fivefilters · fivefilters · Dec 3, 2024 · Dec 3, 2024 · Dec 6, 2024 · Dec 7, 2024
diff --git 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
@@ -27,12 +27,11 @@
         "psr-4": {"fivefilters\\Readability\\Test\\": "test"}
     },
     "require": {
-        "php": ">=8.1",
+        "php": ">=8.4",
         "ext-dom": "*",
         "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
@@ -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 → src/Nodes/DOM/Attr.php b/src/Nodes/DOM/DOMAttr.php → 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 → src/Nodes/DOM/CdataSection.php b/src/Nodes/DOM/DOMNotation.php → 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 → src/Nodes/DOM/CharacterData.php b/src/Nodes/DOM/DOMCdataSection.php → 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 → src/Nodes/DOM/Comment.php b/src/Nodes/DOM/DOMEntity.php → 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
diff --git a/src/Nodes/DOM/DOMDocumentFragment.php b/src/Nodes/DOM/DOMDocumentFragment.php
diff --git a/src/Nodes/DOM/DOMDocumentType.php b/src/Nodes/DOM/DOMDocumentType.php
diff --git a/src/Nodes/DOM/DOMEntityReference.php b/src/Nodes/DOM/DOMEntityReference.php
diff --git a/src/Nodes/DOM/DOMProcessingInstruction.php b/src/Nodes/DOM/DOMProcessingInstruction.php
diff --git a/src/Nodes/DOM/DocumentFragment.php 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
@@ -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 → src/Nodes/DOM/Element.php b/src/Nodes/DOM/DOMElement.php → 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 → src/Nodes/DOM/Entity.php b/src/Nodes/DOM/DOMComment.php → 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 → src/Nodes/DOM/EntityReference.php b/src/Nodes/DOM/DOMCharacterData.php → 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 → src/Nodes/DOM/Node.php b/src/Nodes/DOM/DOMNode.php → 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 → src/Nodes/DOM/NodeList.php b/src/Nodes/DOM/DOMNodeList.php → 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
@@ -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
@@ -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 → src/Nodes/DOM/Text.php b/src/Nodes/DOM/DOMText.php → 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;
 }