From c54b69b6894832e050fe28a3cb7e84a6fbb0db43 Mon Sep 17 00:00:00 2001
From: "FiveFilters.org" <fivefilters@fivefilters.org>
Date: Tue, 3 Dec 2024 16:03:49 +0100
Subject: [PATCH 1/5] Initial version using PHP 8.4's HTML5 parser

---
 composer.json                                 |   1 -
 src/Configuration.php                         |  25 +--
 src/Nodes/DOM/{DOMAttr.php => Attr.php}       |   2 +-
 .../DOM/{DOMNotation.php => CdataSection.php} |   2 +-
 ...{DOMCdataSection.php => CharacterData.php} |   2 +-
 src/Nodes/DOM/{DOMEntity.php => Comment.php}  |   2 +-
 src/Nodes/DOM/DOMDocument.php                 |  30 ---
 src/Nodes/DOM/DOMDocumentFragment.php         |  10 -
 src/Nodes/DOM/DOMDocumentType.php             |  10 -
 src/Nodes/DOM/DOMEntityReference.php          |  10 -
 src/Nodes/DOM/DOMProcessingInstruction.php    |  10 -
 src/Nodes/DOM/DocumentFragment.php            |  10 +
 src/Nodes/DOM/DocumentType.php                |  10 +
 src/Nodes/DOM/{DOMElement.php => Element.php} |   8 +-
 src/Nodes/DOM/{DOMComment.php => Entity.php}  |   2 +-
 ...MCharacterData.php => EntityReference.php} |   2 +-
 src/Nodes/DOM/{DOMNode.php => Node.php}       |   2 +-
 .../DOM/{DOMNodeList.php => NodeList.php}     |   8 +-
 src/Nodes/DOM/Notation.php                    |  10 +
 src/Nodes/DOM/ProcessingInstruction.php       |  10 +
 src/Nodes/DOM/{DOMText.php => Text.php}       |   2 +-
 src/Nodes/NodeTrait.php                       |  56 ++---
 src/Nodes/NodeUtility.php                     |  50 +++--
 src/Readability.php                           | 210 ++++++++----------
 test/ConfigurationTest.php                    |   2 -
 test/ReadabilityTest.php                      |   3 -
 test/test-pages/marketwatch/config.json       |   3 +-
 27 files changed, 221 insertions(+), 271 deletions(-)
 rename src/Nodes/DOM/{DOMAttr.php => Attr.php} (79%)
 rename src/Nodes/DOM/{DOMNotation.php => CdataSection.php} (73%)
 rename src/Nodes/DOM/{DOMCdataSection.php => CharacterData.php} (72%)
 rename src/Nodes/DOM/{DOMEntity.php => Comment.php} (77%)
 delete mode 100644 src/Nodes/DOM/DOMDocument.php
 delete mode 100644 src/Nodes/DOM/DOMDocumentFragment.php
 delete mode 100644 src/Nodes/DOM/DOMDocumentType.php
 delete mode 100644 src/Nodes/DOM/DOMEntityReference.php
 delete mode 100644 src/Nodes/DOM/DOMProcessingInstruction.php
 create mode 100644 src/Nodes/DOM/DocumentFragment.php
 create mode 100644 src/Nodes/DOM/DocumentType.php
 rename src/Nodes/DOM/{DOMElement.php => Element.php} (82%)
 rename src/Nodes/DOM/{DOMComment.php => Entity.php} (76%)
 rename src/Nodes/DOM/{DOMCharacterData.php => EntityReference.php} (70%)
 rename src/Nodes/DOM/{DOMNode.php => Node.php} (86%)
 rename src/Nodes/DOM/{DOMNodeList.php => NodeList.php} (86%)
 create mode 100644 src/Nodes/DOM/Notation.php
 create mode 100644 src/Nodes/DOM/ProcessingInstruction.php
 rename src/Nodes/DOM/{DOMText.php => Text.php} (79%)

diff --git a/composer.json b/composer.json
index c31fc29..f573695 100644
--- a/composer.json
+++ b/composer.json
@@ -32,7 +32,6 @@
         "ext-xml": "*",
         "ext-mbstring": "*",
         "psr/log": "^1.0 || ^2.0 || ^3.0",
-        "masterminds/html5": "^2.0",
         "league/uri": "^7.0"
     },
     "require-dev": {
diff --git a/src/Configuration.php b/src/Configuration.php
index c4fe88f..52222d3 100644
--- a/src/Configuration.php
+++ b/src/Configuration.php
@@ -20,7 +20,6 @@ class Configuration
     protected bool $cleanConditionally = true;
     protected bool $weightClasses = true;
     protected bool $fixRelativeURLs = false;
-    protected bool $substituteEntities = false;
     protected bool $normalizeEntities = false;
     protected bool $summonCthulhu = false;
     protected string $originalURL = 'http://fakehost';
@@ -206,24 +205,6 @@ public function setFixRelativeURLs(bool $fixRelativeURLs): Configuration
         return $this;
     }
 
-    /**
-     * Get substitute entities.
-     */
-    public function getSubstituteEntities(): bool
-    {
-        return $this->substituteEntities;
-    }
-
-    /**
-     * Set substitute entities.
-     */
-    public function setSubstituteEntities(bool $substituteEntities): Configuration
-    {
-        $this->substituteEntities = $substituteEntities;
-
-        return $this;
-    }
-
     /**
      * Get normalize entities.
      */
@@ -273,7 +254,11 @@ public function getParser(): string
      */
     public function setParser(string $parser): Configuration
     {
-        $this->parser = $parser;
+        if ($parser !== 'html5') {
+            throw new \InvalidArgumentException('This version of Readability.php only supports the HTML5 parser introduced in PHP 8.4');
+        } else {
+            $this->parser = $parser;
+        }
 
         return $this;
     }
diff --git a/src/Nodes/DOM/DOMAttr.php b/src/Nodes/DOM/Attr.php
similarity index 79%
rename from src/Nodes/DOM/DOMAttr.php
rename to src/Nodes/DOM/Attr.php
index 1bdf395..ce03daf 100644
--- a/src/Nodes/DOM/DOMAttr.php
+++ b/src/Nodes/DOM/Attr.php
@@ -4,7 +4,7 @@
 
 use fivefilters\Readability\Nodes\NodeTrait;
 
-class DOMAttr extends \DOMAttr
+class Attr extends \DOM\Attr
 {
     use NodeTrait;
 }
diff --git a/src/Nodes/DOM/DOMNotation.php b/src/Nodes/DOM/CdataSection.php
similarity index 73%
rename from src/Nodes/DOM/DOMNotation.php
rename to src/Nodes/DOM/CdataSection.php
index d276e42..39bc0ce 100644
--- a/src/Nodes/DOM/DOMNotation.php
+++ b/src/Nodes/DOM/CdataSection.php
@@ -4,7 +4,7 @@
 
 use fivefilters\Readability\Nodes\NodeTrait;
 
-class DOMNotation extends \DOMNotation
+class CdataSection extends \DOM\CdataSection
 {
     use NodeTrait;
 }
diff --git a/src/Nodes/DOM/DOMCdataSection.php b/src/Nodes/DOM/CharacterData.php
similarity index 72%
rename from src/Nodes/DOM/DOMCdataSection.php
rename to src/Nodes/DOM/CharacterData.php
index 6ac3dcd..ac91426 100644
--- a/src/Nodes/DOM/DOMCdataSection.php
+++ b/src/Nodes/DOM/CharacterData.php
@@ -4,7 +4,7 @@
 
 use fivefilters\Readability\Nodes\NodeTrait;
 
-class DOMCdataSection extends \DOMCdataSection
+class CharacterData extends \DOM\CharacterData
 {
     use NodeTrait;
 }
diff --git a/src/Nodes/DOM/DOMEntity.php b/src/Nodes/DOM/Comment.php
similarity index 77%
rename from src/Nodes/DOM/DOMEntity.php
rename to src/Nodes/DOM/Comment.php
index 751b59c..7415696 100644
--- a/src/Nodes/DOM/DOMEntity.php
+++ b/src/Nodes/DOM/Comment.php
@@ -4,7 +4,7 @@
 
 use fivefilters\Readability\Nodes\NodeTrait;
 
-class DOMEntity extends \DOMEntity
+class Comment extends \DOM\Comment
 {
     use NodeTrait;
 }
diff --git a/src/Nodes/DOM/DOMDocument.php b/src/Nodes/DOM/DOMDocument.php
deleted file mode 100644
index d912338..0000000
--- a/src/Nodes/DOM/DOMDocument.php
+++ /dev/null
@@ -1,30 +0,0 @@
-<?php
-
-namespace fivefilters\Readability\Nodes\DOM;
-
-use fivefilters\Readability\Nodes\NodeTrait;
-
-class DOMDocument extends \DOMDocument
-{
-    use NodeTrait;
-
-    public function __construct($version, $encoding)
-    {
-        parent::__construct($version, $encoding);
-
-        $this->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/src/Nodes/DOM/DOMDocumentFragment.php b/src/Nodes/DOM/DOMDocumentFragment.php
deleted file mode 100644
index 33a3f95..0000000
--- a/src/Nodes/DOM/DOMDocumentFragment.php
+++ /dev/null
@@ -1,10 +0,0 @@
-<?php
-
-namespace fivefilters\Readability\Nodes\DOM;
-
-use fivefilters\Readability\Nodes\NodeTrait;
-
-class DOMDocumentFragment extends \DOMDocumentFragment
-{
-    use NodeTrait;
-}
diff --git a/src/Nodes/DOM/DOMDocumentType.php b/src/Nodes/DOM/DOMDocumentType.php
deleted file mode 100644
index cf3cd0f..0000000
--- a/src/Nodes/DOM/DOMDocumentType.php
+++ /dev/null
@@ -1,10 +0,0 @@
-<?php
-
-namespace fivefilters\Readability\Nodes\DOM;
-
-use fivefilters\Readability\Nodes\NodeTrait;
-
-class DOMDocumentType extends \DOMDocumentType
-{
-    use NodeTrait;
-}
diff --git a/src/Nodes/DOM/DOMEntityReference.php b/src/Nodes/DOM/DOMEntityReference.php
deleted file mode 100644
index d6fd6d4..0000000
--- a/src/Nodes/DOM/DOMEntityReference.php
+++ /dev/null
@@ -1,10 +0,0 @@
-<?php
-
-namespace fivefilters\Readability\Nodes\DOM;
-
-use fivefilters\Readability\Nodes\NodeTrait;
-
-class DOMEntityReference extends \DOMEntityReference
-{
-    use NodeTrait;
-}
diff --git a/src/Nodes/DOM/DOMProcessingInstruction.php b/src/Nodes/DOM/DOMProcessingInstruction.php
deleted file mode 100644
index 82c69cf..0000000
--- a/src/Nodes/DOM/DOMProcessingInstruction.php
+++ /dev/null
@@ -1,10 +0,0 @@
-<?php
-
-namespace fivefilters\Readability\Nodes\DOM;
-
-use fivefilters\Readability\Nodes\NodeTrait;
-
-class DOMProcessingInstruction extends \DOMProcessingInstruction
-{
-    use NodeTrait;
-}
diff --git a/src/Nodes/DOM/DocumentFragment.php b/src/Nodes/DOM/DocumentFragment.php
new file mode 100644
index 0000000..21514ea
--- /dev/null
+++ b/src/Nodes/DOM/DocumentFragment.php
@@ -0,0 +1,10 @@
+<?php
+
+namespace fivefilters\Readability\Nodes\DOM;
+
+use fivefilters\Readability\Nodes\NodeTrait;
+
+class DocumentFragment extends \DOM\DocumentFragment
+{
+    use NodeTrait;
+}
diff --git a/src/Nodes/DOM/DocumentType.php b/src/Nodes/DOM/DocumentType.php
new file mode 100644
index 0000000..b6df856
--- /dev/null
+++ b/src/Nodes/DOM/DocumentType.php
@@ -0,0 +1,10 @@
+<?php
+
+namespace fivefilters\Readability\Nodes\DOM;
+
+use fivefilters\Readability\Nodes\NodeTrait;
+
+class DocumentType extends \DOM\DocumentType
+{
+    use NodeTrait;
+}
diff --git a/src/Nodes/DOM/DOMElement.php b/src/Nodes/DOM/Element.php
similarity index 82%
rename from src/Nodes/DOM/DOMElement.php
rename to src/Nodes/DOM/Element.php
index 0493c41..57c592e 100644
--- a/src/Nodes/DOM/DOMElement.php
+++ b/src/Nodes/DOM/Element.php
@@ -4,7 +4,7 @@
 
 use fivefilters\Readability\Nodes\NodeTrait;
 
-class DOMElement extends \DOMElement
+class Element extends \DOM\HtmlElement
 {
     use NodeTrait;
 
@@ -13,9 +13,9 @@ class DOMElement extends \DOMElement
      *
      * To get all child nodes, including non-element nodes like text and comment nodes, use childNodes.
      */
-    public function children(): DOMNodeList
+    public function children(): NodeList
     {
-        $newList = new DOMNodeList();
+        $newList = new NodeList();
         foreach ($this->childNodes as $node) {
             if ($node->nodeType === XML_ELEMENT_NODE) {
                 $newList->add($node);
@@ -29,7 +29,7 @@ public function children(): DOMNodeList
      *
      * @deprecated Use previousElementSibling instead - introduced in PHP 8.0.
      */
-    public function previousElementSibling(): ?DOMElement
+    public function previousElementSibling(): ?Element
     {
         return $this->previousElementSibling;
     }
diff --git a/src/Nodes/DOM/DOMComment.php b/src/Nodes/DOM/Entity.php
similarity index 76%
rename from src/Nodes/DOM/DOMComment.php
rename to src/Nodes/DOM/Entity.php
index 3b691f4..721be70 100644
--- a/src/Nodes/DOM/DOMComment.php
+++ b/src/Nodes/DOM/Entity.php
@@ -4,7 +4,7 @@
 
 use fivefilters\Readability\Nodes\NodeTrait;
 
-class DOMComment extends \DOMComment
+class Entity extends \DOM\Entity
 {
     use NodeTrait;
 }
diff --git a/src/Nodes/DOM/DOMCharacterData.php b/src/Nodes/DOM/EntityReference.php
similarity index 70%
rename from src/Nodes/DOM/DOMCharacterData.php
rename to src/Nodes/DOM/EntityReference.php
index b196979..286e3f6 100644
--- a/src/Nodes/DOM/DOMCharacterData.php
+++ b/src/Nodes/DOM/EntityReference.php
@@ -4,7 +4,7 @@
 
 use fivefilters\Readability\Nodes\NodeTrait;
 
-class DOMCharacterData extends \DOMCharacterData
+class EntityReference extends \DOM\EntityReference
 {
     use NodeTrait;
 }
diff --git a/src/Nodes/DOM/DOMNode.php b/src/Nodes/DOM/Node.php
similarity index 86%
rename from src/Nodes/DOM/DOMNode.php
rename to src/Nodes/DOM/Node.php
index 4a3ab0d..e879546 100644
--- a/src/Nodes/DOM/DOMNode.php
+++ b/src/Nodes/DOM/Node.php
@@ -8,7 +8,7 @@
  * @method getAttribute($attribute)
  * @method hasAttribute($attribute)
  */
-class DOMNode extends \DOMNode
+class Node extends \DOM\Node
 {
     use NodeTrait;
 }
diff --git a/src/Nodes/DOM/DOMNodeList.php b/src/Nodes/DOM/NodeList.php
similarity index 86%
rename from src/Nodes/DOM/DOMNodeList.php
rename to src/Nodes/DOM/NodeList.php
index 2b34e22..09d2df7 100644
--- a/src/Nodes/DOM/DOMNodeList.php
+++ b/src/Nodes/DOM/NodeList.php
@@ -3,7 +3,7 @@
 namespace fivefilters\Readability\Nodes\DOM;
 
 /**
- * Class DOMNodeList.
+ * Class NodeList.
  *
  * This is a fake DOMNodeList class that allows adding items to the list. The original class is static and the nodes
  * are defined automagically when instantiating it. This fake version behaves exactly the same way but adds the function
@@ -12,7 +12,7 @@
  * It cannot extend the original DOMNodeList class because the functionality behind the property ->length is hidden
  * from the user and cannot be extended, changed, or tweaked.
  */
-class DOMNodeList implements \Countable, \IteratorAggregate
+class NodeList implements \Countable, \IteratorAggregate
 {
     /**
      * @var array
@@ -42,7 +42,7 @@ public function __get($name)
     /**
      * Add node to the list.
      */
-    public function add(DOMNode|DOMElement|DOMText|DOMComment $node): DOMNodeList
+    public function add(Node|Element|Text|Comment $node): NodeList
     {
         $this->items[] = $node;
         $this->length++;
@@ -53,7 +53,7 @@ public function add(DOMNode|DOMElement|DOMText|DOMComment $node): DOMNodeList
     /**
      * Get node.
      */
-    public function item(int $offset): DOMNode|DOMElement|DOMText|DOMComment
+    public function item(int $offset): Node|Element|Text|Comment
     {
         return $this->items[$offset];
     }
diff --git a/src/Nodes/DOM/Notation.php b/src/Nodes/DOM/Notation.php
new file mode 100644
index 0000000..8a50f5c
--- /dev/null
+++ b/src/Nodes/DOM/Notation.php
@@ -0,0 +1,10 @@
+<?php
+
+namespace fivefilters\Readability\Nodes\DOM;
+
+use fivefilters\Readability\Nodes\NodeTrait;
+
+class Notation extends \DOM\Notation
+{
+    use NodeTrait;
+}
diff --git a/src/Nodes/DOM/ProcessingInstruction.php b/src/Nodes/DOM/ProcessingInstruction.php
new file mode 100644
index 0000000..95ff327
--- /dev/null
+++ b/src/Nodes/DOM/ProcessingInstruction.php
@@ -0,0 +1,10 @@
+<?php
+
+namespace fivefilters\Readability\Nodes\DOM;
+
+use fivefilters\Readability\Nodes\NodeTrait;
+
+class ProcessingInstruction extends \DOM\ProcessingInstruction
+{
+    use NodeTrait;
+}
diff --git a/src/Nodes/DOM/DOMText.php b/src/Nodes/DOM/Text.php
similarity index 79%
rename from src/Nodes/DOM/DOMText.php
rename to src/Nodes/DOM/Text.php
index 6a5716c..520488b 100644
--- a/src/Nodes/DOM/DOMText.php
+++ b/src/Nodes/DOM/Text.php
@@ -4,7 +4,7 @@
 
 use fivefilters\Readability\Nodes\NodeTrait;
 
-class DOMText extends \DOMText
+class Text extends \DOM\Text
 {
     use NodeTrait;
 }
diff --git a/src/Nodes/NodeTrait.php b/src/Nodes/NodeTrait.php
index 6ec36a8..1eeb511 100644
--- a/src/Nodes/NodeTrait.php
+++ b/src/Nodes/NodeTrait.php
@@ -2,17 +2,16 @@
 
 namespace fivefilters\Readability\Nodes;
 
-use fivefilters\Readability\Nodes\DOM\DOMDocument;
-use fivefilters\Readability\Nodes\DOM\DOMElement;
-use fivefilters\Readability\Nodes\DOM\DOMNode;
-use fivefilters\Readability\Nodes\DOM\DOMText;
+use fivefilters\Readability\Nodes\DOM\Element;
+use fivefilters\Readability\Nodes\DOM\Node;
+use fivefilters\Readability\Nodes\DOM\Text;
 
 /**
- * @property ?DOMNode $firstChild
- * @property ?DOMNode $lastChild
- * @property ?DOMNode $parentNode
- * @property ?DOMNode $nextSibling
- * @property ?DOMNode $previousSibling
+ * @property ?Node $firstChild
+ * @property ?Node $lastChild
+ * @property ?Node $parentNode
+ * @property ?Node $nextSibling
+ * @property ?Node $previousSibling
  */
 trait NodeTrait
 {
@@ -152,8 +151,8 @@ public function initializeNode(bool $weightClasses): static
      */
     public function getAttribute(string $attributeName): string
     {
-        if (!is_null($this->attributes)) {
-            return parent::getAttribute($attributeName);
+        if ($this instanceof \Dom\HtmlElement) {
+            return parent::getAttribute($attributeName) ?? '';
         }
 
         return '';
@@ -166,7 +165,7 @@ public function getAttribute(string $attributeName): string
      */
     public function hasAttribute(string $attributeName): bool
     {
-        if (!is_null($this->attributes)) {
+        if ($this instanceof \Dom\HtmlElement) {
             return parent::hasAttribute($attributeName);
         }
 
@@ -185,7 +184,7 @@ public function getNodeAncestors(int|bool $maxLevel = 3): array
 
         $node = $this->parentNode;
 
-        while ($node && !($node instanceof DOMDocument)) {
+        while ($node && !($node instanceof \Dom\HtmlDocument)) {
             $ancestors[] = $node;
             $level++;
             if ($level === $maxLevel) {
@@ -221,7 +220,7 @@ public function getLinkDensity(): float
         $links = $this->getAllLinks();
 
         if ($links) {
-            /** @var DOMElement $link */
+            /** @var Element $link */
             foreach ($links as $link) {
                 $href = $link->getAttribute('href');
                 $coefficient = ($href && preg_match(NodeUtility::$regexps['hashUrl'], $href)) ? 0.3 : 1;
@@ -273,12 +272,12 @@ public function getClassWeight(): int
      */
     public function getTextContent(bool $normalize = true): string
     {
-        $nodeValue = trim($this->textContent);
+        $textContent = mb_trim($this->textContent);
         if ($normalize) {
-            $nodeValue = preg_replace(NodeUtility::$regexps['normalize'], ' ', $nodeValue);
+            $textContent = preg_replace(NodeUtility::$regexps['normalize'], ' ', $textContent);
         }
 
-        return $nodeValue;
+        return $textContent;
     }
 
     /**
@@ -289,7 +288,7 @@ public function getRowAndColumnCount(): array
         $rows = $columns = 0;
         $trs = $this->getElementsByTagName('tr');
         foreach ($trs as $tr) {
-            /** @var \DOMElement $tr */
+            /** @var \DOM\Element $tr */
             $rowspan = $tr->getAttribute('rowspan');
             $rows += ($rowspan || 1);
 
@@ -297,7 +296,7 @@ public function getRowAndColumnCount(): array
             $columnsInThisRow = 0;
             $cells = $tr->getElementsByTagName('td');
             foreach ($cells as $cell) {
-                /** @var \DOMElement $cell */
+                /** @var \DOM\Element $cell */
                 $colspan = $cell->getAttribute('colspan');
                 $columnsInThisRow += ($colspan || 1);
             }
@@ -310,10 +309,11 @@ public function getRowAndColumnCount(): array
     /**
      * Creates a new node based on the text content of the original node.
      */
-    public function createNode(DOMNode $originalNode, string $tagName): DOMElement
+    public function createNode(Node $originalNode, string $tagName): Element
     {
         $text = $originalNode->getTextContent(false);
-        $newNode = $originalNode->ownerDocument->createElement($tagName, $text);
+        $newNode = $originalNode->ownerDocument->createElement($tagName);
+        $newNode->appendChild($originalNode->ownerDocument->createTextNode($text));
 
         return $newNode;
     }
@@ -378,7 +378,7 @@ public function hasSingleChildBlockElement(): bool
                     $result = true;
                 } else {
                     // If any of the hasSingleChildBlockElement calls return true, return true then.
-                    /** @var $child DOMElement */
+                    /** @var $child Element */
                     $result = ($result || $child->hasSingleChildBlockElement());
                 }
             }
@@ -392,8 +392,8 @@ public function hasSingleChildBlockElement(): bool
      */
     public function isElementWithoutContent(): bool
     {
-        return $this instanceof DOMElement &&
-            mb_strlen(preg_replace(NodeUtility::$regexps['onlyWhitespace'], '', $this->textContent)) === 0 &&
+        return $this instanceof Element &&
+            mb_strlen(preg_replace(NodeUtility::$regexps['onlyWhitespace'], '', $this->textContent) ?? '') === 0 &&
             ($this->childNodes->length === 0 ||
                 $this->childNodes->length === $this->getElementsByTagName('br')->length + $this->getElementsByTagName('hr')->length
                 /*
@@ -405,7 +405,7 @@ public function isElementWithoutContent(): bool
                  * mb_strlen in this chain of checks).
                  */
                 + count(array_filter(iterator_to_array($this->childNodes), function ($child) {
-                    return $child instanceof DOMText;
+                    return $child instanceof Text;
                 }))
 
             );
@@ -443,7 +443,7 @@ public function isProbablyVisible(): bool
      */
     public function isWhitespace(): bool
     {
-        return ($this->nodeType === XML_TEXT_NODE && $this->isWhitespaceInElementContent()) ||
+        return ($this->nodeType === XML_TEXT_NODE && mb_strlen(mb_trim($this->textContent)) === 0) ||
             ($this->nodeType === XML_ELEMENT_NODE && $this->nodeName === 'br');
     }
 
@@ -487,9 +487,9 @@ public function shiftingAwareGetElementsByTagName(string $tag): \Generator
     /**
      * Git first element child or null
      */
-    public function getFirstElementChild(): ?DOMElement
+    public function getFirstElementChild(): ?Element
     {
-        if ($this->nodeType === XML_ELEMENT_NODE || $this->nodeType === XML_DOCUMENT_NODE) {
+        if ($this->nodeType === XML_ELEMENT_NODE || $this->nodeType === XML_HTML_DOCUMENT_NODE) {
             return $this->firstElementChild;
         }
 
diff --git a/src/Nodes/NodeUtility.php b/src/Nodes/NodeUtility.php
index 2f094f1..a6d63e8 100644
--- a/src/Nodes/NodeUtility.php
+++ b/src/Nodes/NodeUtility.php
@@ -2,12 +2,12 @@
 
 namespace fivefilters\Readability\Nodes;
 
-use fivefilters\Readability\Nodes\DOM\DOMDocument;
-use fivefilters\Readability\Nodes\DOM\DOMElement;
-use fivefilters\Readability\Nodes\DOM\DOMNode;
-use fivefilters\Readability\Nodes\DOM\DOMText;
-use fivefilters\Readability\Nodes\DOM\DOMComment;
-use fivefilters\Readability\Nodes\DOM\DOMNodeList;
+use fivefilters\Readability\Nodes\DOM;
+use fivefilters\Readability\Nodes\DOM\Element;
+use fivefilters\Readability\Nodes\DOM\Node;
+use fivefilters\Readability\Nodes\DOM\Text;
+use fivefilters\Readability\Nodes\DOM\Comment;
+use fivefilters\Readability\Nodes\DOM\NodeList;
 
 /**
  * Class NodeUtility.
@@ -52,7 +52,7 @@ class NodeUtility
      *
      * Imported from the Element class on league\html-to-markdown.
      */
-    public static function nextNode(DOMNode|DOMComment|DOMText|DOMElement|null $node): DOMNode|DOMComment|DOMText|DOMElement|null
+    public static function nextNode(Node|Comment|Text|Element|null $node): Node|Comment|Text|Element|null
     {
         $next = $node;
         while ($next
@@ -68,9 +68,10 @@ public static function nextNode(DOMNode|DOMComment|DOMText|DOMElement|null $node
      * 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.
      */
-    public static function setNodeTag(DOMNode|DOMElement $node, string $value, bool $importAttributes = true): DOMNode|DOMElement
+    public static function setNodeTag(Node|Element $node, string $value, bool $importAttributes = true): Node|Element
     {
-        $new = new DOMDocument('1.0', 'utf-8');
+        $new = \Dom\HtmlDocument::createEmpty();
+        NodeUtility::registerReadabilityNodeClasses($new);
         $new->appendChild($new->createElement($value));
 
         $children = $node->childNodes;
@@ -97,7 +98,7 @@ public static function setNodeTag(DOMNode|DOMElement $node, string $value, bool
     /**
      * Removes the current node and returns the next node to be parsed (child, sibling or parent).
      */
-    public static function removeAndGetNext(DOMNode|DOMComment|DOMText|DOMElement $node): DOMNode|DOMComment|DOMText|DOMElement|null
+    public static function removeAndGetNext(Node|Comment|Text|Element $node): Node|Comment|Text|Element|null
     {
         $nextNode = self::getNextNode($node, true);
         $node->parentNode->removeChild($node);
@@ -108,7 +109,7 @@ public static function removeAndGetNext(DOMNode|DOMComment|DOMText|DOMElement $n
     /**
      * Remove the selected node.
      */
-    public static function removeNode(DOMNode|DOMComment|DOMText|DOMElement $node): void
+    public static function removeNode(Node|Comment|Text|Element $node): void
     {
         $parent = $node->parentNode;
         if ($parent) {
@@ -120,7 +121,7 @@ public static function removeNode(DOMNode|DOMComment|DOMText|DOMElement $node):
      * Returns the next node. First checks for children (if the flag allows it), then for siblings, and finally
      * for parents.
      */
-    public static function getNextNode(DOMNode|DOMComment|DOMText|DOMElement|DOMDocument $originalNode, bool $ignoreSelfAndKids = false): DOMNode|DOMComment|DOMText|DOMElement|DOMDocument|null
+    public static function getNextNode(Node|Comment|Text|Element|\Dom\HtmlDocument $originalNode, bool $ignoreSelfAndKids = false): Node|Comment|Text|Element|\Dom\HtmlDocument|null
     {
         /*
          * Traverse the DOM from node to node, starting at the node passed in.
@@ -153,15 +154,34 @@ public static function getNextNode(DOMNode|DOMComment|DOMText|DOMElement|DOMDocu
     /**
      * Remove all empty DOMNodes from DOMNodeLists.
      */
-    public static function filterTextNodes(\DOMNodeList $list): DOMNodeList
+    public static function filterTextNodes(\Dom\NodeList $list): NodeList
     {
-        $newList = new DOMNodeList();
+        $newList = new NodeList();
         foreach ($list as $node) {
-            if ($node->nodeType !== XML_TEXT_NODE || !$node->isWhitespaceInElementContent()) {
+            if ($node->nodeType !== XML_TEXT_NODE || mb_trim($node->nodeValue) !== '') {
                 $newList->add($node);
             }
         }
 
         return $newList;
     }
+
+    public static function registerReadabilityNodeClasses(\DOM\HtmlDocument $dom): void
+    {
+        $dom->registerNodeClass('DOM\HtmlElement', DOM\Element::class);
+        $dom->registerNodeClass('DOM\Attr', DOM\Attr::class);
+        $dom->registerNodeClass('DOM\CdataSection', DOM\CdataSection::class);
+        $dom->registerNodeClass('DOM\CharacterData', DOM\CharacterData::class);
+        $dom->registerNodeClass('DOM\Comment', DOM\Comment::class);
+        //$dom->registerNodeClass('DOM\HtmlDocument', DOM\HtmlDocument::class);
+        $dom->registerNodeClass('DOM\DocumentFragment', DOM\DocumentFragment::class);
+        $dom->registerNodeClass('DOM\DocumentType', DOM\DocumentType::class);
+        $dom->registerNodeClass('DOM\Element', DOM\Element::class);
+        $dom->registerNodeClass('DOM\Entity', DOM\Entity::class);
+        $dom->registerNodeClass('DOM\EntityReference', DOM\EntityReference::class);
+        $dom->registerNodeClass('DOM\Node', DOM\Node::class);
+        $dom->registerNodeClass('DOM\Notation', DOM\Notation::class);
+        $dom->registerNodeClass('DOM\ProcessingInstruction', DOM\ProcessingInstruction::class);
+        $dom->registerNodeClass('DOM\Text', DOM\Text::class);
+    }
 }
diff --git a/src/Readability.php b/src/Readability.php
index c37c9b6..a2ff53b 100644
--- a/src/Readability.php
+++ b/src/Readability.php
@@ -2,14 +2,12 @@
 
 namespace fivefilters\Readability;
 
-use fivefilters\Readability\Nodes\DOM\DOMDocument;
-use fivefilters\Readability\Nodes\DOM\DOMElement;
-use fivefilters\Readability\Nodes\DOM\DOMNode;
-use fivefilters\Readability\Nodes\DOM\DOMText;
-use fivefilters\Readability\Nodes\DOM\DOMComment;
+use fivefilters\Readability\Nodes\DOM\Element;
+use fivefilters\Readability\Nodes\DOM\Node;
+use fivefilters\Readability\Nodes\DOM\Text;
+use fivefilters\Readability\Nodes\DOM\Comment;
 use fivefilters\Readability\Nodes\NodeUtility;
 use Psr\Log\LoggerInterface;
-use Masterminds\HTML5;
 use League\Uri\BaseUri;
 
 /**
@@ -18,9 +16,9 @@
 class Readability
 {
     /**
-     * Main DOMDocument where all the magic happens.
+     * Main HtmlDocument where all the magic happens.
      */
-    protected DOMDocument $dom;
+    protected \Dom\HtmlDocument $dom;
 
     /**
      * Title of the article.
@@ -28,9 +26,9 @@ class Readability
     protected ?string $title = null;
 
     /**
-     * Final DOMDocument with the fully parsed HTML.
+     * Final HtmlDocument with the fully parsed HTML.
      */
-    protected ?DOMDocument $content = null;
+    protected ?\Dom\HtmlDocument $content = null;
 
     /**
      * Excerpt of the article.
@@ -59,7 +57,7 @@ class Readability
 
     /**
      * Base URI
-     * HTML5PHP doesn't appear to store it in the baseURI property like PHP's DOMDocument does when parsing with libxml
+     * HTML5PHP doesn't appear to store it in the baseURI property like PHP's HtmlDocument does when parsing with libxml
      */
     protected ?string $baseURI = null;
 
@@ -162,6 +160,7 @@ public function parse(?string $html = null): bool
 
             throw new ParseException('Invalid or incomplete HTML.');
         }
+        $root = $this->dom->getElementsByTagName('body')->item(0);
 
         $bodyCache = $root->cloneNode(true);
 
@@ -170,12 +169,11 @@ public function parse(?string $html = null): bool
         $this->getMainImage();
 
         while (true) {
-
-            $this->logger->debug('Starting parse loop');
+            $this->logger->debug('Starting parse loop (#' . count($this->attempts) . ')');
             //$root = $root->firstChild;
 
             $elementsToScore = $this->getNodes($root->firstChild);
-            $this->logger->debug(sprintf('Elements to score: \'%s\'', count($elementsToScore)));
+            $this->logger->debug(sprintf('Elements to score: %d', count($elementsToScore)));
 
             $result = $this->rateNodes($elementsToScore);
 
@@ -187,8 +185,7 @@ public function parse(?string $html = null): bool
              * finding the -right- content.
              */
 
-            $length = !$result ? 0 : mb_strlen(preg_replace(NodeUtility::$regexps['onlyWhitespace'], '', $result->textContent));
-
+            $length = !$result ? 0 : mb_strlen(preg_replace(NodeUtility::$regexps['onlyWhitespace'], '', $result->documentElement->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()) {
@@ -271,14 +268,15 @@ public function loadHTML(string $html): void
         $this->logger->debug('[Loading] Loading HTML...');
 
         // To avoid throwing a gazillion of errors on malformed HTMLs
-        libxml_use_internal_errors(true);
+        //libxml_use_internal_errors(true);
 
         //$html = preg_replace('/(<br[^>]*>[ \n\r\t]*){2,}/i', '</p><p>', $html);
 
         if ($this->configuration->getParser() === 'html5') {
             $this->logger->debug('[Loading] Using HTML5 parser...');
-            $html5 = new HTML5(['disable_html_ns' => true, 'target_document' => new DOMDocument('1.0', 'utf-8')]);
-            $dom = $html5->loadHTML($html);
+            // New DOM class with HTML5 parser introduced in PHP 8.4
+            $dom = \Dom\HtmlDocument::createFromString($html, \Dom\HTML_NO_DEFAULT_NS | LIBXML_NOERROR);
+            NodeUtility::registerReadabilityNodeClasses($dom);
             //TODO: Improve this so it looks inside <html><head><base>, not just any <base>
             $base = $dom->getElementsByTagName('base');
             if ($base->length > 0) {
@@ -289,31 +287,9 @@ public function loadHTML(string $html): void
                 }
             }
         } else {
-            $this->logger->debug('[Loading] Using libxml parser...');
-            $dom = new DOMDocument('1.0', 'utf-8');
-            if ($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->getSubstituteEntities()) {
-            // Keep the original HTML entities
-            $dom->substituteEntities = false;
-        }
-
-        if ($this->configuration->getSummonCthulhu()) {
-            $this->logger->debug('[Loading] Removed script tags via regex H̶͈̩̟̬̱͠E̡̨̬͔̳̜͢͠ ̡̧̯͉̩͙̩̹̞̠͎͈̹̥̠͞ͅͅC̶͉̞̘̖̝̗͓̬̯͍͉̤̬͢͢͞Ò̟̘͉͖͎͉̱̭̣̕M̴̯͈̻̱̱̣̗͈̠̙̲̥͘͞E̷̛͙̼̲͍͕̹͍͇̗̻̬̮̭̱̥͢Ş̛̟͔̙̜̤͇̮͍̙̝̀͘');
-            $html = preg_replace('/<script\b[^>]*>([\s\S]*?)<\/script>/', '', $html);
-        }
-
-        // Prepend the XML tag to avoid having issues with special characters. Should be harmless.
-        if ($this->configuration->getParser() !== 'html5') {
-            $dom->loadHTML('<?xml encoding="UTF-8">' . $html);
-            $this->baseURI = $dom->baseURI;
+            // Throw exception if the parser is not HTML5
+            throw new ParseException('This version of Readability.php only supports the HTML5 parser introduced in PHP 8.4');
         }
-        $dom->encoding = 'UTF-8';
 
         $this->logger->debug('[Loading] Loaded HTML successfully.');
 
@@ -326,7 +302,7 @@ public function loadHTML(string $html): void
      *
      * @return array with any metadata that could be extracted (possibly none)
      */
-    private function getJSONLD(DOMDocument $dom): array
+    private function getJSONLD(\Dom\HtmlDocument $dom): array
     {
         $scripts = $this->_getAllNodesWithTag($dom, ['script']);
 
@@ -418,7 +394,7 @@ private function getMetadata(): void
             /* @var DOMNode $meta */
             $elementName = $meta->getAttribute('name');
             $elementProperty = $meta->getAttribute('property');
-            $content = $meta->getAttribute('content'); 
+            $content = $meta->getAttribute('content');
             $matches = null;
             $name = null;
 
@@ -597,7 +573,7 @@ public function getMainImage(): void
     /**
      * Remove unnecessary nested elements
      */
-    private function simplifyNestedElements(DOMDocument $article): void
+    private function simplifyNestedElements(\Dom\HtmlDocument $article): void
     {
         $node = $article;
 
@@ -635,8 +611,8 @@ private function getArticleTitle(): ?string
             $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;
+                $this->logger->info(sprintf('[Metadata] Using title tag as article title: \'%s\'', $titleTag->item(0)->textContent));
+                $originalTitle = $titleTag->item(0)->textContent;
             }
         }
 
@@ -672,7 +648,7 @@ private function getArticleTitle(): ?string
             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($curTitle)) {
+                    if (trim($hTag->textContent) === trim($curTitle)) {
                         $match = true;
                     }
                 }
@@ -698,7 +674,7 @@ private function getArticleTitle(): ?string
             $hOnes = $this->dom->getElementsByTagName('h1');
 
             if ($hOnes->length === 1) {
-                $curTitle = $hOnes->item(0)->nodeValue;
+                $curTitle = $hOnes->item(0)->textContent;
                 $this->logger->info(sprintf('[Metadata] Using title from an H1 node: \'%s\'', $curTitle));
             }
         }
@@ -795,7 +771,7 @@ public function getPathInfo(string $url): array
     /**
      * Gets nodes from the root element.
      */
-    private function getNodes(DOMNode|DOMComment|DOMText|DOMElement|null $node): array
+    private function getNodes(Node|Comment|Text|Element|null $node): array
     {
         $this->logger->info('[Get Nodes] Retrieving nodes...');
         if ($node === null) {
@@ -817,7 +793,7 @@ private function getNodes(DOMNode|DOMComment|DOMText|DOMElement|null $node): arr
         while ($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)));
+                $this->logger->debug(sprintf('[Get Nodes] Found comment node, removing... Node content was: \'%s\'', substr($node->textContent, 0, 128)));
                 $node = NodeUtility::removeAndGetNext($node);
                 continue;
             }
@@ -832,7 +808,7 @@ private function getNodes(DOMNode|DOMComment|DOMText|DOMElement|null $node): arr
 
             // 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)));
+                $this->logger->debug(sprintf('[Get Nodes] Found byline, removing... Node content was: \'%s\'', substr($node->textContent, 0, 128)));
                 $node = NodeUtility::removeAndGetNext($node);
                 continue;
             }
@@ -854,7 +830,7 @@ private function getNodes(DOMNode|DOMComment|DOMText|DOMElement|null $node): arr
                     $node->nodeName !== 'body' &&
                     $node->nodeName !== 'a'
                 ) {
-                    $this->logger->debug(sprintf('[Get Nodes] Removing unlikely candidate. Node content was: \'%s\'', substr($node->nodeValue, 0, 128)));
+                    $this->logger->debug(sprintf('[Get Nodes] Removing unlikely candidate. Node content was: \'%s\'', substr($node->textContent, 0, 128)));
                     $node = NodeUtility::removeAndGetNext($node);
                     continue;
                 }
@@ -878,7 +854,7 @@ private function getNodes(DOMNode|DOMComment|DOMText|DOMElement|null $node): arr
             }
 
             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)));
+                $this->logger->debug(sprintf('[Get Nodes] Adding node to score list, node content is: \'%s\'', substr($node->textContent, 0, 128)));
                 $elementsToScore[] = $node;
             }
 
@@ -914,13 +890,13 @@ private function getNodes(DOMNode|DOMComment|DOMText|DOMElement|null $node): arr
                  * algorithm with DIVs with are, in practice, paragraphs.
                  */
                 if ($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)));
+                    $this->logger->debug(sprintf('[Get Nodes] Found DIV with a single P node, removing DIV. Node content is: \'%s\'', substr($node->textContent, 0, 128)));
                     $pNode = NodeUtility::filterTextNodes($node->childNodes)->item(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)));
+                    $this->logger->debug(sprintf('[Get Nodes] Found DIV with a single child block element, converting to a P node. Node content is: \'%s\'', substr($node->textContent, 0, 128)));
                     $node = NodeUtility::setNodeTag($node, 'p');
                     $elementsToScore[] = $node;
                 }
@@ -960,7 +936,7 @@ private function textSimilarity(string $textA, string $textB): float
     /**
      * Checks if the node is a byline.
      */
-    private function checkByline(DOMNode|DOMText|DOMElement $node, string $matchString): bool
+    private function checkByline(Node|Text|Element $node, string $matchString): bool
     {
         if (!$this->configuration->getArticleByline()) {
             return false;
@@ -1030,7 +1006,7 @@ private function unescapeHtmlEntities(?string $str): ?string
      * Check if node is image, or if node contains exactly only one image
      * whether as a direct child or as its descendants.
      */
-    private function isSingleImage(DOMElement|DOMNode|DOMText $node): bool
+    private function isSingleImage(Element|Node|Text $node): bool
     {
         if ($node->tagName === 'img') {
             return true;
@@ -1049,7 +1025,7 @@ private function isSingleImage(DOMElement|DOMNode|DOMText $node): bool
      * and remove the <noscript> tag. This improves the quality of the images we use on
      * some sites (e.g. Medium).
      */
-    private function unwrapNoscriptImages(DOMDocument $dom): void
+    private function unwrapNoscriptImages(\Dom\HtmlDocument $dom): void
     {
         // Find img without source or attributes that might contains image, and remove it.
         // This is done to prevent a placeholder img is replaced by img from noscript in next step.
@@ -1125,7 +1101,7 @@ private function unwrapNoscriptImages(DOMDocument $dom): void
     /**
      * Removes all the scripts of the html.
      */
-    private function removeScripts(DOMDocument $dom): void
+    private function removeScripts(\Dom\HtmlDocument $dom): void
     {
         foreach (['script', 'noscript'] as $tag) {
             $nodes = $dom->getElementsByTagName($tag);
@@ -1138,11 +1114,11 @@ private function removeScripts(DOMDocument $dom): void
     /**
      * Prepares the document for parsing.
      */
-    private function prepDocument(DOMDocument $dom): void
+    private function prepDocument(\Dom\HtmlDocument $dom): void
     {
         $this->logger->info('[PrepDocument] Preparing document for parsing...');
 
-        foreach ($dom->shiftingAwareGetElementsByTagName('br') as $br) {
+        foreach ($dom->documentElement->shiftingAwareGetElementsByTagName('br') as $br) {
             $next = $br->nextSibling;
 
             /*
@@ -1225,15 +1201,15 @@ private function prepDocument(DOMDocument $dom): void
     /**
      * Assign scores to each node. Returns full article parsed or false on error.
      */
-    private function rateNodes(array $nodes): DOMDocument|bool
+    private function rateNodes(array $nodes): \Dom\HtmlDocument|bool
     {
         $this->logger->info('[Rating] Rating nodes...');
 
         $candidates = [];
 
-        /** @var DOMElement $node */
+        /** @var Element $node */
         foreach ($nodes as $node) {
-            if (is_null($node->parentNode)) {
+            if (is_null($node->parentElement)) {
                 continue;
             }
 
@@ -1258,9 +1234,9 @@ private function rateNodes(array $nodes): DOMDocument|bool
             // 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)));
+            $this->logger->debug(sprintf('[Rating] Node score %s, content: \'%s\'', $contentScore, substr($node->textContent, 0, 128)));
 
-            /** @var $ancestor DOMElement */
+            /** @var $ancestor Element */
             foreach ($ancestors as $level => $ancestor) {
                 $this->logger->debug('[Rating] Found ancestor, initializing and adding it as a candidate...');
                 if (!$ancestor->isInitialized()) {
@@ -1286,7 +1262,7 @@ private function rateNodes(array $nodes): DOMDocument|bool
                 $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)));
+                $this->logger->debug(sprintf('[Rating] Ancestor score %s, value: \'%s\'', $ancestor->contentScore, substr($ancestor->textContent, 0, 128)));
             }
         }
 
@@ -1331,9 +1307,9 @@ private function rateNodes(array $nodes): DOMDocument|bool
             $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', ''));
+            $topCandidate = \Dom\HtmlDocument::createEmpty();
+            NodeUtility::registerReadabilityNodeClasses($topCandidate);
+            $topCandidate->appendChild($topCandidate->createElement('div'));
             $kids = $this->dom->getElementsByTagName('body')->item(0)->childNodes;
 
             // Cannot be foreached, don't ask me why.
@@ -1342,7 +1318,7 @@ private function rateNodes(array $nodes): DOMDocument|bool
                 $topCandidate->firstChild->appendChild($import);
             }
 
-            // Candidate must be created using firstChild to grab the DOMElement instead of the DOMDocument.
+            // Candidate must be created using firstChild to grab the Element instead of the HtmlDocument.
             $topCandidate = $topCandidate->firstChild;
         } elseif ($topCandidate) {
             $this->logger->info(sprintf('[Rating] Found top candidate, score: %s', $topCandidate->contentScore));
@@ -1361,7 +1337,7 @@ private function rateNodes(array $nodes): DOMDocument|bool
             if (count($alternativeCandidateAncestors) >= $MINIMUM_TOPCANDIDATES) {
                 $parentOfTopCandidate = $topCandidate->parentNode;
 
-                // Check if we are actually dealing with a DOMNode and not a DOMDocument node or higher
+                // Check if we are actually dealing with a DOMNode and not a HtmlDocument node or higher
                 while ($parentOfTopCandidate && $parentOfTopCandidate->nodeName !== 'body' && $parentOfTopCandidate->nodeType === XML_ELEMENT_NODE) {
                     $listsContainingThisAncestor = 0;
                     for ($ancestorIndex = 0; $ancestorIndex < count($alternativeCandidateAncestors) && $listsContainingThisAncestor < $MINIMUM_TOPCANDIDATES; $ancestorIndex++) {
@@ -1391,7 +1367,7 @@ private function rateNodes(array $nodes): DOMDocument|bool
             // The scores shouldn't get too low.
             $scoreThreshold = $lastScore / 3;
 
-            /* @var DOMElement $parentOfTopCandidate */
+            /* @var Element $parentOfTopCandidate */
             while ($parentOfTopCandidate && $parentOfTopCandidate->nodeName !== 'body') {
                 $parentScore = $parentOfTopCandidate->contentScore;
                 if ($parentScore < $scoreThreshold) {
@@ -1425,8 +1401,9 @@ private function rateNodes(array $nodes): DOMDocument|bool
 
         $this->logger->info('[Rating] Creating final article content document...');
 
-        $articleContent = new DOMDocument('1.0', 'utf-8');
-        $articleContent->createElement('div');
+        $articleContent = \Dom\HtmlDocument::createEmpty();
+        NodeUtility::registerReadabilityNodeClasses($articleContent);
+        $articleElement = $articleContent->appendChild($articleContent->createElement('article'));
 
         $siblingScoreThreshold = max(10, $topCandidate->contentScore * 0.2);
         // Keep potential top candidate's parent node to try to get text direction of it later.
@@ -1437,7 +1414,7 @@ private function rateNodes(array $nodes): DOMDocument|bool
 
         $this->logger->info('[Rating] Adding top candidate siblings...');
 
-        /* @var DOMElement $sibling */
+        /* @var Element $sibling */
         // Can't foreach here because down there we might change the tag name and that causes the foreach to skip items
         for ($i = 0; $i < ($siblings->length ?? 0); $i++) {
             $sibling = $siblings[$i];
@@ -1470,7 +1447,7 @@ private function rateNodes(array $nodes): DOMDocument|bool
             }
 
             if ($append) {
-                $this->logger->debug(sprintf('[Rating] Appending sibling to final article, content is: \'%s\'', substr($sibling->nodeValue, 0, 128)));
+                $this->logger->debug(sprintf('[Rating] Appending sibling to final article, content is: \'%s\'', substr($sibling->textContent, 0, 128)));
 
                 $hasContent = true;
 
@@ -1483,11 +1460,11 @@ private function rateNodes(array $nodes): DOMDocument|bool
                 }
 
                 $import = $articleContent->importNode($sibling, true);
-                $articleContent->appendChild($import);
+                $articleElement->appendChild($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
+                 * children of the parent node, instead of using the Element childNodes function, which, when used
                  * along with appendChild, would shift the nodes position and the current foreach will behave in
                  * unpredictable ways.
                  */
@@ -1497,6 +1474,9 @@ private function rateNodes(array $nodes): DOMDocument|bool
         if ($hasContent) {
             $articleContent = $this->prepArticle($articleContent);
             // Find out text direction from ancestors of final top candidate.
+            if ($parentOfTopCandidate->nodeType === XML_HTML_DOCUMENT_NODE) {
+                $parentOfTopCandidate = $parentOfTopCandidate->documentElement;
+            }
             $ancestors = array_merge([$parentOfTopCandidate, $topCandidate], $parentOfTopCandidate->getNodeAncestors());
             foreach ($ancestors as $ancestor) {
                 $articleDir = $ancestor->getAttribute('dir');
@@ -1516,7 +1496,7 @@ private function rateNodes(array $nodes): DOMDocument|bool
     /**
      * Cleans up the final article.
      */
-    public function prepArticle(DOMDocument $article): DOMDocument
+    public function prepArticle(\Dom\HtmlDocument $article): \Dom\HtmlDocument
     {
         $this->logger->info('[PrepArticle] Preparing final article...');
 
@@ -1603,8 +1583,8 @@ public function prepArticle(DOMDocument $article): DOMDocument
         }
 
         // Remove single-cell tables
-        foreach ($article->shiftingAwareGetElementsByTagName('table') as $table) {
-            /** @var DOMElement $table */
+        foreach ($article->documentElement->shiftingAwareGetElementsByTagName('table') as $table) {
+            /** @var Element $table */
             $tbody = $table->hasSingleTagInsideElement('tbody') ? $table->firstElementChild : $table;
             if ($tbody->hasSingleTagInsideElement('tr')) {
                 $row = $tbody->firstElementChild;
@@ -1626,11 +1606,11 @@ public function prepArticle(DOMDocument $article): DOMDocument
      * similar checks as
      * https://dxr.mozilla.org/mozilla-central/rev/71224049c0b52ab190564d3ea0eab089a159a4cf/accessible/html/HTMLTableAccessible.cpp#920.
      */
-    public function _markDataTables(DOMDocument $article): void
+    public function _markDataTables(\Dom\HtmlDocument $article): void
     {
         $tables = $article->getElementsByTagName('table');
         foreach ($tables as $table) {
-            /** @var DOMElement $table */
+            /** @var Element $table */
             $role = $table->getAttribute('role');
             if ($role === 'presentation') {
                 $table->setReadabilityDataTable(false);
@@ -1680,7 +1660,7 @@ public function _markDataTables(DOMDocument $article): void
     /**
      * convert images and figures that have properties like data-src into images that can be loaded without JS
      */
-    public function _fixLazyImages(DOMDocument $article): void
+    public function _fixLazyImages(\Dom\HtmlDocument $article): void
     {
         $images = $this->_getAllNodesWithTag($article, ['img', 'picture', 'figure']);
         foreach ($images as $elem) {
@@ -1754,7 +1734,7 @@ public function _fixLazyImages(DOMDocument $article): void
     /**
      * Remove the style attribute on every e and under.
      **/
-    public function _cleanStyles(DOMDocument|DOMNode|DOMElement|DOMText $node): void
+    public function _cleanStyles(\Dom\HtmlDocument|Node|Element|Text $node): void
     {
         if (property_exists($node, 'tagName') && $node->tagName === 'svg') {
             return;
@@ -1785,10 +1765,10 @@ public function _cleanStyles(DOMDocument|DOMNode|DOMElement|DOMText $node): void
     /**
      * Clean out elements that match the specified conditions
      *
-     * @param $node DOMElement Node to clean
+     * @param $node Element Node to clean
      * @param $filter callable Function determines whether a node should be removed
      **/
-    public function _cleanMatchedNodes(DOMElement $node, callable $filter): void
+    public function _cleanMatchedNodes(Element $node, callable $filter): void
     {
         $endOfSearchMarkerNode = NodeUtility::getNextNode($node, true);
         $next = NodeUtility::getNextNode($node);
@@ -1805,7 +1785,7 @@ public function _cleanMatchedNodes(DOMElement $node, callable $filter): void
     /**
      * Clean extra paragraphs.
      */
-    public function _cleanExtraParagraphs(DOMDocument $article): void
+    public function _cleanExtraParagraphs(\Dom\HtmlDocument $article): void
     {
         $paragraphs = $this->_getAllNodesWithTag($article, ['p']);
         $length = count($paragraphs);
@@ -1843,10 +1823,10 @@ private function getTextDensity($e, array $tags) {
     /**
      * Clean conditionally.
      *
-     * @param DOMDocument $article
+     * @param HtmlDocument $article
      * @param string $tag Tag to clean conditionally
      */
-    public function _cleanConditionally(DOMDocument $article, string $tag): void
+    public function _cleanConditionally(\Dom\HtmlDocument $article, string $tag): void
     {
         if (!$this->configuration->getCleanConditionally()) {
             return;
@@ -1861,7 +1841,7 @@ public function _cleanConditionally(DOMDocument $article, string $tag): void
         $allNodesWithTag = $this->_getAllNodesWithTag($article, [$tag]);
         $length = count($allNodesWithTag);
         for ($i = 0; $i < $length; $i++) {
-            /** @var $node DOMElement */
+            /** @var $node Element */
             $node = $allNodesWithTag[$length - 1 - $i];
 
             $isList = in_array($tag, ['ul', 'ol']);
@@ -1978,10 +1958,10 @@ public function _getAllNodesWithTag($node, array $tagNames): array {
      * Clean a node of all elements of type "tag".
      * (Unless it's a youtube/vimeo video. People love movies.).
      *
-     * @param $article DOMDocument
+     * @param $article HtmlDocument
      * @param $tag string tag to clean
      **/
-    public function _clean(DOMDocument $article, string $tag): void
+    public function _clean(\Dom\HtmlDocument $article, string $tag): void
     {
         $isEmbed = in_array($tag, ['object', 'embed', 'iframe']);
 
@@ -2017,10 +1997,10 @@ public function _clean(DOMDocument $article, string $tag): void
     /**
      * Clean out spurious headers from an Element.
      **/
-    public function _cleanHeaders(DOMDocument $article): void
+    public function _cleanHeaders(\Dom\HtmlDocument $article): void
     {
         $headingNodes = $this->_getAllNodesWithTag($article, ['h1', 'h2']);
-        /** @var $header DOMElement */
+        /** @var $header Element */
         foreach ($headingNodes as $header) {
             $weight = 0;
             if ($this->configuration->getWeightClasses()) {
@@ -2029,7 +2009,7 @@ public function _cleanHeaders(DOMDocument $article): void
             $shouldRemove = $weight < 0;
 
             if ($shouldRemove) {
-                $this->logger->debug(sprintf('[PrepArticle] Removing H node with 0 or less weight. Content was: \'%s\'', substr($header->nodeValue, 0, 128)));
+                $this->logger->debug(sprintf('[PrepArticle] Removing H node with 0 or less weight. Content was: \'%s\'', substr($header->textContent, 0, 128)));
 
                 NodeUtility::removeNode($header);
             }
@@ -2040,10 +2020,10 @@ public function _cleanHeaders(DOMDocument $article): void
      * Check if this node is an H1 or H2 element whose content is mostly
      * the same as the article title.
      *
-     * @param DOMNode the node to check.
+     * @param Node the node to check.
      * @return boolean indicating whether this is a title-like header.
      */
-    private function headerDuplicatesTitle(DOMNode|DOMText|DOMElement $node): bool
+    private function headerDuplicatesTitle(Node|Text|Element $node): bool
     {
         if ($node->nodeName !== 'h1' && $node->nodeName !== 'h2') {
             return false;
@@ -2063,9 +2043,13 @@ private function headerDuplicatesTitle(DOMNode|DOMText|DOMElement $node): bool
      * 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.
      **/
-    public function _cleanClasses(DOMDocument|DOMText|DOMNode|DOMElement $node): void
+    public function _cleanClasses(\Dom\HtmlDocument|Text|Node|Element $node): void
     {
-        if ($node->hasAttribute('class')) {
+        if ($node->nodeType === XML_HTML_DOCUMENT_NODE) {
+            $node = $node->documentElement;
+        }
+
+        if ($node->nodeType === XML_ELEMENT_NODE && $node->hasAttribute('class')) {
             $node->removeAttribute('class');
         }
 
@@ -2077,14 +2061,14 @@ public function _cleanClasses(DOMDocument|DOMText|DOMNode|DOMElement $node): voi
     /**
      * Post process content.
      */
-    public function postProcessContent(DOMDocument $article): DOMDocument
+    public function postProcessContent(\Dom\HtmlDocument $article): \Dom\HtmlDocument
     {
         $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 */
+                /** @var Element $link */
                 $href = $link->getAttribute('href');
                 if ($href) {
                     // Remove links with javascript: URIs, since
@@ -2182,7 +2166,7 @@ public function postProcessContent(DOMDocument $article): DOMDocument
      * @param  array nodeList The NodeList.
      * @param  callable fn    The test function.
      */
-    private function findNode(array $nodeList, callable $fn): DOMNode|DOMText|DOMElement|null
+    private function findNode(array $nodeList, callable $fn): Node|Text|Element|null
     {
         foreach ($nodeList as $node) {
             if ($fn($node)) {
@@ -2221,20 +2205,18 @@ protected function setTitle(?string $title): void
      */
     public function getContent(): ?string
     {
-        if ($this->content instanceof DOMDocument) {
-            $html5 = new HTML5(['disable_html_ns' => true]);
-            // by using childNodes below we make sure HTML5PHP's serialiser
-            // doesn't output the <!DOCTYPE html> string at the start.
-            return $html5->saveHTML($this->content->childNodes);
+        if ($this->content instanceof \Dom\HtmlDocument) {
+            // Does this do HTML5 serialisation?
+            return $this->content->saveHtml();
         } else {
             return null;
         }
     }
 
     /**
-     * Return DOMDocument holding either only the content or the whole document.
+     * Return HtmlDocument holding either only the content or the whole document.
      */
-    public function getDOMDocument(bool $contentOnly = true): ?DOMDocument
+    public function getDOMDocument(bool $contentOnly = true): ?\Dom\HtmlDocument
     {
         if ($contentOnly) {
             return $this->content;
@@ -2246,7 +2228,7 @@ public function getDOMDocument(bool $contentOnly = true): ?DOMDocument
     /**
      * Set content.
      */
-    protected function setContent(DOMDocument $content): void
+    protected function setContent(\Dom\HtmlDocument $content): void
     {
         $this->content = $content;
     }
diff --git a/test/ConfigurationTest.php b/test/ConfigurationTest.php
index 74bcd69..72d2941 100644
--- a/test/ConfigurationTest.php
+++ b/test/ConfigurationTest.php
@@ -45,7 +45,6 @@ private function doEqualsAsserts(Configuration $config, array $options): void
         $this->assertEquals($options['weightClasses'], $config->getWeightClasses());
         $this->assertEquals($options['keepClasses'], $config->getKeepClasses());
         $this->assertEquals($options['fixRelativeURLs'], $config->getFixRelativeURLs());
-        $this->assertEquals($options['substituteEntities'], $config->getSubstituteEntities());
         $this->assertEquals($options['normalizeEntities'], $config->getNormalizeEntities());
         $this->assertEquals($options['originalURL'], $config->getOriginalURL());
         $this->assertEquals($options['summonCthulhu'], $config->getSummonCthulhu());
@@ -66,7 +65,6 @@ public static function getParams(): array
                 'cleanConditionally' => false,
                 'weightClasses' => false,
                 'fixRelativeURLs' => true,
-                'substituteEntities' => true,
                 'normalizeEntities' => true,
                 'originalURL' => 'my.original.url',
                 'summonCthulhu' => false,
diff --git a/test/ReadabilityTest.php b/test/ReadabilityTest.php
index 7073d79..ac8d149 100644
--- a/test/ReadabilityTest.php
+++ b/test/ReadabilityTest.php
@@ -23,7 +23,6 @@ public function testReadabilityParsesHTML(TestPage $testPage): void
     {
         $options = ['OriginalURL' => 'http://fakehost/test/test.html',
             'FixRelativeURLs' => true,
-            'SubstituteEntities' => true,
             'ArticleByline' => true
         ];
 
@@ -67,7 +66,6 @@ public function testReadabilityParsesMetadata(TestPage $testPage): void
     {
         $options = ['OriginalURL' => 'http://fakehost/test/test.html',
             'FixRelativeURLs' => true,
-            'SubstituteEntities' => true,
             'ArticleByline' => true
         ];
 
@@ -112,7 +110,6 @@ public function testHTMLParserParsesImages(TestPage $testPage): void
     {
         $options = ['OriginalURL' => 'http://fakehost/test/test.html',
             'FixRelativeURLs' => true,
-            'SubstituteEntities' => true,
         ];
 
         $configuration = new Configuration(array_merge($testPage->getConfiguration(), $options));
diff --git a/test/test-pages/marketwatch/config.json b/test/test-pages/marketwatch/config.json
index 91162d3..aa64637 100644
--- a/test/test-pages/marketwatch/config.json
+++ b/test/test-pages/marketwatch/config.json
@@ -1,4 +1,3 @@
 {
-  "ArticleByline": true,
-  "SubstituteEntities": true
+  "ArticleByline": true
 }

From d2ce5548b63efc2e373d4ee5c49764e0b15390cb Mon Sep 17 00:00:00 2001
From: "FiveFilters.org" <fivefilters@fivefilters.org>
Date: Tue, 3 Dec 2024 16:10:14 +0100
Subject: [PATCH 2/5] Update versions

---
 .github/workflows/main.yml | 2 +-
 composer.json              | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index b09c9c6..237c7a6 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -22,7 +22,7 @@ jobs:
 
     strategy:
       matrix:
-        php: ['8.1', '8.2', '8.3', '8.4']
+        php: ['8.4']
         libxml: ['2.9.14']
 
     # Steps represent a sequence of tasks that will be executed as part of the job
diff --git a/composer.json b/composer.json
index f573695..fb682c3 100644
--- a/composer.json
+++ b/composer.json
@@ -27,7 +27,7 @@
         "psr-4": {"fivefilters\\Readability\\Test\\": "test"}
     },
     "require": {
-        "php": ">=8.1",
+        "php": ">=8.4",
         "ext-dom": "*",
         "ext-xml": "*",
         "ext-mbstring": "*",

From c8a9e67b0917ae333cdb22e50e535443270b4aea Mon Sep 17 00:00:00 2001
From: "FiveFilters.org" <fivefilters@fivefilters.org>
Date: Sat, 7 Dec 2024 00:52:08 +0100
Subject: [PATCH 3/5] Use constant for arrays that don't change

---
 src/Nodes/NodeTrait.php | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/src/Nodes/NodeTrait.php b/src/Nodes/NodeTrait.php
index 1eeb511..c15ed14 100644
--- a/src/Nodes/NodeTrait.php
+++ b/src/Nodes/NodeTrait.php
@@ -30,7 +30,7 @@ trait NodeTrait
      */
     private bool $readabilityDataTable = false;
 
-    private array $divToPElements = [
+    private const DIV_TO_P_ELEMENTS = [
         'blockquote',
         'dl',
         'div',
@@ -46,7 +46,7 @@ trait NodeTrait
      * The commented out elements qualify as phrasing content but tend to be
      * removed by readability when put into paragraphs, so we ignore them here.
      */
-    private array $phrasing_elems = [
+    private const PHRASING_ELEMS = [
         // 'CANVAS', 'IFRAME', 'SVG', 'VIDEO',
         'abbr', 'audio', 'b', 'bdo', 'br', 'button', 'cite', 'code', 'data',
         'datalist', 'dfn', 'em', 'embed', 'i', 'img', 'input', 'kbd', 'label',
@@ -367,14 +367,14 @@ public function hasSingleTagInsideElement(string $tag): bool
 
     /**
      * Check if the current element has a single child block element.
-     * Block elements are the ones defined in the divToPElements array.
+     * Block elements are the ones defined in the DIV_TO_P_ELEMENTS array.
      */
     public function hasSingleChildBlockElement(): bool
     {
         $result = false;
         if ($this->hasChildNodes()) {
             foreach ($this->childNodes as $child) {
-                if (in_array($child->nodeName, $this->divToPElements)) {
+                if (in_array($child->nodeName, self::DIV_TO_P_ELEMENTS)) {
                     $result = true;
                 } else {
                     // If any of the hasSingleChildBlockElement calls return true, return true then.
@@ -417,7 +417,7 @@ public function isElementWithoutContent(): bool
      */
     public function isPhrasingContent(): bool
     {
-        return $this->nodeType === XML_TEXT_NODE || in_array($this->nodeName, $this->phrasing_elems) !== false ||
+        return $this->nodeType === XML_TEXT_NODE || in_array($this->nodeName, self::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) {

From 8574693e62da84dd10ffedacb9bd10583b020344 Mon Sep 17 00:00:00 2001
From: "FiveFilters.org" <fivefilters@fivefilters.org>
Date: Sat, 7 Dec 2024 02:09:52 +0100
Subject: [PATCH 4/5] Use new rename() method

---
 src/Nodes/NodeUtility.php | 29 +++--------------------------
 1 file changed, 3 insertions(+), 26 deletions(-)

diff --git a/src/Nodes/NodeUtility.php b/src/Nodes/NodeUtility.php
index a6d63e8..90cc079 100644
--- a/src/Nodes/NodeUtility.php
+++ b/src/Nodes/NodeUtility.php
@@ -65,34 +65,11 @@ public static function nextNode(Node|Comment|Text|Element|null $node): Node|Comm
     }
 
     /**
-     * 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.
+     * Not in the DOM spec, but PHP 8.4 introduced rename() for DOM\Element and DOM\Attr
      */
-    public static function setNodeTag(Node|Element $node, string $value, bool $importAttributes = true): Node|Element
+    public static function setNodeTag(Element $element, string $newName): void
     {
-        $new = \Dom\HtmlDocument::createEmpty();
-        NodeUtility::registerReadabilityNodeClasses($new);
-        $new->appendChild($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;
+        $element->rename($element->namespaceURI, $newName);
     }
 
     /**

From d434a01ec8de1806b9551bb63f71be723dbc4b4a Mon Sep 17 00:00:00 2001
From: "FiveFilters.org" <fivefilters@fivefilters.org>
Date: Sat, 7 Dec 2024 02:10:28 +0100
Subject: [PATCH 5/5] Fixes

---
 src/Readability.php | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/src/Readability.php b/src/Readability.php
index a2ff53b..2599a23 100644
--- a/src/Readability.php
+++ b/src/Readability.php
@@ -275,7 +275,7 @@ public function loadHTML(string $html): void
         if ($this->configuration->getParser() === 'html5') {
             $this->logger->debug('[Loading] Using HTML5 parser...');
             // New DOM class with HTML5 parser introduced in PHP 8.4
-            $dom = \Dom\HtmlDocument::createFromString($html, \Dom\HTML_NO_DEFAULT_NS | LIBXML_NOERROR);
+            $dom = \Dom\HtmlDocument::createFromString($html, LIBXML_NOERROR);
             NodeUtility::registerReadabilityNodeClasses($dom);
             //TODO: Improve this so it looks inside <html><head><base>, not just any <base>
             $base = $dom->getElementsByTagName('base');
@@ -897,7 +897,7 @@ private function getNodes(Node|Comment|Text|Element|null $node): array
                     $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->textContent, 0, 128)));
-                    $node = NodeUtility::setNodeTag($node, 'p');
+                    NodeUtility::setNodeTag($node, 'p');
                     $elementsToScore[] = $node;
                 }
             }
@@ -1368,7 +1368,7 @@ private function rateNodes(array $nodes): \Dom\HtmlDocument|bool
             $scoreThreshold = $lastScore / 3;
 
             /* @var Element $parentOfTopCandidate */
-            while ($parentOfTopCandidate && $parentOfTopCandidate->nodeName !== 'body') {
+            while ($parentOfTopCandidate && $parentOfTopCandidate->nodeName !== 'body' && $parentOfTopCandidate->nodeType === XML_ELEMENT_NODE) {
                 $parentScore = $parentOfTopCandidate->contentScore;
                 if ($parentScore < $scoreThreshold) {
                     break;
@@ -1456,7 +1456,7 @@ private function rateNodes(array $nodes): \Dom\HtmlDocument|bool
                      * 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');
+                    NodeUtility::setNodeTag($sibling, 'div');
                 }
 
                 $import = $articleContent->importNode($sibling, true);
@@ -1590,7 +1590,7 @@ public function prepArticle(\Dom\HtmlDocument $article): \Dom\HtmlDocument
                 $row = $tbody->firstElementChild;
                 if ($row->hasSingleTagInsideElement('td')) {
                     $cell = $row->firstElementChild;
-                    $cell = NodeUtility::setNodeTag($cell, (array_reduce(iterator_to_array($cell->childNodes), function ($carry, $node) {
+                    NodeUtility::setNodeTag($cell, (array_reduce(iterator_to_array($cell->childNodes), function ($carry, $node) {
                         return $node->isPhrasingContent() && $carry;
                     }, true)) ? 'p' : 'div');
                     $table->parentNode->replaceChild($cell, $table);