From 751ea8be747a9ccc15b24ed3724ff1374961938f Mon Sep 17 00:00:00 2001
From: osapon <osa@osa-p.net>
Date: Mon, 25 Nov 2024 11:58:17 +0900
Subject: [PATCH] Improved charset tag recognition accuracy.

---
 src/Document.php | 23 ++++++++++++++++++++---
 1 file changed, 20 insertions(+), 3 deletions(-)

diff --git a/src/Document.php b/src/Document.php
index 6db07663..c61e5c62 100644
--- a/src/Document.php
+++ b/src/Document.php
@@ -28,14 +28,31 @@ public function __construct(Extractor $extractor)
 
         $encoding = null;
         $contentType = $extractor->getResponse()->getHeaderLine('content-type');
-        preg_match('/charset="?(.*?)(?=$|\s|;|")/i', $contentType, $match);
+        preg_match('/charset=(?:"|\')?(.*?)(?=$|\s|;|"|\'|>)/i', $contentType, $match);
         if (!empty($match[1])) {
             $encoding = trim($match[1], ',');
-        } elseif (!empty($html)) {
-            preg_match('/charset="?(.*?)(?=$|\s|;|")/i', $html, $match);
+            try {
+                $ret = mb_encoding_aliases($encoding);
+                if ($ret === false) {
+                    $encoding = null;
+                }
+            } catch (\ValueError $exception) {
+                $encoding = null;
+            }
+        }
+        if (is_null($encoding) && !empty($html)) {
+            preg_match('/charset=(?:"|\')?(.*?)(?=$|\s|;|"|\'|>)/i', $html, $match);
             if (!empty($match[1])) {
                 $encoding = trim($match[1], ',');
             }
+            try {
+                $ret = mb_encoding_aliases($encoding);
+                if ($ret === false) {
+                    $encoding = null;
+                }
+            } catch (\ValueError $exception) {
+                $encoding = null;
+            }
         }
         $this->document = !empty($html) ? Parser::parse($html, $encoding) : new DOMDocument();
         $this->initXPath();