diff options
Diffstat (limited to 'src')
-rw-r--r-- | src/Configuration.php | 26 | ||||
-rw-r--r-- | src/NodeClass/DOMDocument.php | 2 | ||||
-rw-r--r-- | src/NodeClass/DOMNode.php | 3 | ||||
-rw-r--r-- | src/NodeClass/NodeClassTrait.php | 9 | ||||
-rw-r--r-- | src/NodeUtility.php | 15 | ||||
-rw-r--r-- | src/Readability.php | 36 |
6 files changed, 46 insertions, 45 deletions
diff --git a/src/Configuration.php b/src/Configuration.php index 6ce8b5b..1a405de 100644 --- a/src/Configuration.php +++ b/src/Configuration.php @@ -3,7 +3,7 @@ namespace andreskrey\Readability; /** - * Class Configuration + * Class Configuration. */ class Configuration { @@ -62,11 +62,13 @@ class Configuration /** * @param int $maxTopCandidates + * * @return $this */ public function setMaxTopCandidates($maxTopCandidates) { $this->maxTopCandidates = $maxTopCandidates; + return $this; } @@ -80,11 +82,13 @@ class Configuration /** * @param int $wordThreshold + * * @return $this */ public function setWordThreshold($wordThreshold) { $this->wordThreshold = $wordThreshold; + return $this; } @@ -98,11 +102,13 @@ class Configuration /** * @param bool $articleByLine + * * @return $this */ public function setArticleByLine($articleByLine) { $this->articleByLine = $articleByLine; + return $this; } @@ -116,11 +122,13 @@ class Configuration /** * @param bool $stripUnlikelyCandidates + * * @return $this */ public function setStripUnlikelyCandidates($stripUnlikelyCandidates) { $this->stripUnlikelyCandidates = $stripUnlikelyCandidates; + return $this; } @@ -134,11 +142,13 @@ class Configuration /** * @param bool $cleanConditionally + * * @return $this */ public function setCleanConditionally($cleanConditionally) { $this->cleanConditionally = $cleanConditionally; + return $this; } @@ -152,11 +162,13 @@ class Configuration /** * @param bool $weightClasses + * * @return $this */ public function setWeightClasses($weightClasses) { $this->weightClasses = $weightClasses; + return $this; } @@ -170,11 +182,13 @@ class Configuration /** * @param bool $removeReadabilityTags + * * @return $this */ public function setRemoveReadabilityTags($removeReadabilityTags) { $this->removeReadabilityTags = $removeReadabilityTags; + return $this; } @@ -188,11 +202,13 @@ class Configuration /** * @param bool $fixRelativeURLs + * * @return $this */ public function setFixRelativeURLs($fixRelativeURLs) { $this->fixRelativeURLs = $fixRelativeURLs; + return $this; } @@ -206,11 +222,13 @@ class Configuration /** * @param bool $substituteEntities + * * @return $this */ public function setSubstituteEntities($substituteEntities) { $this->substituteEntities = $substituteEntities; + return $this; } @@ -224,11 +242,13 @@ class Configuration /** * @param bool $normalizeEntities + * * @return $this */ public function setNormalizeEntities($normalizeEntities) { $this->normalizeEntities = $normalizeEntities; + return $this; } @@ -242,11 +262,13 @@ class Configuration /** * @param string $originalURL + * * @return $this */ public function setOriginalURL($originalURL) { $this->originalURL = $originalURL; + return $this; } @@ -260,11 +282,13 @@ class Configuration /** * @param bool $summonCthulhu + * * @return $this */ public function setSummonCthulhu($summonCthulhu) { $this->summonCthulhu = $summonCthulhu; + return $this; } diff --git a/src/NodeClass/DOMDocument.php b/src/NodeClass/DOMDocument.php index f379268..98b1215 100644 --- a/src/NodeClass/DOMDocument.php +++ b/src/NodeClass/DOMDocument.php @@ -14,7 +14,7 @@ class DOMDocument extends \DOMDocument $this->registerNodeClass('DOMCdataSection', DOMCdataSection::class); $this->registerNodeClass('DOMCharacterData', DOMCharacterData::class); $this->registerNodeClass('DOMComment', DOMComment::class); - $this->registerNodeClass('DOMDocument', DOMDocument::class); + $this->registerNodeClass('DOMDocument', self::class); $this->registerNodeClass('DOMDocumentFragment', DOMDocumentFragment::class); $this->registerNodeClass('DOMDocumentType', DOMDocumentType::class); $this->registerNodeClass('DOMElement', DOMElement::class); diff --git a/src/NodeClass/DOMNode.php b/src/NodeClass/DOMNode.php index c9ed1c3..eb5e93e 100644 --- a/src/NodeClass/DOMNode.php +++ b/src/NodeClass/DOMNode.php @@ -3,10 +3,9 @@ namespace andreskrey\Readability\NodeClass; /** - * Class DOMNode + * Class DOMNode. * * @method getAttribute($attribute) - * @package andreskrey\Readability\NodeClass */ class DOMNode extends \DOMNode { diff --git a/src/NodeClass/NodeClassTrait.php b/src/NodeClass/NodeClassTrait.php index a1382d1..f2df51a 100644 --- a/src/NodeClass/NodeClassTrait.php +++ b/src/NodeClass/NodeClassTrait.php @@ -7,14 +7,14 @@ use andreskrey\Readability\NodeUtility; trait NodeClassTrait { /** - * Content score of the node. Used to determine the value of the content + * Content score of the node. Used to determine the value of the content. * * @var int */ public $contentScore = 0; /** - * Flag for initialized status + * Flag for initialized status. * * @var bool */ @@ -38,7 +38,7 @@ trait NodeClassTrait ]; /** - * initialized getter + * initialized getter. * * @return bool */ @@ -53,6 +53,7 @@ trait NodeClassTrait * @ 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) @@ -181,7 +182,6 @@ trait NodeClassTrait return $linkLength / $textLength; } - /** * Calculates the weight of the class/id of the current element. * @@ -283,7 +283,6 @@ trait NodeClassTrait return ['rows' => $rows, 'columns' => $columns]; } - /** * Creates a new node based on the text content of the original node. * diff --git a/src/NodeUtility.php b/src/NodeUtility.php index d0796dd..7fbdd45 100644 --- a/src/NodeUtility.php +++ b/src/NodeUtility.php @@ -7,14 +7,12 @@ use andreskrey\Readability\NodeClass\DOMElement; use andreskrey\Readability\NodeClass\DOMNode; /** - * Class NodeUtility - * @package andreskrey\Readability + * Class NodeUtility. */ class NodeUtility { - /** - * Collection of regexps to check the node usability + * Collection of regexps to check the node usability. * * @var array */ @@ -36,12 +34,11 @@ class NodeUtility 'onlyWhitespace' => '/\x{00A0}|\s+/u' ]; - /** - * - * Imported from the Element class on league\html-to-markdown + * Imported from the Element class on league\html-to-markdown. * * @param $node + * * @return DOMElement */ public static function nextElement($node) @@ -56,13 +53,13 @@ class NodeUtility 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 string $value * @param bool $importAttributes + * * @return DOMNode */ public static function setNodeTag($node, $value, $importAttributes = false) @@ -72,7 +69,6 @@ class NodeUtility $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); @@ -122,7 +118,6 @@ class NodeUtility } } - /** * Returns the next node. First checks for children (if the flag allows it), then for siblings, and finally * for parents. diff --git a/src/Readability.php b/src/Readability.php index c42e577..282e983 100644 --- a/src/Readability.php +++ b/src/Readability.php @@ -3,21 +3,12 @@ namespace andreskrey\Readability; use andreskrey\Readability\NodeClass\DOMDocument; -use andreskrey\Readability\NodeClass\DOMAttr; -use andreskrey\Readability\NodeClass\DOMCdataSection; -use andreskrey\Readability\NodeClass\DOMCharacterData; -use andreskrey\Readability\NodeClass\DOMComment; -use andreskrey\Readability\NodeClass\DOMDocumentFragment; -use andreskrey\Readability\NodeClass\DOMDocumentType; use andreskrey\Readability\NodeClass\DOMElement; use andreskrey\Readability\NodeClass\DOMNode; -use andreskrey\Readability\NodeClass\DOMNotation; -use andreskrey\Readability\NodeClass\DOMProcessingInstruction; use andreskrey\Readability\NodeClass\DOMText; -use andreskrey\Readability\NodeClass\NodeClassTrait; /** - * Class Readability + * Class Readability. */ class Readability { @@ -29,49 +20,49 @@ class Readability protected $dom; /** - * Title of the article + * Title of the article. * * @var string|null */ protected $title = null; /** - * HTML content article + * HTML content article. * * @var string|null */ protected $content = null; /** - * Excerpt of the article + * Excerpt of the article. * * @var string|null */ protected $excerpt = null; /** - * Main image of the article + * Main image of the article. * * @var string|null */ protected $image = null; /** - * Author of the article. Extracted from the byline tags and other social media properties + * Author of the article. Extracted from the byline tags and other social media properties. * * @var string|null */ protected $author = null; /** - * Direction of the text + * Direction of the text. * * @var string|null */ protected $direction = null; /** - * Configuration object + * Configuration object. * * @var Configuration */ @@ -116,7 +107,7 @@ class Readability } /** - * Main parse function + * Main parse function. * * @param $html * @@ -229,7 +220,7 @@ class Readability } /** - * Tries to guess relevant info from metadata of the html. Sets the results in the Readability properties + * Tries to guess relevant info from metadata of the html. Sets the results in the Readability properties. */ private function getMetadata() { @@ -326,7 +317,6 @@ class Readability return $result; } - /** * Tries to get the main article image. Will only update the metadata if the getMetadata function couldn't * find a correct image. @@ -447,7 +437,6 @@ class Readability return $curTitle; } - private function toAbsoluteURI($uri) { list($pathBase, $scheme, $prePath) = $this->getPathInfo($this->configuration->getOriginalURL()); @@ -495,7 +484,6 @@ class Readability return [$pathBase, $scheme, $prePath]; } - /** * Gets nodes from the root element. * @@ -641,7 +629,6 @@ class Readability return false; } - /** * Removes all the scripts of the html. * @@ -735,7 +722,6 @@ class Readability } } - /** * Assign scores to each node. This function will rate each node and return a DOMElement object for each one. * @@ -1147,7 +1133,6 @@ class Readability } } - /** * Remove the style attribute on every e and under. * TODO: To be moved to Readability. @@ -1417,7 +1402,6 @@ class Readability return $article; } - /** * @return null|string */ |