您将如何更改或优化此 postgres sql 查询以使输出格式保持不变？

Question

原来又贵又慢的查询是这样的：

select 'INTEGRATED' as "STATUS", coalesce(sum(response_count),0) as "COUNT" from f_wfc_transaction_summary_isocc_ft_sum_no_tid('2015-07-22 00:00:00','2015-08-05 01:00:00', null, 'LINK')
where wfc_result in ('TM_OK', 'TM_NO_CHANGE')
union
select 'WFC_FALLOUT' as "STATUS", coalesce(sum(response_count),0) as "COUNT" from f_wfc_transaction_summary_isocc_ft_sum_no_tid('2015-07-22 00:00:00','2015-08-05 01:00:00', null, 'LINK')
where wfc_result LIKE 'WFC%'
union
select 'TM_FALLOUT' as "STATUS", coalesce(sum(response_count),0) as "COUNT" from f_wfc_transaction_summary_isocc_ft_sum_no_tid('2015-07-22 00:00:00','2015-08-05 01:00:00', null, 'LINK')
where wfc_result = 'TM_FAIL'
union
select 'WFC_PENDING' as "STATUS", coalesce(sum(response_count),0) as "COUNT" from f_wfc_transaction_summary_isocc_ft_sum_no_tid('2015-07-22 00:00:00','2015-08-05 01:00:00', null, 'LINK')
where wfc_result = 'PENDING';

输出：

Status       Count
Integrated     40
TM_FALLOUT      50
WFC_PENDING      60
WFC_FALLOUT      70

上面的输出格式是我需要的。但是这个查询需要很多时间。

我想使用的查询如下，因为它花费的时间更少。

select 
sum(CASE WHEN wfc_result IN ('TM_OK','TM_NO_CHANGE') THEN response_count ELSE 0 END) as "INTEGRATED", 
sum(CASE WHEN wfc_result LIKE 'WFC%' THEN response_count ELSE 0 END) as "WFC_FALLOUT",
sum(CASE WHEN wfc_result = 'TM_FAIL' THEN response_count ELSE 0 END) as "TM_FALLOUT",
sum(CASE WHEN wfc_result = 'PENDING' THEN response_count ELSE 0 END) as "WFC_PENDING"
from f_wfc_transaction_summary_isocc_ft_sum_no_tid('2015-07-22 00:00:00','2015-08-05 01:00:00',null, 'LINK')

但是，此查询的输出如下，我需要与第一个查询相同的输出。

Integrated   WFC_FALLOUT   TM_FALLOUT   WFC_PENDING
40             50            60              70

我尝试了几种方法，但找不到如何编辑它？

Answer 1

假设 'wfc_result' WFC% 始终是 WFC_% 你可以执行以下操作：

SELECT status.lbl AS status, coalesce(sum(response_count), 0) AS count
FROM (VALUES
  ('TM_O', 'INTEGRATED'),
  ('TM_N', 'INTEGRATED'),
  ('WFC_', 'WFC_FALLOUT'),
  ('TM_F', 'TM_FALLOUT'),
  ('PEND', 'WFC_PENDING')) status(cls, lbl)
LEFT JOIN f_wfc_transaction_summary_isocc_ft_sum_no_tid('2015-07-22 00:00:00', '2015-08-05 01:00:00', NULL, 'LINK') wfc
  ON substr(wfc.wfc_result, 1, 4) = status.cls
GROUP BY 1;

如果只有几个 'wfc_result' 字符串 LIKE 'WFC%' 你也可以拼出它们（以及其他类）在 VALUES 子句中，然后在 JOIN 子句中删除 substr() 函数。

要得到总输出，查询就简单多了：

SELECT sum(response_count) AS total_count
FROM f_wfc_transaction_summary_isocc_ft_sum_no_tid('2015-07-22 00:00:00', '2015-08-05 01:00:00', NULL, 'LINK');

这当然会再次调用昂贵的函数，但是没有办法在数据库上解决这个问题；在您的客户端应用程序中，您可以简单地总结从上一个查询返回的 "count" 值。

您将如何更改或优化此 postgres sql 查询以使输出格式保持不变？

How would you change or rather optimize this postgres sql query so that the output format remains the same?

postgresql

query-optimization