Not Exists 不能很好地处理多列
Posted
技术标签:
【中文标题】Not Exists 不能很好地处理多列【英文标题】：Not Exists not playing well with multiple columns 【发布时间】：2016-03-18 19:38:30 【问题描述】：这一定是我刚刚错过的一些简单的事情......
我有一个临时表这样说：
CREATE TABLE #tsa
(
AttendeeID int,
SurveyID int,
TrainingAttendeeID int
)
我使用 TOP 1 获得了一条类似这样的记录：
SELECT
TOP 1
@AttendeeID=ta.AttendeeID,
@SurveyID=ts.SurveyID,
@TrainingAttendeeID = ta.TrainingAttendeeID
FROM
TrainingAttendee ta
INNER JOIN
[User] u
ON
u.UserID= ta.AttendeeID
INNER JOIN
[Training] t
ON
t.TrainingID = ta.TrainingID
INNER JOIN
[TrainingSet] ts
ON
ts.TrainingSetID = t.TrainingSetID
WHERE
ta.AttendedTraining = 1
AND ta.ConfirmAttendedEmailOn IS NOT NULL
--only get people who didn't fill out the survey
AND NOT EXISTS (SELECT * FROM SurveyTaken st WHERE st.AddedByUserID = ta.AttendeeID AND st.SurveyID = ts.SurveyID)
ORDER BY
ta.AttendeeID,
ts.SurveyID
一旦我得到这条记录，我就将它存储到我的临时表中：
--insert into our temp table
INSERT INTO #tsa(AttendeeID, SurveyID, TrainingAttendeeID)
VALUES(@AttendeeID, @SurveyID, @TrainingAttendeeID)
然后我需要检查一些数据并发送电子邮件的整个过程......一旦发送电子邮件，我需要拿起下一条记录，不包括我之前的记录......所以没有显示太多代码：
WHILE SomeCondition
BEGIN
--do some thing...
--pick up next one
--grab next one
SELECT
TOP 1
@AttendeeID = ta.AttendeeID,
@SurveyID=ts.SurveyID,
@TrainingAttendeeID=ta.TrainingAttendeeID
FROM
TrainingAttendee ta
INNER JOIN
[User] u
ON
u.UserID= ta.AttendeeID
INNER JOIN
[Training] t
ON
t.TrainingID = ta.TrainingID
INNER JOIN
[TrainingSet] ts
ON
ts.TrainingSetID = t.TrainingSetID
WHERE
(
--same where as original
ta.AttendedTraining = 1
AND (ta.ConfirmAttendedEmailOn IS NOT NULL)
AND (NOT EXISTS (SELECT * FROM SurveyTaken st WHERE st.AddedByUserID = ta.AttendeeID AND st.SurveyID = ts.SurveyID))
--adding the piece such that we compare against the temp table...
AND (NOT EXISTS (SELECT * FROM #tsa tempS WHERE tempS.AttendeeID = ta.AttendeeID AND tempS.SurveyID = ts.SurveyID AND tempS.TrainingAttendeeID = ta.TrainingAttendeeID))
)
ORDER BY
ta.AttendeeID,
ts.SurveyID
--insert into our temp table
INSERT INTO #tsa(AttendeeID, SurveyID, TrainingAttendeeID)
VALUES(@AttendeeID, @SurveyID, @TrainingAttendeeID)
END
注意其中的 where 条件..我又添加了一个 AND...即：
--adding the piece such that we compare against the temp table...
AND (NOT EXISTS (SELECT * FROM #tSa tempS WHERE tempS.AttendeeID = ta.AttendeeID AND tempS.SurveyID = ts.SurveyID AND tempS.TrainingAttendeeID = ta.TrainingAttendeeID))
只是为了确保我不会重复使用我已经在我的临时表中处理过的记录......你会注意到我在最后也重新插入到我的临时表中......
--insert into our temp table
INSERT INTO #tsa(AttendeeID, SurveyID, TrainingAttendeeID)
VALUES(@AttendeeID, @SurveyID, @TrainingAttendeeID)
每次我运行这个存储过程时，它都会无限地运行，所以我相信此时我的情况有问题。我脑子里放了个屁……或者办公室里的噪音太大了。我在这里想念什么？我放置了一个打印语句，它一直在处理相同的记录......所以有些东西告诉我 where 子句中的最后一个条件不正确。
编辑
这是整个过程...我的问题是记录集中只有一条记录...但是存储过程继续处理同一条记录
PROCEDURE ScriptSendTrainingSurveyReminders
AS
BEGIN
-- SET NOCOUNT ON added to prevent extra result sets from
-- interfering with SELECT statements.
SET NOCOUNT ON;
DECLARE @AttendeeID int
DECLARE @TrainingAttendeeID int
DECLARE @SurveyID int
DECLARE @Message nvarchar(MAX)
DECLARE @Subject nvarchar(255)
CREATE TABLE #tSa
(
AttendeeID int,
SurveyID int,
TrainingAttendeeID int
)
SELECT
TOP 1
@AttendeeID=ta.AttendeeID,
@SurveyID=ts.SurveyID,
@TrainingAttendeeID = ta.TrainingAttendeeID
FROM
TrainingAttendee ta
INNER JOIN
[User] u
ON
u.UserID= ta.AttendeeID
INNER JOIN
[Training] t
ON
t.TrainingID = ta.TrainingID
INNER JOIN
[TrainingSet] ts
ON
ts.TrainingSetID = t.TrainingSetID
WHERE
ta.AttendedTraining = 1
AND ta.ConfirmAttendedEmailOn IS NOT NULL
--only get people who didn't fill out the survey
AND NOT EXISTS (SELECT * FROM SurveyTaken st WHERE st.AddedByUserID = ta.AttendeeID AND st.SurveyID = ts.SurveyID)
ORDER BY
ta.TrainingAttendeeID,
ta.AttendeeID,
ts.SurveyID
--insert into our temp table
INSERT INTO #tSa(AttendeeID, SurveyID, TrainingAttendeeID)
VALUES(@AttendeeID, @SurveyID, @TrainingAttendeeID)
WHILE @TrainingAttendeeID IS NOT NULL AND @AttendeeID IS NOT NULL AND @SurveyID IS NOT NULL
BEGIN
DECLARE @TrainingID int
DECLARE @Title nvarchar(50)
DECLARE @StartDateTime nvarchar(50)
DECLARE @EndDateTime nvarchar(50)
DECLARE @FullName nvarchar(255)
DECLARE @EmailAddress nvarchar(255)
DECLARE @Description nvarchar(MAX)
--get the one record we are on...
SELECT
@TrainingID = t.TrainingID,
@Title = t.Title,
@StartDateTime = CAST(CONVERT(DATE, t.StartDate) AS VARCHAR(50)) + ' ' + CAST(t.StartTimeHours AS VARCHAR(50)) + ':' + CAST(CASE WHEN LEN(t.StartTimeMinutes)=1 THEN CAST(t.StartTimeMinutes AS VARCHAR(50)) + '0' ELSE CAST(t.StartTimeMinutes AS VARCHAR(50)) END AS VARCHAR(50)) + ' ' + CAST(t.StartTimeAMorPM AS VARCHAR(50)),
@EndDateTime = CAST(CONVERT(DATE, t.EndDate) AS VARCHAR(50)) + ' ' + CAST(t.EndTimeHours AS VARCHAR(50)) + ':' + CAST(CASE WHEN LEN(t.EndTimeMinutes)=1 THEN CAST(t.EndTimeMinutes AS VARCHAR(50)) + '0' ELSE CAST(t.EndTimeMinutes AS VARCHAR(50)) END AS VARCHAR(50)) + ' ' + CAST(t.EndTimeAMorPM AS VARCHAR(50)),
@Description = t.Descriptionhtml,
@FullName = u.FullName,
@EmailAddress = u.EmailAddress
FROM
Training t
INNER JOIN
TrainingAttendee ta
ON
t.TrainingID = ta.TrainingID
INNER JOIN
[User] u
ON
u.UserID = ta.AttendeeID
WHERE
ta.TrainingAttendeeID = @TrainingAttendeeID
IF @EmailAddress IS NOT NULL
BEGIN
--Email goes out here....
END
--grab next one
SELECT
TOP 1
@AttendeeID = ta.AttendeeID,
@SurveyID=ts.SurveyID,
@TrainingAttendeeID=ta.TrainingAttendeeID
FROM
TrainingAttendee ta
INNER JOIN
[User] u
ON
u.UserID= ta.AttendeeID
INNER JOIN
[Training] t
ON
t.TrainingID = ta.TrainingID
INNER JOIN
[TrainingSet] ts
ON
ts.TrainingSetID = t.TrainingSetID
WHERE
(
--same where as original
(ta.AttendedTraining = 1)
AND (ta.ConfirmAttendedEmailOn IS NOT NULL)
AND (NOT EXISTS (SELECT * FROM SurveyTaken st WHERE st.AddedByUserID = ta.AttendeeID AND st.SurveyID = ts.SurveyID))
--adding the piece such that we compare against the temp table...
AND (NOT EXISTS (SELECT * FROM #tSa tempS WHERE tempS.AttendeeID = ta.AttendeeID AND tempS.SurveyID = ts.SurveyID AND tempS.TrainingAttendeeID = ta.TrainingAttendeeID))
)
ORDER BY
ta.TrainingAttendeeID,
ta.AttendeeID,
ts.SurveyID
PRINT CAST('TrainingAttendeeID: ' + CAST(@TrainingAttendeeID as nvarchar(500)) + ' AttendeeID:' + CAST(@AttendeeID as nvarchar(500)) + ' SurveyID: ' + CAST(@SurveyID as nvarchar(500)) AS nvarchar(4000))
--insert into our temp table
INSERT INTO #tSa(AttendeeID, SurveyID, TrainingAttendeeID)
VALUES(@AttendeeID, @SurveyID, @TrainingAttendeeID)
END
END
GO
【问题讨论】：
标题看不懂……因为NOT EXIST (SELECT *和NOT EXIST (SELECT 1是一样的
通常不这么说。但你有没有想过使用光标？对于select distinct AttendeeID, SurveyID, TrainingAttendeeID
@JamieD77 - 我不想使用光标，我真的很想知道为什么这不能按预期工作...我会发布整个 sproc...
@JuanCarlosOropeza-这不是我的问题，我可以做 SELECT 1...我的问题是为什么它一遍又一遍地重复同一记录的过程..我想我的不存在...会过滤掉它。
按照@Ivan 在他的回答中所说的去做，把SELECT @TrainingAttendeeID = NULL, @AttendeeID = NULL, @SurveyID = NULL 放在WHILE BEGIN 之后
【参考方案1】：
如果 select 不返回任何记录，变量将不会改变。我敢打赌它会循环处理最后一个@AttendeeID。
另一种测试方法 - 向临时表添加唯一约束。一旦没有更多记录可供选择，我认为循环将失败。
修复它的一种方法 - 在每次迭代开始时将 NULL 分配给所有变量（在 while 正文的顶部）。但如果可能的话，我建议将此代码重写为光标（不确定几个选择语句的逻辑是什么）。
请注意，代码块中的变量声明没有“块范围”意义，因为它不是 perl 或 python。
【讨论】：
以上是关于Not Exists 不能很好地处理多列的主要内容，如果未能解决你的问题，请参考以下文章
Plotly 不能很好地处理条形图的 NA。我该在哪里解决这个问题？
通过 talend 从 oracle 加载到 greenplum 时的数据不能很好地处理 CLOB 列
NLTK 正则表达式标记器在正则表达式中不能很好地处理小数点