scala - 如何为每个 uniqueId 获取具有最大时间戳的行

Question

我有一个定义如下的表:

mysql> select * from rules;
+----+----------+-------+--------------+---------------+---------------------+
| id | uniqueId | name  | json         | definition    | createdAt           |
+----+----------+-------+--------------+---------------+---------------------+
|  1 | ABC      | Rule1 | Json Repr v1 | Rule 1 Def v1 | 2016-05-13 09:10:44 |
|  2 | DEF      | Rule2 | Json Repr v1 | Rule 2 Def v1 | 2016-05-13 09:12:04 |
|  3 | ABC      | Rule1 | Json Repr v2 | Rule 1 Def v2 | 2016-05-13 09:14:04 |
|  4 | ABC      | Rule1 | Json Repr v3 | Rule 1 Def v3 | 2016-05-13 09:14:15 |
|  5 | ABC      | Rule1 | Json Repr v4 | Rule 1 Def v4 | 2016-05-13 09:14:22 |
|  6 | DEF      | Rule2 | Json Repr v2 | Rule 2 Def v2 | 2016-05-13 09:14:31 |
+----+----------+-------+--------------+---------------+---------------------+

我知道我要运行的 SQL 查询来获取所需的行:

mysql> select * from rules where createdAt in (select max(createdAt) from rules group by uniqueId);
+----+----------+-------+--------------+---------------+---------------------+
| id | uniqueId | name  | json         | definition    | createdAt           |
+----+----------+-------+--------------+---------------+---------------------+
|  5 | ABC      | Rule1 | Json Repr v4 | Rule 1 Def v4 | 2016-05-13 09:14:22 |
|  6 | DEF      | Rule2 | Json Repr v2 | Rule 2 Def v2 | 2016-05-13 09:14:31 |
+----+----------+-------+--------------+---------------+---------------------+

我试图获得 Slick 查询，但我被卡住了。到目前为止，我可以按如下方式构建子查询:

val latestVersionOfRules = Rules.groupBy(p => p.uniqueId).map{ case(uniqueId, group) => (uniqueId, group.map(_.createdAt).max) }

但是，我无法访问 createdAt 值以与外部表连接。

我不确定如何构建上面的完整查询。

Answer 1

首先:您的查询原样并没有完全按照您的要求执行，因为它只是根据查询中的时间戳进行选择。如果您有多个具有最大时间戳的行，您将选择所有这些行。例如，如果您有:

mysql> select * from rules;
+----+----------+-------+--------------+---------------+---------------------+
| id | uniqueId | name  | json         | definition    | createdAt           |
+----+----------+-------+--------------+---------------+---------------------+
|  1 | ABC      | Rule1 | Json Repr v1 | Rule 1 Def v1 | 2016-05-13 09:10:44 |
|  2 | DEF      | Rule2 | Json Repr v1 | Rule 2 Def v1 | 2016-05-13 09:14:22 |
|  3 | ABC      | Rule1 | Json Repr v2 | Rule 1 Def v2 | 2016-05-13 09:14:04 |
|  4 | ABC      | Rule1 | Json Repr v3 | Rule 1 Def v3 | 2016-05-13 09:14:15 |
|  5 | ABC      | Rule1 | Json Repr v4 | Rule 1 Def v4 | 2016-05-13 09:14:22 |
|  6 | DEF      | Rule2 | Json Repr v2 | Rule 2 Def v2 | 2016-05-13 09:15:31 |
+----+----------+-------+--------------+---------------+---------------------+

请注意，第 2 行的时间戳等于 ABC 的最大时间戳。

你会回来:

mysql> select * from rules where createdAt in (select max(createdAt) from rules group by uniqueId);
+----+----------+-------+--------------+---------------+---------------------+
| id | uniqueId | name  | json         | definition    | createdAt           |
+----+----------+-------+--------------+---------------+---------------------+
|  2 | DEF      | Rule2 | Json Repr v1 | Rule 2 Def v1 | 2016-05-13 09:14:22 |
|  5 | ABC      | Rule1 | Json Repr v4 | Rule 1 Def v4 | 2016-05-13 09:14:22 |
|  6 | DEF      | Rule2 | Json Repr v2 | Rule 2 Def v2 | 2016-05-13 09:15:31 |
+----+----------+-------+--------------+---------------+---------------------+

不是你想要的!在 SQL 中，您将加入一个子选择。这可以在 Slick 中完成:

// Create the query getting you the unique ID and max.
val latestVersionOfRules = Rules.groupBy(p => p.uniqueId).map {
  case(uniqueId, group) => (uniqueId, group.map(_.createdAt).max)
}

// Join your table to the query. This is a DBIO of whatever your Rules
// result type is.
val results = Rules.join(latestVersionOfRules).on {
  case (rulesRow, groupRow) =>
    rulesRow.uniqueId === groupRow._1 &&
      rulesRow.createdAt === groupRow._2
}.map {
  case (rulesRow, _) => rulesRow
}.result

最后要注意的一件事:如果多行具有相同的最大时间戳，这将仍然为每个 uniqueId 返回多行。