feat: add user community openrank interface (#1614)

Signed-off-by: frank-zsy <syzhao1988@126.com>
X-lab2017 · Sep 1, 2024 · dfc2f8d · dfc2f8d
1 parent 1fb895b
commit dfc2f8d
Show file tree

Hide file tree

Showing 3 changed files with 92 additions and 5 deletions.
diff --git a/labeled_data/regions/CN.yml b/labeled_data/regions/CN.yml
@@ -1370,3 +1370,5 @@ data:
           name: xudafeng
         - id: 5351546
           name: Xuanwo
+        - id: 499550
+          name: yyx990803
diff --git a/src/metrics/basic.ts b/src/metrics/basic.ts
@@ -161,7 +161,7 @@ export const getRepoWhereClause = async (config: QueryConfig): Promise<string |
 };
 
 // User
-export const getUserWhereClause = async (config: QueryConfig): Promise<string | null> => {
+export const getUserWhereClause = async (config: QueryConfig, idCol: string = 'actor_id'): Promise<string | null> => {
   const userWhereClauseArray: string[] = [];
 
   // id or names
@@ -171,10 +171,10 @@ export const getUserWhereClause = async (config: QueryConfig): Promise<string |
         // convert user login to id
         const sql = `SELECT any(actor_id) AS id FROM events WHERE actor_login IN [${p.userLogins.map(n => `'${n}'`)}] AND platform='${p.platform}' GROUP BY actor_login`;
         const userIds = (await query<any>(sql, { format: 'JSONEachRow' })).map(r => r.id);
-        userWhereClauseArray.push(`(actor_id IN [${userIds.join(',')}] AND platform='${p.platform}')`);
+        userWhereClauseArray.push(`(${idCol} IN [${userIds.join(',')}] AND platform='${p.platform}')`);
       }
 
-      if (p.userIds && p.userIds.length > 0) userWhereClauseArray.push(`(actor_id IN [${p.userIds.join(',')}] AND platform='${p.platform}')`);
+      if (p.userIds && p.userIds.length > 0) userWhereClauseArray.push(`(${idCol} IN [${p.userIds.join(',')}] AND platform='${p.platform}')`);
     }
   }
 
@@ -184,7 +184,7 @@ export const getUserWhereClause = async (config: QueryConfig): Promise<string |
       const data = getPlatformData([l], config.injectLabelData);
       const arr: string[] = [];
       for (const p of data) {
-        if (p.users.length > 0) arr.push(`(actor_id IN [${p.users.map(u => u.id).join(',')}] AND platform='${p.name}')`);
+        if (p.users.length > 0) arr.push(`(${idCol} IN [${p.users.map(u => u.id).join(',')}] AND platform='${p.name}')`);
       }
       if (arr.length === 0) return null;
       return `(${arr.join(' OR ')})`;
@@ -196,7 +196,7 @@ export const getUserWhereClause = async (config: QueryConfig): Promise<string |
   if (config.labelUnion) {
     const data = getPlatformData(config.labelUnion, config.injectLabelData);
     for (const p of data) {
-      if (p.users.length > 0) userWhereClauseArray.push(`(actor_id IN [${p.users.map(u => u.id).join(',')}] AND platform='${p.name}')`);
+      if (p.users.length > 0) userWhereClauseArray.push(`(${idCol} IN [${p.users.map(u => u.id).join(',')}] AND platform='${p.name}')`);
     }
   }
 

diff --git a/src/metrics/indices.ts b/src/metrics/indices.ts
@@ -168,6 +168,91 @@ ${getOutterOrderAndLimit({ ...config, order: undefined }, 'openrank')}
   return processQueryResult(result, ['openrank']);
 };
 
+interface UserCommunityOpenRankConfig {
+  limit: number;
+  withBot: boolean;
+};
+export const getUserCommunityOpenrank = async (config: QueryConfig<UserCommunityOpenRankConfig>) => {
+  config = getMergedConfig(config);
+  const whereClause: string[] = [];
+  const repoWhereClause = await getRepoWhereClause(config);
+  if (repoWhereClause) whereClause.push(repoWhereClause);
+  const userWhereClause = await getUserWhereClause(config, 'id');
+  const timeRangeClause = getTimeRangeWhereClause(config);
+  if (timeRangeClause) whereClause.push(timeRangeClause);
+  const limit = (config.options?.limit == undefined) ? 30 : config.options.limit;
+  if (config.options?.withBot === false) {
+    const botLabelData = getPlatformData([':bot']);
+    for (const b of botLabelData) {
+      whereClause.push(`(NOT (platform='${b.name}' AND actor_id IN [${b.users.map(u => u.id).join(',')}]))`);
+    }
+    whereClause.push(`(actor_login NOT LIKE '%[bot]')`);
+  }
+
+  const sql = `
+SELECT
+  id,
+  ${getTopLevelPlatform(config)},
+  argMax(name, time) AS name,
+  ${getGroupArrayInsertAtClause(config, { value: 'openrankValue', key: 'openrank' })},
+  ${getGroupArrayInsertAtClause(config, { key: 'openrankDetails', noPrecision: true, defaultValue: '[]' })}
+FROM
+(
+  SELECT
+    id, argMax(name, time) AS name, platform, time,
+    ${limit > 0 ?
+      `arraySlice(groupArray((platform, repo_id, repo_name, openrank)), 1, ${limit}) AS openrankDetails` :
+      `groupArray((platform, repo_id, repo_name, openrank)) AS openrankDetails`},
+    SUM(openrank) AS openrankValue
+  FROM
+    (
+      SELECT
+        ${getGroupIdClause(config, 'user', 'time')},
+        time,
+        platform,
+        repo_id,
+        argMax(repo_name, time) AS repo_name,
+        SUM(openrank) AS openrank
+      FROM
+        (
+          SELECT
+            ${getGroupTimeClause(config, 'g.created_at')},
+            g.platform AS platform,
+            g.repo_id AS repo_id,
+            argMax(g.repo_name, time) AS repo_name,
+            argMax(g.org_id, time) AS org_id,
+            argMax(g.org_login, time) AS org_login,
+            c.actor_id AS actor_id,
+            argMax(c.actor_login, time) AS actor_login,
+            SUM(c.openrank * g.openrank / r.openrank) AS openrank
+          FROM
+            (SELECT * FROM community_openrank WHERE ${whereClause.join(' AND ')}) c,
+            (SELECT * FROM global_openrank WHERE ${whereClause.join(' AND ')}) g,
+            (SELECT repo_id, platform, created_at, SUM(openrank) AS openrank FROM community_openrank WHERE actor_id > 0 AND ${whereClause.join(' AND ')} GROUP BY repo_id, platform, created_at) r
+          WHERE
+            c.actor_id > 0
+            AND c.repo_id = g.repo_id
+            AND c.platform = g.platform
+            AND c.created_at = g.created_at
+            AND g.repo_id = r.repo_id
+            AND g.platform = r.platform
+            AND g.created_at = r.created_at
+          GROUP BY
+            platform, repo_id, actor_id, time
+        ) data
+      GROUP BY id, repo_id, platform, time
+      ORDER BY openrank DESC
+    )
+    GROUP BY id, platform, time
+    ${userWhereClause ? `HAVING ${userWhereClause}` : ''}
+)
+GROUP BY id, platform
+${getOutterOrderAndLimit(config, 'openrank')}
+`;
+  const result = await clickhouse.query(sql);
+  return processQueryResult(result, ['openrank', 'openrankDetails']);
+};
+
 export const basicActivitySqlComponent = `
     if(type='PullRequestEvent' AND action='closed' AND pull_merged=1, issue_author_id, actor_id) AS actor_id,
     argMax(if(type='PullRequestEvent' AND action='closed' AND pull_merged=1, issue_author_login, actor_login), created_at) AS actor_login,