From 2844ef7307e91e8e5650481fde0557dfcfc97c94 Mon Sep 17 00:00:00 2001
From: MusikAnimal <musikanimal@gmail.com>
Date: Tue, 16 Jan 2024 00:20:21 -0500
Subject: [PATCH] Always JOIN on page table to avoid counting orphan revisions

Bug: T355027
---
 src/Repository/EditCounterRepository.php      | 23 ++++++++++++++++++-
 .../SimpleEditCounterRepository.php           |  5 ++--
 2 files changed, 25 insertions(+), 3 deletions(-)

diff --git a/src/Repository/EditCounterRepository.php b/src/Repository/EditCounterRepository.php
index 382e760ed..0c9b71df0 100644
--- a/src/Repository/EditCounterRepository.php
+++ b/src/Repository/EditCounterRepository.php
@@ -57,6 +57,10 @@ public function getPairData(Project $project, User $user): array
         // Prepare the queries and execute them.
         $archiveTable = $project->getTableName('archive');
         $revisionTable = $project->getTableName('revision');
+        $pageTable = $project->getTableName('page');
+
+        // Always JOIN on page, see T355027
+        $pageJoin = "JOIN $pageTable ON rev_page = page_id";
 
         if ($user->isIpRange()) {
             $ipcTable = $project->getTableName('ip_changes');
@@ -86,26 +90,32 @@ public function getPairData(Project $project, User $user): array
 
             -- Revision counts.
             SELECT 'live' AS `key`, COUNT(rev_id) AS val FROM $revisionTable
+                $pageJoin
                 $ipcJoin
                 WHERE $whereClause
             ) UNION (
             SELECT 'day' AS `key`, COUNT(rev_id) AS val FROM $revisionTable
+                $pageJoin
                 $ipcJoin
                 WHERE $whereClause AND rev_timestamp >= DATE_SUB(NOW(), INTERVAL 1 DAY)
             ) UNION (
             SELECT 'week' AS `key`, COUNT(rev_id) AS val FROM $revisionTable
+                $pageJoin
                 $ipcJoin
                 WHERE $whereClause AND rev_timestamp >= DATE_SUB(NOW(), INTERVAL 1 WEEK)
             ) UNION (
             SELECT 'month' AS `key`, COUNT(rev_id) AS val FROM $revisionTable
+                $pageJoin
                 $ipcJoin
                 WHERE $whereClause AND rev_timestamp >= DATE_SUB(NOW(), INTERVAL 1 MONTH)
             ) UNION (
             SELECT 'year' AS `key`, COUNT(rev_id) AS val FROM $revisionTable
+                $pageJoin
                 $ipcJoin
                 WHERE $whereClause AND rev_timestamp >= DATE_SUB(NOW(), INTERVAL 1 YEAR)
             ) UNION (
             SELECT 'minor' AS `key`, COUNT(rev_id) AS val FROM $revisionTable
+                $pageJoin
                 $ipcJoin
                 WHERE $whereClause AND rev_minor_edit = 1
 
@@ -113,11 +123,13 @@ public function getPairData(Project $project, User $user): array
             ) UNION (
             SELECT 'edited-live' AS `key`, COUNT(DISTINCT rev_page) AS `val`
                 FROM $revisionTable
+                $pageJoin
                 $ipcJoin
                 WHERE $whereClause
             ) UNION (
             SELECT 'created-live' AS `key`, COUNT(DISTINCT rev_page) AS `val`
                 FROM $revisionTable
+                $pageJoin
                 $ipcJoin
                 WHERE $whereClause AND rev_parent_id = 0
             )";
@@ -503,17 +515,23 @@ public function getTimeCard(Project $project, User $user): array
         }
 
         $hourInterval = 1;
+        $revisionTable = $project->getTableName('revision');
+        // Always JOIN on page, see T325492
+        $pageTable = $project->getTableName('page');
 
         if ($user->isIpRange()) {
             $column = 'ipc_rev_timestamp';
             $table = $project->getTableName('ip_changes');
             [$params['startIp'], $params['endIp']] = IPUtils::parseRange($user->getUsername());
             $whereClause = 'ipc_hex BETWEEN :startIp AND :endIp';
+            $joinClause = "JOIN $revisionTable ON rev_id = ipc_rev_id
+                JOIN $pageTable ON rev_page = page_id";
         } else {
             $column = 'rev_timestamp';
-            $table = $project->getTableName('revision');
+            $table = $revisionTable;
             $whereClause = 'rev_actor = :actorId';
             $params = ['actorId' => $user->getActorId($project)];
+            $joinClause = "JOIN $pageTable ON rev_page = page_id";
         }
 
         $xCalc = "ROUND(HOUR($column)/$hourInterval) * $hourInterval";
@@ -523,6 +541,7 @@ public function getTimeCard(Project $project, User $user): array
                 $xCalc AS `hour`,
                 COUNT($column) AS `value`
             FROM $table
+            $joinClause
             WHERE $whereClause
             GROUP BY DAYOFWEEK($column), $xCalc";
 
@@ -549,6 +568,7 @@ public function getEditSizeData(Project $project, User $user): array
 
         // Prepare the queries and execute them.
         $revisionTable = $project->getTableName('revision');
+        $pageTable = $project->getTableName('page');
         $ipcJoin = '';
         $whereClause = 'revs.rev_actor = :actorId';
         $params = ['actorId' => $user->getActorId($project)];
@@ -566,6 +586,7 @@ public function getEditSizeData(Project $project, User $user): array
                 FROM (
                     SELECT (CAST(revs.rev_len AS SIGNED) - IFNULL(parentrevs.rev_len, 0)) AS size
                     FROM $revisionTable AS revs
+                    JOIN $pageTable ON revs.rev_page = page_id
                     $ipcJoin
                     LEFT JOIN $revisionTable AS parentrevs ON (revs.rev_parent_id = parentrevs.rev_id)
                     WHERE $whereClause
diff --git a/src/Repository/SimpleEditCounterRepository.php b/src/Repository/SimpleEditCounterRepository.php
index d646852e2..ae168a0df 100644
--- a/src/Repository/SimpleEditCounterRepository.php
+++ b/src/Repository/SimpleEditCounterRepository.php
@@ -65,7 +65,8 @@ private function fetchDataNormal(
         $arDateConditions = $this->getDateConditions($start, $end, false, '', 'ar_timestamp');
         $revDateConditions = $this->getDateConditions($start, $end);
 
-        $revNamespaceJoinSql = 'all' === $namespace ? '' : "JOIN $pageTable ON rev_page = page_id";
+        // Always JOIN on page, see T325492
+        $revNamespaceJoinSql = "JOIN $pageTable ON rev_page = page_id";
         $revNamespaceWhereSql = 'all' === $namespace ? '' : "AND page_namespace = $namespace";
         $arNamespaceWhereSql = 'all' === $namespace ? '' : "AND ar_namespace = $namespace";
 
@@ -126,7 +127,7 @@ private function fetchDataIpRange(
         $sql = "SELECT 'rev' AS source, COUNT(*) AS value
                 FROM $ipcTable
                 $revNamespaceJoinSql
-                WHERE ipc_hex BETWEEN :start AND :end 
+                WHERE ipc_hex BETWEEN :start AND :end
                 $revDateConditions
                 $revNamespaceWhereSql";