[DomCrawler] Added test for supported encodings by mbstring

symfony · Sep 25, 2012 · d7623ae · d7623ae
1 parent 237629a
commit d7623ae
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 1 deletion.
diff --git a/src/Symfony/Component/DomCrawler/Crawler.php b/src/Symfony/Component/DomCrawler/Crawler.php
@@ -132,7 +132,7 @@ public function addHtmlContent($content, $charset = 'UTF-8')
         $dom = new \DOMDocument('1.0', $charset);
         $dom->validateOnParse = true;
 
-        if (function_exists('mb_convert_encoding')) {
+        if (function_exists('mb_convert_encoding') && in_array(strtolower($charset), array_map('strtolower', mb_list_encodings()))) {
             $content = mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
         }
 

diff --git a/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php b/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php
@@ -80,6 +80,17 @@ public function testAddHtmlContentCharset()
         $this->assertEquals('Tiếng Việt', $crawler->filterXPath('//div')->text());
     }
 
+    /**
+     * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent
+     */
+    public function testAddHtmlContentUnsupportedCharset()
+    {
+        $crawler = new Crawler();
+        $crawler->addHtmlContent(file_get_contents(__DIR__.'/Fixtures/windows-1250.html'), 'Windows-1250');
+
+        $this->assertEquals('Žťčýů', $crawler->filterXPath('//p')->text());
+    }
+
     /**
      * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent
      */

diff --git a/src/Symfony/Component/DomCrawler/Tests/Fixtures/windows-1250.html b/src/Symfony/Component/DomCrawler/Tests/Fixtures/windows-1250.html
@@ -0,0 +1,8 @@
+<html>
+    <head>
+        <meta http-equiv="content-type" content="text/html;charset=windows-1250">
+    </head>
+    <body>
+        <p>Žèýù</p>
+    </body>
+</html>