我试图获得某些测试结果的总体通过率。
序列号代表仪器。有一种仪器可以通过107种不同的测试。测试分为两部分：
一种文书将被视为两种方式的传递。
标准方式：它在一次运行中通过107次不同的测试。代码如下：
SELECt序列号 从表 GROUP BY序列号，条目 具有计数（DISTINCT testcriteria）＆gt; = 107
替代方式：它在一次运行中通过107次不同的测试 它在一轮中通过86次其他测试，并通过21次Linux测试 后来回合。
表中的“overalltestrunstatus”列仅表示仪器通过列表中的所有测试，但列表本身可能不完整。 （假设一个乐器得到“整体调整状态”，但是它通过的测试是<107，它仍然是失败的）。有时，如果仪器失败或通过小于一定数量的测试，仪器将重新运行。因此，“输入”代表该仪器运行的回合。
我的目标是找到那些通过测试但不是以标准方式通过的仪器。
我的想法是找到所有那些具有“overalltestrunstatus”等于通过但通过小于107的测试的乐器。然后迭代结果和乐器的每个条目。如果它通过了86个其他测试，我会给它一个检查，如果通过其他21个Linux，我会给它另一个检查。稍后我会计算所有那些有两张支票的乐器。
我想要的是：
+----------------+------------+---------------+------------+
| serial number | entries | others tests | Linux tests|
+----------------+------------+---------------+------------+
| 170119904 | 1 | yes | no |
+----------------+------------+---------------+------------+
| 170119904 | 2 | no | yes |
+----------------+------------+---------------+------------+
| 180117000 | 1 | no | no |
+----------------+------------+---------------+------------+
| 180117000 | 2 | no | no |
+----------------+------------+---------------+------------+
通过检查特定序列号是否已通过其他测试和Linux测试。我能够知道它的通过与否。
我的问题是，每次查询运行20个小时，然后发出错误，如“已断开连接”或“发生某些内部错误”。我永远无法真正理解它。有什么想法吗？我非常感谢你的帮助。
这是表格：
+----------------------+--------------+----------+----------------+------------+---------+
| overalltestrunstatus | serialnumber | passfail | testcriteria | testname | Entries |
+----------------------+--------------+----------+----------------+------------+---------+
| PASS | 170119904 | Pass | Backlight test | Functional | 1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS | 170119904 | Pass | Linux set test | Functional | 1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS | 170119904 | Pass | Factor test | Functional | 1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS | 170119904 | Pass | Voltage test | Functional | 1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS | 170119904 | Pass | Digital test | Functional | 1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS | 170119904 | Pass | Detactor test | Functional | 1 |
+----------------------+--------------+----------+----------------+------------+---------+
这是我的代码：
/* -serialnumber |-- entries--|--Pass1 --|--Pass2---|
--------------|------------|----------|----------|
*/
DECLARE @checklist TABLE
(
serialnumber varchar(100),
entries int,
passother int default 0,
passlinux int default 0
)
DECLARE @temp TABLE
(
serialnumber int,
entries int,
passother int default 0,
passlinux int default 0
)
-- select serial number into the table
-- PCBA 214 ge to be counted
INSERT INTO @checklist (serialnumber, entries)
SELECT serialnumber, entries
FROM dbo.boardtestresults1_full_view t2
WHERE partnumber = 'PCBA-10214-0001'
AND t2.serialnumber IN
(SELECT DISTINCT serialnumber
FROM dbo.boardtestresults1_full_view
WHERE partnumber = 'PCBA-10214-0001' AND overalltestrunstatus = 'Pass' AND serialnumber NOT IN (
SELECT DISTINCT serialnumber
FROM dbo.boardtestresults1_full_view
WHERE partnumber = 'PCBA-10214-0001' AND overalltestrunstatus = 'Pass'
GROUP BY serialnumber, Entries
HAVING COUNT(DISTINCT combined) >= 107))
-- check all for PCBA 10214
DECLARE @MyCursor1 CURSOR;
DECLARE @snumber VARCHAR(100); -- serailnumber
DECLARE @entries VARCHAR(100); -- entries
DECLARE @others int;
DECLARE @linux int;
DECLARE @pass1 int;
DECLARE @pass2 int;
BEGIN
-- get serial numbers for all PCBA 214 from checklist
SET @MyCursor1 = CURSOR FOR
SELECT serialnumber, Entries
FROM @checklist
WHERE serialnumber like 'PCBA-10214-0001%'
-- get the roll
OPEN @MyCursor1
FETCH NEXT FROM @MyCursor1
INTO @snumber, @entries
-- loop through
WHILE @@FETCH_STATUS = 0
BEGIN
/*
YOUR ALGORITHM GOES HERE
*/
-- if pass the most or if pass linux
IF ((SELECT COUNT(DISTINCT combined)FROM dbo.boardtestresults1_full_view WHERE serialnumber = @snumber AND entries = @entries) >= 86)
UPDATE @checklist SET passother = 1 WHERE serialnumber = @snumber AND entries = @entries
ELSE IF ((SELECT COUNT(DISTINCT combined)FROM dbo.boardtestresults1_full_view WHERE serialnumber = @snumber AND entries = @entries AND testname = 'LINUX Test' AND overalltestrunstatus = 'Pass') >= 21)
UPDATE @checklist SET passlinux = 1 WHERE serialnumber = @snumber AND entries = @entries
END;
CLOSE @MyCursor1 ;
DEALLOCATE @MyCursor1;
END;
答案 0 :(得分：1)
这是一个工作示例，PassFail列未被评估，而它在那里它没有查询的限制。我添加了一个名为TestType的列来简化示例，在您的用例中不太清楚Linux测试是如何从样本集派生的，也许是TestCriteria，但这使得它不那么模糊。您应该能够获得测试的二进制文件，如果您需要进一步的帮助，您需要定义所有数据，我们可以提供帮助。
declare @example as table (
ExampleID int identity(1,1) not null primary key clustered
, OverAllStatus nvarchar(255) not null
, SN int not null
, PassFail nvarchar(255) not null
, TestType bit not null
, TestCriteria nvarchar(255) not null
, Entry_ int not null
);
insert into @example (OverAllStatus, SN, PassFail, TestType, TestCriteria, Entry_)
select 'PASS', 170119904, 'Pass', 0, 'Backlight test', 1 union all
select 'PASS', 170119904, 'Pass', 0, 'Factor test', 1 union all
select 'PASS', 170119904, 'Pass', 0, 'Voltage test', 1 union all
select 'PASS', 170119904, 'Pass', 0, 'Detactor test', 1 union all
select 'PASS', 170119904, 'Pass', 0, 'Backlight test', 2 union all
select 'PASS', 170119904, 'Pass', 1, 'Linux set test', 2 union all
select 'PASS', 170119904, 'Pass', 0, 'Factor test', 2 union all
select 'PASS', 170119904, 'Pass', 0, 'Voltage test', 2 union all
select 'PASS', 170119904, 'Fail', 0, 'Digital test', 2 union all
select 'PASS', 170119904, 'Pass', 0, 'Detactor test', 2 union all
select 'PASS', 180117000, 'Pass', 0, 'Backlight test', 1 union all
select 'PASS', 180117000, 'Pass', 1, 'Linux set test', 1 union all
select 'PASS', 180117000, 'Pass', 0, 'Factor test', 1 union all
select 'PASS', 180117000, 'Pass', 0, 'Digital test', 2 union all
select 'PASS', 180117000, 'Pass', 0, 'Detactor test', 1;
--| Control Group, only those that pass but does not have 107 records (OverAllStatus)
;with cte as (
select SN
, Entry_
, count(*) cnt
from @example
where OverAllStatus = 'Pass'
group by SN, Entry_
--| Control count - this is your 107 or the numbers of tests the SN has passed
having count(*) < 7
)
--| We are setting the count of records to the Test Type or the grouping
, cte2 as (
select b.SN
, a.Entry_
, iif(TestType = 1, 'Linux', 'Other') Test
from cte a
join @example b
on a.SN = b.SN
and a.Entry_ = b.Entry_
)
--| Roll up the counts
, cte3 as (
select distinct a.SN
, a.Entry_
, count(*) over(partition by a.SN, a.Test, a.Entry_) cnt
, a.Test
from cte2 a
)
--| Pivot the result set
select SN
, Entry_
, [Other]
, [Linux]
from (
select SN
, Entry_
, cnt
, test
from cte3
) as x
pivot
(
max(cnt)
for Test in ([Other], [Linux])
) as p;
结果集：
SN Entry_ Other Linux
170119904 1 4 NULL
180117000 1 3 1
170119904 2 5 1
180117000 2 1 NULL