Force UTF-8 encoding for better support for non-Latin character sets

Author: stevegrunwell

.gitignore

@@ -1,6 +1,7 @@
 tests/coverage
 vendor
 .phpunit.result.cache
+.vscode
 
 # The composer.lock file is not needed, as this is a library whose dependencies
 # will depend on the version of PHP being used.

src/MarkupAssertionsTrait.php

@@ -222,7 +222,11 @@ public function assertElementNotRegExp($regexp, $selector = '', $markup = '', $m
      */
     protected function executeDomQuery($markup, $query)
     {
-        return Query::execute($query, new Document($markup), Query::TYPE_CSS);
+        return Query::execute(
+            $query,
+            new Document('<?xml encoding="UTF-8">' . $markup, Document::DOC_HTML, 'UTF-8'),
+            Query::TYPE_CSS
+        );
     }
 
     /**
@@ -274,7 +278,7 @@ protected function getInnerHtmlOfMatchedElements($markup, $query)
             $document = new DOMDocument();
             $document->appendChild($document->importNode($result->firstChild, true));
 
-            $contents[] = trim($document->saveHTML());
+            $contents[] = trim(html_entity_decode($document->saveHTML()));
         }
 
         return implode(PHP_EOL, $contents);

tests/MarkupAssertionsTraitTest.php

@@ -153,6 +153,21 @@ public function assertElementContains_should_scope_matches_to_selector()
         );
     }
 
+    /**
+     * @test
+     * @testdox assertElementContains() should handle various character sets
+     * @dataProvider provideGreetingsInDifferentLanguages
+     * @ticket https://github.com/stevegrunwell/phpunit-markup-assertions/issues/31
+     */
+    public function assertElementContains_should_handle_various_character_sets($greeting)
+    {
+        $this->assertElementContains(
+            $greeting,
+            'h1',
+            sprintf('<div><h1>%s</h1></div>', $greeting)
+        );
+    }
+
     /**
      * @test
      * @testdox assertElementNotContains() should be able to search for a selector
@@ -162,7 +177,22 @@ public function assertElementNotContains_can_match_a_selector()
         $this->assertElementNotContains(
             'ipsum',
             '#main',
-            '<header>Foo bar baz</header><div id="main">Some string</div>'
+            '<div>Foo bar baz</div><div id="main">Some string</div>'
+        );
+    }
+
+    /**
+     * @test
+     * @testdox assertElementNotContains() should handle various character sets
+     * @dataProvider provideGreetingsInDifferentLanguages
+     * @ticket https://github.com/stevegrunwell/phpunit-markup-assertions/issues/31
+     */
+    public function assertElementNotContains_should_handle_various_character_sets($greeting)
+    {
+        $this->assertElementNotContains(
+            $greeting,
+            'h1',
+            sprintf('<h1>Translation</h1><p>%s</p>', $greeting)
         );
     }
 
@@ -328,4 +358,23 @@ public function provideSelectorVariants()
             'Tag with href attribute' => ['a[href="https://example.com"]'],
         ];
     }
+
+    /**
+     * Provide a list of strings in various language.
+     *
+     * @return array<string,array<string>>
+     */
+    public function provideGreetingsInDifferentLanguages()
+    {
+        return [
+            'Arabic'    => ['مرحبا!'],
+            'Chinese'   => ['你好'],
+            'English'   => ['Hello'],
+            'Hebrew'    => ['שלום'],
+            'Japanese'  => ['こんにちは'],
+            'Korean'    => ['안녕하십니까'],
+            'Punjabi'   => ['ਸਤ ਸ੍ਰੀ ਅਕਾਲ'],
+            'Ukrainian' => ['Привіт'],
+        ];
+    }
 }