Add xml declaration to force utf-8

sebastianbergmann · Jul 10, 2020 · f6f39fd · f6f39fd
1 parent 988baa0
commit f6f39fd
Show file tree

Hide file tree

Showing 2 changed files with 47 additions and 1 deletion.
diff --git a/src/DOMNodeComparator.php b/src/DOMNodeComparator.php
@@ -13,6 +13,7 @@
 use function strtolower;
 use DOMDocument;
 use DOMNode;
+use DOMEntity;
 
 /**
  * Compares DOMNode instances for equality.
@@ -71,7 +72,11 @@ private function nodeToText(DOMNode $node, bool $canonicalize, bool $ignoreCase)
     {
         if ($canonicalize) {
             $document = new DOMDocument;
-            @$document->loadXML($node->C14N());
+            @$document->loadXML(
+                ''
+                . '<?xml version="1.0" encoding="' . $this->getNodeEncoding($node) . '"?>' . PHP_EOL
+                . $node->C14N()
+            );
 
             $node = $document;
         }
@@ -85,4 +90,20 @@ 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
@@ -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 +177,23 @@ 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', $exception->getActualAsString());
+
+            return;
+        }
+
+        self::fail('Should not happen');
+    }
+
     private function createDOMDocument($content)
     {
         $document                     = new DOMDocument;