Merge pull request #9320 from piwik/website_path

Added support to specify a path in a website URL and Piwik will consider the path during tracking

4 files changed, 163 insertions, 10 deletions

diff --git a/plugins/Referrers/Columns/Base.php b/plugins/Referrers/Columns/Base.php
index 900c7bba63..a7e7eaf1e3 100644
--- a/plugins/Referrers/Columns/Base.php
+++ b/plugins/Referrers/Columns/Base.php
@@ -12,6 +12,8 @@ use Piwik\Common;
 use Piwik\Piwik;
 use Piwik\Plugin\Dimension\VisitDimension;
 use Piwik\Plugins\Referrers\SearchEngine AS SearchEngineDetection;
+use Piwik\Plugins\SitesManager\SiteUrls;
+use Piwik\Tracker\Cache;
 use Piwik\Tracker\PageUrl;
 use Piwik\Tracker\Request;
 use Piwik\Tracker\Visit;
@@ -251,20 +253,34 @@ abstract class Base extends VisitDimension
      */
     protected function detectReferrerDirectEntry()
     {
-        if (!empty($this->referrerHost)) {
-            // is the referrer host the current host?
-            if (isset($this->currentUrlParse['host'])) {
-                $currentHost = Common::mb_strtolower($this->currentUrlParse['host'], 'UTF-8');
-                if ($currentHost == Common::mb_strtolower($this->referrerHost, 'UTF-8')) {
-                    $this->typeReferrerAnalyzed = Common::REFERRER_TYPE_DIRECT_ENTRY;
-                    return true;
-                }
+        if (empty($this->referrerHost)) {
+            return false;
+        }
+
+        $cache = Cache::getCacheGeneral();
+
+        if (!empty($cache['allUrlsByHostAndIdSite'])) {
+            $directEntry   = new SiteUrls();
+            $matchingSites = $directEntry->getIdSitesMatchingUrl($this->referrerUrlParse, $cache['allUrlsByHostAndIdSite']);
+
+            if (isset($matchingSites) && is_array($matchingSites) && in_array($this->idsite, $matchingSites)) {
+                $this->typeReferrerAnalyzed = Common::REFERRER_TYPE_DIRECT_ENTRY;
+                return true;
+            } elseif (isset($matchingSites)) {
+                return false;
             }
-            if (Visit::isHostKnownAliasHost($this->referrerHost, $this->idsite)) {
+        }
+
+        // fallback logic if the referrer domain is not known to any site to not break BC
+        if (isset($this->currentUrlParse['host'])) {
+            // this might be actually buggy if first thing tracked is eg an outlink and referrer is from that site
+            $currentHost = Common::mb_strtolower($this->currentUrlParse['host']);
+            if ($currentHost == Common::mb_strtolower($this->referrerHost)) {
                 $this->typeReferrerAnalyzed = Common::REFERRER_TYPE_DIRECT_ENTRY;
                 return true;
             }
         }
+
         return false;
     }
 
diff --git a/plugins/Referrers/Columns/Campaign.php b/plugins/Referrers/Columns/Campaign.php
index ff2d5c2401..c29b622336 100644
--- a/plugins/Referrers/Columns/Campaign.php
+++ b/plugins/Referrers/Columns/Campaign.php
@@ -37,7 +37,7 @@ class Campaign extends Base
 
     /**
      * If we should create a new visit when the campaign changes, check if the campaign info changed and if so
-     * force the tracker to create a new visit.
+     * force the tracker to create a new visit.i
      *
      * @param Request $request
      * @param Visitor $visitor
diff --git a/plugins/Referrers/Referrers.php b/plugins/Referrers/Referrers.php
index ce8667929b..d262c79224 100644
--- a/plugins/Referrers/Referrers.php
+++ b/plugins/Referrers/Referrers.php
@@ -12,6 +12,7 @@ use Piwik\ArchiveProcessor;
 use Piwik\Common;
 use Piwik\Piwik;
 use Piwik\Plugins\CoreVisualizations\Visualizations\HtmlTable;
+use Piwik\Plugins\SitesManager\SiteUrls;
 
 /**
  * @see plugins/Referrers/functions.php
@@ -31,9 +32,18 @@ class Referrers extends \Piwik\Plugin
             'Insights.addReportToOverview'      => 'addReportToInsightsOverview',
             'Live.getAllVisitorDetails'         => 'extendVisitorDetails',
             'Request.getRenamedModuleAndAction' => 'renameDeprecatedModuleAndAction',
+            'Tracker.setTrackerCacheGeneral'    => 'setTrackerCacheGeneral'
         );
     }
 
+    public function setTrackerCacheGeneral(&$cacheContent)
+    {
+        $siteUrls = new SiteUrls();
+        $urls = $siteUrls->getAllCachedSiteUrls();
+
+        return $cacheContent['allUrlsByHostAndIdSite'] = $siteUrls->groupUrlsByHost($urls);
+    }
+
     public function renameDeprecatedModuleAndAction(&$module, &$action)
     {
         if($module == 'Referers') {
diff --git a/plugins/Referrers/tests/Integration/Columns/ReferrerTypeTest.php b/plugins/Referrers/tests/Integration/Columns/ReferrerTypeTest.php
new file mode 100644
index 0000000000..0d5e0a68ca
--- /dev/null
+++ b/plugins/Referrers/tests/Integration/Columns/ReferrerTypeTest.php
@@ -0,0 +1,127 @@
+<?php
+/**
+ * Piwik - free/libre analytics platform
+ *
+ * @link http://piwik.org
+ * @license http://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later
+ */
+
+namespace Piwik\Plugins\Referrers\tests\Integration\Columns;
+
+use Piwik\Common;
+use Piwik\Plugins\Referrers\Columns\ReferrerType;
+use Piwik\Tests\Framework\Fixture;
+use Piwik\Tests\Framework\TestCase\IntegrationTestCase;
+use Piwik\Tracker\Cache;
+use Piwik\Tracker\Request;
+use Piwik\Tracker\Visit\VisitProperties;
+use Piwik\Tracker\Visitor;
+
+/**
+ * @group Referrers
+ * @group ReferrerTypeTest
+ * @group Plugins
+ */
+class ReferrerTypeTest extends IntegrationTestCase
+{
+    /**
+     * @var ReferrerType
+     */
+    private $referrerType;
+    private $idSite1 = 1;
+    private $idSite2 = 2;
+    private $idSite3 = 3;
+
+    public function setUp()
+    {
+        parent::setUp();
+
+        Cache::clearCacheGeneral();
+
+        $date = '2012-01-01 00:00:00';
+        $ecommerce = false;
+
+        Fixture::createWebsite($date, $ecommerce, $name = 'test1', $url = 'http://piwik.org/foo/bar');
+        Fixture::createWebsite($date, $ecommerce, $name = 'test2', $url = 'http://piwik.org/');
+        Fixture::createWebsite($date, $ecommerce, $name = 'test3', $url = 'http://piwik.pro/');
+
+        $this->referrerType = new ReferrerType();
+    }
+
+    public function tearDown()
+    {
+        // clean up your test here if needed
+        Cache::clearCacheGeneral();
+
+        parent::tearDown();
+    }
+
+    /**
+     * @dataProvider getReferrerUrls
+     */
+    public function test_onNewVisit_shouldDetectCorrectReferrerType($expectedType, $idSite, $url, $referrerUrl)
+    {
+        $request = $this->getRequest(array('idsite' => $idSite, 'url' => $url, 'urlref' => $referrerUrl));
+        $type = $this->referrerType->onNewVisit($request, $this->getNewVisitor(), $action = null);
+
+        $this->assertSame($expectedType, $type);
+    }
+
+    public function getReferrerUrls()
+    {
+        $url = 'http://piwik.org/foo/bar';
+        $referrer = 'http://piwik.org';
+
+        // $expectedType,                             $idSite,        $url, $referrerUrl
+        return array(
+            // domain matches but path does not match for idsite1
+            array(Common::REFERRER_TYPE_WEBSITE,      $this->idSite1, $url, $referrer),
+            array(Common::REFERRER_TYPE_WEBSITE,      $this->idSite1, $url, $referrer . '/'),
+            // idSite2 matches any piwik.org path so this is a direct entry for it
+            array(Common::REFERRER_TYPE_DIRECT_ENTRY, $this->idSite2, $url, $referrer),
+            array(Common::REFERRER_TYPE_DIRECT_ENTRY, $this->idSite2, $url, $referrer . '/'),
+            // idSite3 has different domain so it is coming from different website
+            array(Common::REFERRER_TYPE_WEBSITE,      $this->idSite3, $url, $referrer),
+            array(Common::REFERRER_TYPE_WEBSITE,      $this->idSite3, $url, $referrer . '/'),
+
+            array(Common::REFERRER_TYPE_DIRECT_ENTRY, $this->idSite1, $url, $referrer . '/foo/bar/baz'),
+            array(Common::REFERRER_TYPE_DIRECT_ENTRY, $this->idSite1, $url, $referrer . '/foo/bar/baz/'),
+            array(Common::REFERRER_TYPE_DIRECT_ENTRY, $this->idSite1, $url, $referrer . '/foo/bar/baz?x=5'),
+            // /not/xyz belongs to different website
+            array(Common::REFERRER_TYPE_WEBSITE,      $this->idSite1, $url, $referrer . '/not/xyz'),
+            array(Common::REFERRER_TYPE_DIRECT_ENTRY, $this->idSite2, $url, $referrer . '/not/xyz'),
+
+            // /foo/bar/baz belongs to different website
+            array(Common::REFERRER_TYPE_WEBSITE,      $this->idSite2, $url, $referrer . '/foo/bar/baz'),
+
+            // website as it is from different domain anyway
+            array(Common::REFERRER_TYPE_WEBSITE,      $this->idSite3, $url, $referrer . '/foo/bar/baz'),
+
+            // should detect campaign independent of domain / path
+            array(Common::REFERRER_TYPE_CAMPAIGN,     $this->idSite1, $url . '?pk_campaign=test', $referrer),
+            array(Common::REFERRER_TYPE_CAMPAIGN,     $this->idSite2, $url . '?pk_campaign=test', $referrer),
+            array(Common::REFERRER_TYPE_CAMPAIGN,     $this->idSite3, $url . '?pk_campaign=test', $referrer),
+
+            array(Common::REFERRER_TYPE_SEARCH_ENGINE, $this->idSite3, $url, 'http://google.com/search?q=piwik'),
+
+            // testing case for backwards compatibility where url has same domain as urlref but the domain is not known to any website
+            array(Common::REFERRER_TYPE_DIRECT_ENTRY, $this->idSite3, 'http://example.com/foo', 'http://example.com/bar'),
+            array(Common::REFERRER_TYPE_DIRECT_ENTRY, $this->idSite3, 'http://example.com/foo', 'http://example.com'),
+            array(Common::REFERRER_TYPE_DIRECT_ENTRY, $this->idSite3, 'http://example.com', 'http://example.com/bar'),
+
+            // testing case where domain of referrer is not known to any site but neither is the URL, url != urlref
+            array(Common::REFERRER_TYPE_WEBSITE,      $this->idSite3, 'http://example.org', 'http://example.com/bar'),
+        );
+    }
+
+    private function getRequest($params)
+    {
+        return new Request($params);
+    }
+
+    private function getNewVisitor()
+    {
+        return new Visitor(new VisitProperties());
+    }
+
+}