SQL 窗口函数，用于检测列值的变化

Question

我想检测此（示例）数据库中列值的变化

    WITH events(id, row,event) AS (
    VALUES
    (1,1, 0 )
   ,(1,2, 0 ) 
   ,(1,3, 1 )
   ,(1,4, 0 )
   ,(1,5, 1 )
   ,(2,1, 0 )
   ,(2,2, 1 )
   ,(3,1, 0 )
   ,(3,2, 0 )
   )
   select * from events

我正在寻找的是新列"代码"的代码，该列在 1 之后切换到 1 Sde 事件列显示 1。 在相同的 id 中，代码保持 1。对于此示例，此新列将如下所示

    WITH events2(id, row,event, code) AS (
    VALUES
    (1,1, 0, 0 )
   ,(1,2, 0, 0 ) 
   ,(1,3, 1, 0 )
   ,(1,4, 0, 1 ) -- notice the switch here
   ,(1,5, 1, 1 ) -- 
   ,(2,1, 0, 0 )
   ,(2,2, 1, 0 )
   ,(3,1, 0, 0 )
   ,(3,2, 0, 0 )
   )
   select * from events2

我有一种预感，答案将与这个问题的答案有关：PostgreSQL窗口功能：通过比较进行分区

不知何故，我自己无法弄清楚这一点..

彼得

Answer 1

合并标量子查询：

WITH events(id, zrow, zevent) AS (
    VALUES
    (1,1, 0 ) ,(1,2, 0 ) ,(1,3, 1 ) ,(1,4, 0 ) ,(1,5, 1 )
   ,(2,1, 0 ) ,(2,2, 1 )
   ,(3,1, 0 ) ,(3,2, 0 )
   )
SELECT id, zrow, zevent
        , COALESCE((SELECT 1 FROM events ex WHERE ex.id = ev.id AND ex.zrow < ev.zrow AND ex.zevent> 0),0) AS oevent
 FROM events ev
        ;

---

或者，通过将布尔值 EXISTS（） 类型转换为 INTEGER 来避免 COALESCE（）：

WITH events(id, zrow,event) AS (
    VALUES
    (1,1, 0 ) ,(1,2, 0 ) ,(1,3, 1 ) ,(1,4, 0 ) ,(1,5, 1 )
   ,(2,1, 0 ) ,(2,2, 1 )
   ,(3,1, 0 ) ,(3,2, 0 )
   )
SELECT id, zrow, event
        , EXISTS(SELECT 1 FROM events ex WHERE ex.id = ev.id AND ex.zrow < ev.zrow AND ex.event> 0)::integer AS oevent
FROM events ev
        ;

---

在同一组（帧）中查找先前记录的 MAX（） 值：

WITH events(id, zrow,event) AS (
    VALUES
    (1,1, 0 ) ,(1,2, 0 ) ,(1,3, 1 ) ,(1,4, 0 ) ,(1,5, 1 )
   ,(2,1, 0 ) ,(2,2, 1 )
   ,(3,1, 0 ) ,(3,2, 0 )
   )
, drag AS (
        SELECT id, zrow, event, MAX(event)
                OVER (PARTITION BY id
                        ORDER BY zrow
                        ROWS BETWEEN UNBOUNDED PRECEDING AND 1 PRECEDING
                        ) AS lagged
        FROM events ev
        )
SELECT id, zrow, event
        , COALESCE(lagged,0) AS oevent
        FROM drag dr
        ;

---

没有额外的 CTE 也是如此：

WITH events(id, zrow,event) AS (
    VALUES
    (1,1, 0 ) ,(1,2, 0 ) ,(1,3, 1 ) ,(1,4, 0 ) ,(1,5, 1 )
   ,(2,1, 0 ) ,(2,2, 1 )
   ,(3,1, 0 ) ,(3,2, 0 )
   )
SELECT id, zrow, event, COALESCE(MAX(event) OVER (PARTITION BY id
                                                ORDER BY zrow
                                                ROWS BETWEEN UNBOUNDED PRECEDING AND 1 PRECEDING
                                ),0) AS lagged
        FROM events ev
        ;

---

执行自联接的另一种方法是使用递归查询。