From d6eed6ba699d6edcaed438e60993c079c2833f91 Mon Sep 17 00:00:00 2001 From: Andrew Dolgov Date: Wed, 21 Aug 2019 10:07:21 +0300 Subject: initial --- init.php | 282 ++++ vendor/andreskrey/Readability/Configuration.php | 374 +++++ .../andreskrey/Readability/Nodes/DOM/DOMAttr.php | 10 + .../Readability/Nodes/DOM/DOMCdataSection.php | 10 + .../Readability/Nodes/DOM/DOMCharacterData.php | 10 + .../Readability/Nodes/DOM/DOMComment.php | 10 + .../Readability/Nodes/DOM/DOMDocument.php | 30 + .../Readability/Nodes/DOM/DOMDocumentFragment.php | 10 + .../Readability/Nodes/DOM/DOMDocumentType.php | 10 + .../Readability/Nodes/DOM/DOMElement.php | 10 + .../andreskrey/Readability/Nodes/DOM/DOMEntity.php | 10 + .../Readability/Nodes/DOM/DOMEntityReference.php | 10 + .../andreskrey/Readability/Nodes/DOM/DOMNode.php | 14 + .../Readability/Nodes/DOM/DOMNotation.php | 10 + .../Nodes/DOM/DOMProcessingInstruction.php | 10 + .../andreskrey/Readability/Nodes/DOM/DOMText.php | 10 + vendor/andreskrey/Readability/Nodes/NodeTrait.php | 560 ++++++ .../andreskrey/Readability/Nodes/NodeUtility.php | 160 ++ vendor/andreskrey/Readability/ParseException.php | 7 + vendor/andreskrey/Readability/Readability.php | 1774 ++++++++++++++++++++ 20 files changed, 3321 insertions(+) create mode 100644 init.php create mode 100644 vendor/andreskrey/Readability/Configuration.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMAttr.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMCdataSection.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMCharacterData.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMComment.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMDocument.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMDocumentFragment.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMDocumentType.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMElement.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMEntity.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMEntityReference.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMNode.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMNotation.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMProcessingInstruction.php create mode 100644 vendor/andreskrey/Readability/Nodes/DOM/DOMText.php create mode 100644 vendor/andreskrey/Readability/Nodes/NodeTrait.php create mode 100644 vendor/andreskrey/Readability/Nodes/NodeUtility.php create mode 100644 vendor/andreskrey/Readability/ParseException.php create mode 100644 vendor/andreskrey/Readability/Readability.php diff --git a/init.php b/init.php new file mode 100644 index 0000000..9ffaa00 --- /dev/null +++ b/init.php @@ -0,0 +1,282 @@ + true); + } + + function save() { + $enable_share_anything = checkbox_to_sql_bool($_POST["enable_share_anything"]); + + $this->host->set($this, "enable_share_anything", $enable_share_anything); + + echo __("Data saved."); + } + + function init($host) + { + $this->host = $host; + + if (version_compare(PHP_VERSION, '5.6.0', '<')) { + return; + } + + $host->add_hook($host::HOOK_ARTICLE_FILTER, $this); + $host->add_hook($host::HOOK_PREFS_TAB, $this); + $host->add_hook($host::HOOK_PREFS_EDIT_FEED, $this); + $host->add_hook($host::HOOK_PREFS_SAVE_FEED, $this); + + // Note: we have to install the hook even if disabled because init() is being run before plugin data has loaded + // so we can't check for our storage-set options here + $host->add_hook($host::HOOK_GET_FULL_TEXT, $this); + + $host->add_filter_action($this, "action_inline", __("Inline content")); + } + + function hook_prefs_tab($args) { + if ($args != "prefFeeds") return; + + print "
extension ".__('Readability settings (af_readability)')."\">"; + + if (version_compare(PHP_VERSION, '5.6.0', '<')) { + print_error("This plugin requires PHP version 5.6."); + } else { + + print "

" . __("Global settings") . "

"; + + print_notice("Enable for specific feeds in the feed editor."); + + print "
"; + + print ""; + + print_hidden("op", "pluginhandler"); + print_hidden("method", "save"); + print_hidden("plugin", "af_readability"); + + $enable_share_anything = $this->host->get($this, "enable_share_anything"); + + print "
"; + print ""; + print "
"; + + print print_button("submit", __("Save"), "class='alt-primary'"); + print "
"; + + $enabled_feeds = $this->host->get($this, "enabled_feeds"); + if (!is_array($enabled_feeds)) $enabled_feeds = array(); + + $enabled_feeds = $this->filter_unknown_feeds($enabled_feeds); + $this->host->set($this, "enabled_feeds", $enabled_feeds); + + if (count($enabled_feeds) > 0) { + print "

" . __("Currently enabled for (click to edit):") . "

"; + + print ""; + } + + } + + print "
"; + } + + function hook_prefs_edit_feed($feed_id) { + print "
".__("Readability")."
"; + print "
"; + + $enabled_feeds = $this->host->get($this, "enabled_feeds"); + if (!is_array($enabled_feeds)) $enabled_feeds = array(); + + $key = array_search($feed_id, $enabled_feeds); + $checked = $key !== FALSE ? "checked" : ""; + + print "
"; + + print ""; + + print "
"; + + print "
"; + } + + function hook_prefs_save_feed($feed_id) { + $enabled_feeds = $this->host->get($this, "enabled_feeds"); + if (!is_array($enabled_feeds)) $enabled_feeds = array(); + + $enable = checkbox_to_sql_bool($_POST["af_readability_enabled"]); + $key = array_search($feed_id, $enabled_feeds); + + if ($enable) { + if ($key === FALSE) { + array_push($enabled_feeds, $feed_id); + } + } else { + if ($key !== FALSE) { + unset($enabled_feeds[$key]); + } + } + + $this->host->set($this, "enabled_feeds", $enabled_feeds); + } + + /** + * @SuppressWarnings(PHPMD.UnusedFormalParameter) + */ + function hook_article_filter_action($article, $action) { + return $this->process_article($article); + } + + public function extract_content($url) { + + global $fetch_effective_url; + + $tmp = fetch_file_contents([ + "url" => $url, + "http_accept" => "text/*", + "type" => "text/html"]); + + if ($tmp && mb_strlen($tmp) < 1024 * 500) { + $tmpdoc = new DOMDocument("1.0", "UTF-8"); + + if (!@$tmpdoc->loadHTML($tmp)) + return false; + + // this is the worst hack yet :( + if (strtolower($tmpdoc->encoding) != 'utf-8') { + $tmp = preg_replace("//i", "", $tmp); + $tmp = mb_convert_encoding($tmp, 'utf-8', $tmpdoc->encoding); + } + + try { + $r = new Readability(new Configuration()); + + if ($r->parse($tmp)) { + + $tmpxpath = new DOMXPath($r->getDOMDOcument()); + $entries = $tmpxpath->query('(//a[@href]|//img[@src])'); + + foreach ($entries as $entry) { + if ($entry->hasAttribute("href")) { + $entry->setAttribute("href", + rewrite_relative_url($fetch_effective_url, $entry->getAttribute("href"))); + + } + + if ($entry->hasAttribute("src")) { + $entry->setAttribute("src", + rewrite_relative_url($fetch_effective_url, $entry->getAttribute("src"))); + + } + } + + return $r->getContent(); + } + + } catch (Exception $e) { + return false; + } + } + + return false; + } + + function process_article($article) { + + $extracted_content = $this->extract_content($article["link"]); + + # let's see if there's anything of value in there + $content_test = trim(strip_tags(sanitize($extracted_content))); + + if ($content_test) { + $article["content"] = $extracted_content; + } + + return $article; + } + + function hook_article_filter($article) { + + $enabled_feeds = $this->host->get($this, "enabled_feeds"); + if (!is_array($enabled_feeds)) return $article; + + $key = array_search($article["feed"]["id"], $enabled_feeds); + if ($key === FALSE) return $article; + + return $this->process_article($article); + + } + + function hook_get_full_text($link) + { + $enable_share_anything = $this->host->get($this, "enable_share_anything"); + + if ($enable_share_anything) { + $extracted_content = $this->extract_content($link); + + # let's see if there's anything of value in there + $content_test = trim(strip_tags(sanitize($extracted_content))); + + if ($content_test) { + return $extracted_content; + } + } + + return false; + } + + function api_version() { + return 2; + } + + private function filter_unknown_feeds($enabled_feeds) { + $tmp = array(); + + foreach ($enabled_feeds as $feed) { + + $sth = $this->pdo->prepare("SELECT id FROM ttrss_feeds WHERE id = ? AND owner_uid = ?"); + $sth->execute([$feed, $_SESSION['uid']]); + + if ($row = $sth->fetch()) { + array_push($tmp, $feed); + } + } + + return $tmp; + } + +} diff --git a/vendor/andreskrey/Readability/Configuration.php b/vendor/andreskrey/Readability/Configuration.php new file mode 100644 index 0000000..6c17bc7 --- /dev/null +++ b/vendor/andreskrey/Readability/Configuration.php @@ -0,0 +1,374 @@ + $value) { + $setter = sprintf('set%s', $key); + if (method_exists($this, $setter)) { + call_user_func([$this, $setter], $value); + } + } + } + + /** + * Returns an array-representation of configuration. + * + * @return array + */ + public function toArray() + { + $out = []; + foreach ($this as $key => $value) { + $getter = sprintf('get%s', $key); + if (!is_object($value) && method_exists($this, $getter)) { + $out[$key] = call_user_func([$this, $getter]); + } + } + + return $out; + } + + /** + * @return LoggerInterface + */ + public function getLogger() + { + // If no logger has been set, just return a null logger + if ($this->logger === null) { + return new NullLogger(); + } + + return $this->logger; + } + + /** + * @param LoggerInterface $logger + * + * @return Configuration + */ + public function setLogger(LoggerInterface $logger) + { + $this->logger = $logger; + + return $this; + } + + /** + * @return int + */ + public function getMaxTopCandidates() + { + return $this->maxTopCandidates; + } + + /** + * @param int $maxTopCandidates + * + * @return $this + */ + public function setMaxTopCandidates($maxTopCandidates) + { + $this->maxTopCandidates = $maxTopCandidates; + + return $this; + } + + /** + * @return int + */ + public function getCharThreshold() + { + return $this->charThreshold; + } + + /** + * @param int $charThreshold + * + * @return $this + */ + public function setCharThreshold($charThreshold) + { + $this->charThreshold = $charThreshold; + + return $this; + } + + /** + * @deprecated Use getCharThreshold. Will be removed in version 2.0 + * + * @return int + */ + public function getWordThreshold() + { + @trigger_error('getWordThreshold was replaced with getCharThreshold and will be removed in version 3.0', E_USER_DEPRECATED); + + return $this->charThreshold; + } + + /** + * @param int $charThreshold + * + * @return $this + */ + public function setWordThreshold($charThreshold) + { + @trigger_error('setWordThreshold was replaced with setCharThreshold and will be removed in version 3.0', E_USER_DEPRECATED); + + $this->charThreshold = $charThreshold; + + return $this; + } + + /** + * @return bool + */ + public function getArticleByLine() + { + return $this->articleByLine; + } + + /** + * @param bool $articleByLine + * + * @return $this + */ + public function setArticleByLine($articleByLine) + { + $this->articleByLine = $articleByLine; + + return $this; + } + + /** + * @return bool + */ + public function getStripUnlikelyCandidates() + { + return $this->stripUnlikelyCandidates; + } + + /** + * @param bool $stripUnlikelyCandidates + * + * @return $this + */ + public function setStripUnlikelyCandidates($stripUnlikelyCandidates) + { + $this->stripUnlikelyCandidates = $stripUnlikelyCandidates; + + return $this; + } + + /** + * @return bool + */ + public function getCleanConditionally() + { + return $this->cleanConditionally; + } + + /** + * @param bool $cleanConditionally + * + * @return $this + */ + public function setCleanConditionally($cleanConditionally) + { + $this->cleanConditionally = $cleanConditionally; + + return $this; + } + + /** + * @return bool + */ + public function getWeightClasses() + { + return $this->weightClasses; + } + + /** + * @param bool $weightClasses + * + * @return $this + */ + public function setWeightClasses($weightClasses) + { + $this->weightClasses = $weightClasses; + + return $this; + } + + /** + * @return bool + */ + public function getFixRelativeURLs() + { + return $this->fixRelativeURLs; + } + + /** + * @param bool $fixRelativeURLs + * + * @return $this + */ + public function setFixRelativeURLs($fixRelativeURLs) + { + $this->fixRelativeURLs = $fixRelativeURLs; + + return $this; + } + + /** + * @return bool + */ + public function getSubstituteEntities() + { + return $this->substituteEntities; + } + + /** + * @param bool $substituteEntities + * + * @return $this + */ + public function setSubstituteEntities($substituteEntities) + { + $this->substituteEntities = $substituteEntities; + + return $this; + } + + /** + * @return bool + */ + public function getNormalizeEntities() + { + return $this->normalizeEntities; + } + + /** + * @param bool $normalizeEntities + * + * @return $this + */ + public function setNormalizeEntities($normalizeEntities) + { + $this->normalizeEntities = $normalizeEntities; + + return $this; + } + + /** + * @return string + */ + public function getOriginalURL() + { + return $this->originalURL; + } + + /** + * @param string $originalURL + * + * @return $this + */ + public function setOriginalURL($originalURL) + { + $this->originalURL = $originalURL; + + return $this; + } + + /** + * @return bool + */ + public function getSummonCthulhu() + { + return $this->summonCthulhu; + } + + /** + * @param bool $summonCthulhu + * + * @return $this + */ + public function setSummonCthulhu($summonCthulhu) + { + $this->summonCthulhu = $summonCthulhu; + + return $this; + } +} diff --git a/vendor/andreskrey/Readability/Nodes/DOM/DOMAttr.php b/vendor/andreskrey/Readability/Nodes/DOM/DOMAttr.php new file mode 100644 index 0000000..91729f3 --- /dev/null +++ b/vendor/andreskrey/Readability/Nodes/DOM/DOMAttr.php @@ -0,0 +1,10 @@ +registerNodeClass('DOMAttr', DOMAttr::class); + $this->registerNodeClass('DOMCdataSection', DOMCdataSection::class); + $this->registerNodeClass('DOMCharacterData', DOMCharacterData::class); + $this->registerNodeClass('DOMComment', DOMComment::class); + $this->registerNodeClass('DOMDocument', self::class); + $this->registerNodeClass('DOMDocumentFragment', DOMDocumentFragment::class); + $this->registerNodeClass('DOMDocumentType', DOMDocumentType::class); + $this->registerNodeClass('DOMElement', DOMElement::class); + $this->registerNodeClass('DOMEntity', DOMEntity::class); + $this->registerNodeClass('DOMEntityReference', DOMEntityReference::class); + $this->registerNodeClass('DOMNode', DOMNode::class); + $this->registerNodeClass('DOMNotation', DOMNotation::class); + $this->registerNodeClass('DOMProcessingInstruction', DOMProcessingInstruction::class); + $this->registerNodeClass('DOMText', DOMText::class); + } +} diff --git a/vendor/andreskrey/Readability/Nodes/DOM/DOMDocumentFragment.php b/vendor/andreskrey/Readability/Nodes/DOM/DOMDocumentFragment.php new file mode 100644 index 0000000..bcb8946 --- /dev/null +++ b/vendor/andreskrey/Readability/Nodes/DOM/DOMDocumentFragment.php @@ -0,0 +1,10 @@ +initialized; + } + + /** + * @return bool + */ + public function isReadabilityDataTable() + { + /* + * This is a workaround that I'd like to remove in the future. + * Seems that although we are extending the base DOMElement and adding custom properties (like this one, + * 'readabilityDataTable'), these properties get lost when you search for elements with getElementsByTagName. + * This means that even if we mark the tables in a previous step, when we want to retrieve that information, + * all the custom properties are in their default values. Somehow we need to find a way to make these properties + * permanent across the whole DOM. + * + * @see https://stackoverflow.com/questions/35654709/php-registernodeclass-and-reusing-variable-names + */ + return $this->hasAttribute('readabilityDataTable') + && $this->getAttribute('readabilityDataTable') === '1'; +// return $this->readabilityDataTable; + } + + /** + * @param bool $param + */ + public function setReadabilityDataTable($param) + { + // Can't be "true" because DOMDocument casts it to "1" + $this->setAttribute('readabilityDataTable', $param ? '1' : '0'); +// $this->readabilityDataTable = $param; + } + + /** + * Initializer. Calculates the current score of the node and returns a full Readability object. + * + * @ TODO: I don't like the weightClasses param. How can we get the config here? + * + * @param $weightClasses bool Weight classes? + * + * @return static + */ + public function initializeNode($weightClasses) + { + if (!$this->isInitialized()) { + $contentScore = 0; + + switch ($this->nodeName) { + 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->contentScore = $contentScore + ($weightClasses ? $this->getClassWeight() : 0); + + $this->initialized = true; + } + + return $this; + } + + /** + * Override for native getAttribute method. Some nodes have the getAttribute method, some don't, so we need + * to check first the existence of the attributes property. + * + * @param $attributeName string Attribute to retrieve + * + * @return string + */ + public function getAttribute($attributeName) + { + if (!is_null($this->attributes)) { + return parent::getAttribute($attributeName); + } + + return ''; + } + + /** + * Override for native hasAttribute. + * + * @see getAttribute + * + * @param $attributeName + * + * @return bool + */ + public function hasAttribute($attributeName) + { + if (!is_null($this->attributes)) { + return parent::hasAttribute($attributeName); + } + + return false; + } + + /** + * 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 = $this->parentNode; + + while ($node && !($node instanceof DOMDocument)) { + $ancestors[] = $node; + $level++; + if ($level === $maxLevel) { + break; + } + $node = $node->parentNode; + } + + return $ancestors; + } + + /** + * Returns all links from the current element. + * + * @return array + */ + public function getAllLinks() + { + return iterator_to_array($this->getElementsByTagName('a')); + } + + /** + * 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. + * + * @return int + */ + public function getLinkDensity() + { + $linkLength = 0; + $textLength = mb_strlen($this->getTextContent(true)); + + if (!$textLength) { + return 0; + } + + $links = $this->getAllLinks(); + + if ($links) { + /** @var DOMElement $link */ + foreach ($links as $link) { + $linkLength += mb_strlen($link->getTextContent(true)); + } + } + + return $linkLength / $textLength; + } + + /** + * Calculates the weight of the class/id of the current element. + * + * @return int + */ + public function getClassWeight() + { + $weight = 0; + + // Look for a special classname + $class = $this->getAttribute('class'); + if (trim($class)) { + if (preg_match(NodeUtility::$regexps['negative'], $class)) { + $weight -= 25; + } + + if (preg_match(NodeUtility::$regexps['positive'], $class)) { + $weight += 25; + } + } + + // Look for a special ID + $id = $this->getAttribute('id'); + if (trim($id)) { + if (preg_match(NodeUtility::$regexps['negative'], $id)) { + $weight -= 25; + } + + if (preg_match(NodeUtility::$regexps['positive'], $id)) { + $weight += 25; + } + } + + return $weight; + } + + /** + * Returns the full text of the node. + * + * @param bool $normalize Normalize white space? + * + * @return string + */ + public function getTextContent($normalize = false) + { + $nodeValue = $this->nodeValue; + if ($normalize) { + $nodeValue = trim(preg_replace('/\s{2,}/', ' ', $nodeValue)); + } + + return $nodeValue; + } + + /** + * Returns the children of the current node. + * + * @param bool $filterEmptyDOMText Filter empty DOMText nodes? + * + * @return array + */ + public function getChildren($filterEmptyDOMText = false) + { + $ret = iterator_to_array($this->childNodes); + if ($filterEmptyDOMText) { + // Array values is used to discard the key order. Needs to be 0 to whatever without skipping any number + $ret = array_values(array_filter($ret, function ($node) { + return $node->nodeName !== '#text' || mb_strlen(trim($node->nodeValue)); + })); + } + + return $ret; + } + + /** + * Return an array indicating how many rows and columns this table has. + * + * @return array + */ + public function getRowAndColumnCount() + { + $rows = $columns = 0; + $trs = $this->getElementsByTagName('tr'); + foreach ($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]; + } + + /** + * Creates a new node based on the text content of the original node. + * + * @param $originalNode DOMNode + * @param $tagName string + * + * @return DOMElement + */ + public function createNode($originalNode, $tagName) + { + $text = $originalNode->getTextContent(); + $newNode = $originalNode->ownerDocument->createElement($tagName, $text); + + return $newNode; + } + + /** + * Check if a given node has one of its ancestor tag name matching the + * provided one. + * + * @param string $tagName + * @param int $maxDepth + * @param callable $filterFn + * + * @return bool + */ + public function hasAncestorTag($tagName, $maxDepth = 3, callable $filterFn = null) + { + $depth = 0; + $node = $this; + + while ($node->parentNode) { + if ($maxDepth > 0 && $depth > $maxDepth) { + return false; + } + + if ($node->parentNode->nodeName === $tagName && (!$filterFn || $filterFn($node->parentNode))) { + return true; + } + + $node = $node->parentNode; + $depth++; + } + + return false; + } + + /** + * Check if this node has only whitespace and a single element with given tag + * or if it contains no element with given tag or more than 1 element. + * + * @param $tag string Name of tag + * + * @return bool + */ + public function hasSingleTagInsideElement($tag) + { + // There should be exactly 1 element child with given tag + if (count($children = $this->getChildren(true)) !== 1 || $children[0]->nodeName !== $tag) { + return false; + } + + // And there should be no text nodes with real content + return array_reduce($children, function ($carry, $child) { + if (!$carry === false) { + return false; + } + + /* @var DOMNode $child */ + return !($child->nodeType === XML_TEXT_NODE && !preg_match('/\S$/', $child->getTextContent())); + }); + } + + /** + * Check if the current element has a single child block element. + * Block elements are the ones defined in the divToPElements array. + * + * @return bool + */ + public function hasSingleChildBlockElement() + { + $result = false; + if ($this->hasChildNodes()) { + foreach ($this->getChildren() as $child) { + if (in_array($child->nodeName, $this->divToPElements)) { + $result = true; + } else { + // If any of the hasSingleChildBlockElement calls return true, return true then. + /** @var $child DOMElement */ + $result = ($result || $child->hasSingleChildBlockElement()); + } + } + } + + return $result; + } + + /** + * 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 $this instanceof DOMElement && + mb_strlen(preg_replace(NodeUtility::$regexps['onlyWhitespace'], '', $this->textContent)) === 0 && + ($this->childNodes->length === 0 || + $this->childNodes->length === $this->getElementsByTagName('br')->length + $this->getElementsByTagName('hr')->length + /* + * Special PHP 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 "


) 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($this->childNodes), function ($child) { + return $child instanceof DOMText; + })) + + ); + } + + /** + * Determine if a node qualifies as phrasing content. + * https://developer.mozilla.org/en-US/docs/Web/Guide/HTML/Content_categories#Phrasing_content. + * + * @return bool + */ + public function isPhrasingContent() + { + return $this->nodeType === XML_TEXT_NODE || in_array($this->nodeName, $this->phrasing_elems) !== false || + (!is_null($this->childNodes) && + ($this->nodeName === 'a' || $this->nodeName === 'del' || $this->nodeName === 'ins') && + array_reduce(iterator_to_array($this->childNodes), function ($carry, $node) { + return $node->isPhrasingContent() && $carry; + }, true) + ); + } + + public function isProbablyVisible() + { + /* + * In the original JS project they check if the node has the style display=none, which unfortunately + * in our case we have no way of knowing that. So we just check for the attribute hidden or "display: none". + * + * Might be a good idea to check for classes or other attributes like 'aria-hidden' + */ + + return !preg_match('/display:( )?none/', $this->getAttribute('style')) && !$this->hasAttribute('hidden'); + } + + public function isWhitespace() + { + return ($this->nodeType === XML_TEXT_NODE && mb_strlen(trim($this->textContent)) === 0) || + ($this->nodeType === XML_ELEMENT_NODE && $this->nodeName === 'br'); + } + + /** + * This is a hack that overcomes the issue of node shifting when scanning and removing nodes. + * + * In the JS version of getElementsByTagName, if you remove a node it will not appear during the + * foreach. This does not happen in PHP DOMDocument, because if you remove a node, it will still appear but as an + * orphan node and will give an exception if you try to do anything with it. + * + * Shifting also occurs when converting parent nodes (like a P to a DIV), which in that case the found nodes are + * removed from the foreach "pool" but the internal index of the foreach is not aware and skips over nodes that + * never looped over. (index is at position 5, 2 nodes are removed, next one should be node 3, but the foreach tries + * to access node 6) + * + * This function solves this by searching for the nodes on every loop and keeping track of the count differences. + * Because on every loop we call getElementsByTagName again, this could cause a performance impact and should be + * used only when the results of the search are going to be used to remove the nodes. + * + * @param string $tag + * + * @return \Generator + */ + public function shiftingAwareGetElementsByTagName($tag) + { + /** @var $nodes DOMNodeList */ + $nodes = $this->getElementsByTagName($tag); + $count = $nodes->length; + + for ($i = 0; $i < $count; $i = max(++$i, 0)) { + yield $nodes->item($i); + + // Search for all the nodes again + $nodes = $this->getElementsByTagName($tag); + + // Subtract the amount of nodes removed from the current index + $i -= $count - $nodes->length; + + // Subtract the amount of nodes removed from the current count + $count -= ($count - $nodes->length); + } + } +} diff --git a/vendor/andreskrey/Readability/Nodes/NodeUtility.php b/vendor/andreskrey/Readability/Nodes/NodeUtility.php new file mode 100644 index 0000000..7a1f18e --- /dev/null +++ b/vendor/andreskrey/Readability/Nodes/NodeUtility.php @@ -0,0 +1,160 @@ + '/-ad-|banner|breadcrumbs|combx|comment|community|cover-wrap|disqus|extra|foot|header|legends|menu|related|remark|replies|rss|shoutbox|sidebar|skyscraper|social|sponsor|supplemental|ad-break|agegate|pagination|pager|popup|yom-remote/i', + 'okMaybeItsACandidate' => '/and|article|body|column|main|shadow/i', + 'extraneous' => '/print|archive|comment|discuss|e[\-]?mail|share|reply|all|login|sign|single|utility/i', + 'byline' => '/byline|author|dateline|writtenby|p-author/i', + 'replaceFonts' => '/<(\/?)font[^>]*>/gi', + 'normalize' => '/\s{2,}/', + 'videos' => '/\/\/(www\.)?((dailymotion|youtube|youtube-nocookie|player\.vimeo|v\.qq)\.com|(archive|upload\.wikimedia)\.org|player\.twitch\.tv)/i', + 'nextLink' => '/(next|weiter|continue|>([^\|]|$)|»([^\|]|$))/i', + 'prevLink' => '/(prev|earl|old|new|<|«)/i', + 'whitespace' => '/^\s*$/', + 'hasContent' => '/\S$/', + 'positive' => '/article|body|content|entry|hentry|h-entry|main|page|pagination|post|text|blog|story/i', + 'negative' => '/hidden|^hid$| hid$| hid |^hid |banner|combx|comment|com-|contact|foot|footer|footnote|masthead|media|meta|outbrain|promo|related|scroll|share|shoutbox|sidebar|skyscraper|sponsor|shopping|tags|tool|widget/i', + // \x{00A0} is the unicode version of   + 'onlyWhitespace' => '/\x{00A0}|\s+/u' + ]; + + /** + * Imported from the Element class on league\html-to-markdown. + * + * @param $node + * + * @return DOMElement + */ + public static function nextElement($node) + { + $next = $node; + while ($next + && $next->nodeType !== XML_ELEMENT_NODE + && $next->isWhitespace()) { + $next = $next->nextSibling; + } + + return $next; + } + + /** + * 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 DOMNode $node + * @param string $value + * @param bool $importAttributes + * + * @return DOMNode + */ + public static function setNodeTag($node, $value, $importAttributes = true) + { + $new = new DOMDocument('1.0', 'utf-8'); + $new->appendChild($new->createElement($value)); + + $children = $node->childNodes; + /** @var $children \DOMNodeList $i */ + for ($i = 0; $i < $children->length; $i++) { + $import = $new->importNode($children->item($i), true); + $new->firstChild->appendChild($import); + } + + if ($importAttributes) { + // Import attributes from the original node. + foreach ($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 = $node->ownerDocument->importNode($new->firstChild, true); + $node->parentNode->replaceChild($import, $node); + + return $import; + } + + /** + * Removes the current node and returns the next node to be parsed (child, sibling or parent). + * + * @param DOMNode $node + * + * @return DOMNode + */ + public static function removeAndGetNext($node) + { + $nextNode = self::getNextNode($node, true); + $node->parentNode->removeChild($node); + + return $nextNode; + } + + /** + * Remove the selected node. + * + * @param $node DOMElement + * + * @return void + **/ + public static function removeNode($node) + { + $parent = $node->parentNode; + if ($parent) { + $parent->removeChild($node); + } + } + + /** + * Returns the next node. First checks for children (if the flag allows it), then for siblings, and finally + * for parents. + * + * @param DOMNode $originalNode + * @param bool $ignoreSelfAndKids + * + * @return DOMNode + */ + public static function getNextNode($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->firstChild) { + return $originalNode->firstChild; + } + + // Then for siblings... + if ($originalNode->nextSibling) { + return $originalNode->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 = $originalNode->parentNode; + } while ($originalNode && !$originalNode->nextSibling); + + return ($originalNode) ? $originalNode->nextSibling : $originalNode; + } +} diff --git a/vendor/andreskrey/Readability/ParseException.php b/vendor/andreskrey/Readability/ParseException.php new file mode 100644 index 0000000..646c04a --- /dev/null +++ b/vendor/andreskrey/Readability/ParseException.php @@ -0,0 +1,7 @@ +configuration = $configuration; + $this->logger = $this->configuration->getLogger(); + } + + /** + * Main parse function. + * + * @param $html + * + * @throws ParseException + * + * @return bool + */ + public function parse($html) + { + $this->logger->info('*** Starting parse process...'); + + $this->dom = $this->loadHTML($html); + + // Checking for minimum HTML to work with. + if (!($root = $this->dom->getElementsByTagName('body')->item(0)) || !$root->firstChild) { + $this->logger->emergency('No body tag present or body tag empty'); + + throw new ParseException('Invalid or incomplete HTML.'); + } + + $this->getMetadata(); + + $this->getMainImage(); + + while (true) { + $root = $root->firstChild; + + $elementsToScore = $this->getNodes($root); + $this->logger->debug(sprintf('Elements to score: \'%s\'', count($elementsToScore))); + + $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. + */ + + $length = mb_strlen(preg_replace(NodeUtility::$regexps['onlyWhitespace'], '', $result->textContent)); + + $this->logger->info(sprintf('[Parsing] Article parsed. Amount of words: %s. Current threshold is: %s', $length, $this->configuration->getCharThreshold())); + + if ($result && $length < $this->configuration->getCharThreshold()) { + $this->dom = $this->loadHTML($html); + $root = $this->dom->getElementsByTagName('body')->item(0); + + if ($this->configuration->getStripUnlikelyCandidates()) { + $this->logger->debug('[Parsing] Threshold not met, trying again setting StripUnlikelyCandidates as false'); + $this->configuration->setStripUnlikelyCandidates(false); + $this->attempts[] = ['articleContent' => $result, 'textLength' => $length]; + } elseif ($this->configuration->getWeightClasses()) { + $this->logger->debug('[Parsing] Threshold not met, trying again setting WeightClasses as false'); + $this->configuration->setWeightClasses(false); + $this->attempts[] = ['articleContent' => $result, 'textLength' => $length]; + } elseif ($this->configuration->getCleanConditionally()) { + $this->logger->debug('[Parsing] Threshold not met, trying again setting CleanConditionally as false'); + $this->configuration->setCleanConditionally(false); + $this->attempts[] = ['articleContent' => $result, 'textLength' => $length]; + } else { + $this->logger->debug('[Parsing] Threshold not met, searching across attempts for some content.'); + $this->attempts[] = ['articleContent' => $result, 'textLength' => $length]; + + // No luck after removing flags, just return the longest text we found during the different loops + usort($this->attempts, function ($a, $b) { + return $a['textLength'] < $b['textLength']; + }); + + // But first check if we actually have something + if (!$this->attempts[0]['textLength']) { + $this->logger->emergency('[Parsing] Could not parse text, giving up :('); + + throw new ParseException('Could not parse text.'); + } + + $this->logger->debug('[Parsing] Threshold not met, but found some content in previous attempts.'); + + $result = $this->attempts[0]['articleContent']; + break; + } + } else { + break; + } + } + + $result = $this->postProcessContent($result); + + // If we haven't found an excerpt in the article's metadata, use the article's + // first paragraph as the excerpt. This can be used for displaying a preview of + // the article's content. + if (!$this->getExcerpt()) { + $this->logger->debug('[Parsing] No excerpt text found on metadata, extracting first p node and using it as excerpt.'); + $paragraphs = $result->getElementsByTagName('p'); + if ($paragraphs->length > 0) { + $this->setExcerpt(trim($paragraphs->item(0)->textContent)); + } + } + + $this->setContent($result); + + $this->logger->info('*** Parse successful :)'); + + return true; + } + + /** + * 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) + { + $this->logger->debug('[Loading] Loading HTML...'); + + // To avoid throwing a gazillion of errors on malformed HTMLs + libxml_use_internal_errors(true); + + $dom = new DOMDocument('1.0', 'utf-8'); + + if (!$this->configuration->getSubstituteEntities()) { + // Keep the original HTML entities + $dom->substituteEntities = false; + } + + if ($this->configuration->getNormalizeEntities()) { + $this->logger->debug('[Loading] Normalized entities via mb_convert_encoding.'); + // 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->configuration->getSummonCthulhu()) { + $this->logger->debug('[Loading] Removed script tags via regex H̶͈̩̟̬̱͠E̡̨̬͔̳̜͢͠ ̡̧̯͉̩͙̩̹̞̠͎͈̹̥̠͞ͅͅC̶͉̞̘̖̝̗͓̬̯͍͉̤̬͢͢͞Ò̟̘͉͖͎͉̱̭̣̕M̴̯͈̻̱̱̣̗͈̠̙̲̥͘͞E̷̛͙̼̲͍͕̹͍͇̗̻̬̮̭̱̥͢Ş̛̟͔̙̜̤͇̮͍̙̝̀͘'); + $html = preg_replace('/]*>([\s\S]*?)<\/script>/', '', $html); + } + + // Prepend the XML tag to avoid having issues with special characters. Should be harmless. + $dom->loadHTML('' . $html); + $dom->encoding = 'UTF-8'; + + $this->removeScripts($dom); + + $this->prepDocument($dom); + + $this->logger->debug('[Loading] Loaded HTML successfully.'); + + return $dom; + } + + /** + * Tries to guess relevant info from metadata of the html. Sets the results in the Readability properties. + */ + private function getMetadata() + { + $this->logger->debug('[Metadata] Retrieving metadata...'); + + $values = []; + // property is a space-separated list of values + $propertyPattern = '/\s*(dc|dcterm|og|twitter)\s*:\s*(author|creator|description|title|image)\s*/i'; + + // name is a single value + $namePattern = '/^\s*(?:(dc|dcterm|og|twitter|weibo:(article|webpage))\s*[\.:]\s*)?(author|creator|description|title|image)\s*$/i'; + + // Find description tags. + foreach ($this->dom->getElementsByTagName('meta') as $meta) { + /* @var DOMNode $meta */ + $elementName = $meta->getAttribute('name'); + $elementProperty = $meta->getAttribute('property'); + $content = $meta->getAttribute('content'); + $matches = null; + $name = null; + + if ($elementProperty) { + if (preg_match($propertyPattern, $elementProperty, $matches)) { + for ($i = count($matches) - 1; $i >= 0; $i--) { + // Convert to lowercase, and remove any whitespace + // so we can match below. + $name = preg_replace('/\s/', '', mb_strtolower($matches[$i])); + // multiple authors + $values[$name] = trim($content); + } + } + } + + if (!$matches && $elementName && preg_match($namePattern, $elementName)) { + $name = $elementName; + if ($content) { + // Convert to lowercase, remove any whitespace, and convert dots + // to colons so we can match below. + $name = preg_replace(['/\s/', '/\./'], ['', ':'], mb_strtolower($name)); + $values[$name] = trim($content); + } + } + } + + // get title + /* + * This is a very convoluted way of extracting the first matching key of the $values array + * against a set of options. + * + * This could be easily replaced with an ugly set of isset($values['key']) or a bunch of ??s. + * Will probably replace it with ??s after dropping support of PHP5.6 + */ + + $key = current(array_intersect([ + 'dc:title', + 'dcterm:title', + 'og:title', + 'weibo:article:title', + 'weibo:webpage:title', + 'title', + 'twitter:title' + ], array_keys($values))); + + $this->setTitle(isset($values[$key]) ? trim($values[$key]) : null); + + if (!$this->getTitle()) { + $this->setTitle($this->getArticleTitle()); + } + + // get author + $key = current(array_intersect([ + 'dc:creator', + 'dcterm:creator', + 'author' + ], array_keys($values))); + + $this->setAuthor(isset($values[$key]) ? $values[$key] : null); + + // get description + $key = current(array_intersect([ + 'dc:description', + 'dcterm:description', + 'og:description', + 'weibo:article:description', + 'weibo:webpage:description', + 'description', + 'twitter:description' + ], array_keys($values))); + + $this->setExcerpt(isset($values[$key]) ? $values[$key] : null); + + // get main image + $key = current(array_intersect([ + 'og:image', + 'twitter:image' + ], array_keys($values))); + + $this->setImage(isset($values[$key]) ? $values[$key] : null); + } + + /** + * Returns all the images of the parsed article. + * + * @return array + */ + public function getImages() + { + $result = []; + if ($this->getImage()) { + $result[] = $this->getImage(); + } + + if (null == $this->getDOMDocument()) { + return $result; + } + + foreach ($this->getDOMDocument()->getElementsByTagName('img') as $img) { + if ($src = $img->getAttribute('src')) { + $result[] = $src; + } + } + + if ($this->configuration->getFixRelativeURLs()) { + foreach ($result as &$imgSrc) { + $imgSrc = $this->toAbsoluteURI($imgSrc); + } + } + + $result = array_unique(array_filter($result)); + + return $result; + } + + /** + * Tries to get the main article image. Will only update the metadata if the getMetadata function couldn't + * find a correct image. + */ + public function getMainImage() + { + $imgUrl = false; + + if ($this->getImage() !== null) { + $imgUrl = $this->getImage(); + } + + if (!$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->configuration->getFixRelativeURLs()) { + $this->setImage($this->toAbsoluteURI($imgUrl)); + } + } + + /** + * Returns the title of the html. Prioritizes the title from the metadata against the title tag. + * + * @return string|null + */ + private function getArticleTitle() + { + $originalTitle = null; + + if ($this->getTitle()) { + $originalTitle = $this->getTitle(); + } else { + $this->logger->debug('[Metadata] Could not find title in metadata, searching for the title tag...'); + $titleTag = $this->dom->getElementsByTagName('title'); + if ($titleTag->length > 0) { + $this->logger->info(sprintf('[Metadata] Using title tag as article title: \'%s\'', $titleTag->item(0)->nodeValue)); + $originalTitle = $titleTag->item(0)->nodeValue; + } + } + + if ($originalTitle === null) { + return null; + } + + $curTitle = $originalTitle = trim($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); + + $this->logger->info(sprintf('[Metadata] Found hierarchical separators in title, new title is: \'%s\'', $curTitle)); + + // 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); + $this->logger->info(sprintf('[Metadata] Title too short, using the first part of the title instead: \'%s\'', $curTitle)); + } + } elseif (strpos($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); + + $this->logger->info(sprintf('[Metadata] Title has a colon in the middle, new title is: \'%s\'', $curTitle)); + + // 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); + $this->logger->info(sprintf('[Metadata] Title too short, using the first part of the title instead: \'%s\'', $curTitle)); + } elseif (count(preg_split('/\s+/', substr($curTitle, 0, strpos($curTitle, ':')))) > 5) { + // But if we have too many words before the colon there's something weird + // with the titles and the H tags so let's just use the original title instead + $curTitle = $originalTitle; + } + } + } elseif (mb_strlen($curTitle) > 150 || mb_strlen($curTitle) < 15) { + $hOnes = $this->dom->getElementsByTagName('h1'); + + if ($hOnes->length === 1) { + $curTitle = $hOnes->item(0)->nodeValue; + $this->logger->info(sprintf('[Metadata] Using title from an H1 node: \'%s\'', $curTitle)); + } + } + + $curTitle = trim($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; + + $this->logger->info(sprintf('Using title from an H1 node: \'%s\'', $curTitle)); + } + + return $curTitle; + } + + /** + * Convert URI to an absolute URI. + * + * @param $uri string URI to convert + * + * @return string + */ + private function toAbsoluteURI($uri) + { + list($pathBase, $scheme, $prePath) = $this->getPathInfo($this->configuration->getOriginalURL()); + + // 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; + } + + /** + * Returns full path info of an URL. + * + * @param string $url + * + * @return array [$pathBase, $scheme, $prePath] + */ + public function getPathInfo($url) + { + // Check for base URLs + if ($this->dom->baseURI !== null) { + if (substr($this->dom->baseURI, 0, 1) === '/') { + // URLs starting with '/' override completely the URL defined in the link + $pathBase = parse_url($url, PHP_URL_SCHEME) . '://' . parse_url($url, PHP_URL_HOST) . $this->dom->baseURI; + } else { + // Otherwise just prepend the base to the actual path + $pathBase = parse_url($url, PHP_URL_SCHEME) . '://' . parse_url($url, PHP_URL_HOST) . dirname(parse_url($url, PHP_URL_PATH)) . '/' . rtrim($this->dom->baseURI, '/') . '/'; + } + } else { + $pathBase = parse_url($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]; + } + + /** + * Gets nodes from the root element. + * + * @param $node DOMNode|DOMText + * + * @return array + */ + private function getNodes($node) + { + $this->logger->info('[Get Nodes] Retrieving nodes...'); + + $stripUnlikelyCandidates = $this->configuration->getStripUnlikelyCandidates(); + + $elementsToScore = []; + + /* + * First, node prepping. Trash nodes that look cruddy (like ones with the + * class name "comment", etc), and turn divs into P tags where they have been + * used inappropriately (as in, where they contain no other block level elements.) + */ + + while ($node) { + // Remove DOMComments nodes as we don't need them and mess up children counting + if ($node->nodeType === XML_COMMENT_NODE) { + $this->logger->debug(sprintf('[Get Nodes] Found comment node, removing... Node content was: \'%s\'', substr($node->nodeValue, 0, 128))); + $node = NodeUtility::removeAndGetNext($node); + continue; + } + + $matchString = $node->getAttribute('class') . ' ' . $node->getAttribute('id'); + + if (!$node->isProbablyVisible()) { + $this->logger->debug(sprintf('[Get Nodes] Removing hidden node... Match string was: \'%s\'', $matchString)); + $node = NodeUtility::removeAndGetNext($node); + continue; + } + + // Check to see if this node is a byline, and remove it if it is. + if ($this->checkByline($node, $matchString)) { + $this->logger->debug(sprintf('[Get Nodes] Found byline, removing... Node content was: \'%s\'', substr($node->nodeValue, 0, 128))); + $node = NodeUtility::removeAndGetNext($node); + continue; + } + + // Remove unlikely candidates + if ($stripUnlikelyCandidates) { + if ( + preg_match(NodeUtility::$regexps['unlikelyCandidates'], $matchString) && + !preg_match(NodeUtility::$regexps['okMaybeItsACandidate'], $matchString) && + $node->nodeName !== 'body' && + $node->nodeName !== 'a' + ) { + $this->logger->debug(sprintf('[Get Nodes] Removing unlikely candidate. Node content was: \'%s\'', substr($node->nodeValue, 0, 128))); + $node = NodeUtility::removeAndGetNext($node); + continue; + } + } + + // Remove DIV, SECTION, and HEADER nodes without any content(e.g. text, image, video, or iframe). + if (($node->nodeName === 'div' || $node->nodeName === 'section' || $node->nodeName === 'header' || + $node->nodeName === 'h1' || $node->nodeName === 'h2' || $node->nodeName === 'h3' || + $node->nodeName === 'h4' || $node->nodeName === 'h5' || $node->nodeName === 'h6' || + $node->nodeName === 'p') && + $node->isElementWithoutContent()) { + $this->logger->debug(sprintf('[Get Nodes] Removing empty \'%s\' node.', $node->nodeName)); + $node = NodeUtility::removeAndGetNext($node); + continue; + } + + if (in_array(strtolower($node->nodeName), $this->defaultTagsToScore)) { + $this->logger->debug(sprintf('[Get Nodes] Adding node to score list, node content is: \'%s\'', substr($node->nodeValue, 0, 128))); + $elementsToScore[] = $node; + } + + // Turn all divs that don't have children block level elements into p's + if ($node->nodeName === 'div') { + // Put phrasing content into paragraphs. + $p = null; + $childNode = $node->firstChild; + while ($childNode) { + $nextSibling = $childNode->nextSibling; + if ($childNode->isPhrasingContent()) { + if ($p !== null) { + $p->appendChild($childNode); + } elseif (!$childNode->isWhitespace()) { + $p = $this->dom->createElement('p'); + $node->replaceChild($p, $childNode); + $p->appendChild($childNode); + } + } elseif ($p !== null) { + while ($p->lastChild && $p->lastChild->isWhitespace()) { + $p->removeChild($p->lastChild); + } + $p = null; + } + $childNode = $nextSibling; + } + + /* + * Sites like http://mobile.slate.com encloses each paragraph with a DIV + * element. DIVs with only a P element inside and no text content can be + * safely converted into plain P elements to avoid confusing the scoring + * algorithm with DIVs with are, in practice, paragraphs. + */ + if ($node->hasSingleTagInsideElement('p') && $node->getLinkDensity() < 0.25) { + $this->logger->debug(sprintf('[Get Nodes] Found DIV with a single P node, removing DIV. Node content is: \'%s\'', substr($node->nodeValue, 0, 128))); + $pNode = $node->getChildren(true)[0]; + $node->parentNode->replaceChild($pNode, $node); + $node = $pNode; + $elementsToScore[] = $node; + } elseif (!$node->hasSingleChildBlockElement()) { + $this->logger->debug(sprintf('[Get Nodes] Found DIV with a single child block element, converting to a P node. Node content is: \'%s\'', substr($node->nodeValue, 0, 128))); + $node = NodeUtility::setNodeTag($node, 'p'); + $elementsToScore[] = $node; + } + } + + $node = NodeUtility::getNextNode($node); + } + + return $elementsToScore; + } + + /** + * Checks if the node is a byline. + * + * @param DOMNode $node + * @param string $matchString + * + * @return bool + */ + private function checkByline($node, $matchString) + { + if (!$this->configuration->getArticleByLine()) { + return false; + } + + /* + * Check if the byline is already set + */ + if ($this->getAuthor()) { + return false; + } + + $rel = $node->getAttribute('rel'); + + if ($rel === 'author' || preg_match(NodeUtility::$regexps['byline'], $matchString) && $this->isValidByline($node->getTextContent())) { + $this->logger->info(sprintf('[Metadata] Found article author: \'%s\'', $node->getTextContent())); + $this->setAuthor(trim($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($byline) > 0) && (mb_strlen($byline) < 100); + } + + return false; + } + + /** + * Removes all the scripts of the html. + * + * @param DOMDocument $dom + */ + private function removeScripts(DOMDocument $dom) + { + foreach (['script', 'noscript'] as $tag) { + $nodes = $dom->getElementsByTagName($tag); + foreach (iterator_to_array($nodes) as $node) { + NodeUtility::removeNode($node); + } + } + } + + /** + * Prepares the document for parsing. + * + * @param DOMDocument $dom + */ + private function prepDocument(DOMDocument $dom) + { + $this->logger->info('[PrepDocument] Preparing document for parsing...'); + + foreach ($dom->shiftingAwareGetElementsByTagName('br') as $br) { + $next = $br->nextSibling; + + /* + * Whether 2 or more
elements have been found and replaced with a + *

block. + */ + $replaced = false; + + /* + * If we find a
chain, remove the
s until we hit another element + * or non-whitespace. This leaves behind the first
in the chain + * (which will be replaced with a

later). + */ + while (($next = NodeUtility::nextElement($next)) && ($next->nodeName === 'br')) { + $this->logger->debug('[PrepDocument] Removing chain of BR nodes...'); + + $replaced = true; + $brSibling = $next->nextSibling; + $next->parentNode->removeChild($next); + $next = $brSibling; + } + + /* + * If we removed a
chain, replace the remaining
with a

. Add + * all sibling nodes as children of the

until we hit another
+ * chain. + */ + + if ($replaced) { + $p = $dom->createElement('p'); + $br->parentNode->replaceChild($p, $br); + + $next = $p->nextSibling; + while ($next) { + // If we've hit another

, we're done adding children to this

. + if ($next->nodeName === 'br') { + $nextElem = NodeUtility::nextElement($next->nextSibling); + if ($nextElem && $nextElem->nodeName === 'br') { + break; + } + } + + if (!$next->isPhrasingContent()) { + break; + } + + $this->logger->debug('[PrepDocument] Replacing BR with a P node...'); + + // Otherwise, make this node a child of the new

. + $sibling = $next->nextSibling; + $p->appendChild($next); + $next = $sibling; + } + + while ($p->lastChild && $p->lastChild->isWhitespace()) { + $p->removeChild($p->lastChild); + } + + if ($p->parentNode->tagName === 'p') { + NodeUtility::setNodeTag($p->parentNode, 'div'); + } + } + } + + // Replace font tags with span + $fonts = $dom->getElementsByTagName('font'); + $length = $fonts->length; + for ($i = 0; $i < $length; $i++) { + $this->logger->debug('[PrepDocument] Converting font tag into a span tag.'); + $font = $fonts->item($length - 1 - $i); + NodeUtility::setNodeTag($font, 'span'); + } + } + + /** + * Assign scores to each node. Returns full article parsed or false on error. + * + * @param array $nodes + * + * @return DOMDocument|bool + */ + private function rateNodes($nodes) + { + $this->logger->info('[Rating] Rating nodes...'); + + $candidates = []; + + /** @var DOMElement $node */ + foreach ($nodes as $node) { + if (is_null($node->parentNode)) { + 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); + + $this->logger->debug(sprintf('[Rating] Node score %s, content: \'%s\'', $contentScore, substr($node->nodeValue, 0, 128))); + + /** @var $ancestor DOMElement */ + foreach ($ancestors as $level => $ancestor) { + $this->logger->debug('[Rating] Found ancestor, initializing and adding it as a candidate...'); + if (!$ancestor->isInitialized()) { + $ancestor->initializeNode($this->configuration->getWeightClasses()); + $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->contentScore; + $ancestor->contentScore = $currentScore + ($contentScore / $scoreDivider); + + $this->logger->debug(sprintf('[Rating] Ancestor score %s, value: \'%s\'', $ancestor->contentScore, substr($ancestor->nodeValue, 0, 128))); + } + } + + /* + * 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->contentScore = $candidate->contentScore * (1 - $candidate->getLinkDensity()); + + for ($i = 0; $i < $this->configuration->getMaxTopCandidates(); $i++) { + $aTopCandidate = isset($topCandidates[$i]) ? $topCandidates[$i] : null; + + if (!$aTopCandidate || $candidate->contentScore > $aTopCandidate->contentScore) { + array_splice($topCandidates, $i, 0, [$candidate]); + if (count($topCandidates) > $this->configuration->getMaxTopCandidates()) { + array_pop($topCandidates); + } + break; + } + } + } + + $topCandidate = isset($topCandidates[0]) ? $topCandidates[0] : null; + $parentOfTopCandidate = null; + + /* + * If we still have no top candidate, just use the body as a last resort. + * We also have to copy the body node so it is something we can modify. + */ + + if ($topCandidate === null || $topCandidate->nodeName === 'body') { + $this->logger->info('[Rating] No top candidate found or top candidate is the body tag. Moving all child nodes to a new DIV node.'); + + // 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); + } + + // Candidate must be created using firstChild to grab the DOMElement instead of the DOMDocument. + $topCandidate = $topCandidate->firstChild; + } elseif ($topCandidate) { + $this->logger->info(sprintf('[Rating] Found top candidate, score: %s', $topCandidate->contentScore)); + // 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); $i++) { + // In some cases we may end up with a top candidate with zero content score. To avoid dividing by zero + // we have to use max() and replace zero with a low value like 0.1 + if ($topCandidates[$i]->contentScore / max($topCandidate->contentScore, 0.1) >= 0.75) { + array_push($alternativeCandidateAncestors, $topCandidates[$i]->getNodeAncestors(false)); + } + } + + $MINIMUM_TOPCANDIDATES = 3; + if (count($alternativeCandidateAncestors) >= $MINIMUM_TOPCANDIDATES) { + $parentOfTopCandidate = $topCandidate->parentNode; + + // Check if we are actually dealing with a DOMNode and not a DOMDocument node or higher + while ($parentOfTopCandidate->nodeName !== 'body' && $parentOfTopCandidate->nodeType === XML_ELEMENT_NODE) { + $listsContainingThisAncestor = 0; + for ($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->parentNode; + } + } + + /* + * 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->parentNode; + $lastScore = $topCandidate->contentScore; + + // The scores shouldn't get too low. + $scoreThreshold = $lastScore / 3; + + /* @var DOMElement $parentOfTopCandidate */ + while ($parentOfTopCandidate->nodeName !== 'body') { + $parentScore = $parentOfTopCandidate->contentScore; + if ($parentScore < $scoreThreshold) { + break; + } + + if ($parentScore > $lastScore) { + // Alright! We found a better parent to use. + $topCandidate = $parentOfTopCandidate; + $this->logger->info('[Rating] Found a better top candidate.'); + break; + } + $lastScore = $parentOfTopCandidate->contentScore; + $parentOfTopCandidate = $parentOfTopCandidate->parentNode; + } + + // 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 = $topCandidate->parentNode; + while ($parentOfTopCandidate->nodeName !== 'body' && count($parentOfTopCandidate->getChildren(true)) === 1) { + $topCandidate = $parentOfTopCandidate; + $parentOfTopCandidate = $topCandidate->parentNode; + } + } + + /* + * 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. + */ + + $this->logger->info('[Rating] Creating final article content document...'); + + $articleContent = new DOMDocument('1.0', 'utf-8'); + $articleContent->createElement('div'); + + $siblingScoreThreshold = max(10, $topCandidate->contentScore * 0.2); + // Keep potential top candidate's parent node to try to get text direction of it later. + $parentOfTopCandidate = $topCandidate->parentNode; + $siblings = $parentOfTopCandidate->getChildren(); + + $hasContent = false; + + $this->logger->info('[Rating] Adding top candidate siblings...'); + + /** @var DOMElement $sibling */ + foreach ($siblings as $sibling) { + $append = false; + + if ($sibling === $topCandidate) { + $this->logger->debug('[Rating] Sibling is equal to the top candidate, adding to the final article...'); + + $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 += $topCandidate->contentScore * 0.2; + } + if ($sibling->contentScore + $contentBonus >= $siblingScoreThreshold) { + $append = true; + } elseif ($sibling->nodeName === 'p') { + $linkDensity = $sibling->getLinkDensity(); + $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) { + $this->logger->debug(sprintf('[Rating] Appending sibling to final article, content is: \'%s\'', substr($sibling->nodeValue, 0, 128))); + + $hasContent = true; + + if (!in_array(strtolower($sibling->nodeName), $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. + */ + + $sibling = NodeUtility::setNodeTag($sibling, 'div'); + } + + $import = $articleContent->importNode($sibling, 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) { + $this->setDirection($articleDir); + $this->logger->debug(sprintf('[Rating] Found article direction: %s', $articleDir)); + break; + } + } + + return $articleContent; + } else { + return false; + } + } + + /** + * Cleans up the final article. + * + * @param DOMDocument $article + * + * @return DOMDocument + */ + public function prepArticle(DOMDocument $article) + { + $this->logger->info('[PrepArticle] Preparing final 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'); + $this->_clean($article, 'link'); + $this->_clean($article, 'aside'); + + // 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($this->getTitle())) / max(mb_strlen($this->getTitle()), 1); + + if (abs($lengthSimilarRate) < 0.5) { + if ($lengthSimilarRate > 0) { + $titlesMatch = strpos($h2->item(0)->textContent, $this->getTitle()) !== false; + } else { + $titlesMatch = strpos($this->getTitle(), $h2->item(0)->textContent) !== false; + } + if ($titlesMatch) { + $this->logger->info('[PrepArticle] Found title repeated in an H2 node, removing...'); + $this->_clean($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); + + foreach (iterator_to_array($article->getElementsByTagName('br')) as $br) { + $next = $br->nextSibling; + if ($next && $next->nodeName === 'p') { + $this->logger->debug('[PrepArticle] Removing br node next to a p node.'); + $br->parentNode->removeChild($br); + } + } + + // Remove single-cell tables + foreach ($article->shiftingAwareGetElementsByTagName('table') as $table) { + /** @var DOMNode $table */ + $tbody = $table->hasSingleTagInsideElement('tbody') ? $table->childNodes[0] : $table; + if ($tbody->hasSingleTagInsideElement('tr')) { + $row = $tbody->firstChild; + if ($row->hasSingleTagInsideElement('td')) { + $cell = $row->firstChild; + $cell = NodeUtility::setNodeTag($cell, (array_reduce(iterator_to_array($cell->childNodes), function ($carry, $node) { + return $node->isPhrasingContent() && $carry; + }, true)) ? 'p' : 'div'); + $table->parentNode->replaceChild($cell, $table); + } + } + } + + 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. + * + * @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->setReadabilityDataTable(false); + continue; + } + $datatable = $table->getAttribute('datatable'); + if ($datatable == '0') { + $table->setReadabilityDataTable(false); + continue; + } + $summary = $table->getAttribute('summary'); + if ($summary) { + $table->setReadabilityDataTable(true); + continue; + } + + $caption = $table->getElementsByTagName('caption'); + if ($caption->length > 0 && $caption->item(0)->childNodes->length > 0) { + $table->setReadabilityDataTable(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) { + $table->setReadabilityDataTable(true); + continue 2; + } + } + + // Nested tables indicate a layout table: + if ($table->getElementsByTagName('table')->length > 0) { + $table->setReadabilityDataTable(false); + continue; + } + + $sizeInfo = $table->getRowAndColumnCount(); + if ($sizeInfo['rows'] >= 10 || $sizeInfo['columns'] > 4) { + $table->setReadabilityDataTable(true); + continue; + } + // Now just go by size entirely: + $table->setReadabilityDataTable($sizeInfo['rows'] * $sizeInfo['columns'] > 10); + } + } + + /** + * Remove the style attribute on every e and under. + * + * @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. + * + * @param $node DOMElement Node to clean + * @param $regex string Match id/class combination. + * + * @return void + **/ + public function _cleanMatchedNodes($node, $regex) + { + $endOfSearchMarkerNode = NodeUtility::getNextNode($node, true); + $next = NodeUtility::getNextNode($node); + while ($next && $next !== $endOfSearchMarkerNode) { + if (preg_match($regex, sprintf('%s %s', $next->getAttribute('class'), $next->getAttribute('id')))) { + $this->logger->debug(sprintf('Removing matched node with regex: \'%s\', node class was: \'%s\', id: \'%s\'', $regex, $next->getAttribute('class'), $next->getAttribute('id'))); + $next = NodeUtility::removeAndGetNext($next); + } else { + $next = NodeUtility::getNextNode($next); + } + } + } + + /** + * @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(NodeUtility::$regexps['onlyWhitespace'], '', $paragraph->textContent)) { + $this->logger->debug(sprintf('[PrepArticle] Removing extra paragraph. Text content was: \'%s\'', substr($paragraph->textContent, 0, 128))); + $paragraph->parentNode->removeChild($paragraph); + } + } + } + + /** + * @param DOMDocument $article + * @param string $tag Tag to clean conditionally + * + * @return void + */ + public function _cleanConditionally(DOMDocument $article, $tag) + { + if (!$this->configuration->getCleanConditionally()) { + 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++) { + /** @var $node DOMElement */ + $node = $DOMNodeList->item($length - 1 - $i); + + // First check if we're in a data table, in which case don't remove us. + if ($node->hasAncestorTag('table', -1, function ($node) { + return $node->isReadabilityDataTable(); + })) { + continue; + } + + $weight = 0; + if ($this->configuration->getWeightClasses()) { + $weight = $node->getClassWeight(); + } + + if ($weight < 0) { + $this->logger->debug(sprintf('[PrepArticle] Removing tag \'%s\' with 0 or less weight', $tag)); + + NodeUtility::removeNode($node); + 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. + */ + + $p = $node->getElementsByTagName('p')->length; + $img = $node->getElementsByTagName('img')->length; + $li = $node->getElementsByTagName('li')->length - 100; + $input = $node->getElementsByTagName('input')->length; + + $embedCount = 0; + $embeds = $node->getElementsByTagName('embed'); + + foreach ($embeds as $embedNode) { + if (preg_match(NodeUtility::$regexps['videos'], $embedNode->C14N())) { + $embedCount++; + } + } + + $linkDensity = $node->getLinkDensity(); + $contentLength = mb_strlen($node->getTextContent(true)); + + $haveToRemove = + ($img > 1 && $p / $img < 0.5 && !$node->hasAncestorTag('figure')) || + (!$isList && $li > $p) || + ($input > floor($p / 3)) || + (!$isList && $contentLength < 25 && ($img === 0 || $img > 2) && !$node->hasAncestorTag('figure')) || + (!$isList && $weight < 25 && $linkDensity > 0.2) || + ($weight >= 25 && $linkDensity > 0.5) || + (($embedCount === 1 && $contentLength < 75) || $embedCount > 1); + + if ($haveToRemove) { + $this->logger->debug(sprintf('[PrepArticle] Removing tag \'%s\'.', $tag)); + + NodeUtility::removeNode($node); + } + } + } + } + + /** + * Clean a node of all elements of type "tag". + * (Unless it's a youtube/vimeo video. People love movies.). + * + * @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 $value) { + $attributeValues[] = $value->nodeValue; + } + $attributeValues = implode('|', $attributeValues); + + // First, check the elements attributes to see if any of them contain youtube or vimeo + if (preg_match(NodeUtility::$regexps['videos'], $attributeValues)) { + continue; + } + + // Then check the elements inside this element for the same. + if (preg_match(NodeUtility::$regexps['videos'], $item->C14N())) { + continue; + } + } + $this->logger->debug(sprintf('[PrepArticle] Removing node \'%s\'.', $item->tagName)); + + NodeUtility::removeNode($item); + } + } + + /** + * Clean out spurious headers from an Element. Checks things like classnames and link density. + * + * @param DOMDocument $article + * + * @return void + **/ + public function _cleanHeaders(DOMDocument $article) + { + for ($headerIndex = 1; $headerIndex < 3; $headerIndex++) { + $headers = $article->getElementsByTagName('h' . $headerIndex); + /** @var $header DOMElement */ + foreach ($headers as $header) { + $weight = 0; + if ($this->configuration->getWeightClasses()) { + $weight = $header->getClassWeight(); + } + + if ($weight < 0) { + $this->logger->debug(sprintf('[PrepArticle] Removing H node with 0 or less weight. Content was: \'%s\'', substr($header->nodeValue, 0, 128))); + + NodeUtility::removeNode($header); + } + } + } + } + + /** + * Removes the class="" attribute from every element in the given + * subtree. + * + * Readability.js has a special filter to avoid cleaning the classes that the algorithm adds. We don't add classes + * here so no need to filter those. + * + * @param DOMDocument|DOMNode $node + * + * @return void + **/ + public function _cleanClasses($node) + { + if ($node->getAttribute('class') !== '') { + $node->removeAttribute('class'); + } + + for ($node = $node->firstChild; $node !== null; $node = $node->nextSibling) { + $this->_cleanClasses($node); + } + } + + /** + * @param DOMDocument $article + * + * @return DOMDocument + */ + public function postProcessContent(DOMDocument $article) + { + $this->logger->info('[PostProcess] PostProcessing content...'); + + // Readability cannot open relative uris so we convert them to absolute uris. + if ($this->configuration->getFixRelativeURLs()) { + 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) { + $this->logger->debug(sprintf('[PostProcess] Removing \'javascript:\' link. Content is: \'%s\'', substr($link->textContent, 0, 128))); + + $text = $article->createTextNode($link->textContent); + $link->parentNode->replaceChild($text, $link); + } else { + $this->logger->debug(sprintf('[PostProcess] Converting link to absolute URI: \'%s\'', substr($href, 0, 128))); + + $link->setAttribute('href', $this->toAbsoluteURI($href)); + } + } + } + + foreach ($article->getElementsByTagName('img') as $img) { + /** @var DOMElement $img */ + /* + * Extract all possible sources of img url and select the first one on the list. + */ + $url = [ + $img->getAttribute('src'), + $img->getAttribute('data-src'), + $img->getAttribute('data-original'), + $img->getAttribute('data-orig'), + $img->getAttribute('data-url') + ]; + + $src = array_filter($url); + $src = reset($src); + if ($src) { + $this->logger->debug(sprintf('[PostProcess] Converting image URL to absolute URI: \'%s\'', substr($src, 0, 128))); + + $img->setAttribute('src', $this->toAbsoluteURI($src)); + } + } + } + + $this->_cleanClasses($article); + + return $article; + } + + /** + * @return null|string + */ + public function __toString() + { + return sprintf('

%s

%s', $this->getTitle(), $this->getContent()); + } + + /** + * @return string|null + */ + public function getTitle() + { + return $this->title; + } + + /** + * @param string $title + */ + protected function setTitle($title) + { + $this->title = $title; + } + + /** + * @return string|null + */ + public function getContent() + { + return ($this->content instanceof DOMDocument) ? $this->content->C14N() : null; + } + + /** + * @return DOMDocument|null + */ + public function getDOMDocument() + { + return $this->content; + } + + /** + * @param DOMDocument $content + */ + protected function setContent(DOMDocument $content) + { + $this->content = $content; + } + + /** + * @return null|string + */ + public function getExcerpt() + { + return $this->excerpt; + } + + /** + * @param null|string $excerpt + */ + public function setExcerpt($excerpt) + { + $this->excerpt = $excerpt; + } + + /** + * @return string|null + */ + public function getImage() + { + return $this->image; + } + + /** + * @param string $image + */ + protected function setImage($image) + { + $this->image = $image; + } + + /** + * @return string|null + */ + public function getAuthor() + { + return $this->author; + } + + /** + * @param string $author + */ + protected function setAuthor($author) + { + $this->author = $author; + } + + /** + * @return null|string + */ + public function getDirection() + { + return $this->direction; + } + + /** + * @param null|string $direction + */ + public function setDirection($direction) + { + $this->direction = $direction; + } +} -- cgit v1.2.3