Overwrite direct entry referrer information if campaign referrer is found in later request. (#14273)

* Overwrite direct entry referrer information if campaign referrer is found in later request.

* Update for non-campaign referrers as well.

* test comment tweaks

* Add current tracking URL into site urls in case website has no main url.

* Fix a couple issues and start adding tests.

* More tests and a fix.

* Apply review feedback.

* Fix couple tests.

* Fix referrer tests.

* try to fix random failure

* Add note to README.

1 files changed, 42 insertions, 18 deletions

diff --git a/plugins/SitesManager/SiteUrls.php b/plugins/SitesManager/SiteUrls.php
index 33e2e4844a..54153d42e0 100644
--- a/plugins/SitesManager/SiteUrls.php
+++ b/plugins/SitesManager/SiteUrls.php
@@ -10,6 +10,8 @@ namespace Piwik\Plugins\SitesManager;
 
 use Piwik\Cache;
 use Piwik\Common;
+use Piwik\Tracker\Request;
+use Piwik\Tracker\Visitor;
 
 class SiteUrls
 {
@@ -51,35 +53,49 @@ class SiteUrls
         foreach ($siteUrls as $idSite => $urls) {
             $idSite = (int) $idSite;
             foreach ($urls as $url) {
-                $urlParsed = @parse_url($url);
+                $this->addUrlByHost($allUrls, $idSite, $url);
+            }
+        }
 
-                if ($urlParsed === false || !isset($urlParsed['host'])) {
-                    continue;
-                }
+        $this->sortUrlsByHost($allUrls);
 
-                $host = $this->toCanonicalHost($urlParsed['host']);
-                $path = $this->getCanonicalPathFromParsedUrl($urlParsed);
+        return $allUrls;
+    }
 
-                if (!isset($allUrls[$host])) {
-                    $allUrls[$host] = array();
-                }
+    private function addUrlByHost(&$allUrls, $idSite, $url, $addPath = true)
+    {
+        $urlParsed = @parse_url($url);
 
-                if (!isset($allUrls[$host][$path])) {
-                    $allUrls[$host][$path] = array();
-                }
+        if ($urlParsed === false || !isset($urlParsed['host'])) {
+            return;
+        }
 
-                if (!in_array($idSite, $allUrls[$host][$path])) {
-                    $allUrls[$host][$path][] = $idSite;
-                }
-            }
+        $host = $this->toCanonicalHost($urlParsed['host']);
+        $path = $this->getCanonicalPathFromParsedUrl($urlParsed);
+
+        if (!isset($allUrls[$host])) {
+            $allUrls[$host] = array();
+        }
+
+        if (!$addPath) {
+            $path = '/';
+        }
+
+        if (!isset($allUrls[$host][$path])) {
+            $allUrls[$host][$path] = array();
         }
 
+        if (!in_array($idSite, $allUrls[$host][$path])) {
+            $allUrls[$host][$path][] = $idSite;
+        }
+    }
+
+    private function sortUrlsByHost(&$allUrls)
+    {
         foreach ($allUrls as $host => $paths) {
             uksort($paths, array($this, 'sortByPathDepth'));
             $allUrls[$host] = $paths;
         }
-
-        return $allUrls;
     }
 
     public function getIdSitesMatchingUrl($parsedUrl, $urlsGroupedByHost)
@@ -172,6 +188,14 @@ class SiteUrls
         return Cache::getLazyCache();
     }
 
+    public function addRequestUrlToSiteUrls(&$allUrls, Request $request)
+    {
+        $idSite = $request->getIdSite();
+        $url = $request->getParam('url');
+
+        $this->addUrlByHost($allUrls, $idSite, $url, $addPath = false);
+    }
+
     private function sortByPathDepth($pathA, $pathB)
     {
         // list first the paths with most '/' , and list path = '/' last