Allow LogAggregator to work with multiple site IDs, compute unique visitors for archives of multiple sites and add option to control whether unique visitors for multiple sites are processed.

3 files changed, 44 insertions, 12 deletions

diff --git a/config/global.ini.php b/config/global.ini.php
index 649b1964ba..ff987583d2 100644
--- a/config/global.ini.php
+++ b/config/global.ini.php
@@ -111,6 +111,11 @@ enable_processing_unique_visitors_month = 1
 enable_processing_unique_visitors_year = 0
 enable_processing_unique_visitors_range = 0
 
+; controls whether Unique Visitors will be processed for groups of websites. these metrics describe the number
+; of unique visitors across the entire set of websites, so if a visitor visited two websites in the group, she
+; would still only be counted as one. only relevant when using plugins that group sites together
+enable_processing_unique_visitors_multiple_sites = 0
+
 ; The list of periods that are available in the Piwik calendar
 ; Example use case: custom date range requests are processed in real time,
 ; so they may take a few minutes on very high traffic website: you may remove "range" below to disable this period
diff --git a/core/ArchiveProcessor.php b/core/ArchiveProcessor.php
index 800db50a11..9cc67f6271 100644
--- a/core/ArchiveProcessor.php
+++ b/core/ArchiveProcessor.php
@@ -99,13 +99,28 @@ class ArchiveProcessor
      * @var int
      */
     protected $numberOfVisits = false;
+
     protected $numberOfVisitsConverted = false;
 
+    /**
+     * If true, unique visitors are not calculated when we are aggregating data for multiple sites.
+     * The `[General] enable_processing_unique_visitors_multiple_sites` INI config option controls
+     * the value of this variable.
+     *
+     * @var bool
+     */
+    private $skipUniqueVisitorsCalculationForMultipleSites = true;
+
+    const SKIP_UNIQUE_VISITORS_FOR_MULTIPLE_SITES = 'enable_processing_unique_visitors_multiple_sites';
+
     public function __construct(Parameters $params, ArchiveWriter $archiveWriter)
     {
         $this->params = $params;
         $this->logAggregator = new LogAggregator($params);
         $this->archiveWriter = $archiveWriter;
+
+        $this->skipUniqueVisitorsCalculationForMultipleSites =
+            Config::getInstance()->General['enable_processing_unique_visitors_multiple_sites'] == 1;
     }
 
     protected function getArchive()
@@ -365,12 +380,15 @@ class ArchiveProcessor
 
     protected function enrichWithUniqueVisitorsMetric(Row $row)
     {
-        if(!$this->getParams()->isSingleSite() ) {
-            // we only compute unique visitors for a single site
+        // skip unique visitors metrics calculation if calculating for multiple sites is disabled
+        if (!$this->getParams()->isSingleSite()
+            && $this->skipUniqueVisitorsCalculationForMultipleSites
+        ) {
             return;
         }
-        if ( $row->getColumn('nb_uniq_visitors') !== false
-            || $row->getColumn('nb_users') !== false) {
+        if ($row->getColumn('nb_uniq_visitors') !== false
+            || $row->getColumn('nb_users') !== false
+        ) {
             if (SettingsPiwik::isUniqueVisitorsEnabled($this->getParams()->getPeriod()->getLabel())) {
                 $metrics = array(Metrics::INDEX_NB_UNIQ_VISITORS, Metrics::INDEX_NB_USERS);
                 $uniques = $this->computeNbUniques( $metrics );
diff --git a/core/DataAccess/LogAggregator.php b/core/DataAccess/LogAggregator.php
index 0170097fe6..95a7603176 100644
--- a/core/DataAccess/LogAggregator.php
+++ b/core/DataAccess/LogAggregator.php
@@ -9,6 +9,7 @@
 namespace Piwik\DataAccess;
 
 use Piwik\ArchiveProcessor\Parameters;
+use Piwik\Common;
 use Piwik\DataArray;
 use Piwik\Db;
 use Piwik\Metrics;
@@ -128,8 +129,8 @@ class LogAggregator
     /** @var \Piwik\Date */
     protected $dateEnd;
 
-    /** @var \Piwik\Site */
-    protected $site;
+    /** @var int[] */
+    protected $sites;
 
     /** @var \Piwik\Segment */
     protected $segment;
@@ -144,12 +145,12 @@ class LogAggregator
         $this->dateStart = $params->getDateStart();
         $this->dateEnd = $params->getDateEnd();
         $this->segment = $params->getSegment();
-        $this->site = $params->getSite();
+        $this->sites = $params->getIdSites();
     }
 
     public function generateQuery($select, $from, $where, $groupBy, $orderBy)
     {
-        $bind = $this->getBindDatetimeSite();
+        $bind = $this->getGeneralQueryBindParams();
         $query = $this->segment->getSelectQuery($select, $from, $where, $bind, $orderBy, $groupBy);
         return $query;
     }
@@ -438,7 +439,7 @@ class LogAggregator
     {
         $where = "$tableName.$datetimeField >= ?
 				AND $tableName.$datetimeField <= ?
-				AND $tableName.idsite = ?";
+				AND $tableName.idsite IN (". Common::getSqlStringFieldsArray($this->sites) . ")";
         if (!empty($extraWhere)) {
             $extraWhere = sprintf($extraWhere, $tableName, $tableName);
             $where .= ' AND ' . $extraWhere;
@@ -453,9 +454,17 @@ class LogAggregator
         return $groupBy;
     }
 
-    protected function getBindDatetimeSite()
+    /**
+     * Returns general bind parameters for all log aggregation queries. This includes the datetime
+     * start of entities, datetime end of entities and IDs of all sites.
+     *
+     * @return array
+     */
+    protected function getGeneralQueryBindParams()
     {
-        return array($this->dateStart->getDateStartUTC(), $this->dateEnd->getDateEndUTC(), $this->site->getId());
+        $bind = array($this->dateStart->getDateStartUTC(), $this->dateEnd->getDateEndUTC());
+        $bind = array_merge($bind, $this->sites);
+        return $bind;
     }
 
     /**
@@ -545,7 +554,7 @@ class LogAggregator
                 array(
                     'log_conversion_item.server_time >= ?',
                     'log_conversion_item.server_time <= ?',
-                    'log_conversion_item.idsite = ?',
+                    'log_conversion_item.idsite IN (' . Common::getSqlStringFieldsArray($this->sites) . ')',
                     'log_conversion_item.deleted = 0'
                 )
             ),