SQL 子查询行作为 GROUP BY 列

Question

我想在每个task_type 的响应中都有列，计数按date_trunc('day') 和user_id 分组。因此，一旦整个查询运行，它将返回 task_type_1 列，字段值将是给定用户在给定日期具有该类型的任务数。

到目前为止，我有这个运行但不确定如何将 task_type 分组添加到此查询：

SELECT users.id AS user_id,
  date_trunc('day', workforce_assigned_tasks.created_at) AS day,
  SUM(workforce_assigned_tasks.duration) AS duration,
  SUM(workforce_assigned_tasks.earnings_cents) AS earnings_cents,
  SUM(workforce_assigned_tasks.subtask_count) AS subtask_count,
  WHAT GOES HERE?
FROM users
JOIN workforce_assigned_tasks ON workforce_assigned_tasks.user_id = users.id
JOIN workforce_tasks ON workforce_assigned_tasks.workforce_task_id = workforce_tasks.id
GROUP BY date_trunc('day', workforce_assigned_tasks.created_at), users.id;

Answer 1

您可以使用条件聚合，在 Postgres 中使用 FILTER 子句：

SELECT u.id AS user_id, date_trunc('day', wat.created_at) AS day,
       SUM(wat.duration) AS duration,
       SUM(wat.earnings_cents) AS earnings_cents,
       SUM(wat.subtask_count) AS subtask_count,
       COUNT(*) FILTER (WHERE wt.task_type_1 = 1)
FROM users u JOIN
     workforce_assigned_tasks wat
     ON wat.user_id = u.id JOIN
     workforce_tasks wt
     ON wat.workforce_task_id = wt.id
GROUP BY date_trunc('day', wat.created_at), u.id;

我猜task_type 在workforce_tasks 中。

请注意，表别名的使用使查询更易于编写和阅读。