infong/php-readability
1
0
Fork 0
mirror of https://github.com/j0k3r/php-readability.git synced 2026-09-26 22:16:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #73 from Kdecherf/fix/impr

Browse Source 
Fix `isPhrasingContent` conditions, text node replacement
This commit is contained in:
Jérémy Benoist
2022-02-16 00:03:37 +01:00
committed by GitHub
parent 8af69ad68c 2ab87d7445
commit 0c0653dad6
1 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/Readability.php
+5 -5
View File
@@ -26,7 +26,7 @@ class Readability implements LoggerAwareInterface
public const MIN_NODE_LENGTH = 80; public const MIN_NODE_LENGTH = 80;
public const MAX_LINK_DENSITY = 0.25; public const MAX_LINK_DENSITY = 0.25;
public $convertLinksToFootnotes = false; public $convertLinksToFootnotes = false;
public $revertForcedParagraphElements = true; public $revertForcedParagraphElements = false;
public $articleTitle; public $articleTitle;
public $articleContent; public $articleContent;
public $original_html; public $original_html;
@@ -991,9 +991,9 @@ class Readability implements LoggerAwareInterface
$p->appendChild($childNode); $p->appendChild($childNode);
} elseif ('' !== $this->getInnerText($childNode, true, true)) { } elseif ('' !== $this->getInnerText($childNode, true, true)) {
$p = $this->dom->createElement('p'); $p = $this->dom->createElement('p');
$p->setInnerHtml($childNode->nodeValue);
$p->setAttribute('data-readability-styled', 'true'); $p->setAttribute('data-readability-styled', 'true');
$childNode->parentNode->replaceChild($p, $childNode); $node->replaceChild($p, $childNode);
$p->appendChild($childNode);
} }
} elseif (null !== $p) { } elseif (null !== $p) {
while ($p->lastChild && '' === $this->getInnerText($p->lastChild, true, true)) { while ($p->lastChild && '' === $this->getInnerText($p->lastChild, true, true)) {
@@ -1465,8 +1465,8 @@ class Readability implements LoggerAwareInterface
private function isPhrasingContent($node): bool private function isPhrasingContent($node): bool
{ {
return \XML_TEXT_NODE === $node->nodeType return \XML_TEXT_NODE === $node->nodeType
|| \in_array($node->nodeName, $this->phrasingElements, true) || \in_array(strtoupper($node->nodeName), $this->phrasingElements, true)
|| (\in_array($node->nodeName, ['a', 'del', 'ins'], true) && !\in_array(false, array_map(function ($c) { || (\in_array(strtoupper($node->nodeName), ['A', 'DEL', 'INS'], true) && !\in_array(false, array_map(function ($c) {
return $this->isPhrasingContent($c); return $this->isPhrasingContent($c);
}, iterator_to_array($node->childNodes)), true)); }, iterator_to_array($node->childNodes)), true));
} }