From 2af5077bc86b290ef54fd35a233354d672d10b97 Mon Sep 17 00:00:00 2001
From: Sander de Jong <sander@recras.nl>
Date: Tue, 20 Dec 2022 12:08:19 +0100
Subject: [PATCH] Fix multibyte string conversion

Refs #237
---
 src/CssToInlineStyles.php       |  2 +-
 tests/CssToInlineStylesTest.php | 40 +++++++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 1 deletion(-)

diff --git a/src/CssToInlineStyles.php b/src/CssToInlineStyles.php
index 6efc06f..e40ae25 100644
--- a/src/CssToInlineStyles.php
+++ b/src/CssToInlineStyles.php
@@ -113,7 +113,7 @@ protected function createDomDocumentFromHtml($html)
     {
         $document = new \DOMDocument('1.0', 'UTF-8');
         $internalErrors = libxml_use_internal_errors(true);
-        $document->loadHTML(mb_encode_numericentity($html, [0x80, 0xFFFF, 0, 0xFFFF], 'UTF-8'));
+        $document->loadHTML(mb_encode_numericentity($html, [0x80, 0x10FFFF, 0, 0x1FFFFF], 'UTF-8'));
         libxml_use_internal_errors($internalErrors);
         $document->formatOutput = true;
 
diff --git a/tests/CssToInlineStylesTest.php b/tests/CssToInlineStylesTest.php
index ed3e573..5c77815 100644
--- a/tests/CssToInlineStylesTest.php
+++ b/tests/CssToInlineStylesTest.php
@@ -267,6 +267,46 @@ public function testSelfClosingTags()
         $this->assertCorrectConversion($expected, $html, $css);
     }
 
+    public function testConversionAsciiRegular()
+    {
+        $html = '~';
+        $css = '';
+        $expected = "<p>{$html}</p>";
+        $this->assertCorrectConversion($expected, $html, $css);
+    }
+
+    public function testConversionAsciiDelete()
+    {
+        $html = "\u{007F}";
+        $css = '';
+        $expected = "<p>{$html}</p>";
+        $this->assertCorrectConversion($expected, $html, $css);
+    }
+
+    public function testConversionLowestCodepoint()
+    {
+        $html = "\u{0080}";
+        $css = '';
+        $expected = "<p>{$html}</p>";
+        $this->assertCorrectConversion($expected, $html, $css);
+    }
+
+    public function testConversionHighestCodepoint()
+    {
+        $html = "\u{10FFFF}";
+        $css = '';
+        $expected = "<p>{$html}</p>";
+        $this->assertCorrectConversion($expected, $html, $css);
+    }
+
+    public function testMB4character()
+    {
+        $html = '🇳🇱';
+        $css = '';
+        $expected = "<p>{$html}</p>";
+        $this->assertCorrectConversion($expected, $html, $css);
+    }
+
     private function assertCorrectConversion($expected, $html, $css = null)
     {
         $this->assertEquals(