[DomCrawler] added a way to get parsing errors for Crawler::addHtmlCo…

…ntent() and Crawler::addXmlContent() via libxml functions
symfony · Sep 28, 2011 · a57a4af · a57a4af
1 parent 382a421
commit a57a4af
Show file tree

Hide file tree

Showing 3 changed files with 75 additions and 0 deletions.
diff --git a/CHANGELOG-2.1.md b/CHANGELOG-2.1.md
@@ -35,6 +35,7 @@ To get the diff between two versions, go to https://github.com/symfony/symfony/c
 
 ### DomCrawler
 
+ * added a way to get parsing errors for Crawler::addHtmlContent() and Crawler::addXmlContent() via libxml functions
  * added support for submitting a form without a submit button
 
 ### Finder

diff --git a/src/Symfony/Component/DomCrawler/Crawler.php b/src/Symfony/Component/DomCrawler/Crawler.php
@@ -107,6 +107,13 @@ public function addContent($content, $type = null)
     /**
      * Adds an HTML content to the list of nodes.
      *
+     * The libxml errors are disabled when the content is parsed.
+     *
+     * If you want to get parsing errors, be sure to enable
+     * internal errors via libxml_use_internal_errors(true)
+     * and then, get the errors via libxml_get_errors(). Be
+     * sure to clear errors with libxml_clear_errors() afterward.
+     *
      * @param string $content The HTML content
      * @param string $charset The charset
      *
@@ -117,7 +124,10 @@ public function addHtmlContent($content, $charset = 'UTF-8')
         $dom = new \DOMDocument('1.0', $charset);
         $dom->validateOnParse = true;
 
+        $current = libxml_use_internal_errors(true);
         @$dom->loadHTML($content);
+        libxml_use_internal_errors($current);
+
         $this->addDocument($dom);
 
         $base = $this->filter('base')->extract(array('href'));
@@ -130,6 +140,13 @@ public function addHtmlContent($content, $charset = 'UTF-8')
     /**
      * Adds an XML content to the list of nodes.
      *
+     * The libxml errors are disabled when the content is parsed.
+     *
+     * If you want to get parsing errors, be sure to enable
+     * internal errors via libxml_use_internal_errors(true)
+     * and then, get the errors via libxml_get_errors(). Be
+     * sure to clear errors with libxml_clear_errors() afterward.
+     *
      * @param string $content The XML content
      * @param string $charset The charset
      *
@@ -141,7 +158,10 @@ public function addXmlContent($content, $charset = 'UTF-8')
         $dom->validateOnParse = true;
 
         // remove the default namespace to make XPath expressions simpler
+        $current = libxml_use_internal_errors(true);
         @$dom->loadXML(str_replace('xmlns', 'ns', $content));
+        libxml_use_internal_errors($current);
+
         $this->addDocument($dom);
     }
 

diff --git a/tests/Symfony/Tests/Component/DomCrawler/CrawlerTest.php b/tests/Symfony/Tests/Component/DomCrawler/CrawlerTest.php
@@ -69,6 +69,34 @@ public function testAddHtmlContent()
         $this->assertEquals('http://symfony.com/contact', $crawler->filter('a')->link()->getUri(), '->addHtmlContent() adds nodes from an HTML string');
     }
 
+    /**
+     * @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent
+     */
+    public function testAddHtmlContentWithErrors()
+    {
+        libxml_use_internal_errors(true);
+
+        $crawler = new Crawler();
+        $crawler->addHtmlContent(<<<EOF
+<!DOCTYPE html>
+<html>
+    <head>
+    </head>
+    <body>
+        <nav><a href="#"><a href="#"></nav>
+    </body>
+</html>
+EOF
+        , 'UTF-8');
+
+        $errors = libxml_get_errors();
+        $this->assertEquals(1, count($errors));
+        $this->assertEquals("Tag nav invalid\n", $errors[0]->message);
+
+        libxml_clear_errors();
+        libxml_use_internal_errors(false);
+    }
+
     /**
      * @covers Symfony\Component\DomCrawler\Crawler::addXmlContent
      */
@@ -80,6 +108,32 @@ public function testAddXmlContent()
         $this->assertEquals('foo', $crawler->filter('div')->attr('class'), '->addXmlContent() adds nodes from an XML string');
     }
 
+    /**
+     * @covers Symfony\Component\DomCrawler\Crawler::addXmlContent
+     */
+    public function testAddXmlContentWithErrors()
+    {
+        libxml_use_internal_errors(true);
+
+        $crawler = new Crawler();
+        $crawler->addXmlContent(<<<EOF
+<!DOCTYPE html>
+<html>
+    <head>
+    </head>
+    <body>
+        <nav><a href="#"><a href="#"></nav>
+    </body>
+</html>
+EOF
+        , 'UTF-8');
+
+        $this->assertTrue(count(libxml_get_errors()) > 1);
+
+        libxml_clear_errors();
+        libxml_use_internal_errors(false);
+    }
+
     /**
      * @covers Symfony\Component\DomCrawler\Crawler::addContent
      */