diff options
-rw-r--r-- | src/Configuration.php | 60 | ||||
-rw-r--r-- | src/Environment.php | 36 | ||||
-rw-r--r-- | src/HTMLParser.php | 1598 | ||||
-rw-r--r-- | src/Readability.php | 548 | ||||
-rw-r--r-- | src/ReadabilityInterface.php | 92 |
5 files changed, 0 insertions, 2334 deletions
diff --git a/src/Configuration.php b/src/Configuration.php index d2fa6a7..e69de29 100644 --- a/src/Configuration.php +++ b/src/Configuration.php @@ -1,60 +0,0 @@ -<?php - -namespace andreskrey\Readability; - -class Configuration -{ - protected $config; - - /** - * @param array $config - */ - public function __construct(array $config = []) - { - $this->config = $config; - } - - /** - * @param array $config - */ - public function merge(array $config = []) - { - $this->config = array_replace_recursive($this->config, $config); - } - - /** - * @param array $config - */ - public function replace(array $config = []) - { - $this->config = $config; - } - - /** - * @param string $key - * @param mixed $value - */ - public function setOption($key, $value) - { - $this->config[$key] = $value; - } - - /** - * @param string|null $key - * @param mixed|null $default - * - * @return mixed|null - */ - public function getOption($key = null, $default = null) - { - if ($key === null) { - return $this->config; - } - - if (!isset($this->config[$key])) { - return $default; - } - - return $this->config[$key]; - } -} diff --git a/src/Environment.php b/src/Environment.php deleted file mode 100644 index 6e88783..0000000 --- a/src/Environment.php +++ /dev/null @@ -1,36 +0,0 @@ -<?php - -namespace andreskrey\Readability; - -final class Environment -{ - /** - * @var Configuration - */ - protected $config; - - public function __construct(array $config = []) - { - $this->config = new Configuration($config); - } - - /** - * @return Configuration - */ - public function getConfig() - { - return $this->config; - } - - /** - * @param array $config - * - * @return Environment - */ - public static function createDefaultEnvironment(array $config = []) - { - $environment = new static($config); - - return $environment; - } -} diff --git a/src/HTMLParser.php b/src/HTMLParser.php deleted file mode 100644 index e4ed4bb..0000000 --- a/src/HTMLParser.php +++ /dev/null @@ -1,1598 +0,0 @@ -<?php - -namespace andreskrey\Readability; - -use DOMDocument; - -/** - * Class HTMLParser. - * - * A helper class to parse HTML and get a Readability object. - */ -class HTMLParser -{ - /** - * @var DOMDocument - */ - private $dom = null; - - /** - * TODO Make this an object? Instead of a dumb array. - * - * @var array - */ - private $metadata = []; - - /** - * @var array - */ - private $regexps = [ - 'unlikelyCandidates' => '/banner|breadcrumbs|combx|comment|community|cover-wrap|disqus|extra|foot|header|legends|menu|modal|related|remark|replies|rss|shoutbox|sidebar|skyscraper|social|sponsor|supplemental|ad-break|agegate|pagination|pager|popup|yom-remote/i', - 'okMaybeItsACandidate' => '/and|article|body|column|main|shadow/i', - 'extraneous' => '/print|archive|comment|discuss|e[\-]?mail|share|reply|all|login|sign|single|utility/i', - 'byline' => '/byline|author|dateline|writtenby|p-author/i', - 'replaceFonts' => '/<(\/?)font[^>]*>/gi', - 'normalize' => '/\s{2,}/', - 'videos' => '/\/\/(www\.)?(dailymotion|youtube|youtube-nocookie|player\.vimeo)\.com/i', - 'nextLink' => '/(next|weiter|continue|>([^\|]|$)|»([^\|]|$))/i', - 'prevLink' => '/(prev|earl|old|new|<|«)/i', - 'whitespace' => '/^\s*$/', - 'hasContent' => '/\S$/', - // \x{00A0} is the unicode version of - 'onlyWhitespace' => '/\x{00A0}|\s+/u' - ]; - - private $defaultTagsToScore = [ - 'section', - 'h2', - 'h3', - 'h4', - 'h5', - 'h6', - 'p', - 'td', - 'pre', - ]; - - /** - * @var array - */ - private $alterToDIVExceptions = [ - 'div', - 'article', - 'section', - 'p', - // TODO, check if this is correct, #text elements do not exist in js - '#text', - ]; - - /** - * @var array - */ - private $divToPElements = [ - 'a', - 'blockquote', - 'dl', - 'div', - 'img', - 'ol', - 'p', - 'pre', - 'table', - 'ul', - 'select', - ]; - - /** - * Constructor. - * - * @param array $options Options to override the default ones - */ - public function __construct(array $options = []) - { - $defaults = [ - 'maxTopCandidates' => 5, - 'wordThreshold' => 500, - 'articleByLine' => false, - 'stripUnlikelyCandidates' => true, - 'cleanConditionally' => true, - 'weightClasses' => true, - 'removeReadabilityTags' => true, - 'fixRelativeURLs' => false, - 'substituteEntities' => true, - 'normalizeEntities' => false, - 'summonCthulhu' => false, - 'originalURL' => 'http://fakehost', - ]; - - $this->environment = Environment::createDefaultEnvironment($defaults); - - $this->environment->getConfig()->merge($options); - - // To avoid having a gazillion of errors on malformed HTMLs - libxml_use_internal_errors(true); - } - - /** - * Parse the html. This is the main entry point of the HTMLParser. - * - * @param string $html Full html of the website, page, etc. - * - * #return ? TBD - */ - public function parse($html) - { - $this->dom = $this->loadHTML($html); - - $this->metadata = $this->getMetadata(); - - $this->metadata['image'] = $this->getMainImage(); - - // Checking for minimum HTML to work with. - if (!($root = $this->dom->getElementsByTagName('body')->item(0)) || !$root->firstChild) { - return false; - } - - $parseSuccessful = true; - while (true) { - $root = new Readability($root->firstChild); - - $elementsToScore = $this->getNodes($root); - - $result = $this->rateNodes($elementsToScore); - - /* - * 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. - */ - - // TODO Better way to count resulting text. Textcontent usually has alt titles and that stuff - // that doesn't really count to the quality of the result. - $length = 0; - foreach ($result->getElementsByTagName('p') as $p) { - $length += mb_strlen($p->textContent); - } - if ($result && mb_strlen(preg_replace('/\s/', '', $result->textContent)) < $this->getConfig()->getOption('wordThreshold')) { - $this->dom = $this->loadHTML($html); - $root = $this->dom->getElementsByTagName('body')->item(0); - - if ($this->getConfig()->getOption('stripUnlikelyCandidates')) { - $this->getConfig()->setOption('stripUnlikelyCandidates', false); - } elseif ($this->getConfig()->getOption('weightClasses')) { - $this->getConfig()->setOption('weightClasses', false); - } elseif ($this->getConfig()->getOption('cleanConditionally')) { - $this->getConfig()->setOption('cleanConditionally', false); - } else { - $parseSuccessful = false; - break; - } - } else { - break; - } - } - - if (!$parseSuccessful) { - return false; - } - - $result = $this->postProcessContent($result); - - // Todo, fix return, check for values, maybe create a function to create the return object - return [ - 'title' => isset($this->metadata['title']) ? $this->metadata['title'] : null, - 'author' => isset($this->metadata['author']) ? $this->metadata['author'] : null, - 'image' => isset($this->metadata['image']) ? $this->metadata['image'] : null, - 'images' => $this->getImages(), - 'article' => $result, - 'html' => $result->C14N(), - 'dir' => isset($this->metadata['articleDir']) ? $this->metadata['articleDir'] : null, - ]; - } - - /** - * Creates a DOM Document object and loads the provided HTML on it. - * - * Used for the first load of Readability and subsequent reloads (when disabling flags and rescanning the text) - * Previous versions of Readability used this method one time and cloned the DOM to keep a backup. This caused bugs - * because cloning the DOM object keeps a relation between the clone and the original one, doing changes in both - * objects and ruining the backup. - * - * @param string $html - * - * @return DOMDocument - */ - private function loadHTML($html) - { - $dom = new DOMDocument('1.0', 'utf-8'); - - if (!$this->getConfig()->getOption('substituteEntities')) { - // Keep the original HTML entities - $dom->substituteEntities = false; - } - - if ($this->getConfig()->getOption('normalizeEntities')) { - // Replace UTF-8 characters with the HTML Entity equivalent. Useful to fix html with mixed content - $html = mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'); - } - - if ($this->getConfig()->getOption('summonCthulhu')) { - $html = preg_replace('/<script\b[^>]*>([\s\S]*?)<\/script>/', '', $html); - } - - // Prepend the XML tag to avoid having issues with special characters. Should be harmless. - $dom->loadHTML('<?xml encoding="UTF-8">' . $html); - $dom->encoding = 'UTF-8'; - - $this->removeScripts($dom); - - $this->prepDocument($dom); - - return $dom; - } - - /** - * @return Configuration - */ - public function getConfig() - { - return $this->environment->getConfig(); - } - - /** - * Removes all the scripts of the html. - * - * @param DOMDocument $dom - */ - private function removeScripts(DOMDocument $dom) - { - $toRemove = ['script', 'noscript']; - - foreach ($toRemove as $tag) { - while ($script = $dom->getElementsByTagName($tag)) { - if ($script->item(0)) { - $script->item(0)->parentNode->removeChild($script->item(0)); - } else { - break; - } - } - } - } - - /** - * Prepares the document for parsing. - * - * @param DOMDocument $dom - */ - private function prepDocument(DOMDocument $dom) - { - /* - * DOMNodeList must be converted to an array before looping over it. - * This is done to avoid node shifting when removing nodes. - * - * Reverse traversing cannot be done here because we need to find brs that are right next to other brs. - * (If we go the other way around we need to search for previous nodes forcing the creation of new functions - * that will be used only here) - */ - foreach (iterator_to_array($dom->getElementsByTagName('br')) as $br) { - $next = $br->nextSibling; - - /* - * Whether 2 or more <br> elements have been found and replaced with a - * <p> block. - */ - $replaced = false; - - /* - * If we find a <br> chain, remove the <br>s until we hit another element - * or non-whitespace. This leaves behind the first <br> in the chain - * (which will be replaced with a <p> later). - */ - while (($next = $this->nextElement($next)) && ($next->nodeName === 'br')) { - $replaced = true; - $brSibling = $next->nextSibling; - $next->parentNode->removeChild($next); - $next = $brSibling; - } - - /* - * If we removed a <br> chain, replace the remaining <br> with a <p>. Add - * all sibling nodes as children of the <p> until we hit another <br> - * chain. - */ - - if ($replaced) { - $p = $dom->createElement('p'); - $br->parentNode->replaceChild($p, $br); - - $next = $p->nextSibling; - while ($next) { - // If we've hit another <br><br>, we're done adding children to this <p>. - if ($next->nodeName === 'br') { - $nextElem = $this->nextElement($next); - if ($nextElem && $nextElem->nodeName === 'br') { - break; - } - } - - // Otherwise, make this node a child of the new <p>. - $sibling = $next->nextSibling; - $p->appendChild($next); - $next = $sibling; - } - } - } - - // Replace font tags with span - $fonts = $dom->getElementsByTagName('font'); - $length = $fonts->length; - for ($i = 0; $i < $length; $i++) { - $font = $fonts->item($length - 1 - $i); - $span = new Readability($font); - $span->setNodeTag('span', true); - } - } - - public function postProcessContent(DOMDocument $article) - { - // Readability cannot open relative uris so we convert them to absolute uris. - if ($this->getConfig()->getOption('fixRelativeURLs')) { - foreach (iterator_to_array($article->getElementsByTagName('a')) as $link) { - /** @var \DOMElement $link */ - $href = $link->getAttribute('href'); - if ($href) { - // Replace links with javascript: URIs with text content, since - // they won't work after scripts have been removed from the page. - if (strpos($href, 'javascript:') === 0) { - $text = $article->createTextNode($link->textContent); - $link->parentNode->replaceChild($text, $link); - } else { - $link->setAttribute('href', $this->toAbsoluteURI($href)); - } - } - } - - foreach ($article->getElementsByTagName('img') as $img) { - /** @var \DOMElement $img */ - $src = $img->getAttribute('src'); - if ($src) { - $img->setAttribute('src', $this->toAbsoluteURI($src)); - } - } - } - - return $article; - } - - private function toAbsoluteURI($uri) - { - list($pathBase, $scheme, $prePath) = $this->getPathInfo($this->getConfig()->getOption('originalURL')); - - // If this is already an absolute URI, return it. - if (preg_match('/^[a-zA-Z][a-zA-Z0-9\+\-\.]*:/', $uri)) { - return $uri; - } - - // Scheme-rooted relative URI. - if (substr($uri, 0, 2) === '//') { - return $scheme . '://' . substr($uri, 2); - } - - // Prepath-rooted relative URI. - if (substr($uri, 0, 1) === '/') { - return $prePath . $uri; - } - - // Dotslash relative URI. - if (strpos($uri, './') === 0) { - return $pathBase . substr($uri, 2); - } - // Ignore hash URIs: - if (substr($uri, 0, 1) === '#') { - return $uri; - } - - // Standard relative URI; add entire path. pathBase already includes a - // trailing "/". - return $pathBase . $uri; - } - - /** - * @param string $url - * - * @return array [$pathBase, $scheme, $prePath] - */ - public function getPathInfo($url) - { - $pathBase = parse_url($url, PHP_URL_SCHEME) . '://' . parse_url($url, PHP_URL_HOST) . dirname(parse_url($url, PHP_URL_PATH)) . '/'; - $scheme = parse_url($pathBase, PHP_URL_SCHEME); - $prePath = $scheme . '://' . parse_url($pathBase, PHP_URL_HOST); - - return [$pathBase, $scheme, $prePath]; - } - - private function nextElement($node) - { - $next = $node; - while ($next - && $next->nodeName !== '#text' - && trim($next->textContent)) { - $next = $next->nextSibling; - } - - return $next; - } - - /** - * Tries to guess relevant info from metadata of the html. - * - * @return array Metadata info. May have title, excerpt and or byline. - */ - private function getMetadata() - { - $metadata = $values = []; - // Match "description", or Twitter's "twitter:description" (Cards) - // in name attribute. - $namePattern = '/^\s*((twitter)\s*:\s*)?(description|title|image)\s*$/i'; - - // Match Facebook's Open Graph title & description properties. - $propertyPattern = '/^\s*og\s*:\s*(description|title|image)\s*$/i'; - - foreach ($this->dom->getElementsByTagName('meta') as $meta) { - /* @var Readability $meta */ - $elementName = $meta->getAttribute('name'); - $elementProperty = $meta->getAttribute('property'); - - if (in_array('author', [$elementName, $elementProperty])) { - $metadata['byline'] = $meta->getAttribute('content'); - continue; - } - - $name = null; - if (preg_match($namePattern, $elementName)) { - $name = $elementName; - } elseif (preg_match($propertyPattern, $elementProperty)) { - $name = $elementProperty; - } - - if ($name) { - $content = $meta->getAttribute('content'); - if ($content) { - // Convert to lowercase and remove any whitespace - // so we can match below. - $name = preg_replace('/\s/', '', strtolower($name)); - $values[$name] = trim($content); - } - } - } - if (array_key_exists('description', $values)) { - $metadata['excerpt'] = $values['description']; - } elseif (array_key_exists('og:description', $values)) { - // Use facebook open graph description. - $metadata['excerpt'] = $values['og:description']; - } elseif (array_key_exists('twitter:description', $values)) { - // Use twitter cards description. - $metadata['excerpt'] = $values['twitter:description']; - } - - $metadata['title'] = $this->getTitle(); - - if (!$metadata['title']) { - if (array_key_exists('og:title', $values)) { - // Use facebook open graph title. - $metadata['title'] = $values['og:title']; - } elseif (array_key_exists('twitter:title', $values)) { - // Use twitter cards title. - $metadata['title'] = $values['twitter:title']; - } - } - - if (array_key_exists('og:image', $values) || array_key_exists('twitter:image', $values)) { - $metadata['image'] = array_key_exists('og:image', $values) ? $values['og:image'] : $values['twitter:image']; - } else { - $metadata['image'] = null; - } - - return $metadata; - } - - /** - * Tries to get the main article image. Will only update the metadata if the getMetadata function couldn't - * find a correct image. - * - * @return bool|string URL of the top image or false if unsuccessful. - */ - public function getMainImage() - { - $imgUrl = false; - - if ($this->metadata['image'] !== null) { - $imgUrl = $this->metadata['image']; - } - - if (!$imgUrl) { - foreach ($this->dom->getElementsByTagName('link') as $link) { - /** @var \DOMElement $link */ - /* - * Check for the rel attribute, then check if the rel attribute is either img_src or image_src, and - * finally check for the existence of the href attribute, which should hold the image url. - */ - if ($link->hasAttribute('rel') && ($link->getAttribute('rel') === 'img_src' || $link->getAttribute('rel') === 'image_src') && $link->hasAttribute('href')) { - $imgUrl = $link->getAttribute('href'); - break; - } - } - } - - if (!empty($imgUrl) && $this->getConfig()->getOption('fixRelativeURLs')) { - $imgUrl = $this->toAbsoluteURI($imgUrl); - } - - return $imgUrl; - } - - /** - * @return array - */ - public function getImages() - { - $result = []; - if (!empty($this->metadata['image'])) { - $result[] = $this->metadata['image']; - } - if (null == $this->dom) { - return $result; - } - - foreach ($this->dom->getElementsByTagName('img') as $img) { - if ($src = $img->getAttribute('src')) { - $result[] = $src; - } - } - - if ($this->getConfig()->getOption('fixRelativeURLs')) { - foreach ($result as &$imgSrc) { - $imgSrc = $this->toAbsoluteURI($imgSrc); - } - } - - $result = array_unique(array_filter($result)); - - return $result; - } - - /** - * 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 Readability $readability - * - * @return int - */ - public function getLinkDensity($readability) - { - $linkLength = 0; - $textLength = mb_strlen($readability->getTextContent(true)); - - if (!$textLength) { - return 0; - } - - $links = $readability->getAllLinks(); - - if ($links) { - /** @var Readability $link */ - foreach ($links as $link) { - $linkLength += mb_strlen($link->getTextContent(true)); - } - } - - return $linkLength / $textLength; - } - - /** - * Returns the title of the html. Prioritizes the title from the metadata against the title tag. - * - * @return string|null - */ - private function getTitle() - { - $originalTitle = null; - - if (isset($this->metadata['title'])) { - $originalTitle = $this->metadata['title']; - } else { - $titleTag = $this->dom->getElementsByTagName('title'); - if ($titleTag->length > 0) { - $originalTitle = $titleTag->item(0)->nodeValue; - } - } - - if ($originalTitle === null) { - return null; - } - - $curTitle = $originalTitle; - $titleHadHierarchicalSeparators = false; - - /* - * If there's a separator in the title, first remove the final part - * - * Sanity warning: if you eval this match in PHPStorm's "Evaluate expression" box, it will return false - * I can assure you it works properly if you let the code run. - */ - if (preg_match('/ [\|\-\\\\\/>»] /i', $curTitle)) { - $titleHadHierarchicalSeparators = (bool)preg_match('/ [\\\\\/>»] /', $curTitle); - $curTitle = preg_replace('/(.*)[\|\-\\\\\/>»] .*/i', '$1', $originalTitle); - - // If the resulting title is too short (3 words or fewer), remove - // the first part instead: - if (count(preg_split('/\s+/', $curTitle)) < 3) { - $curTitle = preg_replace('/[^\|\-\\\\\/>»]*[\|\-\\\\\/>»](.*)/i', '$1', $originalTitle); - } - } elseif (strpos($curTitle, ': ') !== false) { - // Check if we have an heading containing this exact string, so we - // could assume it's the full title. - $match = false; - for ($i = 1; $i <= 2; $i++) { - foreach ($this->dom->getElementsByTagName('h' . $i) as $hTag) { - // Trim texts to avoid having false negatives when the title is surrounded by spaces or tabs - if (trim($hTag->nodeValue) === trim($curTitle)) { - $match = true; - } - } - } - - // If we don't, let's extract the title out of the original title string. - if (!$match) { - $curTitle = substr($originalTitle, strrpos($originalTitle, ':') + 1); - - // If the title is now too short, try the first colon instead: - if (count(preg_split('/\s+/', $curTitle)) < 3) { - $curTitle = substr($originalTitle, strpos($originalTitle, ':') + 1); - } - } - } elseif (mb_strlen($curTitle) > 150 || mb_strlen($curTitle) < 15) { - $hOnes = $this->dom->getElementsByTagName('h1'); - - if ($hOnes->length === 1) { - $curTitle = $hOnes->item(0)->nodeValue; - } - } - - $curTitle = trim($curTitle); - - /* - * If we now have 4 words or fewer as our title, and either no - * 'hierarchical' separators (\, /, > or ») were found in the original - * title or we decreased the number of words by more than 1 word, use - * the original title. - */ - $curTitleWordCount = count(preg_split('/\s+/', $curTitle)); - $originalTitleWordCount = count(preg_split('/\s+/', preg_replace('/[\|\-\\\\\/>»]+/', '', $originalTitle))) - 1; - - if ($curTitleWordCount <= 4 && - (!$titleHadHierarchicalSeparators || $curTitleWordCount !== $originalTitleWordCount)) { - $curTitle = $originalTitle; - } - - return $curTitle; - } - - /** - * Gets nodes from the root element. - * - * @param $node Readability - * - * @return array - */ - private function getNodes(Readability $node) - { - $stripUnlikelyCandidates = $this->getConfig()->getOption('stripUnlikelyCandidates'); - - $elementsToScore = []; - - /* - * First, node prepping. Trash nodes that look cruddy (like ones with the - * class name "comment", etc), and turn divs into P tags where they have been - * used inappropriately (as in, where they contain no other block level elements.) - */ - - while ($node) { - $matchString = $node->getAttribute('class') . ' ' . $node->getAttribute('id'); - - // Remove DOMComments nodes as we don't need them and mess up children counting - if ($node->nodeTypeEqualsTo(XML_COMMENT_NODE)) { - $node = $node->removeAndGetNext($node); - continue; - } - - // Check to see if this node is a byline, and remove it if it is. - if ($this->checkByline($node, $matchString)) { - $node = $node->removeAndGetNext($node); - continue; - } - - // Remove unlikely candidates - if ($stripUnlikelyCandidates) { - if ( - preg_match($this->regexps['unlikelyCandidates'], $matchString) && - !preg_match($this->regexps['okMaybeItsACandidate'], $matchString) && - !$node->tagNameEqualsTo('body') && - !$node->tagNameEqualsTo('a') - ) { - $node = $node->removeAndGetNext($node); - continue; - } - } - - // Remove DIV, SECTION, and HEADER nodes without any content(e.g. text, image, video, or iframe). - if (($node->tagNameEqualsTo('div') || $node->tagNameEqualsTo('section') || $node->tagNameEqualsTo('header') || - $node->tagNameEqualsTo('h1') || $node->tagNameEqualsTo('h2') || $node->tagNameEqualsTo('h3') || - $node->tagNameEqualsTo('h4') || $node->tagNameEqualsTo('h5') || $node->tagNameEqualsTo('h6') || - $node->tagNameEqualsTo('p')) && - $node->isElementWithoutContent()) { - $node = $node->removeAndGetNext($node); - continue; - } - - if (in_array(strtolower($node->getTagName()), $this->defaultTagsToScore)) { - $elementsToScore[] = $node; - } - - // Turn all divs that don't have children block level elements into p's - if ($node->tagNameEqualsTo('div')) { - /* - * Sites like http://mobile.slate.com encloses each paragraph with a DIV - * element. DIVs with only a P element inside and no text content can be - * safely converted into plain P elements to avoid confusing the scoring - * algorithm with DIVs with are, in practice, paragraphs. - */ - if ($this->hasSinglePNode($node)) { - $pNode = $node->getChildren(true)[0]; - $node->replaceChild($pNode); - $node = $pNode; - $elementsToScore[] = $node; - } elseif (!$this->hasSingleChildBlockElement($node)) { - $node->setNodeTag('p'); - $elementsToScore[] = $node; - } else { - // EXPERIMENTAL - foreach ($node->getChildren() as $child) { - /** @var Readability $child */ - if ($child->isText() && mb_strlen(trim($child->getTextContent())) > 0) { - $newNode = $node->createNode($child, 'p'); - $child->replaceChild($newNode); - } - } - } - } - - $node = $node->getNextNode($node); - } - - return $elementsToScore; - } - - /** - * Assign scores to each node. This function will rate each node and return a Readability object for each one. - * - * @param array $nodes - * - * @return DOMDocument|bool - */ - private function rateNodes($nodes) - { - $candidates = []; - - /** @var Readability $node */ - foreach ($nodes as $node) { - if (!$node->getParent()) { - continue; - } - // Discard nodes with less than 25 characters, without blank space - if (mb_strlen($node->getTextContent(true)) < 25) { - continue; - } - - $ancestors = $node->getNodeAncestors(); - - // Exclude nodes with no ancestor - if (count($ancestors) === 0) { - continue; - } - - // Start with a point for the paragraph itself as a base. - $contentScore = 1; - - // Add points for any commas within this paragraph. - $contentScore += count(explode(',', $node->getTextContent(true))); - - // For every 100 characters in this paragraph, add another point. Up to 3 points. - $contentScore += min(floor(mb_strlen($node->getTextContent(true)) / 100), 3); - - // Initialize and score ancestors. - /** @var Readability $ancestor */ - foreach ($ancestors as $level => $ancestor) { - if (!$ancestor->isInitialized()) { - $ancestor->initializeNode(); - $candidates[] = $ancestor; - } - - /* - * Node score divider: - * - parent: 1 (no division) - * - grandparent: 2 - * - great grandparent+: ancestor level * 3 - */ - - if ($level === 0) { - $scoreDivider = 1; - } elseif ($level === 1) { - $scoreDivider = 2; - } else { - $scoreDivider = $level * 3; - } - - $currentScore = $ancestor->getContentScore(); - $ancestor->setContentScore($currentScore + ($contentScore / $scoreDivider)); - } - } - - /* - * TODO This is an horrible hack because I don't know how to properly pass by reference. - * When candidates are added to the $candidates array, they lose the reference to the original object - * and on each loop, the object inside $candidates doesn't get updated. This function restores the score - * by getting it of the data-readability tag. This should be fixed using proper references and good coding - * practices (which I lack) - */ - - foreach ($candidates as $candidate) { - $candidate->reloadScore(); - } - - /* - * After we've calculated scores, loop through all of the possible - * candidate nodes we found and find the one with the highest score. - */ - - $topCandidates = []; - foreach ($candidates as $candidate) { - - /* - * 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. - */ - - $candidate->setContentScore($candidate->getContentScore() * (1 - $this->getLinkDensity($candidate))); - - for ($i = 0; $i < $this->getConfig()->getOption('maxTopCandidates'); $i++) { - $aTopCandidate = isset($topCandidates[$i]) ? $topCandidates[$i] : null; - - if (!$aTopCandidate || $candidate->getContentScore() > $aTopCandidate->getContentScore()) { - array_splice($topCandidates, $i, 0, [$candidate]); - if (count($topCandidates) > $this->getConfig()->getOption('maxTopCandidates')) { - array_pop($topCandidates); - } - break; - } - } - } - - $topCandidate = isset($topCandidates[0]) ? $topCandidates[0] : null; - $neededToCreateTopCandidate = false; - $parentOfTopCandidate = null; - - /* - * If we still have no top candidate, just use the body as a last resort. - * We also have to copy the body node so it is something we can modify. - */ - - if ($topCandidate === null || $topCandidate->tagNameEqualsTo('body')) { - // Move all of the page's children into topCandidate - $topCandidate = new DOMDocument('1.0', 'utf-8'); - $topCandidate->encoding = 'UTF-8'; - $topCandidate->appendChild($topCandidate->createElement('div', '')); - $kids = $this->dom->getElementsByTagName('body')->item(0)->childNodes; - - // Cannot be foreached, don't ask me why. - for ($i = 0; $i < $kids->length; $i++) { - $import = $topCandidate->importNode($kids->item($i), true); - $topCandidate->firstChild->appendChild($import); - } - - // Readability must be created using firstChild to grab the DOMElement instead of the DOMDocument. - $topCandidate = new Readability($topCandidate->firstChild); - $topCandidate->initializeNode(); - - //TODO on the original code, $topCandidate is added to the page variable, which holds the whole HTML - // Should be done this here also? (line 823 in readability.js) - } elseif ($topCandidate) { - // Find a better top candidate node if it contains (at least three) nodes which belong to `topCandidates` array - // and whose scores are quite closed with current `topCandidate` node. - $alternativeCandidateAncestors = []; - for ($i = 1; $i < count($topCandidates); $i++) { - if ($topCandidates[$i]->getContentScore() / $topCandidate->getContentScore() >= 0.75) { - array_push($alternativeCandidateAncestors, $topCandidates[$i]->getNodeAncestors(false)); - } - } - - $MINIMUM_TOPCANDIDATES = 3; - if (count($alternativeCandidateAncestors) >= $MINIMUM_TOPCANDIDATES) { - $parentOfTopCandidate = $topCandidate->getParent(); - while (!$parentOfTopCandidate->tagNameEqualsTo('body')) { - $listsContainingThisAncestor = 0; - for ($ancestorIndex = 0; $ancestorIndex < count($alternativeCandidateAncestors) && $listsContainingThisAncestor < $MINIMUM_TOPCANDIDATES; $ancestorIndex++) { - $listsContainingThisAncestor += (int)in_array($parentOfTopCandidate, $alternativeCandidateAncestors[$ancestorIndex]); - } - if ($listsContainingThisAncestor >= $MINIMUM_TOPCANDIDATES) { - $topCandidate = $parentOfTopCandidate; - break; - } - $parentOfTopCandidate = $parentOfTopCandidate->getParent(); - } - } - - /* - * Because of our bonus system, parents of candidates might have scores - * themselves. They get half of the node. There won't be nodes with higher - * scores than our topCandidate, but if we see the score going *up* in the first - * few steps up the tree, that's a decent sign that there might be more content - * lurking in other places that we want to unify in. The sibling stuff - * below does some of that - but only if we've looked high enough up the DOM - * tree. - */ - - $parentOfTopCandidate = $topCandidate->getParent(); - $lastScore = $topCandidate->getContentScore(); - - // The scores shouldn't get too low. - $scoreThreshold = $lastScore / 3; - - /* @var Readability $parentOfTopCandidate */ - while (!$parentOfTopCandidate->tagNameEqualsTo('body')) { - $parentScore = $parentOfTopCandidate->getContentScore(); - if ($parentScore < $scoreThreshold) { - break; - } - - if ($parentScore > $lastScore) { - // Alright! We found a better parent to use. - $topCandidate = $parentOfTopCandidate; - break; - } - $lastScore = $parentOfTopCandidate->getContentScore(); - $parentOfTopCandidate = $parentOfTopCandidate->getParent(); - } - - // If the top candidate is the only child, use parent instead. This will help sibling - // joining logic when adjacent content is actually located in parent's sibling node. - $parentOfTopCandidate = $topCandidate->getParent(); - while (!$parentOfTopCandidate->tagNameEqualsTo('body') && count($parentOfTopCandidate->getChildren(true)) === 1) { - $topCandidate = $parentOfTopCandidate; - $parentOfTopCandidate = $topCandidate->getParent(); - } - } - - /* - * 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 = new DOMDocument('1.0', 'utf-8'); - $articleContent->createElement('div'); - - $siblingScoreThreshold = max(10, $topCandidate->getContentScore() * 0.2); - // Keep potential top candidate's parent node to try to get text direction of it later. - $parentOfTopCandidate = $topCandidate->getParent(); - $siblings = $parentOfTopCandidate->getChildren(); - - $hasContent = false; - - /** @var Readability $sibling */ - foreach ($siblings as $sibling) { - $append = false; - - if ($sibling->compareNodes($sibling, $topCandidate)) { - $append = true; - } else { - $contentBonus = 0; - - // Give a bonus if sibling nodes and top candidates have the example same classname - if ($sibling->getAttribute('class') === $topCandidate->getAttribute('class') && $topCandidate->getAttribute('class') !== '') { - $contentBonus += $topCandidate->getContentScore() * 0.2; - } - if ($sibling->getContentScore() + $contentBonus >= $siblingScoreThreshold) { - $append = true; - } elseif ($sibling->tagNameEqualsTo('p')) { - $linkDensity = $this->getLinkDensity($sibling); - $nodeContent = $sibling->getTextContent(true); - - if (mb_strlen($nodeContent) > 80 && $linkDensity < 0.25) { - $append = true; - } elseif ($nodeContent && mb_strlen($nodeContent) < 80 && $linkDensity === 0 && preg_match('/\.( |$)/', $nodeContent)) { - $append = true; - } - } - } - - if ($append) { - $hasContent = true; - - if (!in_array(strtolower($sibling->getTagName()), $this->alterToDIVExceptions)) { - /* - * 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. - */ - - $sibling->setNodeTag('div'); - } - - $import = $articleContent->importNode($sibling->getDOMNode(), true); - $articleContent->appendChild($import); - - /* - * No node shifting needs to be check because when calling getChildren, an array is made with the - * children of the parent node, instead of using the DOMElement childNodes function, which, when used - * along with appendChild, would shift the nodes position and the current foreach will behave in - * unpredictable ways. - */ - } - } - - $articleContent = $this->prepArticle($articleContent); - - if ($hasContent) { - // Find out text direction from ancestors of final top candidate. - $ancestors = array_merge([$parentOfTopCandidate, $topCandidate], $parentOfTopCandidate->getNodeAncestors()); - foreach ($ancestors as $ancestor) { - $articleDir = $ancestor->getAttribute('dir'); - if ($articleDir) { - $this->metadata['articleDir'] = $articleDir; - break; - } - } - - return $articleContent; - } else { - return false; - } - } - - /** - * TODO To be moved to Readability. - * - * @param DOMDocument $article - * - * @return DOMDocument - */ - public function prepArticle(DOMDocument $article) - { - $this->_cleanStyles($article); - $this->_clean($article, 'style'); - - // Check for data tables before we continue, to avoid removing items in - // those tables, which will often be isolated even though they're - // visually linked to other content-ful elements (text, images, etc.). - $this->_markDataTables($article); - - // Clean out junk from the article content - $this->_cleanConditionally($article, 'form'); - $this->_cleanConditionally($article, 'fieldset'); - $this->_clean($article, 'object'); - $this->_clean($article, 'embed'); - $this->_clean($article, 'h1'); - $this->_clean($article, 'footer'); - - // Clean out elements have "share" in their id/class combinations from final top candidates, - // which means we don't remove the top candidates even they have "share". - foreach ($article->childNodes as $child) { - $this->_cleanMatchedNodes($child, '/share/i'); - } - - /* - * If there is only one h2 and its text content substantially equals article title, - * they are probably using it as a header and not a subheader, - * so remove it since we already extract the title separately. - */ - $h2 = $article->getElementsByTagName('h2'); - if ($h2->length === 1) { - $lengthSimilarRate = (mb_strlen($h2->item(0)->textContent) - mb_strlen($this->metadata['title'])) / max(mb_strlen($this->metadata['title']), 1); - - if (abs($lengthSimilarRate) < 0.5) { - if ($lengthSimilarRate > 0) { - $titlesMatch = strpos($h2->item(0)->textContent, $this->metadata['title']) !== false; - } else { - $titlesMatch = strpos($this->metadata['title'], $h2->item(0)->textContent) !== false; - } - if ($titlesMatch) { - $this->_clean($article, 'h2'); - } - } - } - - $this->_clean($article, 'iframe'); - $this->_clean($article, 'input'); - $this->_clean($article, 'textarea'); - $this->_clean($article, 'select'); - $this->_clean($article, 'button'); - $this->_cleanHeaders($article); - - // Do these last as the previous stuff may have removed junk - // that will affect these - $this->_cleanConditionally($article, 'table'); - $this->_cleanConditionally($article, 'ul'); - $this->_cleanConditionally($article, 'div'); - - $this->_cleanExtraParagraphs($article); - - $this->_cleanReadabilityTags($article); - - foreach (iterator_to_array($article->getElementsByTagName('br')) as $br) { - $next = $br->nextSibling; - if ($next && $next->nodeName === 'p') { - $br->parentNode->removeChild($br); - } - } - - return $article; - } - - /** - * Look for 'data' (as opposed to 'layout') tables, for which we use - * similar checks as - * https://dxr.mozilla.org/mozilla-central/rev/71224049c0b52ab190564d3ea0eab089a159a4cf/accessible/html/HTMLTableAccessible.cpp#920. - * - * TODO To be moved to Readability. WARNING: check if we actually keep the "readabilityDataTable" param and - * maybe switch to a readability data-tag? - * - * @param DOMDocument $article - * - * @return void - */ - public function _markDataTables(DOMDocument $article) - { - $tables = $article->getElementsByTagName('table'); - foreach ($tables as $table) { - /** @var \DOMElement $table */ - $role = $table->getAttribute('role'); - if ($role === 'presentation') { - $table->readabilityDataTable = false; - continue; - } - $datatable = $table->getAttribute('datatable'); - if ($datatable == '0') { - $table->readabilityDataTable = false; - continue; - } - $summary = $table->getAttribute('summary'); - if ($summary) { - $table->readabilityDataTable = true; - continue; - } - - $caption = $table->getElementsByTagName('caption'); - if ($caption->length > 0 && $caption->item(0)->childNodes->length > 0) { - $table->readabilityDataTable = true; - continue; - } - - // If the table has a descendant with any of these tags, consider a data table: - foreach (['col', 'colgroup', 'tfoot', 'thead', 'th'] as $dataTableDescendants) { - if ($table->getElementsByTagName($dataTableDescendants)->length > 0) { - $table->readabilityDataTable = true; - continue 2; - } - } - - // Nested tables indicate a layout table: - if ($table->getElementsByTagName('table')->length > 0) { - $table->readabilityDataTable = false; - continue; - } - - $sizeInfo = $this->_getRowAndColumnCount($table); - if ($sizeInfo['rows'] >= 10 || $sizeInfo['columns'] > 4) { - $table->readabilityDataTable = true; - continue; - } - // Now just go by size entirely: - $table->readabilityDataTable = $sizeInfo['rows'] * $sizeInfo['columns'] > 10; - } - } - - /** - * Return an array indicating how many rows and columns this table has. - * - * @param \DOMElement $table - * - * @return array - */ - public function _getRowAndColumnCount(\DOMElement $table) - { - $rows = $columns = 0; - $trs = $table->getElementsByTagName('tr'); - foreach ($trs as $tr) { - /** @var \DOMElement $tr */ - $rowspan = $tr->getAttribute('rowspan'); - $rows += ($rowspan || 1); - - // Now look for column-related info - $columnsInThisRow = 0; - $cells = $tr->getElementsByTagName('td'); - foreach ($cells as $cell) { - /** @var \DOMElement $cell */ - $colspan = $cell->getAttribute('colspan'); - $columnsInThisRow += ($colspan || 1); - } - $columns = max($columns, $columnsInThisRow); - } - - return ['rows' => $rows, 'columns' => $columns]; - } - - /** - * TODO To be moved to Readability. - * - * @param DOMDocument $article - * - * @return void - */ - public function _cleanReadabilityTags(DOMDocument $article) - { - if ($this->getConfig()->getOption('removeReadabilityTags')) { - foreach ($article->getElementsByTagName('*') as $tag) { - if ($tag->hasAttribute('data-readability')) { - $tag->removeAttribute('data-readability'); - } - } - } - } - - /** - * Remove the style attribute on every e and under. - * TODO: To be moved to Readability. - * - * @param $node \DOMDocument|\DOMNode - **/ - public function _cleanStyles($node) - { - if (property_exists($node, 'tagName') && $node->tagName === 'svg') { - return; - } - - // Do not bother if there's no method to remove an attribute - if (method_exists($node, 'removeAttribute')) { - $presentational_attributes = ['align', 'background', 'bgcolor', 'border', 'cellpadding', 'cellspacing', 'frame', 'hspace', 'rules', 'style', 'valign', 'vspace']; - // Remove `style` and deprecated presentational attributes - foreach ($presentational_attributes as $presentational_attribute) { - $node->removeAttribute($presentational_attribute); - } - - $deprecated_size_attribute_elems = ['table', 'th', 'td', 'hr', 'pre']; - if (property_exists($node, 'tagName') && in_array($node->tagName, $deprecated_size_attribute_elems)) { - $node->removeAttribute('width'); - $node->removeAttribute('height'); - } - } - - $cur = $node->firstChild; - while ($cur !== null) { - $this->_cleanStyles($cur); - $cur = $cur->nextSibling; - } - } - - /** - * Clean out elements whose id/class combinations match specific string. - * - * TODO To be moved to readability - * - * @param string $regex Match id/class combination. - * - * @return void - **/ - public function _cleanMatchedNodes($node, $regex) - { - $node = new Readability($node); - $endOfSearchMarkerNode = $node->getNextNode($node, true); - $next = $node->getNextNode($node); - while ($next && $next !== $endOfSearchMarkerNode) { - if (preg_match($regex, sprintf('%s %s', $next->getAttribute('class'), $next->getAttribute('id')))) { - $next = $next->removeAndGetNext($next); - } else { - $next = $next->getNextNode($next); - } - } - } - - /** - * TODO To be moved to Readability. - * - * @param DOMDocument $article - * - * @return void - */ - public function _cleanExtraParagraphs(DOMDocument $article) - { - $paragraphs = $article->getElementsByTagName('p'); - $length = $paragraphs->length; - - for ($i = 0; $i < $length; $i++) { - $paragraph = $paragraphs->item($length - 1 - $i); - - $imgCount = $paragraph->getElementsByTagName('img')->length; - $embedCount = $paragraph->getElementsByTagName('embed')->length; - $objectCount = $paragraph->getElementsByTagName('object')->length; - // At this point, nasty iframes have been removed, only remain embedded video ones. - $iframeCount = $paragraph->getElementsByTagName('iframe')->length; - $totalCount = $imgCount + $embedCount + $objectCount + $iframeCount; - - if ($totalCount === 0 && !preg_replace($this->regexps['onlyWhitespace'], '', $paragraph->textContent)) { - // TODO must be done via readability - $paragraph->parentNode->removeChild($paragraph); - } - } - } - - /** - * TODO To be moved to Readability. - * - * @param DOMDocument $article - * - * @return void - */ - public function _cleanConditionally(DOMDocument $article, $tag) - { - if (!$this->getConfig()->getOption('cleanConditionally')) { - return; - } - - $isList = in_array($tag, ['ul', 'ol']); - - /* - * Gather counts for other typical elements embedded within. - * Traverse backwards so we can remove nodes at the same time - * without effecting the traversal. - */ - - $DOMNodeList = $article->getElementsByTagName($tag); - $length = $DOMNodeList->length; - for ($i = 0; $i < $length; $i++) { - $node = $DOMNodeList->item($length - 1 - $i); - - $node = new Readability($node); - - // First check if we're in a data table, in which case don't remove us. - if ($node->hasAncestorTag($node, 'table', -1) && isset($node->readabilityDataTable)) { - continue; - } - - $weight = $node->getClassWeight(); - - if ($weight < 0) { - $this->removeNode($node->getDOMNode()); - continue; - } - - if (substr_count($node->getTextContent(), ',') < 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. - */ - - // TODO Horrible hack, must be removed once this function is inside Readability - $p = $node->getDOMNode()->getElementsByTagName('p')->length; - $img = $node->getDOMNode()->getElementsByTagName('img')->length; - $li = $node->getDOMNode()->getElementsByTagName('li')->length - 100; - $input = $node->getDOMNode()->getElementsByTagName('input')->length; - - $embedCount = 0; - $embeds = $node->getDOMNode()->getElementsByTagName('embed'); - - foreach ($embeds as $embedNode) { - if (preg_match($this->regexps['videos'], $embedNode->C14N())) { - $embedCount++; - } - } - - $linkDensity = $this->getLinkDensity($node); - $contentLength = mb_strlen($node->getTextContent(true)); - - $haveToRemove = - ($img > 1 && $p / $img < 0.5 && !$node->hasAncestorTag($node, 'figure')) || - (!$isList && $li > $p) || - ($input > floor($p / 3)) || - (!$isList && $contentLength < 25 && ($img === 0 || $img > 2) && !$node->hasAncestorTag($node, 'figure')) || - (!$isList && $weight < 25 && $linkDensity > 0.2) || - ($weight >= 25 && $linkDensity > 0.5) || - (($embedCount === 1 && $contentLength < 75) || $embedCount > 1); - - if ($haveToRemove) { - $this->removeNode($node->getDOMNode()); - } - } - } - } - - /** - * Clean a node of all elements of type "tag". - * (Unless it's a youtube/vimeo video. People love movies.). - * - * TODO To be moved to Readability - * - * @param $article DOMDocument - * @param $tag string tag to clean - * - * @return void - **/ - public function _clean(DOMDocument $article, $tag) - { - $isEmbed = in_array($tag, ['object', 'embed', 'iframe']); - - $DOMNodeList = $article->getElementsByTagName($tag); - $length = $DOMNodeList->length; - for ($i = 0; $i < $length; $i++) { - $item = $DOMNodeList->item($length - 1 - $i); - - // Allow youtube and vimeo videos through as people usually want to see those. - if ($isEmbed) { - $attributeValues = []; - foreach ($item->attributes as $name => $value) { - $attributeValues[] = $value->nodeValue; - } - $attributeValues = implode('|', $attributeValues); - - // First, check the elements attributes to see if any of them contain youtube or vimeo - if (preg_match($this->regexps['videos'], $attributeValues)) { - continue; - } - - // Then check the elements inside this element for the same. - if (preg_match($this->regexps['videos'], $item->C14N())) { - continue; - } - } - $this->removeNode($item); - } - } - - /** - * Clean out spurious headers from an Element. Checks things like classnames and link density. - * - * TODO To be moved to Readability - * - * @param DOMDocument $article - * - * @return void - **/ - public function _cleanHeaders(DOMDocument $article) - { - for ($headerIndex = 1; $headerIndex < 3; $headerIndex++) { - $headers = $article->getElementsByTagName('h' . $headerIndex); - foreach ($headers as $header) { - $header = new Readability($header); - if ($header->getClassWeight() < 0) { - $this->removeNode($header->getDOMNode()); - } - } - } - } - - /** - * Remove the passed node. - * - * TODO To be moved to Readability - * - * @param \DOMNode $node - * - * @return void - **/ - public function removeNode(\DOMNode $node) - { - $parent = $node->parentNode; - if ($parent) { - $parent->removeChild($node); - } - } - - /** - * Checks if the node is a byline. - * - * @param Readability $node - * @param string $matchString - * - * @return bool - */ - private function checkByline($node, $matchString) - { - if (!$this->getConfig()->getOption('articleByLine')) { - return false; - } - - /* - * Check if the byline is already set - */ - if (isset($this->metadata['byline'])) { - return false; - } - - $rel = $node->getAttribute('rel'); - - if ($rel === 'author' || preg_match($this->regexps['byline'], $matchString) && $this->isValidByline($node->getTextContent())) { - $this->metadata['byline'] = trim($node->getTextContent()); - - return true; - } - - return false; - } - - /** - * Checks the validity of a byLine. Based on string length. - * - * @param string $text - * - * @return bool - */ - private function isValidByline($text) - { - if (gettype($text) == 'string') { - $byline = trim($text); - - return (mb_strlen($byline) > 0) && (mb_strlen($text) < 100); - } - - return false; - } - - /** - * Checks if the current node has a single child and if that child is a P node. - * Useful to convert <div><p> nodes to a single <p> node and avoid confusing the scoring system since div with p - * tags are, in practice, paragraphs. - * - * @param Readability $node - * - * @return bool - */ - private function hasSinglePNode(Readability $node) - { - // There should be exactly 1 element child which is a P: - if (count($children = $node->getChildren(true)) !== 1 || !$children[0]->tagNameEqualsTo('p')) { - return false; - } - - // And there should be no text nodes with real content (param true on ->getChildren) - foreach ($children as $child) { - /** @var $child Readability */ - if ($child->nodeTypeEqualsTo(XML_TEXT_NODE) && !preg_match('/\S$/', $child->getTextContent())) { - return false; - } - } - - return true; - } - - private function hasSingleChildBlockElement(Readability $node) - { - $result = false; - if ($node->hasChildren()) { - /** @var Readability $child */ - foreach ($node->getChildren() as $child) { - if (in_array($child->getTagName(), $this->divToPElements)) { - $result = true; - } else { - // If any of the hasSingleChildBlockElement calls return true, return true then. - $result = ($result || $this->hasSingleChildBlockElement($child)); - } - } - } - - return $result; - } -} diff --git a/src/Readability.php b/src/Readability.php index bf8cdf8..e69de29 100644 --- a/src/Readability.php +++ b/src/Readability.php @@ -1,548 +0,0 @@ -<?php - -namespace andreskrey\Readability; - -use League\HTMLToMarkdown\Element; - -/** - * Class DOMElement. - * - * This is a extension of the original Element class from League\HTMLToMarkdown\Element. - * This class adds functions specific to Readability.php and overloads some of them to fit the purpose of this project. - */ -class Readability extends Element implements ReadabilityInterface -{ - /** - * @var \DOMNode|\DOMElement - */ - protected $node; - - /** - * @var int - */ - protected $contentScore = 0; - - /** - * @var int - */ - protected $initialized = false; - - /** - * @var array - */ - private $regexps = [ - 'positive' => '/article|body|content|entry|hentry|h-entry|main|page|pagination|post|text|blog|story/i', - 'negative' => '/hidden|^hid$| hid$| hid |^hid |banner|combx|comment|com-|contact|foot|footer|footnote|masthead|media|meta|modal|outbrain|promo|related|scroll|share|shoutbox|sidebar|skyscraper|sponsor|shopping|tags|tool|widget/i', - ]; - - /** - * Constructor. - * - * @param \DOMNode $node Selected element from DOMDocument - */ - public function __construct(\DOMNode $node) - { - parent::__construct($node); - - /* - * Restore the score if the object has been already scored. - * - * An if must be added before calling the getAttribute function, because if we reach the DOMDocument - * by getting the node parents we'll get a undefined function fatal error - */ - if (method_exists($node, 'getAttribute')) { - if ($node->hasAttribute('data-readability')) { - // Node was initialized previously. Restoring score and setting flag. - $this->initialized = true; - $score = $node->getAttribute('data-readability'); - $this->setContentScore($score); - } - } - } - - /** - * Checks for the tag name. Case insensitive. - * - * @param string $value Name to compare to the current tag - * - * @return bool - */ - public function tagNameEqualsTo($value) - { - $tagName = $this->getTagName(); - if (strtolower($value) === strtolower($tagName)) { - return true; - } - - return false; - } - - /** - * Checks for the node type. - * - * @param string $value Type of node to compare to - * - * @return bool - */ - public function nodeTypeEqualsTo($value) - { - return $this->node->nodeType === $value; - } - - /** - * Get the ancestors of the current node. - * - * @param int|bool $maxLevel Max amount of ancestors to get. False for all of them - * - * @return array - */ - public function getNodeAncestors($maxLevel = 3) - { - $ancestors = []; - $level = 0; - - $node = $this->getParent(); - - while ($node) { - $ancestors[] = $node; - $level++; - if ($level === $maxLevel) { - break; - } - $node = $node->getParent(); - } - - return $ancestors; - } - - /** - * Overloading the getParent function from League\HTMLToMarkdown\Element due to a bug when there are no more parents - * on the selected element. - * - * @return Readability|null - */ - public function getParent() - { - $node = $this->node->parentNode; - - return ($node) ? new self($node) : null; - } - - /** - * Returns all links from the current element. - * - * @return array|null - */ - public function getAllLinks() - { - if (($this->isText())) { - return null; - } else { - $links = []; - foreach ($this->node->getElementsByTagName('a') as $link) { - $links[] = new self($link); - } - - return $links; - } - } - - /** - * Initializer. Calculates the current score of the node and returns a full Readability object. - * - * @return Readability - */ - public function initializeNode() - { - if (!$this->initialized) { - $contentScore = 0; - - switch ($this->getTagName()) { - case 'div': - $contentScore += 5; - break; - - case 'pre': - case 'td': - case 'blockquote': - $contentScore += 3; - break; - - case 'address': - case 'ol': - case 'ul': - case 'dl': - case 'dd': - case 'dt': - case 'li': - case 'form': - $contentScore -= 3; - break; - - case 'h1': - case 'h2': - case 'h3': - case 'h4': - case 'h5': - case 'h6': - case 'th': - $contentScore -= 5; - break; - } - - $this->setContentScore($contentScore + $this->getClassWeight()); - - $this->initialized = true; - } - - return $this; - } - - /** - * Calculates the weight of the class/id of the current element. - * - * @todo check for flag that lets this function run or not - * - * @return int - */ - public function getClassWeight() - { - // TODO To implement. How to get config from html parser from readability -// if ($this->getConfig()->getOption('weightClasses')) { -// return 0; -// } -// - $weight = 0; - - // Look for a special classname - $class = $this->getAttribute('class'); - if (trim($class)) { - if (preg_match($this->regexps['negative'], $class)) { - $weight -= 25; - } - - if (preg_match($this->regexps['positive'], $class)) { - $weight += 25; - } - } - - // Look for a special ID - $id = $this->getAttribute('id'); - if (trim($id)) { - if (preg_match($this->regexps['negative'], $id)) { - $weight -= 25; - } - - if (preg_match($this->regexps['positive'], $id)) { - $weight += 25; - } - } - - return $weight; - } - - /** - * Returns the current score of the Readability object. - * - * @return int - */ - public function getContentScore() - { - return $this->contentScore; - } - - /** - * Returns the current score of the Readability object. - * - * @param int $score - * - * @return int - */ - public function setContentScore($score) - { - // Check if the setAttribute method exists, as some elements lack of it (and calling it anyway throws an exception) - if (method_exists($this->node, 'setAttribute')) { - $this->contentScore = (float)$score; - - // Set score in an attribute of the tag to prevent losing it while creating new Readability objects. - $this->node->setAttribute('data-readability', $this->contentScore); - - return $this->contentScore; - } - - return 0; - } - - /** - * Returns the full text of the node. - * - * @param bool $normalize Normalize white space? - * - * @return string - */ - public function getTextContent($normalize = false) - { - $nodeValue = $this->node->nodeValue; - if ($normalize) { - $nodeValue = trim(preg_replace('/\s{2,}/', ' ', $nodeValue)); - } - - return $nodeValue; - } - - /** - * Changes the node tag name. Since tagName on DOMElement is a read only value, this must be done creating a new - * element with the new tag name and importing it to the main DOMDocument. - * - * @param string $value - * @param bool $importAttributes - */ - public function setNodeTag($value, $importAttributes = false) - { - $new = new \DOMDocument(); - $new->appendChild($new->createElement($value)); - - $childs = $this->node->childNodes; - for ($i = 0; $i < $childs->length; $i++) { - $import = $new->importNode($childs->item($i), true); - $new->firstChild->appendChild($import); - } - - if ($importAttributes) { - // Import attributes from the original node. - foreach ($this->node->attributes as $attribute) { - $new->firstChild->setAttribute($attribute->nodeName, $attribute->nodeValue); - } - } - - // The import must be done on the firstChild of $new, since $new is a DOMDocument and not a DOMElement. - $import = $this->node->ownerDocument->importNode($new->firstChild, true); - $this->node->parentNode->replaceChild($import, $this->node); - - $this->node = $import; - } - - /** - * Returns the current DOMNode. - * - * @return \DOMNode - */ - public function getDOMNode() - { - return $this->node; - } - - /** - * Removes the current node and returns the next node to be parsed (child, sibling or parent). - * - * @param Readability $node - * - * @return Readability - */ - public function removeAndGetNext($node) - { - $nextNode = $this->getNextNode($node, true); - $node->node->parentNode->removeChild($node->node); - - return $nextNode; - } - - /** - * Returns the next node. First checks for childs (if the flag allows it), then for siblings, and finally - * for parents. - * - * @param Readability $originalNode - * @param bool $ignoreSelfAndKids - * - * @return Readability - */ - public function getNextNode($originalNode, $ignoreSelfAndKids = false) - { - /* - * Traverse the DOM from node to node, starting at the node passed in. - * Pass true for the second parameter to indicate this node itself - * (and its kids) are going away, and we want the next node over. - * - * Calling this in a loop will traverse the DOM depth-first. - */ - - // First check for kids if those aren't being ignored - if (!$ignoreSelfAndKids && $originalNode->node->firstChild) { - return new self($originalNode->node->firstChild); - } - - // Then for siblings... - if ($originalNode->node->nextSibling) { - return new self($originalNode->node->nextSibling); - } - - // And finally, move up the parent chain *and* find a sibling - // (because this is depth-first traversal, we will have already - // seen the parent nodes themselves). - do { - $originalNode = $originalNode->getParent(); - } while ($originalNode && !$originalNode->node->nextSibling); - - return ($originalNode) ? new self($originalNode->node->nextSibling) : $originalNode; - } - - /** - * Compares nodes. Checks for tag name and text content. - * - * It's a replacement of the original JS code, which looked like this: - * - * $node1 == $node2 - * - * I'm not sure this works the same in PHP, so I created a mock function to check the actual content of the node. - * Should serve the same porpuse as the original comparison. - * - * @param Readability $node1 - * @param Readability $node2 - * - * @return bool - */ - public function compareNodes($node1, $node2) - { - if ($node1->getTagName() !== $node2->getTagName()) { - return false; - } - - if ($node1->getTextContent(true) !== $node2->getTextContent(true)) { - return false; - } - - return true; - } - - /** - * Replaces child node with a new one. - * - * @param Readability $newNode - */ - public function replaceChild(self $newNode) - { - $this->node->parentNode->replaceChild($newNode->node, $this->node); - } - - /** - * Creates a new node based on the text content of the original node. - * - * @param Readability $originalNode - * @param string $tagName - * - * @return Readability - */ - public function createNode(self $originalNode, $tagName) - { - $text = $originalNode->getTextContent(); - $newNode = $originalNode->node->ownerDocument->createElement($tagName, $text); - - return new static($newNode); - } - - /** - * Checks if the object is initialized. - * - * @return bool - */ - public function isInitialized() - { - return $this->initialized; - } - - /** - * Reloads the score stores in the data-readability tag. - * - * @return int|bool - */ - public function reloadScore() - { - if (method_exists($this->node, 'getAttribute')) { - if ($this->node->hasAttribute('data-readability')) { - $this->initialized = true; - $score = $this->node->getAttribute('data-readability'); - $this->setContentScore($score); - - return $score; - } - } - - return false; - } - - /** - * Check if a given node has one of its ancestor tag name matching the - * provided one. - * - * @param Readability $node - * @param string $tagName - * @param int $maxDepth - * - * @return bool - */ - public function hasAncestorTag(self $node, $tagName, $maxDepth = 3) - { - $depth = 0; - while ($node->getParent()) { - if ($maxDepth > 0 && $depth > $maxDepth) { - return false; - } - if ($node->getParent()->tagNameEqualsTo($tagName)) { - return true; - } - $node = $node->getParent(); - $depth++; - } - - return false; - } - - /** - * Returns the children of the current node. - * - * @param bool $filterEmptyDOMText Filter empty DOMText nodes? - * - * @return array - */ - public function getChildren($filterEmptyDOMText = false) - { - $ret = []; - /** @var \DOMNode $node */ - foreach ($this->node->childNodes as $node) { - if ($filterEmptyDOMText && $node->nodeName === '#text' && !trim($node->nodeValue)) { - continue; - } - - $ret[] = new static($node); - } - - return $ret; - } - - /** - * Determines if a node has no content or it is just a bunch of dividing lines and/or whitespace. - * - * @return bool - */ - public function isElementWithoutContent() - { - return $this->node instanceof \DOMElement && - // /\x{00A0}|\s+/u TODO to be replaced with regexps array - mb_strlen(preg_replace('/\x{00A0}|\s+/u', '', $this->node->textContent)) === 0 && - ($this->node->childNodes->length === 0 || - $this->node->childNodes->length === $this->node->getElementsByTagName('br')->length + $this->node->getElementsByTagName('hr')->length - /* - * Special DOMDocument case: We also need to count how many DOMText we have inside the node. - * If there's an empty tag with an space inside and a BR (for example "<p> <br/></p>) counting only BRs and - * HRs will will say that the example has 2 nodes, instead of one. This happens because in DOMDocument, - * DOMTexts are also nodes (which doesn't happen in JS). So we need to also count how many DOMText we - * are dealing with (And at this point we know they are empty or are just whitespace, because of the - * mb_strlen in this chain of checks). - */ - + count(array_filter(iterator_to_array($this->node->childNodes), function ($child) { - return $child instanceof \DOMText; - })) - - ); - } -} diff --git a/src/ReadabilityInterface.php b/src/ReadabilityInterface.php deleted file mode 100644 index 0dee01b..0000000 --- a/src/ReadabilityInterface.php +++ /dev/null @@ -1,92 +0,0 @@ -<?php - -namespace andreskrey\Readability; - -use League\HTMLToMarkdown\ElementInterface; - -interface ReadabilityInterface extends ElementInterface -{ - /** - * @param string $value - * - * @return bool - */ - public function tagNameEqualsTo($value); - - /** - * @return int - */ - public function getNodeAncestors(); - - /** - * @return Readability|null - */ - public function getAllLinks(); - - /** - * @return int - */ - public function getContentScore(); - - /** - * @return Readability - */ - public function initializeNode(); - - /** - * @return int - */ - public function getClassWeight(); - - /** - * @param int $score - * - * @return int - */ - public function setContentScore($score); - - /** - * @param bool $normalize Normalize white space? - * - * @return string - */ - public function getTextContent($normalize); - - /** - * @param string $value - */ - public function setNodeTag($value); - - /** - * @return \DOMNode - */ - public function getDOMNode(); - - /** - * @param Readability $node - * - * @return Readability - */ - public function removeAndGetNext($node); - - /** - * @param Readability $originalNode - * @param bool $ignoreSelfAndKids - * - * @return Readability - */ - public function getNextNode($originalNode, $ignoreSelfAndKids = false); - - /** - * @param Readability $node1 - * @param Readability $node2 - * - * @return bool - */ - public function compareNodes($node1, $node2); - - /** - * @param Readability $newNode - */ - public function replaceChild(Readability $newNode); -} |