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

1 files changed, 141 insertions, 7 deletions

diff --git a/plugins/SitesManager/SiteUrls.php b/plugins/SitesManager/SiteUrls.php
index a1fac67a89..059521a4a9 100644
--- a/plugins/SitesManager/SiteUrls.php
+++ b/plugins/SitesManager/SiteUrls.php
@@ -9,6 +9,7 @@
 namespace Piwik\Plugins\SitesManager;
 
 use Piwik\Cache;
+use Piwik\Common;
 
 class SiteUrls
 {
@@ -19,6 +20,96 @@ class SiteUrls
         self::getCache()->delete(self::$cacheId);
     }
 
+    /**
+     * Groups all URLs by host, path and idsite.
+     *
+     * @param array $urls  An array containing URLs by idsite,
+     *                     eg array(array($idSite = 1 => array('apache.piwik', 'apache2.piwik'), 2 => array(), ...))
+     *                     as returned by {@link getAllCachedSiteUrls()} and {@link getAllSiteUrls}
+     * @return array All urls grouped by host => path => idSites. Path having the most '/' will be listed first
+     * array(
+         'apache.piwik' => array(
+             '/test/two' => $idsite = array(3),
+             '/test' => $idsite = array(1),
+             '/' => $idsite = array(2),
+         ),
+         'test.apache.piwik' => array(
+             '/test/two' => $idsite = array(3),
+             '/test' => $idsite = array(1),
+             '/' => $idsite = array(2, 3),
+         ),
+       );
+     */
+    public function groupUrlsByHost($siteUrls)
+    {
+        if (empty($siteUrls)) {
+            return array();
+        }
+
+        $allUrls = array();
+
+        foreach ($siteUrls as $idSite => $urls) {
+            $idSite = (int) $idSite;
+            foreach ($urls as $url) {
+                $urlParsed = @parse_url($url);
+
+                if ($urlParsed === false || !isset($urlParsed['host'])) {
+                    continue;
+                }
+
+                $host = $this->toCanonicalHost($urlParsed['host']);
+                $path = $this->getCanonicalPathFromParsedUrl($urlParsed);
+
+                if (!isset($allUrls[$host])) {
+                    $allUrls[$host] = array();
+                }
+
+                if (!isset($allUrls[$host][$path])) {
+                    $allUrls[$host][$path] = array();
+                }
+
+                if (!in_array($idSite, $allUrls[$host][$path])) {
+                    $allUrls[$host][$path][] = $idSite;
+                }
+            }
+        }
+
+        foreach ($allUrls as $host => $paths) {
+            uksort($paths, array($this, 'sortByPathDepth'));
+            $allUrls[$host] = $paths;
+        }
+
+        return $allUrls;
+    }
+
+    public function getIdSitesMatchingUrl($parsedUrl, $urlsGroupedByHost)
+    {
+        if (empty($parsedUrl['host'])) {
+            return null;
+        }
+
+        $urlHost = $this->toCanonicalHost($parsedUrl['host']);
+        $urlPath = $this->getCanonicalPathFromParsedUrl($parsedUrl);
+
+        $matchingSites = null;
+        if (isset($urlsGroupedByHost[$urlHost])) {
+            $paths = $urlsGroupedByHost[$urlHost];
+
+            foreach ($paths as $path => $idSites) {
+                if (0 === strpos($urlPath, $path)) {
+                    $matchingSites = $idSites;
+                    break;
+                }
+            }
+
+            if (!isset($matchingSites) && isset($paths['/'])) {
+                $matchingSites = $paths['/'];
+            }
+        }
+
+        return $matchingSites;
+    }
+
     public function getAllCachedSiteUrls()
     {
         $cache    = $this->getCache();
@@ -35,23 +126,66 @@ class SiteUrls
     public function getAllSiteUrls()
     {
         $model    = new Model();
-        $siteIds  = $model->getSitesId();
-        $siteUrls = array();
+        $siteUrls = $model->getAllKnownUrlsForAllSites();
 
-        if (empty($siteIds)) {
+        if (empty($siteUrls)) {
             return array();
         }
 
-        foreach ($siteIds as $siteId) {
-            $siteId = (int) $siteId;
-            $siteUrls[$siteId] = $model->getSiteUrlsFromId($siteId);
+        $urls = array();
+        foreach ($siteUrls as $siteUrl) {
+            $siteId = (int) $siteUrl['idsite'];
+
+            if (!isset($urls[$siteId])) {
+                $urls[$siteId] = array();
+            }
+
+            $urls[$siteId][] = $siteUrl['url'];
         }
 
-        return $siteUrls;
+        return $urls;
     }
 
     private static function getCache()
     {
         return Cache::getLazyCache();
     }
+
+    private function sortByPathDepth($pathA, $pathB)
+    {
+        // list first the paths with most '/' , and list path = '/' last
+        $numSlashA = substr_count($pathA, '/');
+        $numSlashB = substr_count($pathB, '/');
+
+        if ($numSlashA === $numSlashB) {
+            return -1 * strcmp($pathA, $pathB);
+        }
+
+        return $numSlashA > $numSlashB ? -1 : 1;
+    }
+
+    private function toCanonicalHost($host)
+    {
+        $host = Common::mb_strtolower($host);
+        if (strpos($host, 'www.') === 0) {
+            $host = substr($host, 4);
+        }
+
+        return $host;
+    }
+
+    private function getCanonicalPathFromParsedUrl($urlParsed)
+    {
+        $path = '/';
+
+        if (isset($urlParsed['path'])) {
+            $path = Common::mb_strtolower($urlParsed['path']);
+            if (!Common::stringEndsWith($path, '/')) {
+                $path .= '/';
+            }
+        }
+
+        return $path;
+    }
+
 }