bug #13145 [DomCrawler] Fix behaviour with <base> tag (dkop, WouterJ)

This PR was merged into the 2.3 branch. Discussion ---------- [DomCrawler] Fix behaviour with <base> tag Finishes #12283 | Q | A | ------------- | --- | Bug fix? | yes | New feature? | no | BC breaks? | no | Deprecations? | no | Tests pass? | yes | Fixed tickets | #12283, #12143, #12144 | License | MIT | Doc PR | - Commits ------- 91447e8 Make fabbot happy 1d35e48 Clean up testing 61f22d7 [DomCrawler] fixed bug #12143
symfony · Dec 29, 2014 · 0469ea8 · 0469ea8
2 parents 459b8b6 + 91447e8
commit 0469ea8
Show file tree

Hide file tree

Showing 2 changed files with 62 additions and 30 deletions.
diff --git a/src/Symfony/Component/DomCrawler/Crawler.php b/src/Symfony/Component/DomCrawler/Crawler.php
@@ -23,21 +23,27 @@
 class Crawler extends \SplObjectStorage
 {
     /**
-     * @var string The current URI or the base href value
+     * @var string The current URI
      */
     protected $uri;
 
+    /**
+     * @var string The base href value
+     */
+    private $baseHref;
+
     /**
      * Constructor.
      *
      * @param mixed  $node A Node to use as the base for the crawling
-     * @param string $uri  The current URI or the base href value
-     *
+     * @param string $currentUri The current URI
+     * @param string $baseHref The base href value
      * @api
      */
-    public function __construct($node = null, $uri = null)
+    public function __construct($node = null, $currentUri = null, $baseHref = null)
     {
-        $this->uri = $uri;
+        $this->uri = $currentUri;
+        $this->baseHref = $baseHref ?: $currentUri;
 
         $this->add($node);
     }
@@ -176,13 +182,13 @@ public function addHtmlContent($content, $charset = 'UTF-8')
 
         $baseHref = current($base);
         if (count($base) && !empty($baseHref)) {
-            if ($this->uri) {
+            if ($this->baseHref) {
                 $linkNode = $dom->createElement('a');
                 $linkNode->setAttribute('href', $baseHref);
-                $link = new Link($linkNode, $this->uri);
-                $this->uri = $link->getUri();
+                $link = new Link($linkNode, $this->baseHref);
+                $this->baseHref = $link->getUri();
             } else {
-                $this->uri = $baseHref;
+                $this->baseHref = $baseHref;
             }
         }
     }
@@ -294,11 +300,11 @@ public function eq($position)
     {
         foreach ($this as $i => $node) {
             if ($i == $position) {
-                return new static($node, $this->uri);
+                return new static($node, $this->uri, $this->baseHref);
             }
         }
 
-        return new static(null, $this->uri);
+        return new static(null, $this->uri, $this->baseHref);
     }
 
     /**
@@ -323,7 +329,7 @@ public function each(\Closure $closure)
     {
         $data = array();
         foreach ($this as $i => $node) {
-            $data[] = $closure(new static($node, $this->uri), $i);
+            $data[] = $closure(new static($node, $this->uri, $this->baseHref), $i);
         }
 
         return $data;
@@ -344,12 +350,12 @@ public function reduce(\Closure $closure)
     {
         $nodes = array();
         foreach ($this as $i => $node) {
-            if (false !== $closure(new static($node, $this->uri), $i)) {
+            if (false !== $closure(new static($node, $this->uri, $this->baseHref), $i)) {
                 $nodes[] = $node;
             }
         }
 
-        return new static($nodes, $this->uri);
+        return new static($nodes, $this->uri, $this->baseHref);
     }
 
     /**
@@ -391,7 +397,7 @@ public function siblings()
             throw new \InvalidArgumentException('The current node list is empty.');
         }
 
-        return new static($this->sibling($this->getNode(0)->parentNode->firstChild), $this->uri);
+        return new static($this->sibling($this->getNode(0)->parentNode->firstChild), $this->uri, $this->baseHref);
     }
 
     /**
@@ -409,7 +415,7 @@ public function nextAll()
             throw new \InvalidArgumentException('The current node list is empty.');
         }
 
-        return new static($this->sibling($this->getNode(0)), $this->uri);
+        return new static($this->sibling($this->getNode(0)), $this->uri, $this->baseHref);
     }
 
     /**
@@ -427,7 +433,7 @@ public function previousAll()
             throw new \InvalidArgumentException('The current node list is empty.');
         }
 
-        return new static($this->sibling($this->getNode(0), 'previousSibling'), $this->uri);
+        return new static($this->sibling($this->getNode(0), 'previousSibling'), $this->uri, $this->baseHref);
     }
 
     /**
@@ -454,7 +460,7 @@ public function parents()
             }
         }
 
-        return new static($nodes, $this->uri);
+        return new static($nodes, $this->uri, $this->baseHref);
     }
 
     /**
@@ -474,7 +480,7 @@ public function children()
 
         $node = $this->getNode(0)->firstChild;
 
-        return new static($node ? $this->sibling($node) : array(), $this->uri);
+        return new static($node ? $this->sibling($node) : array(), $this->uri, $this->baseHref);
     }
 
     /**
@@ -601,7 +607,7 @@ public function filterXPath($xpath)
 
         // If we dropped all expressions in the XPath while preparing it, there would be no match
         if ('' === $xpath) {
-            return new static(null, $this->uri);
+            return new static(null, $this->uri, $this->baseHref);
         }
 
         return $this->filterRelativeXPath($xpath);
@@ -687,7 +693,7 @@ public function link($method = 'get')
 
         $node = $this->getNode(0);
 
-        return new Link($node, $this->uri, $method);
+        return new Link($node, $this->baseHref, $method);
     }
 
     /**
@@ -701,7 +707,7 @@ public function links()
     {
         $links = array();
         foreach ($this as $node) {
-            $links[] = new Link($node, $this->uri, 'get');
+            $links[] = new Link($node, $this->baseHref, 'get');
         }
 
         return $links;
@@ -792,7 +798,7 @@ public static function xpathLiteral($s)
      */
     private function filterRelativeXPath($xpath)
     {
-        $crawler = new static(null, $this->uri);
+        $crawler = new static(null, $this->uri, $this->baseHref);
 
         foreach ($this as $node) {
             $domxpath = new \DOMXPath($node->ownerDocument);

diff --git a/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php b/src/Symfony/Component/DomCrawler/Tests/CrawlerTest.php
@@ -823,16 +823,42 @@ public function testParents()
         }
     }
 
-    public function testBaseTag()
+    /**
+     * @dataProvider getBaseTagData
+     */
+    public function testBaseTag($baseValue, $linkValue, $expectedUri, $currentUri = null, $description = null)
     {
-        $crawler = new Crawler('<html><base href="http://base.com"><a href="link"></a></html>');
-        $this->assertEquals('http://base.com/link', $crawler->filterXPath('//a')->link()->getUri());
+        $crawler = new Crawler('<html><base href="'.$baseValue.'"><a href="'.$linkValue.'"></a></html>', $currentUri);
+        $this->assertEquals($expectedUri, $crawler->filterXPath('//a')->link()->getUri(), $description);
+    }
 
-        $crawler = new Crawler('<html><base href="//base.com"><a href="link"></a></html>', 'https://domain.com');
-        $this->assertEquals('https://base.com/link', $crawler->filterXPath('//a')->link()->getUri(), '<base> tag can use a schema-less URL');
+    public function getBaseTagData()
+    {
+        return array(
+            array('http://base.com', 'link', 'http://base.com/link'),
+            array('//base.com', 'link', 'https://base.com/link', 'https://domain.com', '<base> tag can use a schema-less URL'),
+            array('path/', 'link', 'https://domain.com/path/link', 'https://domain.com', '<base> tag can set a path'),
+            array('http://base.com', '#', 'http://base.com#', 'http://domain.com/path/link', '<base> tag does work with links to an anchor'),
+            array('http://base.com', '', 'http://base.com', 'http://domain.com/path/link', '<base> tag does work with empty links'),
+        );
+    }
 
-        $crawler = new Crawler('<html><base href="path/"><a href="link"></a></html>', 'https://domain.com');
-        $this->assertEquals('https://domain.com/path/link', $crawler->filterXPath('//a')->link()->getUri(), '<base> tag can set a path');
+    /**
+     * @dataProvider getBaseTagWithFormData
+     */
+    public function testBaseTagWithForm($baseValue, $actionValue, $expectedUri, $currentUri = null, $description = null)
+    {
+        $crawler = new Crawler('<html><base href="'.$baseValue.'"><form method="post" action="'.$actionValue.'"><button type="submit" name="submit"/></form></html>', $currentUri);
+        $this->assertEquals($expectedUri, $crawler->filterXPath('//button')->form()->getUri(), $description);
+    }
+
+    public function getBaseTagWithFormData()
+    {
+        return array(
+            array('/basepath', '/registration', 'http://domain.com/registration', 'http://domain.com/registration', '<base> tag does work with a path and form action'),
+            array('/basepath', '', 'http://domain.com/registration', 'http://domain.com/registration', '<base> tag does work with a path and empty form action'),
+            array('http://base.com', '', 'http://domain.com/path/form', 'http://domain.com/path/form', '<base> tag does work with a URL and an empty form action'),
+        );
     }
 
     public function createTestCrawler($uri = null)