ORACLE SQL基於不同的基於時間的事件連接表

Question

我有2個表：

• PROCESSOR_EVENT（ID，時間，機，名稱，狀態）
• COMM_EVENT（ID，時間，機，名稱，類型）

我想檢索PROCESSOR_EVENT.Time和下一個PROCESSOR_EVENT.Time之前的下一個COMM_EVENT之間的時間差。

即在下一個PROCESSOR_EVENT之前發生的PROCESSOR_EVENT之後的第一個COMM_EVENT事件。

示例數據：

PROCESSOR_EVENT(Id, Time, Machine, Name, State) 
1, 01:00:00, TRX4, Ignition, Heated 
2, 01:00:03, TRX1, Movement, Triangulating 
3, 01:00:23, TRX4, Movement, Heated 
4, 01:00:32, TRX4, Direction Change, Stable 
5, 01:00:56, TRX4, Stopping, Heated 

COMM_EVENT(Id, Time, Machine, Name, Type) 
1, 01:00:02, TRX4, Direction request, Request 
2, 01:00:15, TRX4, Direction acknowledgement, Acknowledgement 
3, 01:00:16, TRX1, Position change, Command 
4, 01:00:34, TRX4, Direction request, Request 
5, 01:01:02, TRX4, Position change, Command 

預期回報：

PROCESSOR_EVENT.Time, PROCESSOR_EVENT.Machine, PROCESSOR_EVENT.Name, COMM_EVENT.Time, COMM_EVENT.Type, TimeDifference 
01:00:00, TRX4, Ignition, 01:00:02, Request, 2 
01:00:03, TRX1, Movement, 01:00:16, Command, 13 
01:00:23, TRX4, Movement, null, null, null 
01:00:32, TRX4, Direction Change, 01:00:34, Request, 11 
01:00:56, TRX4, Stopping, 01:01:02, Command, 6 

如何去這樣做將是巨大的任何想法，在此先感謝您的時間。

Answer 1

嘗試此查詢：

SELECT 
    TO_CHAR(t.Time, 'HH24:MI:SS') AS "PROCESSOR_EVENT.Time", 
    t.Machine AS "PROCESSOR_EVENT.Machine", 
    t.Name AS "PROCESSOR_EVENT.Name", 
    TO_CHAR(t.ce_time, 'HH24:MI:SS') AS "COMM_EVENT.Time", 
    t.ce_type AS "COMM_EVENT.Type", 
    (t.ce_time - t.Time)*24*60*60 AS TimeDifference 
FROM (
    SELECT pe.*, 
    ROW_NUMBER() OVER (PARTITION BY pe.ID ORDER BY ce.Time) AS rn, 
    ce.Time AS ce_time, 
    ce.Type AS ce_type 
    FROM (
    SELECT PROCESSOR_EVENT.*, 
     LEAD(Time) OVER (PARTITION BY Machine ORDER BY Time) AS next_pe_time 
    FROM PROCESSOR_EVENT 
    -- here put WHERE with filtering conditions for PROCESSOR_EVENT table 
) pe 
    LEFT OUTER JOIN COMM_EVENT ce 
    ON ce.Machine = pe.Machine 
    AND ce.Time > pe.Time 
    AND (pe.next_pe_time IS NULL OR ce.Time < pe.next_pe_time) 
) t 
WHERE t.rn = 1 
ORDER BY t.Time 

測試它SQL Fiddle

Answer 2

有多種方法可以解決這個問題。在完成兩個表的結合之後，這是一種多次使用lead()的方法。基本數據是：

select pc.*, 
     lead(time) over (partition by machine order by time) as nexttime, 
     lead(which) over (partition by machine order by time) as nextwhich, 
     lead(type) over (partition by machine order by time) as nexttype 
from (select pe.time, pe.machine, pe.machine, pe.name, 'pe' as which 
     from processor_event pe 
     union all 
     select ce.time, ce.machine, ce.name, 'ce' as which 
     from comm_event ce 
    ) pc; 

接下來，我們只需要申請要這個數據的邏輯：

with pc as (
     select pc.*, 
      lead(time) over (partition by machine order by time) as nexttime, 
      lead(which) over (partition by machine order by time) as nextwhich, 
      lead(type) over (partition by machine order by time) as nexttype 
     from (select pe.time, pe.machine, pe.name, NULL as type, 'pe' as which 
      from processor_event pe 
      union all 
      select ce.time, ce.machine, ce.name, ce.type, 'ce' as which 
      from comm_event ce 
      ) pc 
    ) 
select pc.time, pc.machine, pc.name, 
     (case when nextwhich = 'ce' then nexttime end) as comm_time, 
     (case when nextwhich = 'ce' then nexttype end) as comm_type, 
     ((case when nextwhich = 'ce' then nexttime end) - time) as diff 
from pc 
where pc.which = 'pe';