[DomCrawler] fixed encoding when using addHtmlContent() (fixes #3881)

fabpot · fabpot · commit c9ebe67d6588 · 2012-05-07T14:20:03.000+02:00
diff --git a/src/Symfony/Component/DomCrawler/Crawler.php b/src/Symfony/Component/DomCrawler/Crawler.php
@@ -129,6 +129,10 @@ public function addHtmlContent($content, $charset = 'UTF-8')
         $dom = new \DOMDocument('1.0', $charset);
         $dom->validateOnParse = true;
 
+        if (function_exists('mb_convert_encoding')) {
+            $content = mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
+        }
+
         $current = libxml_use_internal_errors(true);
         @$dom->loadHTML($content);
         libxml_use_internal_errors($current);
diff --git a/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php b/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php
@@ -69,6 +69,17 @@ public function testAddHtmlContent()
         $this->assertEquals('http://symfony.com/contact', $crawler->filterXPath('//a')->link()->getUri(), '->addHtmlContent() adds nodes from an HTML string');
     }
 
+    /**
+     * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent
+     */
+    public function testAddHtmlContentCharset()
+    {
+        $crawler = new Crawler();
+        $crawler->addHtmlContent('<html><div class="foo">Tiếng Việt</html>', 'UTF-8');
+
+        $this->assertEquals('Tiếng Việt', $crawler->filterXPath('//div')->text());
+    }
+
     /**
      * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent
      */
@@ -108,6 +119,17 @@ public function testAddXmlContent()
         $this->assertEquals('foo', $crawler->filterXPath('//div')->attr('class'), '->addXmlContent() adds nodes from an XML string');
     }
 
+    /**
+     * @covers Symfony\Component\DomCrawler\Crawler::addXmlContent
+     */
+    public function testAddXmlContentCharset()
+    {
+        $crawler = new Crawler();
+        $crawler->addXmlContent('<html><div class="foo">Tiếng Việt</div></html>', 'UTF-8');
+
+        $this->assertEquals('Tiếng Việt', $crawler->filterXPath('//div')->text());
+    }
+
     /**
      * @covers Symfony\Component\DomCrawler\Crawler::addXmlContent
      */
