From 952588ee1f883ec6ac0993a5916747ee30c42046 Mon Sep 17 00:00:00 2001
From: Jeremy Benoist <jeremy.benoist@gmail.com>
Date: Fri, 27 Mar 2026 20:34:45 +0100
Subject: [PATCH] Fix bad encoding for `tidy_repair_string`

Tidy on PHP 8.5 is more restrictive on what can be given as input encoding. Before, it worked as an unknown value was converted to `utf8`.
---
 src/Readability.php       | 2 +-
 tests/ReadabilityTest.php | 1 +
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/Readability.php b/src/Readability.php
index bb9e9f6..92314c5 100644
--- a/src/Readability.php
+++ b/src/Readability.php
@@ -118,7 +118,7 @@ class Readability implements LoggerAwareInterface
         'enclose-text' => true,
         'merge-divs' => true,
         // 'merge-spans' => true,
-        'input-encoding' => '????',
+        'input-encoding' => 'utf8',
         'output-encoding' => 'utf8',
         'hide-comments' => true,
     ];
diff --git a/tests/ReadabilityTest.php b/tests/ReadabilityTest.php
index df8a13b..6e6e448 100644
--- a/tests/ReadabilityTest.php
+++ b/tests/ReadabilityTest.php
@@ -17,6 +17,7 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
     public function testConstructDefault()
     {
         $readability = $this->getReadability('');
+        $this->assertSame('utf8', $readability->tidy_config['input-encoding']);
 
         $this->assertNull($readability->url);
         $this->assertInstanceOf(\DOMDocument::class, $readability->dom);