From 963c3e1abac4a39733d75059e1dc36813e861387 Mon Sep 17 00:00:00 2001
From: Anatoly Pashin <anatoly.pashin@gmail.com>
Date: Fri, 10 Jul 2020 18:09:23 +1000
Subject: [PATCH] More tries to detect the encoding + add the test

---
 src/DOMNodeComparator.php       | 19 ++++++++++++++++++-
 tests/DOMNodeComparatorTest.php | 26 ++++++++++++++++++++++++++
 2 files changed, 44 insertions(+), 1 deletion(-)

diff --git a/src/DOMNodeComparator.php b/src/DOMNodeComparator.php
index ae5da1d..a6c9155 100644
--- a/src/DOMNodeComparator.php
+++ b/src/DOMNodeComparator.php
@@ -9,6 +9,7 @@
  */
 namespace SebastianBergmann\Comparator;
 
+use DOMEntity;
 use function sprintf;
 use function strtolower;
 use DOMDocument;
@@ -72,7 +73,8 @@ private function nodeToText(DOMNode $node, bool $canonicalize, bool $ignoreCase)
         if ($canonicalize) {
             $document = new DOMDocument;
             @$document->loadXML(
-                '<?xml version="1.0" encoding="' . ($node->ownerDocument->encoding ?: 'UTF-8') . '"?>' . \PHP_EOL
+                ''
+                . '<?xml version="1.0" encoding="' . $this->getNodeEncoding($node) . '"?>' . \PHP_EOL
                 . $node->C14N()
             );
 
@@ -88,4 +90,19 @@ private function nodeToText(DOMNode $node, bool $canonicalize, bool $ignoreCase)
 
         return $ignoreCase ? strtolower($text) : $text;
     }
+
+    /**
+     * Tries to get the encoding from DOMNode or falls back to UTF-8
+     */
+    private function getNodeEncoding(DOMNode $node): string
+    {
+        $encoding = '';
+        if ($node->ownerDocument) {
+            $encoding = $node->ownerDocument->xmlEncoding ?: $node->ownerDocument->encoding;
+        } elseif ($node instanceof DOMEntity) {
+            $encoding = $node->encoding;
+        }
+
+        return $encoding ?: 'UTF-8';
+    }
 }
diff --git a/tests/DOMNodeComparatorTest.php b/tests/DOMNodeComparatorTest.php
index 8d52fc2..5dc9b7a 100644
--- a/tests/DOMNodeComparatorTest.php
+++ b/tests/DOMNodeComparatorTest.php
@@ -84,6 +84,14 @@ public function assertEqualsSucceedsProvider()
                 $this->createDOMDocument("<a x='' a=''/>"),
                 $this->createDOMDocument("<a a='' x=''/>"),
             ],
+            [
+                $this->createDOMDocument('<root>Тест кириллицы</root>'),
+                $this->createDOMDocument('<root>Тест кириллицы</root>'),
+            ],
+            [
+                $this->createDOMDocument('<root>Тест кириллицы</root>'),
+                $this->createDOMDocument('<root>&#x422;&#x435;&#x441;&#x442; &#x43A;&#x438;&#x440;&#x438;&#x43B;&#x43B;&#x438;&#x446;&#x44B;</root>'),
+            ],
         ];
     }
 
@@ -169,6 +177,24 @@ public function testAssertEqualsFails($expected, $actual): void
         $this->comparator->assertEquals($expected, $actual);
     }
 
+    public function testCyrillicText(): void
+    {
+        try {
+            $this->comparator->assertEquals(
+                $this->createDOMDocument('<root>Тест кириллицы</root>'),
+                $this->createDOMDocument('<root>Тест кириллицы 1</root>'),
+            );
+        } catch (ComparisonFailure $exception) {
+            self::assertStringContainsString('Тест кириллицы', $exception->getExpectedAsString());
+            self::assertStringContainsString('Тест кириллицы', $exception->getActualAsString());
+
+            return;
+        }
+
+        // should not happen
+        self::assertFalse(false);
+    }
+
     private function createDOMDocument($content)
     {
         $document                     = new DOMDocument;