无法在子查询中使用 conditional_true_event 的结果

Question

我的桌子是

         ts          | symbol | bid
---------------------+--------+------
 2009-01-01 03:00:00 | XYZ    |   10
 2009-01-01 03:00:03 | XYZ    |   11
 2009-01-01 03:00:06 | XYZ    | 10.5
 2009-01-01 03:00:09 | XYZ    |   11
 2009-01-01 03:00:00 | ABC    |   10
 2009-01-01 03:00:03 | ABC    |   11
 2009-01-01 03:00:06 | ABC    |   11
 2009-01-01 03:00:09 | ABC    |   11

我想运行一个查询，首先获取符号为ABC 的所有行，然后我想根据条件将它们放在窗口中，然后选择第二个窗口。

子查询有效（如果不使用子查询）

select * 
from 
    (select 
         ts, symbol, bid, 
         conditional_true_event(bid > 10.6) over(order by ts) as window 
     from Tickstore3 
     where symbol = 'ABC')  as t3;

输出：

         ts          | symbol | bid | window
---------------------+--------+-----+--------
 2009-01-01 03:00:00 | ABC    |  10 |      0
 2009-01-01 03:00:03 | ABC    |  11 |      1 <-- I want to get only this row
 2009-01-01 03:00:06 | ABC    |  11 |      2
 2009-01-01 03:00:09 | ABC    |  11 |      3 

但我不能只得到第二行

select * 
from 
    (select 
         ts, symbol, bid, 
         conditional_true_event(bid > 10.6) over(order by ts) as window 
     from Tickstore3 
     where symbol = 'ABC') as t3 
where window = 1;

我收到一个错误：

错误 4856：在字符 151 处或附近的“窗口”处出现语法错误
LINE 1: ...w from Tickstore3 where symbol='ABC') as t3 where window=1;

Answer 1

发现了错误。我不能推荐window。我必须将它与临时结果名称一起使用

select * from (select ts, symbol, bid, conditional_true_event(bid > 10.6) over(order by ts) as
 window from Tickstore3 where symbol='ABC') as t3 where t3.window=1 ; <-- note t3.window
         ts          | symbol | bid | window
---------------------+--------+-----+--------
 2009-01-01 03:00:03 | ABC    |  11 |      1
(1 row)

Answer 2

我认为CONDITIONAL_TRUE_EVENT() 在这里完全是矫枉过正。

有条件的运行总和（使用CASE WHEN ... 表达式）会做得很好。

而且，只是为了好玩，我添加了 named window 子句，以说明为什么它是保留字 - 我认为 counter 是您想要的更合适的名称过滤...

如果您明智地选择，正确的程序员的话可以使任何文档或注释变得多余...

WITH                                                                                                                                                                                                    
-- your input ...
tickstore(ts,symbol,bid) AS (
          SELECT TIMESTAMP '2009-01-01 03:00:00','XYZ',10
UNION ALL SELECT TIMESTAMP '2009-01-01 03:00:03','XYZ',11
UNION ALL SELECT TIMESTAMP '2009-01-01 03:00:06','XYZ',10.5
UNION ALL SELECT TIMESTAMP '2009-01-01 03:00:09','XYZ',11
UNION ALL SELECT TIMESTAMP '2009-01-01 03:00:00','ABC',10
UNION ALL SELECT TIMESTAMP '2009-01-01 03:00:03','ABC',11
UNION ALL SELECT TIMESTAMP '2009-01-01 03:00:06','ABC',11
UNION ALL SELECT TIMESTAMP '2009-01-01 03:00:09','ABC',11
)
-- real query starts here, replace following comma with "WITH" ...
,
w_counter AS (
  SELECT
    *
  , SUM(CASE WHEN bid > 10.6 THEN 1 END) OVER w AS counter
  FROM tickstore
  WHERE symbol='ABC'
  WINDOW w AS (ORDER BY ts)
)
SELECT
  ts
, symbol
, bid
, counter
FROM w_counter
WHERE counter=1
;
-- out          ts          | symbol | bid  | counter 
-- out ---------------------+--------+------+---------
-- out  2009-01-01 03:00:03 | ABC    | 11.0 |       1