TopEdits and PagesCreated: add list of ten top wikiprojects (#521)

DEWISOTT. (Figuring out the queries took a damn long time.)

On performance: same for topedits (just added an instant subquery), and we're at about +1 second for a user with 20K edits for pages. (Had to make new query.) If you've got an idea of how to integrate this into another query, feel free to. Just have bashed my head at SQL for too long today to be in a state to figure it out. Maybe will tomorrow. Didn't do for editcounter, as I don't know where we'd put it; will possibly also do tomorrow.

Bug: T344464

Author: agoujot

assets/css/application.scss

@@ -730,3 +730,12 @@ a.help-icon {
 .reverted-edit {
     background: #fcf8e3 !important;
 }
+
+.side-to-side > div {
+    display: inline-block;
+    vertical-align: top;
+}
+
+.side-to-side {
+    clear: both;
+}

i18n/en.json

@@ -514,6 +514,7 @@
 	"wiki": "Wiki",
 	"wiki-set-change": "Wiki set change",
 	"wikiproject": "WikiProject",
+	"top-wikiprojects": "Top WikiProjects",
 	"wikitext": "Wikitext",
 	"with-summary": "With summary",
 	"words": "Words",

i18n/qqq.json

@@ -533,6 +533,7 @@
 	"wiki": "Short label for Wiki\n{{Identical|Wiki}}",
 	"wiki-set-change": "Label for MediaWiki action done by stewards that changes the wiki sets. See [[meta:Special:WikiSets]] for possible translations.",
 	"wikiproject": "{{Identical|WikiProject}}",
+	"top-wikiprojects": "Caption for table of the top WikiProjects (by number of edits or pages). This message should be short.",
 	"wikitext": "Name of the markup language used on wiki pages. This message is shown as one of the download options.\n{{Identical|Wikitext}}",
 	"with-summary": "Chart label for edits with summary. This message should be short",
 	"words": "Label for the number of words that are in an article.\n{{Identical|Word}}",

public/build/app.da10c09b.css public/build/app.b59bcd25.csspublic/build/app.da10c09b.css renamed to public/build/app.b59bcd25.css

@@ -7,7 +7,7 @@
         "/build/app.65582b0e.js"
       ],
       "css": [
-        "/build/app.da10c09b.css"
+        "/build/app.b59bcd25.css"
       ]
     }
   }

public/build/entrypoints.json

@@ -1,5 +1,5 @@
 {
-  "build/app.css": "/build/app.da10c09b.css",
+  "build/app.css": "/build/app.b59bcd25.css",
   "build/app.js": "/build/app.65582b0e.js",
   "build/runtime.js": "/build/runtime.c217f8c4.js",
   "build/95.7a87ed1a.js": "/build/95.7a87ed1a.js",

public/build/manifest.json

@@ -301,6 +301,42 @@ public function getAssessmentCounts(): array
         return $counts;
     }
 
+    /**
+     * Get the number of pages the user created by WikiProject.
+     * @return array Keys are the WikiProject name, values are the counts.
+     */
+    public function getWikiprojectCounts(): array
+    {
+        if ($this->getNumPages() > $this->resultsPerPage()) {
+            $counts = $this->repository->getWikiprojectCounts(
+                $this->project,
+                $this->user,
+                $this->namespace,
+                $this->redirects,
+                $this->start,
+                $this->end,
+            );
+        } else {
+            $counts_tmp = [];
+            foreach ($this->pages as $nsPages) {
+                foreach ($nsPages as $page) {
+                    foreach ($page['assessment']['projects'] as $project) {
+                        $counts_tmp[$project] ??= 0;
+                        $counts_tmp[$project]++;
+                    }
+                }
+            }
+            arsort($counts_tmp);
+            $counts_tmp = array_slice($counts_tmp, 0, 10);
+            $counts = [];
+            foreach ($counts_tmp as $project => $count) {
+                $counts[] = [ "pap_project_title" => $project, "count" => $count ];
+            }
+        }
+
+        return $counts;
+    }
+
     /**
      * Number of results to show, depending on the namespace.
      * @param bool $all Whether to get *all* results. This should only be used for
@@ -452,6 +488,7 @@ private function formatPages(array $pages): array
                         ->getBadgeURL($row['pa_class'] ?: 'Unknown'),
                     'color' => $attrs['color'],
                     'category' => $attrs['category'],
+                    'projects' => json_decode($row['pap_project_title'] ?? '[]'),
                 ];
             }
 

src/Model/Pages.php

@@ -155,6 +155,45 @@ public function getNumTopEdits(): int
         return count($this->topEdits);
     }
 
+    /**
+     * Get the WikiProject totals.
+     * @param int Namespace ID.
+     * @return string[]|int
+     */
+    public function getProjectTotals(int $ns) : array
+    {
+        if ($this->getNumPagesAnyNamespace($ns) > $this->limit) {
+            $projectTotals = $this->repository->getProjectTotals(
+                $this->project,
+                $this->user,
+                $ns,
+                $this->start,
+                $this->end
+            );
+        } else {
+            $counts_tmp = [];
+            // List of pages for this namespace
+            $rows = $this->topEdits[$ns];
+            foreach ($rows as $row) {
+                $num = $row["count"];
+                // May be null or nonexistent for assessment-less pages
+                $titles = $row["pap_project_title"] ?? "{}";
+                // Had to use json to pass multiple values in SQL select
+                foreach (json_decode($titles) as $projectName) {
+                    $counts_tmp[$projectName] ??= 0;
+                    $counts_tmp[$projectName] += $num;
+                }
+            }
+            arsort($counts_tmp);
+            $counts_tmp = array_slice($projectTotals, 0, 10);
+            $projectTotals = [];
+            foreach ($counts_tmp as $project => $count) {
+                $counts[] = [ "pap_project_title" => $project, "count" => $count ];
+            }
+        }
+        return $projectTotals;
+    }
+
     /**
      * Get the average time between edits (in days).
      * @return float

src/Model/TopEdits.php

@@ -118,13 +118,22 @@ public function getPagesCreated(
         $hasPageAssessments = $this->isWMF && $project->hasPageAssessments($namespace);
         if ($hasPageAssessments) {
             $pageAssessmentsTable = $project->getTableName('page_assessments');
-            $conditions['paSelects'] = ", (SELECT pa_class
-                        FROM $pageAssessmentsTable
-                        WHERE rev_page = pa_page_id
-                        AND pa_class != ''
-                        LIMIT 1
-                    ) AS pa_class";
-            $conditions['paSelectsArchive'] = ', NULL AS pa_class';
+            $paProjectsTable = $project->getTableName('page_assessments_projects');
+            $conditions['paSelects'] = ",
+                (SELECT pa_class
+                    FROM $pageAssessmentsTable
+                    WHERE rev_page = pa_page_id
+                    AND pa_class != ''
+                    LIMIT 1
+                ) AS pa_class,
+                (SELECT JSON_ARRAYAGG(pap_project_title)
+                    FROM $pageAssessmentsTable
+                    JOIN $paProjectsTable
+                    ON pa_project_id = pap_project_id
+                    WHERE pa_page_id = page_id
+                ) AS pap_project_title";
+            $conditions['paSelectsArchive'] = ', NULL AS pa_class, NULL as pap_project_title';
+            $conditions['revPageGroupBy'] = 'GROUP BY rev_page';
         }
 
         $wasRedirect = $this->getWasRedirectClause($redirects, $deleted);
@@ -350,6 +359,62 @@ public function getAssessmentCounts(
         return $this->setCache($cacheKey, $assessments);
     }
 
+    /**
+     * Get the number of pages the user created by WikiProject.
+     * Max 10 projects.
+     * @param Project $project
+     * @param User $user
+     * @param int|string $namespace
+     * @param string $redirects One of the Pages::REDIR_ constants.
+     * @param int|false $start Start date as Unix timestamp.
+     * @param int|false $end End date as Unix timestamp.
+     * @return array Each element is an array with keys pap_project_title and count.
+     */
+    public function getWikiprojectCounts(
+        Project $project,
+        User $user,
+        $namespace,
+        string $redirects,
+        $start = false,
+        $end = false
+    ): array {
+        $cacheKey = $this->getCacheKey(func_get_args(), 'user_pages_created_wikiprojects');
+        if ($this->cache->hasItem($cacheKey)) {
+            return $this->cache->getItem($cacheKey)->get();
+        }
+
+        $pageTable = $project->getTableName('page');
+        $revisionTable = $project->getTableName('revision');
+        $pageAssessmentsTable = $project->getTableName('page_assessments');
+        $paProjectsTable = $project->getTableName('page_assessments_projects');
+
+        $conditions = array_merge(
+            $this->getNamespaceRedirectAndDeletedPagesConditions($namespace, $redirects),
+            $this->getUserConditions('' !== $start.$end)
+        );
+        $revDateConditions = $this->getDateConditions($start, $end);
+
+        $sql = "SELECT pap_project_title, count(pap_project_title) as `count`
+                FROM $pageTable
+                LEFT JOIN $revisionTable ON page_id = rev_page
+                JOIN $pageAssessmentsTable ON page_id = pa_page_id
+                JOIN $paProjectsTable ON pa_project_id = pap_project_id
+                WHERE ".$conditions['whereRev']."
+                    AND rev_parent_id = '0'".
+                    $conditions['namespaceRev'].
+                    $conditions['redirects'].
+                    $revDateConditions."
+                GROUP BY pap_project_title
+                ORDER BY `count` DESC
+                LIMIT 10";
+
+        $totals = $this->executeQuery($sql, $project, $user, $namespace)
+            ->fetchAllAssociative();
+
+        // Cache and return.
+        return $this->setCache($cacheKey, $totals);
+    }
+
     /**
      * Fetch the closest 'delete' event as of the time of the given $offset.
      *

src/Repository/PagesRepository.php

@@ -122,6 +122,16 @@ public function getTopEditsNamespace(
                     LIMIT 1
                 ) AS pa_class"
             : '';
+        $paProjectsTable = $project->getTableName('page_assessments_projects');
+        $paProjectsSelect = $hasPageAssessments
+            ? ", (
+                    SELECT JSON_ARRAYAGG(pap_project_title)
+                    FROM $paTable
+                    JOIN $paProjectsTable
+                    ON pa_project_id = pap_project_id
+                    WHERE pa_page_id = page_id
+                ) AS pap_project_title"
+            : '';
 
         $ipcJoin = '';
         $whereClause = 'rev_actor = :actorId';
@@ -137,7 +147,9 @@ public function getTopEditsNamespace(
         $sql = "SELECT page_namespace AS `namespace`, page_title,
                     page_is_redirect AS `redirect`, COUNT(page_title) AS `count`
                     $paSelect
+                    $paProjectsSelect
                 FROM $pageTable
+
                 JOIN $revisionTable ON page_id = rev_page
                 $ipcJoin
                 WHERE $whereClause
@@ -200,6 +212,66 @@ public function countEditsNamespace(Project $project, User $user, $namespace, $s
         return $this->setCache($cacheKey, $resultQuery->fetchOne());
     }
 
+    /**
+     * Get the 10 Wikiprojects within which the user has the most edits.
+     * @param Project $project
+     * @param User $user
+     * @param int $ns
+     * @param int|false $start
+     * @param int|false $end
+     */
+    public function getProjectTotals(
+        Project $project,
+        User $user,
+        int $ns,
+        $start = false,
+        $end = false
+    ): array {
+        $cacheKey = $this->getCacheKey(func_get_args(), 'top_edits_wikiprojects');
+        if ($this->cache->hasItem($cacheKey)) {
+            return $this->cache->getItem($cacheKey)->get();
+        }
+
+        $revDateConditions = $this->getDateConditions($start, $end);
+        $pageTable = $project->getTableName('page');
+        $revisionTable = $project->getTableName('revision');
+        $pageAssessmentsTable = $project->getTableName('page_assessments');
+        $paProjectsTable = $project->getTableName('page_assessments_projects');
+
+        $ipcJoin = '';
+        $whereClause = 'rev_actor = :actorId';
+        $params = [];
+        if ($user->isIpRange()) {
+            $ipcTable = $project->getTableName('ip_changes');
+            $ipcJoin = "JOIN $ipcTable ON rev_id = ipc_rev_id";
+            $whereClause = 'ipc_hex BETWEEN :startIp AND :endIp';
+            [$params['startIp'], $params['endIp']] = IPUtils::parseRange($user->getUsername());
+        }
+
+        $sql = "SELECT pap_project_title, SUM(`edit_count`) AS `count`
+                FROM (
+                    SELECT page_id, COUNT(page_id) AS `edit_count`
+                    FROM $revisionTable
+                    $ipcJoin
+                    JOIN $pageTable ON page_id = rev_page
+                    WHERE $whereClause
+                    AND page_namespace = :namespace
+                    $revDateConditions
+                    GROUP BY page_id
+                ) a
+                JOIN $pageAssessmentsTable ON pa_page_id = page_id
+                JOIN $paProjectsTable ON pa_project_id = pap_project_id
+                GROUP BY pap_project_title
+                ORDER BY `count` DESC
+                LIMIT 10";
+
+        $totals = $this->executeQuery($sql, $project, $user, $ns)
+            ->fetchAllAssociative();
+
+        // Cache and return.
+        return $this->setCache($cacheKey, $totals);
+    }
+
     /**
      * Get the top edits by a user across all namespaces.
      * @param Project $project
@@ -236,6 +308,17 @@ public function getTopEditsAllNamespaces(
                     LIMIT 1
                 ) AS pa_class"
             : '';
+        $paProjectsTable = $project->getTableName('page_assessments_projects');
+        $paProjectsSelect = $hasPageAssessments
+            ? ", (
+                    SELECT JSON_ARRAYAGG(pap_project_title)
+                    FROM $pageAssessmentsTable
+                    JOIN $paProjectsTable
+                    ON pa_project_id = pap_project_id
+                    WHERE pa_page_id = e.page_id
+                ) AS pap_project_title"
+            : '';
+
 
         $ipcJoin = '';
         $whereClause = 'rev_actor = :actorId';
@@ -248,7 +331,7 @@ public function getTopEditsAllNamespaces(
         }
 
         $sql = "SELECT c.page_namespace AS `namespace`, e.page_title,
-                    c.page_is_redirect AS `redirect`, c.count $paSelect
+                    c.page_is_redirect AS `redirect`, c.count $paSelect $paProjectsSelect
                 FROM
                 (
                     SELECT b.page_namespace, b.page_is_redirect, b.rev_page, b.count