sql - 尝试从 SQL 中的组中获取第一条记录

Question

我正在尝试查询候选人、项目、spark SQL 合同的最早时间戳。

spark.sql(
      """
        |SELECT
        | DISTICT
        | timestamp,
        | candidate_id,
        | project_id,
        | contract_id
        |FROM candidatesHistory
        |GROUP BY timestamp, candidate_id, project_id, contract_id
        |ORDER BY timestamp DESC
        |LIMIT 1
        |""".stripMargin)

这段代码不这样做，它只是获取一条记录 - 我如何获得合同项目候选人的最旧时间戳？

任何帮助表示赞赏

Answer 1

如果表中只有 4 列，则可以使用聚合：

select candidate_id, project_id, contract_id, min(timestamp) first_timestamp
from candidateshistory
group by candidate_id, project_id, contract_id

如果有更多列并且您想将所有列都带进来，那么您可以使用row_number()来过滤表：

select ch.*
from (
    select ch.*,
        row_number() over(partition by candidate_id, project_id, contract_id order by timestamp) rn
    from candidateshistory ch
) ch
where rn = 1

对于每个(candidate_id, project_id, contract_id)元组，这将为您提供最早的行timestamp。

Answer 2

这应该可行，但不知道这是否是最好的方法：

SELECT candidate_id
, project_id
, contract_id
, timestamp
FROM (
    SELECT RANK() OVER (PARTITION BY candidate_id ORDER BY timestamp) AS RNK
    , candidate_id
    , project_id
    , contract_id
    FROM candidatesHistory
    ) as CH
WHERE CH.RNK = 1;