[DomCrawler] Added support for an automatic default namespace registration.

jakzal · jakzal · commit c6fbb13938e1 · 2013-09-22T23:05:57.000+01:00
diff --git a/src/Symfony/Component/DomCrawler/Crawler.php b/src/Symfony/Component/DomCrawler/Crawler.php
@@ -577,16 +577,8 @@ public function filterXPath($xpath)
             $root->appendChild($document->importNode($node, true));
         }
 
-        $domxpath = new \DOMXPath($document);
-        if (preg_match_all('/(?P<prefix>[a-zA-Z_][a-zA-Z_0-9\-\.]+):[^:]/', $xpath, $matches)) {
-            foreach ($matches['prefix'] as $prefix) {
-                // ask for one namespace, otherwise we'd get a collection with an item for each node
-                $namespaces = $domxpath->query(sprintf('(//namespace::*[name()="%s"])[last()]', $prefix));
-                foreach ($namespaces as $node) {
-                    $domxpath->registerNamespace($node->prefix, $node->nodeValue);
-                }
-            }
-        }
+        $prefixes = $this->findNamespacePrefixes($xpath);
+        $domxpath = $this->createDOMXPath($document, $prefixes);
 
         return new static($domxpath->query($xpath), $this->uri);
     }
@@ -799,4 +791,43 @@ protected function sibling($node, $siblingDir = 'nextSibling')
 
         return $nodes;
     }
+
+    /**
+     * @param \DOMDocument $document
+     * @param array        $prefixes
+     *
+     * @return \DOMXPath
+     *
+     * @throws \InvalidArgumentException
+     */
+    private function createDOMXPath(\DOMDocument $document, array $prefixes = array())
+    {
+        $domxpath = new \DOMXPath($document);
+
+        foreach ($prefixes as $prefix) {
+            // ask for one namespace, otherwise we'd get a collection with an item for each node
+            $namespaces = $domxpath->query(sprintf('(//namespace::*[name()="%s"])[last()]', 'default' === $prefix ? '' : $prefix));
+            if ($node = $namespaces->item(0)) {
+                $domxpath->registerNamespace($prefix, $node->nodeValue);
+            } else {
+                throw new \InvalidArgumentException(sprintf('Could not find a namespace for the prefix: "%s"', $prefix));
+            }
+        }
+
+        return $domxpath;
+    }
+
+    /**
+     * @param $xpath
+     *
+     * @return array
+     */
+    private function findNamespacePrefixes($xpath)
+    {
+        if (preg_match_all('/(?P<prefix>[a-zA-Z_][a-zA-Z_0-9\-\.]+):[^:]/', $xpath, $matches)) {
+            return array_unique($matches['prefix']);
+        }
+
+        return array();
+    }
 }
diff --git a/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php b/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php
@@ -373,7 +373,7 @@ public function testFilterXPath()
 
     public function testFilterXPathWithDefaultNamespace()
     {
-        $crawler = $this->createTestXmlCrawler()->filterXPath('//entry/id');
+        $crawler = $this->createTestXmlCrawler()->filterXPath('//default:entry/default:id');
         $this->assertCount(1, $crawler, '->filterXPath() automatically registers a namespace');
         $this->assertSame('tag:youtube.com,2008:video:kgZRZmEc9j4', $crawler->text());
     }
@@ -391,6 +391,15 @@ public function testFilterXPathWithMultipleNamespaces()
         $this->assertSame('widescreen', $crawler->text());
     }
 
+    /**
+     * @expectedException        \InvalidArgumentException
+     * @expectedExceptionMessage Could not find a namespace for the prefix: "foo"
+     */
+    public function testFilterXPathWithAnInvalidNamespace()
+    {
+        $this->createTestXmlCrawler()->filterXPath('//media:group/foo:aspectRatio');
+    }
+
     /**
      * @covers Symfony\Component\DomCrawler\Crawler::filter
      */
@@ -411,7 +420,7 @@ public function testFilterWithDefaultNamespace()
     {
         $this->markSkippedIfCssSelectorNotPresent();
 
-        $crawler = $this->createTestXmlCrawler()->filter('entry id');
+        $crawler = $this->createTestXmlCrawler()->filter('default|entry default|id');
         $this->assertCount(1, $crawler, '->filter() automatically registers namespaces');
         $this->assertSame('tag:youtube.com,2008:video:kgZRZmEc9j4', $crawler->text());
     }

Original file line number	Diff line number	Diff line change
`@@ -373,7 +373,7 @@ public function testFilterXPath()`
`373`	`373`
`374`	`374`	`public function testFilterXPathWithDefaultNamespace()`
`375`	`375`	`{`
`376`		`- $crawler = $this->createTestXmlCrawler()->filterXPath('//entry/id');`
	`376`	`+ $crawler = $this->createTestXmlCrawler()->filterXPath('//default:entry/default:id');`
`377`	`377`	`$this->assertCount(1, $crawler, '->filterXPath() automatically registers a namespace');`
`378`	`378`	`$this->assertSame('tag:youtube.com,2008:video:kgZRZmEc9j4', $crawler->text());`
`379`	`379`	`}`
`@@ -391,6 +391,15 @@ public function testFilterXPathWithMultipleNamespaces()`
`391`	`391`	`$this->assertSame('widescreen', $crawler->text());`
`392`	`392`	`}`
`393`	`393`
	`394`	`+ /**`
	`395`	`+ * @expectedException \InvalidArgumentException`
	`396`	`+ * @expectedExceptionMessage Could not find a namespace for the prefix: "foo"`
	`397`	`+ */`
	`398`	`+ public function testFilterXPathWithAnInvalidNamespace()`
	`399`	`+ {`
	`400`	`+ $this->createTestXmlCrawler()->filterXPath('//media:group/foo:aspectRatio');`
	`401`	`+ }`
	`402`	`+`
`394`	`403`	`/**`
`395`	`404`	`* @covers Symfony\Component\DomCrawler\Crawler::filter`
`396`	`405`	`*/`
`@@ -411,7 +420,7 @@ public function testFilterWithDefaultNamespace()`
`411`	`420`	`{`
`412`	`421`	`$this->markSkippedIfCssSelectorNotPresent();`
`413`	`422`
`414`		`- $crawler = $this->createTestXmlCrawler()->filter('entry id');`
	`423`	`+ $crawler = $this->createTestXmlCrawler()->filter('default\|entry default\|id');`
`415`	`424`	`$this->assertCount(1, $crawler, '->filter() automatically registers namespaces');`
`416`	`425`	`$this->assertSame('tag:youtube.com,2008:video:kgZRZmEc9j4', $crawler->text());`
`417`	`426`	`}`