElementsByTagName('h1');
			if($hOnes->length == 1)
			{
				$curTitle = $this->getInnerText($hOnes->item(0));
			}
		}

		$curTitle = trim($curTitle);

		if (count(explode(' ', $curTitle)) <= 4) {
			$curTitle = $origTitle;
		}
		
		$articleTitle = $this->dom->createElement('h1');
		$articleTitle->innerHTML = $curTitle;
		
		return $articleTitle;
	}
	
	/**
	* Prepare the HTML document for readability to scrape it.
	* This includes things like stripping javascript, CSS, and handling terrible markup.
	* 
	* @return void
	**/
	protected function prepDocument() {
		/**
		* In some cases a body element can't be found (if the HTML is totally hosed for example)
		* so we create a new body node and append it to the document.
		*/
		if ($this->body == null)
		{
			$this->body = $this->dom->createElement('body');
			$this->dom->documentElement->appendChild($this->body);
		}
		$this->body->setAttribute('id', 'readabilityBody');

		/* Remove all style tags in head */
		$styleTags = $this->dom->getElementsByTagName('style');
		for ($i = $styleTags->length-1; $i >= 0; $i--)
		{
			$styleTags->item($i)->parentNode->removeChild($styleTags->item($i));
		}

		/* Turn all double br's into p's */
		/* Note, this is pretty costly as far as processing goes. Maybe optimize later. */
		//document.body.innerHTML = document.body.innerHTML.replace(readability.regexps.replaceBrs, '</p><p>').replace(readability.regexps.replaceFonts, '<$1span>');
		// We do this in the constructor for PHP as that's when we have raw HTML - before parsing it into a DOM tree.
		// Manipulating innerHTML as it's done in JS is not possible in PHP.
	}

	/**
	* For easier reading, convert this document to have footnotes at the bottom rather than inline links.
	* @see http://www.roughtype.com/archives/2010/05/experiments_in.php
	*
	* @return void
	**/
	public function addFootnotes($articleContent) {
		$footnotesWrapper = $this->dom->createElement('div');
		$footnotesWrapper->setAttribute('id', 'readability-footnotes');
		$footnotesWrapper->innerHTML = '<h3>References</h3>';
		
		$articleFootnotes = $this->dom->createElement('ol');
		$articleFootnotes->setAttribute('id', 'readability-footnotes-list');
		$footnotesWrapper->appendChild($articleFootnotes);
		
		$articleLinks = $articleContent->getElementsByTagName('a');
		
		$linkCount = 0;
		for ($i = 0; $i < $articleLinks->length; $i++)
		{
			$articleLink  = $articleLinks->item($i);
			$footnoteLink = $articleLink->cloneNode(true);
			$refLink      = $this->dom->createElement('a');
			$footnote     = $this->dom->createElement('li');
			$linkDomain   = @parse_url($footnoteLink->getAttribute('href'), PHP_URL_HOST);
			if (!$linkDomain && isset($this->url)) $linkDomain = @parse_url($this->url, PHP_URL_HOST);
			//linkDomain   = footnoteLink.host ? footnoteLink.host : document.location.host,
			$linkText     = $this->getInnerText($articleLink);
			
			if ((strpos($articleLink->getAttribute('class'), 'readability-DoNotFootnote') !== false) || preg_match($this->regexps['skipFootnoteLink'], $linkText)) {
				continue;
			}
			
			$linkCount++;

			/** Add a superscript reference after the article link */
			$refLink->setAttribute('href', '#readabilityFootnoteLink-' . $linkCount);
			$refLink->innerHTML = '<small><sup>[' . $linkCount . ']</sup></small>';
			$refLink->setAttribute('class', 'readability-DoNotFootnote');
			$refLink->setAttribute('style', 'color: inherit;');
			
			//TODO: does this work or should we use DOMNode.isSameNode()?
			if ($articleLink->parentNode->lastChild == $articleLink) {
				$articleLink->parentNode->appendChild($refLink);
			} else {
				$articleLink->parentNode->insertBefore($refLink, $articleLink->nextSibling);
			}

			$articleLink->setAttribute('style', 'color: inherit; text-decoration: none;');
			$articleLink->setAttribute('name', 'readabilityLink-' . $linkCount);

			$footnote->innerHTML = '<small><sup><a href="#readabilityLink-' . $linkCount . '" title="Jump to Link in Article">^</a></sup></small> ';

			$footnoteLink->innerHTML = ($footnoteLink->getAttribute('title') != '' ? $footnoteLink->getAttribute('title') : $linkText);
			$footnoteLink->setAttribute('name', 'readabilityFootnoteLink-' . $linkCount);
			
			$footnote->appendChild($footnoteLink);
			if ($linkDomain) $footnote->innerHTML = $footnote->innerHTML . '<small> (' . $linkDomain . ')</small>';
			
			$articleFootnotes->appendChild($footnote);
		}

		if ($linkCount > 0) {
			$articleContent->appendChild($footnotesWrapper);           
		}
	}

	/**
	* Reverts P elements with class 'readability-styled'
	* to text nodes - which is what they were before.
	*
	* @param DOMElement
	* @return void
	*/
	function revertReadabilityStyledElements($articleContent) {
		$xpath = new DOMXPath($articleContent->ownerDocument);
		$elems = $xpath->query('.//p[@class="readability-styled"]', $articleContent);
		//$elems = $articleContent->getElementsByTagName('p');
		for ($i = $elems->length-1; $i >= 0; $i--) {
			$e = $elems->item($i);
			$e->parentNode->replaceChild($articleContent->ownerDocument->createTextNode($e->textContent), $e);
			//if ($e->hasAttribute('class') && $e->getAttribute('class') == 'readability-styled') {
			//	$e->parentNode->replaceChild($this->dom->createTextNode($e->textContent), $e);
			//}
		}
	}
	
	/**
	* Prepare the article node for display. Clean out any inline styles,
	* iframes, forms, strip extraneous <p> tags, etc.
	*
	* @param DOMElement
	* @return void
	*/
	function prepArticle($articleContent) {
		$this->cleanStyles($articleContent);
		$this->killBreaks($articleContent);
		if ($this->revertForcedParagraphElements) {
			$this->revertReadabilityStyledElements($articleContent);
		}

		/* Clean out junk from the article content */
		$this->cleanConditionally($articleContent, 'form');
		$this->clean($articleContent, 'object');
		$this->clean($articleContent, 'h1');

		/**
		* If there is only one h2, they are probably using it
		* as a header and not a subheader, so remove it since we already have a header.
		***/
		if (!$this->lightClean && ($articleContent->getElementsByTagName('h2')->length == 1)) {
			$this->clean($articleContent, 'h2'); 
		}
		$this->clean($articleContent, 'iframe');

		$this->cleanHeaders($articleContent);

		/* Do these last as the previous stuff may have removed junk that will affect these */
		$this->cleanConditionally($articleContent, 'table');
		$this->cleanConditionally($articleContent, 'ul');
		$this->cleanConditionally($articleContent, 'div');

		/* Remove extra paragraphs */
		$articleParagraphs = $articleContent->getElementsByTagName('p');
		for ($i = $articleParagraphs->length-1; $i >= 0; $i--)
		{
			$imgCount    = $articleParagraphs->item($i)->getElementsByTagName('img')->length;
			$embedCount  = $articleParagraphs->item($i)->getElementsByTagName('embed')->length;
			$objectCount = $articleParagraphs->item($i)->getElementsByTagName('object')->length;
			$iframeCount = $articleParagraphs->item($i)->getElementsByTagName('iframe')->length;
			
			if ($imgCount === 0 && $embedCount === 0 && $objectCount === 0 && $iframeCount === 0 && $this->getInnerText($articleParagraphs->item($i), false) == '')
			{
				$articleParagraphs->item($i)->parentNode->removeChild($articleParagraphs->item($i));
			}
		}

		try {
			$articleContent->innerHTML = preg_replace('/<br[^>]*>\s*<p/i', '<p', $articleContent->innerHTML);
			//articleContent.innerHTML = articleContent.innerHTML.replace(/<br[^>]*>\s*<p/gi, '<p');      
		}
		catch (Exception $e) {
			$this->dbg("Cleaning innerHTML of breaks failed. This is an IE strict-block-elements bug. Ignoring.: " . $e);
		}
	}
	
	/**
	* Initialize a node with the readability object. Also checks the
	* className/id for special names to add to its score.
	*
	* @param Element
	* @return void
	**/
	protected function initializeNode($node) {
		$readability = $this->dom->createAttribute('readability');
		$readability->value = 0; // this is our contentScore
		$node->setAttributeNode($readability);		         

		switch (strtoupper($node->tagName)) { // unsure if strtoupper is needed, but using it just in case
			case 'DIV':
				$readability->value += 5;
				break;

			case 'PRE':
			case 'TD':
			case 'BLOCKQUOTE':
				$readability->value += 3;
				break;
				
			case 'ADDRESS':
			case 'OL':
			case 'UL':
			case 'DL':
			case 'DD':
			case 'DT':
			case 'LI':
			case 'FORM':
				$readability->value -= 3;
				break;

			case 'H1':
			case 'H2':
			case 'H3':
			case 'H4':
			case 'H5':
			case 'H6':
			case 'TH':
				$readability->value -= 5;
				break;
		}
		$readability->value += $this->getClassWeight($node);
	}
	
	/***
	* grabArticle - Using a variety of metrics (content score, classname, element types), find the content that is
	*               most likely to be the stuff a user wants to read. Then return it wrapped up in a div.
	*
	* @return DOMElement
	**/
	protected function grabArticle($page=null) {
		$stripUnlikelyCandidates = $this->flagIsActive(self::FLAG_STRIP_UNLIKELYS);
		if (!$page) $page = $this->dom;
		$allElements = $page->getElementsByTagName('*');
		/**
		* First, node prepping. Trash nodes that look cruddy (like ones with the class name "comment", etc), and turn divs
		* into P tags where they have been used inappropriately (as in, where they contain no other block level elements.)
		*
		* Note: Assignment from index for performance. See http://www.peachpit.com/articles/article.aspx?p=31567&seqNum=5
		* TODO: Shouldn't this be a reverse traversal?
		**/
		$node = null;
		$nodesToScore = array();
		for ($nodeIndex = 0; ($node = $allElements->item($nodeIndex)); $nodeIndex++) {
		//for ($nodeIndex=$targetList->length-1; $nodeIndex >= 0; $nodeIndex--) {
			//$node = $targetList->item($nodeIndex);
			$tagName = strtoupper($node->tagName);
			/* Remove unlikely candidates */
			if ($stripUnlikelyCandidates) {
				$unlikelyMatchString = $node->getAttribute('class') . $node->getAttribute('id');
				if (
					preg_match($this->regexps['unlikelyCandidates'], $unlikelyMatchString) &&
					!preg_match($this->regexps['okMaybeItsACandidate'], $unlikelyMatchString) &&
					$tagName != 'BODY'
				)
				{
					$this->dbg('Removing unlikely candidate - ' . $unlikelyMatchString);
					//$nodesToRemove[] = $node;
					$node->parentNode->removeChild($node);
					$nodeIndex--;
					continue;
				}               
			}

			if ($tagName == 'P' || $tagName == 'TD' || $tagName == 'PRE') {
				$nodesToScore[] = $node;
			}

			/* Turn all divs that don't have children block level elements into p's */
			if ($tagName == 'DIV') {
				if (!preg_match($this->regexps['divToPElements'], $node->innerHTML)) {
					//$this->dbg('Altering div to p');
					$newNode = $this->dom->createElement('p');
					try {
						$newNode->innerHTML = $node->innerHTML;
						//$nodesToReplace[] = array('new'=>$newNode, 'old'=>$node);
						$node->parentNode->replaceChild($newNode, $node);
						$nodeIndex--;
						$nodesToScore[] = $node; // or $newNode?
					}
					catch(Exception $e) {
						$this->dbg('Could not alter div to p, reverting back to div.: ' . $e);
					}
				}
				else
				{
					/* EXPERIMENTAL */
					// TODO: change these p elements back to text nodes after processing
					for ($i = 0, $il = $node->childNodes->length; $i < $il; $i++) {
						$childNode = $node->childNodes->item($i);
						if ($childNode->nodeType == 3) { // XML_TEXT_NODE
							//$this->dbg('replacing text node with a p tag with the same content.');
							$p = $this->dom->createElement('p');
							$p->innerHTML = $childNode->nodeValue;
							$p->setAttribute('style', 'display: inline;');
							$p->setAttribute('class', 'readability-styled');
							$childNode->parentNode->replaceChild($p, $childNode);
						}
					}
				}
			}
		}
		
		/**
		* Loop through all paragraphs, and assign a score to them based on how content-y they look.
		* Then add their score to their parent node.
		*
		* A score is determined by things like number of commas, class names, etc. Maybe eventually link density.
		**/
		$candidates = array();
		for ($pt=0; $pt < count($nodesToScore); $pt++) {
			$parentNode      = $nodesToScore[$pt]->parentNode;
			// $grandParentNode = $parentNode ? $parentNode->parentNode : null;
			$grandParentNode = !$parentNode ? null : (($parentNode->parentNode instanceof DOMElement) ? $parentNode->parentNode : null);
			$innerText       = $this->getInnerText($nodesToScore[$pt]);

			if (!$parentNode || !isset($parentNode->tagName)) {
				continue;
			}

			/* If this paragraph is less than 25 characters, don't even count it. */
			if(strlen($innerText) < 25) {
				continue;
			}

			/* Initialize readability data for the parent. */
			if (!$parentNode->hasAttribute('readability')) 
			{
				$this->initializeNode($parentNode);
				$candidates[] = $parentNode;
			}

			/* Initialize readability data for the grandparent. */
			if ($grandParentNode && !$grandParentNode->hasAttribute('readability') && isset($grandParentNode->tagName))
			{
				$this->initializeNode($grandParentNode);
				$candidates[] = $grandParentNode;
			}

			$contentScore = 0;

			/* Add a point for the paragraph itself as a base. */
			$contentScore++;

			/* Add points for any commas within this paragraph */
			$contentScore += count(explode(',', $innerText));
			
			/* For every 100 characters in this paragraph, add another point. Up to 3 points. */
			$contentScore += min(floor(strlen($innerText) / 100), 3);
			
			/* Add the score to the parent. The grandparent gets half. */
			$parentNode->getAttributeNode('readability')->value += $contentScore;

			if ($grandParentNode) {
				$grandParentNode->getAttributeNode('readability')->value += $contentScore/2;             
			}
		}

		/**
		* After we've calculated scores, loop through all of the possible candidate nodes we found
		* and find the one with the highest score.
		**/
		$topCandidate = null;
		for ($c=0, $cl=count($candidates); $c < $cl; $c++)
		{
			/**
			* Scale the final candidates score based on link density. Good content should have a
			* relatively small link density (5% or less) and be mostly unaffected by this operation.
			**/
			$readability = $candidates[$c]->getAttributeNode('readability');
			$readability->value = $readability->value * (1-$this->getLinkDensity($candidates[$c]));

			$this->dbg('Candidate: ' . $candidates[$c]->tagName . ' (' . $candidates[$c]->getAttribute('class') . ':' . $candidates[$c]->getAttribute('id') . ') with score ' . $readability->value);

			if (!$topCandidate || $readability->value > (int)$topCandidate->getAttribute('readability')) {
				$topCandidate = $candidates[$c];
			}
		}

		/**
		* If we still have no top candidate, just use the body as a last resort.
		* We also have to copy the body node so it is something we can modify.
		**/
		if ($topCandidate === null || strtoupper($topCandidate->tagName) == 'BODY')
		{
			$topCandidate = $this->dom->createElement('div');
			if ($page instanceof DOMDocument) {
				if (!isset($page->documentElement)) {
					// we don't have a body either? what a mess! :)
				} else {
					$topCandidate->innerHTML = $page->documentElement->innerHTML;
					$page->documentElement->innerHTML = '';
					$this->reinitBody();
					$page->documentElement->appendChild($topCandidate);
				}
			} else {
				$topCandidate->innerHTML = $page->innerHTML;
				$page->innerHTML = '';
				$page->appendChild($topCandidate);
			}
			$this->initializeNode($topCandidate);
		}

		/**
		* Now that we have the top candidate, look through its siblings for content that might also be related.
		* Things like preambles, content split by ads that we removed, etc.
		**/
		$articleContent        = $this->dom->createElement('div');
		$articleContent->setAttribute('id', 'readability-content');
		$siblingScoreThreshold = max(10, ((int)$topCandidate->getAttribute('readability')) * 0.2);
		$siblingNodes          = @$topCandidate->parentNode->childNodes;
		if (!isset($siblingNodes)) {
			$siblingNodes = new stdClass;
			$siblingNodes->length = 0;
		}

		for ($s=0, $sl=$siblingNodes->length; $s < $sl; $s++)
		{
			$siblingNode = $siblingNodes->item($s);
			$append      = false;

			$this->dbg('Looking at sibling node: ' . $siblingNode->nodeName . (($siblingNode->nodeType === XML_ELEMENT_NODE && $siblingNode->hasAttribute('readability')) ? (' with score ' . $siblingNode->getAttribute('readability')) : ''));

			//dbg('Sibling has score ' . ($siblingNode->readability ? siblingNode.readability.contentScore : 'Unknown'));

			if ($siblingNode === $topCandidate)
			// or if ($siblingNode->isSameNode($topCandidate))
			{
				$append = true;
			}

			$contentBonus = 0;
			/* Give a bonus if sibling nodes and top candidates have the example same classname */
			if ($siblingNode->nodeType === XML_ELEMENT_NODE && $siblingNode->getAttribute('class') == $topCandidate->getAttribute('class') && $topCandidate->getAttribute('class') != '') {
				$contentBonus += ((int)$topCandidate->getAttribute('readability')) * 0.2;
			}

			if ($siblingNode->nodeType === XML_ELEMENT_NODE && $siblingNode->hasAttribute('readability') && (((int)$siblingNode->getAttribute('readability')) + $contentBonus) >= $siblingScoreThreshold)
			{
				$append = true;
			}
			
			if (strtoupper($siblingNode->nodeName) == 'P') {
				$linkDensity = $this->getLinkDensity($siblingNode);
				$nodeContent = $this->getInnerText($siblingNode);
				$nodeLength  = strlen($nodeContent);
				
				if ($nodeLength > 80 && $linkDensity < 0.25)
				{
					$append = true;
				}
				else if ($nodeLength < 80 && $linkDensity === 0 && preg_match('/\.( |$)/', $nodeContent))
				{
					$append = true;
				}
			}

			if ($append)
			{
				$this->dbg('Appending node: ' . $siblingNode->nodeName);

				$nodeToAppend = null;
				$sibNodeName = strtoupper($siblingNode->nodeName);
				if ($sibNodeName != 'DIV' && $sibNodeName != 'P') {
					/* We have a node that isn't a common block level element, like a form or td tag. Turn it into a div so it doesn't get filtered out later by accident. */
					
					$this->dbg('Altering siblingNode of ' . $sibNodeName . ' to div.');
					$nodeToAppend = $this->dom->createElement('div');
					try {
						$nodeToAppend->setAttribute('id', $siblingNode->getAttribute('id'));
						$nodeToAppend->innerHTML = $siblingNode->innerHTML;
					}
					catch(Exception $e)
					{
						$this->dbg('Could not alter siblingNode to div, reverting back to original.');
						$nodeToAppend = $siblingNode;
						$s--;
						$sl--;
					}
				} else {
					$nodeToAppend = $siblingNode;
					$s--;
					$sl--;
				}
				
				/* To ensure a node does not interfere with readability styles, remove its classnames */
				$nodeToAppend->removeAttribute('class');

				/* Append sibling and subtract from our list because it removes the node when you append to another node */
				$articleContent->appendChild($nodeToAppend);
			}
		}

		/**
		* So we have all of the content that we need. Now we clean it up for presentation.
		**/
		$this->prepArticle($articleContent);

		/**
		* Now that we've gone through the full algorithm, check to see if we got any meaningful content.
		* If we didn't, we may need to re-run grabArticle with different flags set. This gives us a higher
		* likelihood of finding the content, and the sieve approach gives us a higher likelihood of
		* finding the -right- content.
		**/
		if (strlen($this->getInnerText($articleContent, false)) < 250)
		{
			// TODO: find out why element disappears sometimes, e.g. for this URL http://www.businessinsider.com/6-hedge-fund-etfs-for-average-investors-2011-7
			// in the meantime, we check and create an empty element if it's not there.
			$this->reinitBody();
			
			if ($this->flagIsActive(self::FLAG_STRIP_UNLIKELYS)) {
				$this->removeFlag(self::FLAG_STRIP_UNLIKELYS);
				return $this->grabArticle($this->body);
			}
			else if ($this->flagIsActive(self::FLAG_WEIGHT_CLASSES)) {
				$this->removeFlag(self::FLAG_WEIGHT_CLASSES);
				return $this->grabArticle($this->body);              
			}
			else if ($this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) {
				$this->removeFlag(self::FLAG_CLEAN_CONDITIONALLY);
				return $this->grabArticle($this->body);
			}
			else {
				return false;
			}
		}
		return $articleContent;
	}
	
	/**
	* Remove script tags from document
	*
	* @param DOMElement
	* @return void
	*/
	public function removeScripts($doc) {
		$scripts = $doc->getElementsByTagName('script');
		for($i = $scripts->length-1; $i >= 0; $i--)
		{
			$scripts->item($i)->parentNode->removeChild($scripts->item($i));
		}
	}
	
	/**
	* Get the inner text of a node.
	* This also strips out any excess whitespace to be found.
	*
	* @param DOMElement $
	* @param boolean $normalizeSpaces (default: true)
	* @return string
	**/
	public function getInnerText($e, $normalizeSpaces=true) {
		$textContent = '';

		if (!isset($e->textContent) || $e->textContent == '') {
			return '';
		}

		$textContent = trim($e->textContent);

		if ($normalizeSpaces) {
			return preg_replace($this->regexps['normalize'], ' ', $textContent);
		} else {
			return $textContent;
		}
	}

	/**
	* Get the number of times a string $s appears in the node $e.
	*
	* @param DOMElement $e
	* @param string - what to count. Default is ","
	* @return number (integer)
	**/
	public function getCharCount($e, $s=',') {
		return substr_count($this->getInnerText($e), $s);
	}

	/**
	* Remove the style attribute on every $e and under.
	*
	* @param DOMElement $e
	* @return void
	*/
	public function cleanStyles($e) {
		if (!is_object($e)) return;
		$elems = $e->getElementsByTagName('*');
		foreach ($elems as $elem) {
			$elem->removeAttribute('style');
		}
	}
	
	/**
	* Get the density of links as a percentage of the content
	* This is the amount of text that is inside a link divided by the total text in the node.
	* 
	* @param DOMElement $e
	* @return number (float)
	*/
	public function getLinkDensity($e) {
		$links      = $e->getElementsByTagName('a');
		$textLength = strlen($this->getInnerText($e));
		$linkLength = 0;
		for ($i=0, $il=$links->length; $i < $il; $i++)
		{
			$linkLength += strlen($this->getInnerText($links->item($i)));
		}
		if ($textLength > 0) {
			return $linkLength / $textLength;
		} else {
			return 0;
		}
	}
	
	/**
	* Get an elements class/id weight. Uses regular expressions to tell if this 
	* element looks good or bad.
	*
	* @param DOMElement $e
	* @return number (Integer)
	*/
	public function getClassWeight($e) {
		if(!$this->flagIsActive(self::FLAG_WEIGHT_CLASSES)) {
			return 0;
		}

		$weight = 0;

		/* Look for a special classname */
		if ($e->hasAttribute('class') && $e->getAttribute('class') != '')
		{
			if (preg_match($this->regexps['negative'], $e->getAttribute('class'))) {
				$weight -= 25;
			}
			if (preg_match($this->regexps['positive'], $e->getAttribute('class'))) {
				$weight += 25;
			}
		}

		/* Look for a special ID */
		if ($e->hasAttribute('id') && $e->getAttribute('id') != '')
		{
			if (preg_match($this->regexps['negative'], $e->getAttribute('id'))) {
				$weight -= 25;
			}
			if (preg_match($this->regexps['positive'], $e->getAttribute('id'))) {
				$weight += 25;
			}
		}
		return $weight;
	}

	/**
	* Remove extraneous break tags from a node.
	*
	* @param DOMElement $node
	* @return void
	*/
	public function killBreaks($node) {
		$html = $node->innerHTML;
		$html = preg_replace($this->regexps['killBreaks'], '<br />', $html);
		$node->innerHTML = $html;
	}

	/**
	* Clean a node of all elements of type "tag".
	* (Unless it's a youtube/vimeo video. People love movies.)
	*
	* Updated 2012-09-18 to preserve youtube/vimeo iframes
	*
	* @param DOMElement $e
	* @param string $tag
	* @return void
	*/
	public function clean($e, $tag) {
		$targetList = $e->getElementsByTagName($tag);
		$isEmbed = ($tag == 'iframe' || $tag == 'object' || $tag == 'embed');
		
		for ($y=$targetList->length-1; $y >= 0; $y--) {
			/* Allow youtube and vimeo videos through as people usually want to see those. */
			if ($isEmbed) {
				$attributeValues = '';
				for ($i=0, $il=$targetList->item($y)->attributes->length; $i < $il; $i++) {
					$attributeValues .= $targetList->item($y)->attributes->item($i)->value . '|'; // DOMAttr? (TODO: test)
				}
				
				/* First, check the elements attributes to see if any of them contain youtube or vimeo */
				if (preg_match($this->regexps['video'], $attributeValues)) {
					continue;
				}

				/* Then check the elements inside this element for the same. */
				if (preg_match($this->regexps['video'], $targetList->item($y)->innerHTML)) {
					continue;
				}
			}
			$targetList->item($y)->parentNode->removeChild($targetList->item($y));
		}
	}
	
	/**
	* Clean an element of all tags of type "tag" if they look fishy.
	* "Fishy" is an algorithm based on content length, classnames, 
	* link density, number of images & embeds, etc.
	*
	* @param DOMElement $e
	* @param string $tag
	* @return void
	*/
	public function cleanConditionally($e, $tag) {
		if (!$this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) {
			return;
		}

		$tagsList = $e->getElementsByTagName($tag);
		$curTagsLength = $tagsList->length;

		/**
		* Gather counts for other typical elements embedded within.
		* Traverse backwards so we can remove nodes at the same time without effecting the traversal.
		*
		* TODO: Consider taking into account original contentScore here.
		*/
		for ($i=$curTagsLength-1; $i >= 0; $i--) {
			$weight = $this->getClassWeight($tagsList->item($i));
			$contentScore = ($tagsList->item($i)->hasAttribute('readability')) ? (int)$tagsList->item($i)->getAttribute('readability') : 0;
			
			$this->dbg('Cleaning Conditionally ' . $tagsList->item($i)->tagName . ' (' . $tagsList->item($i)->getAttribute('class') . ':' . $tagsList->item($i)->getAttribute('id') . ')' . (($tagsList->item($i)->hasAttribute('readability')) ? (' with score ' . $tagsList->item($i)->getAttribute('readability')) : ''));

			if ($weight + $contentScore < 0) {
				$tagsList->item($i)->parentNode->removeChild($tagsList->item($i));
			}
			else if ( $this->getCharCount($tagsList->item($i), ',') < 10) {
				/**
				* If there are not very many commas, and the number of
				* non-paragraph elements is more than paragraphs or other ominous signs, remove the element.
				**/
				$p      = $tagsList->item($i)->getElementsByTagName('p')->length;
				$img    = $tagsList->item($i)->getElementsByTagName('img')->length;
				$li     = $tagsList->item($i)->getElementsByTagName('li')->length-100;
				$input  = $tagsList->item($i)->getElementsByTagName('input')->length;
				$a 		= $tagsList->item($i)->getElementsByTagName('a')->length;

				$embedCount = 0;
				$embeds = $tagsList->item($i)->getElementsByTagName('embed');
				for ($ei=0, $il=$embeds->length; $ei < $il; $ei++) {
					if (preg_match($this->regexps['video'], $embeds->item($ei)->getAttribute('src'))) {
						$embedCount++; 
					}
				}
				$embeds = $tagsList->item($i)->getElementsByTagName('iframe');
				for ($ei=0, $il=$embeds->length; $ei < $il; $ei++) {
					if (preg_match($this->regexps['video'], $embeds->item($ei)->getAttribute('src'))) {
						$embedCount++; 
					}
				}

				$linkDensity   = $this->getLinkDensity($tagsList->item($i));
				$contentLength = strlen($this->getInnerText($tagsList->item($i)));
				$toRemove      = false;

				if ($this->lightClean) {
					$this->dbg('Light clean...');
					if ( ($img > $p) && ($img > 4) ) {
						$this->dbg(' more than 4 images and more image elements than paragraph elements');
						$toRemove = true;
					} else if ($li > $p && $tag != 'ul' && $tag != 'ol') {
						$this->dbg(' too many <li> elements, and parent is not <ul> or <ol>');
						$toRemove = true;
					} else if ( $input > floor($p/3) ) {
						$this->dbg(' too many <input> elements');
						$toRemove = true; 
					} else if ($contentLength < 10 && ($embedCount === 0 && ($img === 0 || $img > 2))) {
						$this->dbg(' content length less than 10 chars, 0 embeds and either 0 images or more than 2 images');
						$toRemove = true;
					} else if($weight < 25 && $linkDensity > 0.2) {
						$this->dbg(' weight smaller than 25 and link density above 0.2');
						$toRemove = true;
					} else if($a > 2 && ($weight >= 25 && $linkDensity > 0.5)) {
						$this->dbg(' more than 2 links and weight above 25 but link density greater than 0.5');
						$toRemove = true;
					} else if($embedCount > 3) {
						$this->dbg(' more than 3 embeds');
						$toRemove = true;
					}
				} else {
					$this->dbg('Standard clean...');
					if ( $img > $p ) {
						$this->dbg(' more image elements than paragraph elements');
						$toRemove = true;
					} else if ($li > $p && $tag != 'ul' && $tag != 'ol') {
						$this->dbg(' too many <li> elements, and parent is not <ul> or <ol>');
						$toRemove = true;
					} else if ( $input > floor($p/3) ) {
						$this->dbg(' too many <input> elements');
						$toRemove = true; 
					} else if ($contentLength < 25 && ($img === 0 || $img > 2) ) {
						$this->dbg(' content length less than 25 chars and 0 images, or more than 2 images');
						$toRemove = true;
					} else if($weight < 25 && $linkDensity > 0.2) {
						$this->dbg(' weight smaller than 25 and link density above 0.2');
						$toRemove = true;
					} else if($weight >= 25 && $linkDensity > 0.5) {
						$this->dbg(' weight above 25 but link density greater than 0.5');
						$toRemove = true;
					} else if(($embedCount == 1 && $contentLength < 75) || $embedCount > 1) {
						$this->dbg(' 1 embed and content length smaller than 75 chars, or more than one embed');
						$toRemove = true;
					}
				}

				if ($toRemove) {
					//$this->dbg('Removing: '.$tagsList->item($i)->innerHTML);
					$tagsList->item($i)->parentNode->removeChild($tagsList->item($i));
				}
			}
		}
	}

	/**
	* Clean out spurious headers from an Element. Checks things like classnames and link density.
	*
	* @param DOMElement $e
	* @return void
	*/
	public function cleanHeaders($e) {
		for ($headerIndex = 1; $headerIndex < 3; $headerIndex++) {
			$headers = $e->getElementsByTagName('h' . $headerIndex);
			for ($i=$headers->length-1; $i >=0; $i--) {
				if ($this->getClassWeight($headers->item($i)) < 0 || $this->getLinkDensity($headers->item($i)) > 0.33) {
					$headers->item($i)->parentNode->removeChild($headers->item($i));
				}
			}
		}
	}

	public function flagIsActive($flag) {
		return ($this->flags & $flag) > 0;
	}
	
	public function addFlag($flag) {
		$this->flags = $this->flags | $flag;
	}
	
	public function removeFlag($flag) {
		$this->flags = $this->flags & ~$flag;
	}
	
	/**
	 * Will recreate previously deleted body property
	 *
	 * @return void
	 */
	protected function reinitBody() {
		if (!isset($this->body->childNodes)) {
			$this->body = $this->dom->createElement('body');
			$this->body->innerHTML = $this->bodyCache;
		}
	}
		
}
?>