@ -2,7 +2,6 @@
namespace Readability; namespace Readability;
use DOMElement;
use Masterminds\HTML5; use Masterminds\HTML5;
use Psr\Log\LoggerAwareInterface; use Psr\Log\LoggerAwareInterface;
use Psr\Log\LoggerInterface; use Psr\Log\LoggerInterface;
@ -115,7 +114,7 @@ class Readability implements LoggerAwareInterface
// HACK: replace linebreaks plus br's with p's // HACK: replace linebreaks plus br's with p's
'!(< br [ ^ > ]*>[ \r\n\s]*){2,}!i' => '< / p > < p > ', '!(< br [ ^ > ]*>[ \r\n\s]*){2,}!i' => '< / p > < p > ',
// replace noscripts // replace noscripts
//'!< /?noscript>!is' => '', // '!< /?noscript>!is' => '',
// replace fonts to spans // replace fonts to spans
'!< (/?)font[^>]*>!is' => '< \\1span>', '!< (/?)font[^>]*>!is' => '< \\1span>',
]; ];
@ -126,8 +125,8 @@ class Readability implements LoggerAwareInterface
// replace empty tags that break layouts // replace empty tags that break layouts
'!< (?:a|div|p|figure)[^>]+/>!is' => '', '!< (?:a|div|p|figure)[^>]+/>!is' => '',
// remove all attributes on text tags // remove all attributes on text tags
//'!< (\s*/?\s*(?:blockquote|br|hr|code|div|article|span|footer|aside|p|pre|dl|li|ul|ol)) [^>]+>!is' => "< \\1>", // '!< (\s*/?\s*(?:blockquote|br|hr|code|div|article|span|footer|aside|p|pre|dl|li|ul|ol)) [^>]+>!is' => "< \\1>",
//single newlines cleanup // single newlines cleanup
"/\n+/" => "\n", "/\n+/" => "\n",
// modern web... // modern web...
'!< pre [ ^ > ]*>\s*< code ! is ' = > '< pre ' , '!< pre [ ^ > ]*>\s*< code ! is ' = > '< pre ' ,
@ -161,7 +160,7 @@ class Readability implements LoggerAwareInterface
/** /**
* Get article title element. * Get article title element.
* *
* @return DOMElement * @return \ DOMElement
*/ */
public function getTitle() public function getTitle()
{ {
@ -171,7 +170,7 @@ class Readability implements LoggerAwareInterface
/** /**
* Get article content element. * Get article content element.
* *
* @return DOMElement * @return \ DOMElement
*/ */
public function getContent() public function getContent()
{ {
@ -280,7 +279,7 @@ class Readability implements LoggerAwareInterface
/** /**
* Run any post-process modifications to article content as necessary. * Run any post-process modifications to article content as necessary.
*/ */
public function postProcessContent(DOMElement $articleContent): void public function postProcessContent(\ DOMElement $articleContent): void
{ {
if ($this->convertLinksToFootnotes & & !preg_match('/\bwiki/', $this->url)) { if ($this->convertLinksToFootnotes & & !preg_match('/\bwiki/', $this->url)) {
$this->addFootnotes($articleContent); $this->addFootnotes($articleContent);
@ -292,7 +291,7 @@ class Readability implements LoggerAwareInterface
* *
* @see http://www.roughtype.com/archives/2010/05/experiments_in.php * @see http://www.roughtype.com/archives/2010/05/experiments_in.php
*/ */
public function addFootnotes(DOMElement $articleContent): void public function addFootnotes(\ DOMElement $articleContent): void
{ {
$footnotesWrapper = $this->dom->createElement('footer'); $footnotesWrapper = $this->dom->createElement('footer');
$footnotesWrapper->setAttribute('class', 'readability-footnotes'); $footnotesWrapper->setAttribute('class', 'readability-footnotes');
@ -335,7 +334,7 @@ class Readability implements LoggerAwareInterface
$articleLink->setAttribute('style', 'color: inherit; text-decoration: none;'); $articleLink->setAttribute('style', 'color: inherit; text-decoration: none;');
$articleLink->setAttribute('name', 'readabilityLink-' . $linkCount); $articleLink->setAttribute('name', 'readabilityLink-' . $linkCount);
$footnote->setInnerHtml('< small > < sup > < a href = "#readabilityLink-' . $linkCount . '" title = "Jump to Link in Article" > ^< / a > < / sup > < / small > '); $footnote->setInnerHtml('< small > < sup > < a href = "#readabilityLink-' . $linkCount . '" title = "Jump to Link in Article" > ^< / a > < / sup > < / small > ');
$footnoteLink->setInnerHtml(( '' !== $footnoteLink->getAttribute('title') ? $footnoteLink->getAttribute('title') : $linkText) ); $footnoteLink->setInnerHtml('' !== $footnoteLink->getAttribute('title') ? $footnoteLink->getAttribute('title') : $linkText);
$footnoteLink->setAttribute('name', 'readabilityFootnoteLink-' . $linkCount); $footnoteLink->setAttribute('name', 'readabilityFootnoteLink-' . $linkCount);
$footnote->appendChild($footnoteLink); $footnote->appendChild($footnoteLink);
@ -356,7 +355,7 @@ class Readability implements LoggerAwareInterface
*/ */
public function prepArticle(\DOMNode $articleContent): void public function prepArticle(\DOMNode $articleContent): void
{ {
if (!$articleContent instanceof DOMElement) { if (!$articleContent instanceof \ DOMElement) {
return; return;
} }
@ -456,9 +455,9 @@ class Readability implements LoggerAwareInterface
* Get the inner text of a node. * Get the inner text of a node.
* This also strips out any excess whitespace to be found. * This also strips out any excess whitespace to be found.
* *
* @param DOMElement $e * @param \ DOMElement $e
* @param bool $normalizeSpaces (default: true) * @param bool $normalizeSpaces (default: true)
* @param bool $flattenLines (default: false) * @param bool $flattenLines (default: false)
*/ */
public function getInnerText($e, bool $normalizeSpaces = true, bool $flattenLines = false): string public function getInnerText($e, bool $normalizeSpaces = true, bool $flattenLines = false): string
{ {
@ -482,7 +481,7 @@ class Readability implements LoggerAwareInterface
/** /**
* Remove the style attribute on every $e and under. * Remove the style attribute on every $e and under.
*/ */
public function cleanStyles(DOMElement $e): void public function cleanStyles(\ DOMElement $e): void
{ {
if (\is_object($e)) { if (\is_object($e)) {
$elems = $e->getElementsByTagName('*'); $elems = $e->getElementsByTagName('*');
@ -515,7 +514,7 @@ class Readability implements LoggerAwareInterface
* This is the amount of text that is inside a link divided by the total text in the node. * This is the amount of text that is inside a link divided by the total text in the node.
* Can exclude external references to differentiate between simple text and menus/infoblocks. * Can exclude external references to differentiate between simple text and menus/infoblocks.
*/ */
public function getLinkDensity(DOMElement $e, bool $excludeExternal = false): float public function getLinkDensity(\ DOMElement $e, bool $excludeExternal = false): float
{ {
$links = $e->getElementsByTagName('a'); $links = $e->getElementsByTagName('a');
$textLength = mb_strlen($this->getInnerText($e, true, true)); $textLength = mb_strlen($this->getInnerText($e, true, true));
@ -538,7 +537,7 @@ class Readability implements LoggerAwareInterface
/** /**
* Get an element relative weight. * Get an element relative weight.
*/ */
public function getWeight(DOMElement $e): int public function getWeight(\ DOMElement $e): int
{ {
if (!$this->flagIsActive(self::FLAG_WEIGHT_ATTRIBUTES)) { if (!$this->flagIsActive(self::FLAG_WEIGHT_ATTRIBUTES)) {
return 0; return 0;
@ -556,7 +555,7 @@ class Readability implements LoggerAwareInterface
/** /**
* Remove extraneous break tags from a node. * Remove extraneous break tags from a node.
*/ */
public function killBreaks(DOMElement $node): void public function killBreaks(\ DOMElement $node): void
{ {
$html = $node->getInnerHTML(); $html = $node->getInnerHTML();
$html = preg_replace($this->regexps['killBreaks'], '< br / > ', $html); $html = preg_replace($this->regexps['killBreaks'], '< br / > ', $html);
@ -569,7 +568,7 @@ class Readability implements LoggerAwareInterface
* *
* Updated 2012-09-18 to preserve youtube/vimeo iframes * Updated 2012-09-18 to preserve youtube/vimeo iframes
*/ */
public function clean(DOMElement $e, string $tag): void public function clean(\ DOMElement $e, string $tag): void
{ {
$targetList = $e->getElementsByTagName($tag); $targetList = $e->getElementsByTagName($tag);
$isEmbed = ('audio' === $tag || 'video' === $tag || 'iframe' === $tag || 'object' === $tag || 'embed' === $tag); $isEmbed = ('audio' === $tag || 'video' === $tag || 'iframe' === $tag || 'object' === $tag || 'embed' === $tag);
@ -601,7 +600,7 @@ class Readability implements LoggerAwareInterface
* "Fishy" is an algorithm based on content length, classnames, * "Fishy" is an algorithm based on content length, classnames,
* link density, number of images & embeds, etc. * link density, number of images & embeds, etc.
*/ */
public function cleanConditionally(DOMElement $e, string $tag): void public function cleanConditionally(\ DOMElement $e, string $tag): void
{ {
if (!$this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) { if (!$this->flagIsActive(self::FLAG_CLEAN_CONDITIONALLY)) {
return; return;
@ -714,7 +713,7 @@ class Readability implements LoggerAwareInterface
/** /**
* Clean out spurious headers from an Element. Checks things like classnames and link density. * Clean out spurious headers from an Element. Checks things like classnames and link density.
*/ */
public function cleanHeaders(DOMElement $e): void public function cleanHeaders(\ DOMElement $e): void
{ {
for ($headerIndex = 1; $headerIndex < 3 ; + + $ headerIndex ) { for ($headerIndex = 1; $headerIndex < 3 ; + + $ headerIndex ) {
$headers = $e->getElementsByTagName('h' . $headerIndex); $headers = $e->getElementsByTagName('h' . $headerIndex);
@ -754,7 +753,7 @@ class Readability implements LoggerAwareInterface
/** /**
* Get the article title as an H1. * Get the article title as an H1.
* *
* @return DOMElement * @return \ DOMElement
*/ */
protected function getArticleTitle() protected function getArticleTitle()
{ {
@ -826,7 +825,7 @@ class Readability implements LoggerAwareInterface
* Initialize a node with the readability object. Also checks the * Initialize a node with the readability object. Also checks the
* className/id for special names to add to its score. * className/id for special names to add to its score.
*/ */
protected function initializeNode(DOMElement $node): void protected function initializeNode(\ DOMElement $node): void
{ {
if (!isset($node->tagName)) { if (!isset($node->tagName)) {
return; return;
@ -894,11 +893,11 @@ class Readability implements LoggerAwareInterface
* Using a variety of metrics (content score, classname, element types), find the content that is * Using a variety of metrics (content score, classname, element types), find the content that is
* most likely to be the stuff a user wants to read. Then return it wrapped up in a div. * most likely to be the stuff a user wants to read. Then return it wrapped up in a div.
* *
* @param DOMElement $page * @param \ DOMElement $page
* *
* @return DOMElement|false * @return \ DOMElement|false
*/ */
protected function grabArticle(DOMElement $page = null) protected function grabArticle(\ DOMElement $page = null)
{ {
if (!$page) { if (!$page) {
$page = $this->dom; $page = $this->dom;
@ -1040,7 +1039,7 @@ class Readability implements LoggerAwareInterface
// For every SCORE_CHARS_IN_PARAGRAPH (default:100) characters in this paragraph, add another point. Up to 3 points. // For every SCORE_CHARS_IN_PARAGRAPH (default:100) characters in this paragraph, add another point. Up to 3 points.
$contentScore += min(floor(mb_strlen($innerText) / self::SCORE_CHARS_IN_PARAGRAPH), 3); $contentScore += min(floor(mb_strlen($innerText) / self::SCORE_CHARS_IN_PARAGRAPH), 3);
// For every SCORE_WORDS_IN_PARAGRAPH (default:20) words in this paragraph, add another point. Up to 3 points. // For every SCORE_WORDS_IN_PARAGRAPH (default:20) words in this paragraph, add another point. Up to 3 points.
//$contentScore += min(floor($this->getWordCount($innerText) / self::SCORE_WORDS_IN_PARAGRAPH), 3); // $contentScore += min(floor($this->getWordCount($innerText) / self::SCORE_WORDS_IN_PARAGRAPH), 3);
foreach ($ancestors as $level => $ancestor) { foreach ($ancestors as $level => $ancestor) {
if (!$ancestor->nodeName || !$ancestor->parentNode) { if (!$ancestor->nodeName || !$ancestor->parentNode) {
@ -1211,7 +1210,7 @@ class Readability implements LoggerAwareInterface
if (0 === strcasecmp($tagName, 'td') || 0 === strcasecmp($tagName, 'tr')) { if (0 === strcasecmp($tagName, 'td') || 0 === strcasecmp($tagName, 'tr')) {
$up = $topCandidate; $up = $topCandidate;
if ($up->parentNode instanceof DOMElement) { if ($up->parentNode instanceof \ DOMElement) {
$up = $up->parentNode; $up = $up->parentNode;
if (0 === strcasecmp($up->tagName, 'table')) { if (0 === strcasecmp($up->tagName, 'table')) {
@ -1292,8 +1291,8 @@ class Readability implements LoggerAwareInterface
// To ensure a node does not interfere with readability styles, remove its classnames & ids. // To ensure a node does not interfere with readability styles, remove its classnames & ids.
// Now done via RegExp post_filter. // Now done via RegExp post_filter.
//$nodeToAppend->removeAttribute('class'); // $nodeToAppend->removeAttribute('class');
//$nodeToAppend->removeAttribute('id'); // $nodeToAppend->removeAttribute('id');
// Append sibling and subtract from our list as appending removes a node. // Append sibling and subtract from our list as appending removes a node.
$articleContent->appendChild($nodeToAppend); $articleContent->appendChild($nodeToAppend);
} }
@ -1340,7 +1339,7 @@ class Readability implements LoggerAwareInterface
* Get an element weight by attribute. * Get an element weight by attribute.
* Uses regular expressions to tell if this element looks good or bad. * Uses regular expressions to tell if this element looks good or bad.
*/ */
protected function weightAttribute(DOMElement $element, string $attribute): int protected function weightAttribute(\ DOMElement $element, string $attribute): int
{ {
if (!$element->hasAttribute($attribute)) { if (!$element->hasAttribute($attribute)) {
return 0; return 0;
@ -1427,7 +1426,7 @@ class Readability implements LoggerAwareInterface
unset($tidy); unset($tidy);
} }
$this->html = mb_convert_encoding((string) $this->html, 'HTML-ENTITIES', 'UTF-8') ; $this->html = '< meta charset = "utf-8" > ' . (string) $this->html ;
if ('html5lib' === $this->parser || 'html5' === $this->parser) { if ('html5lib' === $this->parser || 'html5' === $this->parser) {
$this->dom = (new HTML5())->loadHTML($this->html); $this->dom = (new HTML5())->loadHTML($this->html);
@ -1443,14 +1442,14 @@ class Readability implements LoggerAwareInterface
libxml_use_internal_errors(false); libxml_use_internal_errors(false);
} }
$this->dom->registerNodeClass(DOMElement::class, \Readability\JSLikeHTMLElement::class); $this->dom->registerNodeClass(\ DOMElement::class, \Readability\JSLikeHTMLElement::class);
} }
private function getAncestors(DOMElement $node, int $maxDepth = 0): array private function getAncestors(\ DOMElement $node, int $maxDepth = 0): array
{ {
$ancestors = []; $ancestors = [];
$i = 0; $i = 0;
while ($node->parentNode instanceof DOMElement) { while ($node->parentNode instanceof \ DOMElement) {
$ancestors[] = $node->parentNode; $ancestors[] = $node->parentNode;
if (++$i === $maxDepth) { if (++$i === $maxDepth) {
break; break;
@ -1470,7 +1469,7 @@ class Readability implements LoggerAwareInterface
}, iterator_to_array($node->childNodes)), true)); }, iterator_to_array($node->childNodes)), true));
} }
private function hasSingleTagInsideElement(DOMElement $node, string $tag): bool private function hasSingleTagInsideElement(\ DOMElement $node, string $tag): bool
{ {
if (1 !== $node->childNodes->length || $node->childNodes->item(0)->nodeName !== $tag) { if (1 !== $node->childNodes->length || $node->childNodes->item(0)->nodeName !== $tag) {
return false; return false;
@ -1490,11 +1489,11 @@ class Readability implements LoggerAwareInterface
* Tidy must be configured to not clean the input for this function to * Tidy must be configured to not clean the input for this function to
* work as expected, see $this->tidy_config['clean'] * work as expected, see $this->tidy_config['clean']
*/ */
private function isNodeVisible(DOMElement $node): bool private function isNodeVisible(\ DOMElement $node): bool
{ {
return !($node->hasAttribute('style') return !($node->hasAttribute('style')
& & preg_match($this->regexps['isNotVisible'], $node->getAttribute('style')) & & preg_match($this->regexps['isNotVisible'], $node->getAttribute('style'))
) )
& & !$node->hasAttribute('hidden'); & & !$node->hasAttribute('hidden');
} }
} }
Jérémy Benoist
3 years ago
committed by
GitHub