From fe5d2d1554097861cb5535ee70595bd1e2ba6530 Mon Sep 17 00:00:00 2001
From: Benjamin Grandfond <benjamin.grandfond@gmail.com>
Date: Sat, 21 Jun 2014 22:15:52 +0200
Subject: [PATCH] [DomCrawler] Remove the query string and the anchor of the
 uri of a link

---
 src/Symfony/Component/DomCrawler/Link.php     | 67 ++++++++++++++-----
 .../Component/DomCrawler/Tests/LinkTest.php   |  4 ++
 2 files changed, 54 insertions(+), 17 deletions(-)

diff --git a/src/Symfony/Component/DomCrawler/Link.php b/src/Symfony/Component/DomCrawler/Link.php
index 05d5e927a2..57cf2b4f21 100644
--- a/src/Symfony/Component/DomCrawler/Link.php
+++ b/src/Symfony/Component/DomCrawler/Link.php
@@ -98,34 +98,23 @@ class Link
             return $this->currentUri;
         }
 
-        // only an anchor
+        // an anchor
         if ('#' === $uri[0]) {
-            $baseUri = $this->currentUri;
-            if (false !== $pos = strpos($baseUri, '#')) {
-                $baseUri = substr($baseUri, 0, $pos);
-            }
-
-            return $baseUri.$uri;
+            return $this->cleanupAnchor($this->currentUri).$uri;
         }
 
-        // only a query string
+        $baseUri = $this->cleanupUri($this->currentUri);
+
         if ('?' === $uri[0]) {
-            $baseUri = $this->currentUri;
-
-            // remove the query string from the current URI
-            if (false !== $pos = strpos($baseUri, '?')) {
-                $baseUri = substr($baseUri, 0, $pos);
-            }
-
             return $baseUri.$uri;
         }
 
         // absolute URL with relative schema
         if (0 === strpos($uri, '//')) {
-            return preg_replace('#^([^/]*)//.*$#', '$1', $this->currentUri).$uri;
+            return preg_replace('#^([^/]*)//.*$#', '$1', $baseUri).$uri;
         }
 
-        $baseUri = preg_replace('#^(.*?//[^/]*)(?:\/.*)?$#', '$1', $this->currentUri);
+        $baseUri = preg_replace('#^(.*?//[^/]*)(?:\/.*)?$#', '$1', $baseUri);
 
         // absolute path
         if ('/' === $uri[0]) {
@@ -194,4 +183,48 @@ class Link
 
         $this->node = $node;
     }
+
+    /**
+     * Removes the query string and the anchor from the given uri.
+     *
+     * @param  string $uri The uri to clean
+     *
+     * @return string
+     */
+    private function cleanupUri($uri)
+    {
+        return $this->cleanupQuery($this->cleanupAnchor($uri));
+    }
+
+    /**
+     * Remove the query string from the uri.
+     *
+     * @param $uri
+     *
+     * @return array
+     */
+    private function cleanupQuery($uri)
+    {
+        if (false !== $pos = strpos($uri, '?')) {
+            return substr($uri, 0, $pos);
+        }
+
+        return $uri;
+    }
+
+    /**
+     * Remove the anchor from the uri.
+     *
+     * @param $uri
+     *
+     * @return string
+     */
+    private function cleanupAnchor($uri)
+    {
+        if (false !== $pos = strpos($uri, '#')) {
+            return substr($uri, 0, $pos);
+        }
+
+        return $uri;
+    }
 }
diff --git a/src/Symfony/Component/DomCrawler/Tests/LinkTest.php b/src/Symfony/Component/DomCrawler/Tests/LinkTest.php
index a17142830f..941e4b2f6d 100644
--- a/src/Symfony/Component/DomCrawler/Tests/LinkTest.php
+++ b/src/Symfony/Component/DomCrawler/Tests/LinkTest.php
@@ -101,7 +101,9 @@ class LinkTest extends \PHPUnit_Framework_TestCase
 
             array('', 'http://localhost/bar/', 'http://localhost/bar/'),
             array('#', 'http://localhost/bar/', 'http://localhost/bar/#'),
+            array('#bar', 'http://localhost/bar?a=b', 'http://localhost/bar?a=b#bar'),
             array('#bar', 'http://localhost/bar/#foo', 'http://localhost/bar/#bar'),
+            array('?a=b', 'http://localhost/bar#foo', 'http://localhost/bar?a=b'),
             array('?a=b', 'http://localhost/bar/', 'http://localhost/bar/?a=b'),
 
             array('http://login.foo.com/foo', 'http://localhost/bar/', 'http://login.foo.com/foo'),
@@ -135,6 +137,8 @@ class LinkTest extends \PHPUnit_Framework_TestCase
             array('../../', 'http://localhost/', 'http://localhost/'),
             array('../../', 'http://localhost', 'http://localhost/'),
 
+            array('/foo', 'http://localhost?bar=1', 'http://localhost/foo'),
+            array('/foo', 'http://localhost#bar', 'http://localhost/foo'),
             array('/foo', 'file:///', 'file:///foo'),
             array('/foo', 'file:///bar/baz', 'file:///foo'),
             array('foo', 'file:///', 'file:///foo'),