fix #9321 Crawler::addHtmlContent add gbk encoding support

bronze1man · fabpot · commit acb2df08428b · 2013-12-29T19:30:30.000+01:00
diff --git a/src/Symfony/Component/DomCrawler/Crawler.php b/src/Symfony/Component/DomCrawler/Crawler.php
@@ -147,8 +147,18 @@ public function addHtmlContent($content, $charset = 'UTF-8')
         $dom = new \DOMDocument('1.0', $charset);
         $dom->validateOnParse = true;
 
-        if (function_exists('mb_convert_encoding') && in_array(strtolower($charset), array_map('strtolower', mb_list_encodings()))) {
-            $content = mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
+        if (function_exists('mb_convert_encoding')) {
+            $has_error = false;
+            $previous = set_error_handler(function()use(&$has_error){
+                $has_error = true;
+            });
+            $tmpContent = @mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
+
+            set_error_handler($previous);
+
+            if (!$has_error) {
+                $content = $tmpContent;
+            }
         }
 
         @$dom->loadHTML($content);
diff --git a/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php b/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php
@@ -112,6 +112,18 @@ public function testAddHtmlContentUnsupportedCharset()
         $this->assertEquals('Žťčýů', $crawler->filterXPath('//p')->text());
     }
 
+    /**
+     * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent
+     */
+    public function testAddHtmlContentCharsetGbk()
+    {
+        $crawler = new Crawler();
+        //gbk encode of <html><p>中文</p></html>
+        $crawler->addHtmlContent(base64_decode('PGh0bWw+PHA+1tDOxDwvcD48L2h0bWw+'), 'gbk');
+
+        $this->assertEquals('中文', $crawler->filterXPath('//p')->text());
+    }
+
     /**
      * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent
      */
