I am looking for a way to increment a uniqueidentifier by 1 in TSQL. For example, if the id is A6BC60AD-A4D9-46F4-A7D3-98B2A7237A9E, I'd like to be able to select A6BC60AD-A4D9-46F4-A7D3-98B2A7237A9F.
#rein It's for a data import. We have an intermediate table with IDs that we're generating records from, and we join on those IDs later in the import. Unfortunately, now some of those records generate a couple of records in the next table, so we need a new id that is reproducible.
The way you want to increment Guid is not correct for SQL Server as Guid is a structure with different byte order in the byte groups, please have a look at:
http://sqlblog.com/blogs/alberto_ferrari/archive/2007/08/31/how-are-guids-sorted-by-sql-server.aspx
and notice the following:
Now, when I run modified Alberto's query, I'm getting the following sequence:
3, 2, 1, 0, 5, 4, 7, 6, 9, 8, 15, 14, 13, 12, 11, 10
That means, that GUID's byte #3 is the least significant and GUID's byte #10 is the most significant [from SQL Server ORDER BY clause perspective].
Here is simple function to increment a uniqueidentifier accounting for this:
create function [dbo].[IncrementGuid](#guid uniqueidentifier)
returns uniqueidentifier
as
begin
declare #guid_binary binary(16), #b03 binary(4), #b45 binary(2), #b67 binary(2), #b89 binary(2), #bAF binary(6)
select #guid_binary = #guid
select #b03 = convert(binary(4), reverse(substring(#guid_binary,1,4)))
select #b45 = convert(binary(2), reverse(substring(#guid_binary,5,2)))
select #b67 = convert(binary(2), reverse(substring(#guid_binary,7,2)))
select #b89 = convert(binary(2), substring(#guid_binary,9,2))
select #bAF = convert(binary(6), substring(#guid_binary,11,6))
if (#b03 < 'FFFFFFFF')
begin
select #b03 = convert(binary(4), cast(#b03 as int) + 1)
end
else if (#b45 < 'FFFF')
begin
select #b45 = convert(binary(2), cast(#b45 as int) + 1)
end
else if (#b89 < 'FFFF')
begin
select #b89 = convert(binary(2), cast(#b89 as int) + 1)
end
else
begin
select #bAF = convert(binary(6), cast(#bAF as bigint) + 1)
end
return convert(binary(16), reverse(convert(char(4),#b03)) + reverse(convert(char(2),#b45)) + reverse(convert(char(2),#b67)) + convert(char(2),#b89) + convert(char(6),#bAF))
end
Note that bytes 6 and 7 are not incremented as they contain the Guid version bits.
But as others has pointed you really should not be doing this. In your case it might be better if you create a temp table for these Guids (with two columns: one integer as index and second one with generated Guids).
Here is one way I've come up with, but I'm hoping there is a better way.
LEFT([ID], 19) + RIGHT(CONVERT(uniqueidentifier, CONVERT(binary(16), CONVERT(binary(16), [ID]) + CONVERT(bigint, 1))), 17) AS 'MyNewID'
You can do this approach, but I'm not accounting for the case of overflowing lower 8 bytes.
declare #guid uniqueidentifier, #binaryUpper8 binary(8), #binaryLower8 binary(8), #binary16 binary(16), #bigint bigint
set #guid = 'A6BC60AD-A4D9-46F4-A7D3-98B2A7237A9E'
set #binary16 = cast(#guid as binary(16))
--harvest lower 8 bytes
select #binaryUpper8= substring(#binary16, 1, 8)
,#binaryLower8 = substring(#binary16, 9, 8)
set #bigint = cast(#binaryLower8 as bigint)
--increment
set #bigint = #bigint + 1
--convert back
set #binaryLower8 = cast(#bigint as binary(8))
set #binary16 = #binaryUpper8 + #binaryLower8
set #guid = cast(#binary16 as uniqueidentifier)
select #guid
Related
I try to create dynamic forecast for 18(!) months depend on previous columns (months) and i am stuck:
I have three columns:
Stock
SafetyStock
Need for production - another select with clause WHERE date = getdate()
what i need to achieve:
Index, Stock- Current month, SafetyStock-Current month, Need for production (select * from Nfp where date = getdate()), Stock - Current month + 1, Safetystock - Current Month + 1, Need for Production - Current Month + 1 ... etc till 18 months
calculations:
Stock - Current month + 1 = Stock previous month + SafetyStock previous month - Needs for production of current month
there is any possibility to create something like this ? it has to be dynamic and get calculation for current date and next 18 months. So now i have to calculate from 2020-10 till let's say 2022-04
What i have tried:
I prepared 18 cte and joins everything. Then i do calculations - it works but it slow and i think it is not profesional.
I have tried to do dynamic sql, below you can see my code but i have stucked when i wanted to do computed column depended on previous computed column:
------------------- CODE -------------------------
if object_id('tempdb..#tmp') is not null
drop table #tmp
if object_id('tempdb..#tmp2') is not null
drop table #tmp2
declare #cols as int
declare #iteration as int
declare #Mth as nvarchar(30)
declare #data as date
declare #sql as nvarchar(max)
declare #sql2 as nvarchar(max)
set #cols = 18
set #iteration = 0
set #Mth = month(getdate())
set #data = cast(getdate() as date)
select
10 as SS,
12 as Stock
into #tmp
WHILE #iteration < #cols
begin
set #iteration = #iteration + 1
set #sql =
'
alter table #tmp
add [StockUwzgledniajacSS - ' + cast(concat(year(DATEADD(Month, #Iteration, #data)),'-', month(DATEADD(Month, #Iteration, #data))) as nvarchar(max)) +'] as (Stock - SS)
'
exec (#sql)
set #Mth= #Mth+ 1
set #sql2 =
'
alter table #tmp
add [StockUwzgledniajacSS - ' + #Mth +'] as ([StockUwzgledniajacSS - ' + #Mth +'])
'
end
select * from #tmp
thanks in advance!
Update 1 note: I wrote this before you posted your data. This still holds I believe but, of course, stock levels are way different. Given that your NFP data is by day, and your report is by month, I suggest adding something to preprocess that data into months e.g., sum of NPS values, grouped by month.
Update 2 (next day) note: From the OPs comments below, I've tried to integrate this with what was written and more directly answering the question e.g., creating a reporting table #tmp.
Given that the OP also mentions millions of rows, I imagine each row represents a specific part/item - I've included this as a field called StockNum.
I have done something that probably doesn't do your calculations properly, but demonstrates the approach and should get you over your current hurdle. Indeed, if you haven't used these before, then updating this code with your own calculations will help you to understand how it works so you can maintain it.
I'm assuming the key issue here for calculation is that this month's stock is based on last month's stock and then new stock minus old stock for this month.
It is possible to calculate this in 18 separate statements (update table set col2 = some function of col1, then update table set col3 = some function of col2, etc). However, updating the same table multiple times is often an anti-pattern causing poor performance - especially if you need to read the base data again and again.
Instead, something like this is often best calculated using a Recusive CTE (here's an example description), where it 'builds' a set of data based on previous results.
The key difference in this approach is that it
Creates the reporting table (without any data/calculations going in)
Calculates the data as a separate step - but with columns/fields that can be used to link to the reporting table
Inserts the data from calculations into the reporting table as a single insert statement.
I have used temporary tables/etc liberally, to help demonstrate the process.
You haven't explained what safety stock is, nor how you measure what's coming in, so for the example below, I have assumed safety stock is the amount produced and is 5 per month. I've then assumed that NFP is amount going out each month (e.g., forward estimates of sales). The key result will be stock at the end of month (e.g., which you could then review whether it's too high or too low).
As you want to store it in a table that has each month as columns, the first step is to create a list with the relevant buckets (months). These include fields used for matching in later calculations/etc. Note I have included some date fields (startdate and enddate) which may be useful when you customise the code. This part of the SQL is designed to be as straightforward as possible.
We then create the scratch table that has our reference data for stock movements, replacing your SELECT * FROM NFP WHERE date = getdate()
/* SET UP BUCKET LIST TO HELP CALCULATION */
CREATE TABLE #RepBuckets (BucketNum int, BucketName nvarchar(30), BucketStartDate datetime, BucketEndDate datetime)
INSERT INTO #RepBuckets (BucketNum) VALUES
(0),(1),(2),(3),(4),(5),(6),(7),(8),(9),(10),
(11),(12),(13),(14),(15),(16),(17),(18)
DECLARE #CurrentBucketStart date
SET #CurrentBucketStart = DATEFROMPARTS(YEAR(getdate()), MONTH(getdate()), 1)
UPDATE #RepBuckets
SET BucketName = 'StockAtEnd_' + FORMAT(DATEADD(month, BucketNum, #CurrentBucketStart), 'MMM_yy'),
BucketStartDate = DATEADD(month, BucketNum, #CurrentBucketStart),
BucketEndDate = DATEADD(month, BucketNum + 1, #CurrentBucketStart)
/* CREATE BASE DATA */
-- Current stock
CREATE TABLE #Stock (StockNum int, MonthNum int, StockAtStart int, SafetyStock int, NFP int, StockAtEnd int, PRIMARY KEY(StockNum, MonthNum))
INSERT INTO #Stock (StockNum, MonthNum, StockAtStart, SafetyStock, NFP, StockAtEnd) VALUES
(12422, 0, NULL, NULL, NULL, 10)
-- Simulates SELECT * FROM NFP WHERE date = getdate()
CREATE TABLE #NFP_by_month (StockNum int, MonthNum int, StockNFP int, PRIMARY KEY(StockNum, MonthNum))
INSERT INTO #NFP_by_month (StockNum, MonthNum, StockNFP) VALUES
(12422, 1, 4), (12422, 7, 4), (12422, 13, 4),
(12422, 2, 5), (12422, 8, 5), (12422, 14, 5),
(12422, 3, 2), (12422, 9, 2), (12422, 15, 2),
(12422, 4, 7), (12422, 10, 7), (12422, 16, 7),
(12422, 5, 9), (12422, 11, 9), (12422, 17, 9),
(12422, 6, 3), (12422, 12, 3), (12422, 18, 3)
We then use the recursive CTE to get calculate our data. It stores these in table #StockProjections.
What this does is
Start with your current stock (last row in the #Stock table). Note that the only value that matters in that is the stock at end of month.
Uses that stock level at the end of last month, as the stock level at the start of the new month
Adds the safety stock, minuses the NFP, and calculates your stock at end.
Note that within the recursive part of the CTE, 'SBM' (StockByMonth) refers to last month's data). This is then used with whatever external data (e.g., #NFP) to calculate new data.
These calculations create a table with
StockNum (the ID number of the relevant stock item - for this example, I've used one stock item 12422)
MonthNum (I've used integers this rather than dates, for clarity/simplicity)
BucketName (an nvarchar representing the month, used for column names)
Stock at start of month
Safety stock (which I assume is incoming stock, 5 per month)
NFP (which I assume is outgoing stock, varies by month and comes from a scratch table here - you'll need to adjust this to your select)
Stock at end of month
/* CALCULATE PROJECTIONS */
CREATE TABLE #StockProjections (StockNum int, BucketName nvarchar(30), MonthNum int, StockAtStart int, SafetyStock int, NFP int, StockAtEnd int, PRIMARY KEY (StockNum, BucketName))
; WITH StockByMonth AS
(-- Anchor
SELECT TOP 1 StockNum, MonthNum, StockAtStart, SafetyStock, NFP, StockAtEnd
FROM #Stock S
ORDER BY MonthNum DESC
-- Recursion
UNION ALL
SELECT NFP.StockNum,
SBM.MonthNum + 1 AS MonthNum,
SBM.StockAtEnd AS NewStockAtStart,
5 AS Safety_Stock,
NFP.StockNFP,
SBM.StockAtEnd + 5 - NFP.StockNFP AS NewStockAtEnd
FROM StockByMonth SBM
INNER JOIN #NFP_by_month NFP ON NFP.MonthNum = SBM.MonthNum + 1
WHERE NFP.MonthNum <= 18
)
INSERT INTO #StockProjections (StockNum, BucketName, MonthNum, StockAtStart, SafetyStock, NFP, StockAtEnd)
SELECT StockNum, BucketName, MonthNum, StockAtStart, SafetyStock, NFP, StockAtEnd
FROM StockByMonth
INNER JOIN #RepBuckets ON StockByMonth.MonthNum = #RepBuckets.BucketNum
Now we have the data, we set up a table for reporting purposes. Note that this table has the month names embedded into the column names (e.g., StockAtEnd_Jun_21). It would be easier to use a generic name (e.g., StockAtEnd_Month4) but I've gone for the slightly more complex case here for demonstration.
/* SET UP TABLE FOR REPORTING */
DECLARE #cols int = 18
DECLARE #iteration int = 0
DECLARE #colname nvarchar(30)
DECLARE #sql2 as nvarchar(max)
CREATE TABLE #tmp (StockNum int PRIMARY KEY)
WHILE #iteration <= #cols
BEGIN
SET #colname = (SELECT TOP 1 BucketName FROM #RepBuckets WHERE BucketNum = #iteration)
SET #sql2 = 'ALTER TABLE #tmp ADD ' + QUOTENAME(#colname) + ' int'
EXEC (#sql2)
SET #iteration = #iteration + 1
END
The last step is to add the data to your reporting table. I've used a pivot here but feel free to use whatever you like.
/* POPULATE TABLE */
DECLARE #columnList nvarchar(max) = N'';
SELECT #columnList += QUOTENAME(BucketName) + N' ' FROM #RepBuckets
SET #columnList = REPLACE(RTRIM(#columnList), ' ', ', ')
DECLARE #sql3 nvarchar(max)
SET #sql3 = N'
;WITH StockPivotCTE AS
(SELECT *
FROM (SELECT StockNum, BucketName, StockAtEnd
FROM #StockProjections
) StockSummary
PIVOT
(SUM(StockAtEnd)
FOR [BucketName]
IN (' + #columnList + N')
) AS StockPivot
)
INSERT INTO #tmp (StockNum, ' + #columnList + N')
SELECT StockNum, ' + #columnList + N'
FROM StockPivotCTE'
EXEC (#sql3)
Here's a DB<>fiddle showing it running with results of each sub-step.
I found my SQL Server job gets stuck occasionally, about once every two months. Since I am not from DBA background, I need some to help to rectify the issue.
So far, I have tried to pinpoint the issue by checking the activity monitor. I found the issue is caused by one of my stored procedures which it will create a temp table to collect data, then the data will be inserted into one of my transaction tables. This table have 400 millions of records.
Whenever this issue occur, I stop the job and:
I rerun the job, the stored procedure can complete
I execute the stored procedure manually, the stored procedure completes
I implemented the SP_BlitzCache, and execute it. I can see it suggest DBCC FREEPROCCACHE (0x0...) on the stored procedure.
CREATE TABLE #dtResult
(
RunningNumber INTEGER,
, AlphaID BIGINT
, BetaID BIGINT
, Content varchar(100)
, X varchar(10)
, Y varchar(10)
)
INSERT INTO #dtResult ( RunningNumber, ...)
SELECT RowId AS RunningNumber,
...
FROM
...
/*** Based on activity monitor, the highest CPU caused by this statement ***/
INSERT INTO tblTransaction ( ... )
SELECT DISTINCT
RES.AlphaID
, b.UnitId
, RES.BetaID
, CASE WHEN RES.BinData IS NULL THEN [dbo].[fnGetCode](B.Data, RES.X, RES.Y) ELSE RES.Content END
, CONVERT(DATETIME, SUBSTRING(RES.Timestamp, 1, 4) + '-' + SUBSTRING(RES.Timestamp, 5, 2) + '-' + SUBSTRING(RES.Timestamp, 7, 2) + ' ' + SUBSTRING(RES.Timestamp, 9, 2) + ':' + SUBSTRING(RES.Timestamp, 11, 2) + ':' + SUBSTRING(RES.Timestamp, 13, 2) + '.' + SUBSTRING(RES.Timestamp, 15, 3), 121)
FROM
#dtResult RES
INNER JOIN
tblA a with(nolock) ON RES.AlphaID = a.AlphaID
INNER JOIN
tblB b with(nolock) ON a.UnitId = b.UnitId AND CAST(RES.X AS INTEGER) = b.X AND CAST(RES.Y AS INTEGER) = b.Y
INNER JOIN
tblC c with(nolock) ON RES.BetaID = c.BetaID
LEFT OUTER JOIN
tblTransaction t with(nolock) ON RES.AlphaID = t.AlphaID AND RES.BetaID = t.BetaID AND t.UnitId = b.UnitId
WHERE
t.BetaID IS NULL
/* FUNCTION */
CREATE FUNCTION [dbo].[fnGetCode]
(
#Data VARCHAR(MAX),
#SearchX INT,
#SearchY INT
)
RETURNS CHAR(4)
WITH ENCRYPTION
AS
BEGIN
DECLARE #SearchResult CHAR(4)
DECLARE #Pos INT
SET #Pos = (#SearchY * #SearchX) + 1
SET #SearchResult = CONVERT(char(1),SUBSTRING(#Data,#Pos,1), 1)
RETURN #SearchResult
END
I am being passed the following parameter to my stored procedure -
#AddOns = 23:2,33:1,13:5
I need to split the string by the commas using this -
SET #Addons = #Addons + ','
set #pos = 0
set #len - 0
While CHARINDEX(',', #Addons, #pos+1)>0
Begin
SET #len = CHARINDEX(','), #Addons, #pos+1) - #pos
SET #value = SUBSTRING(#Addons, #pos, #len)
So now #value = 23:2 and I need to get 23 which is my ID and 2 which is my quantity. Here is the rest of my code -
INSERT INTO TABLE(ID, Qty)
VALUES(#ID, #QTY)
set #pos = CHARINDEX(',', #Addons, #pos+#len) + 1
END
So what is the best way to get the values of 23 and 2 in separate fields to us in the INSERT statement?
First you would split the sets of key-value pairs into rows (and it looks like you already got that far), and then you get the position of the colon and use that to do two SUBSTRING operations to split the key and value apart.
Also, this can be done much more efficiently than storing each row's key and value into separate variables just to get inserted into a table. If you INSERT from the SELECT that breaks this data apart, it will be a set-based operation instead of row-by-row.
For example:
DECLARE #AddOns VARCHAR(1000) = N'23:2,33:1,13:5,999:45';
;WITH pairs AS
(
SELECT [SplitVal] AS [Value], CHARINDEX(N':', [SplitVal]) AS [ColonIndex]
FROM SQL#.String_Split(#AddOns, N',', 1) -- https://SQLsharp.com/
)
SELECT *,
SUBSTRING(pairs.[Value], 1, pairs.[ColonIndex] - 1) AS [ID],
SUBSTRING(pairs.[Value], pairs.[ColonIndex] + 1, 1000) AS [QTY]
FROM pairs;
/*
Value ColonIndex ID QTY
23:2 3 23 2
33:1 3 33 1
13:5 3 13 5
999:45 4 999 45
*/
GO
For that example I am using a SQLCLR string splitter found in the SQL# library (that I am the author of), which is available in the Free version. You can use whatever splitter you like, including the built-in STRING_SPLIT that was introduced in SQL Server 2016.
It would be used as follows:
DECLARE #AddOns VARCHAR(1000) = N'23:2,33:1,13:5,999:45';
;WITH pairs AS
(
SELECT [value] AS [Value], CHARINDEX(N':', [value]) AS [ColonIndex]
FROM STRING_SPLIT(#AddOns, N',') -- built-in function starting in SQL Server 2016
)
INSERT INTO dbo.TableName (ID, QTY)
SELECT SUBSTRING(pairs.[Value], 1, pairs.[ColonIndex] - 1) AS [ID],
SUBSTRING(pairs.[Value], pairs.[ColonIndex] + 1, 1000) AS [QTY]
FROM pairs;
Of course, the Full (i.e. paid) version of SQL# includes an additional splitter designed to handle key-value pairs. It's called String_SplitKeyValuePairs and works as follows:
DECLARE #AddOns VARCHAR(1000) = N'23:2,33:1,13:5,999:45';
SELECT *
FROM SQL#.String_SplitKeyValuePairs(#AddOns, N',', N':', 1, NULL, NULL, NULL);
/*
KeyID Key Value
1 23 2
2 33 1
3 13 5
4 999 45
*/
GO
So, it would be used as follows:
DECLARE #AddOns VARCHAR(1000) = N'23:2,33:1,13:5,999:45';
INSERT INTO dbo.[TableName] ([Key], [Value])
SELECT kvp.[Key], kvp.[Value]
FROM SQL#.String_SplitKeyValuePairs(#AddOns, N',', N':', 1, NULL, NULL, NULL) kvp;
Check out this blog post...
http://www.sqlservercentral.com/blogs/querying-microsoft-sql-server/2013/09/19/how-to-split-a-string-by-delimited-char-in-sql-server/
Noel
I am going to make another attempt at this inspired by the answer given by #gofr1 on this question...
How to insert bulk of column data to temp table?
That answer showed how to use an XML variable and the nodes method to split comma separated data and insert it into individual columns in a table. It seemed to me to be very similar to what you were trying to do here.
Check out this SQL. It certainly isn't has concise as just having "split" function, but it seems better than chopping up the string based on position of the colon.
Noel
This question already has answers here:
TSQL Generate 5 character length string, all digits [0-9] that doesn't already exist in database
(6 answers)
Closed 7 years ago.
i want to create a coupon code generator by using SQL database but i don't know how to generate 1000 of random number without repeating them. so can someone help me it's important. thanks
Records in a relational database tables are unordered by nature.
therefor, you can simply create a table that has all the values between #First and #Last (0 and 9999 in your case), and then use a random order by when selecting from that table. you can also use a simple int in the database table and just format it when you select the data from the table.
Since my main database is Sql server, and I have no experience with sqlite, I will use Sql Server syntax in my code example, and leave it up to you to find the sqllite equivalent.
First, create the table:
CREATE TABLE Tbl
(
IntValue int PRIMARY KEY,
IsUsed bit NOT NULL DEFAULT 0
)
Then, populate it with numbers between 0 and 9999:
;With CTE AS (
SELECT 0 As IntValue
UNION ALL
SELECT IntValue + 1
FROM CTE
WHERE IntValue + 1 < 10000
)
INSERT INTO Tbl (IntValue)
SELECT IntValue
FROM CTE
OPTION(MAXRECURSION 0)
Then, you want to select multiple values each time, so I would write a stored procedure like this:
CREATE PROCEDURE stp_GetCouponCodes
(
#Number int = 5 -- or whatever number is default
)
AS
BEGIN
DECLARE #UsedValues AS TABLE
(
IntValue int
)
BEGIN TRY
BEGIN TRANSACTION
INSERT INTO #UsedValues
SELECT TOP(#Number) IntValue
FROM Tbl
WHERE IsUsed = 0
ORDER BY NEWID()
UPDATE Tbl
SET IsUsed = 1
FROM Tbl
INNER JOIN
#UsedValues uv ON(Tbl.IntValue = uv.IntValue)
SELECT RIGHT('00000' + CAST(IntValue as varchar), 5)
FROM #UsedValues
COMMIT TRANSACTION
END TRY
BEGIN CATCH
IF ##TRANCOUNT > 0
ROLLBACK TRANSACTION
END CATCH
END
Then, when ever you want to generate coupons, simply execute the stored procedure with the number of coupons you want:
EXEC stp_GetCouponCodes 10;
See working fiddle example here.
The code below uses a quick method to generate 100 random 5-character strings based on the alphabet provided. You'll still need to perform duplicate checking, but this should get you started.
DECLARE #Quantity INT = 1000
DECLARE #Alphabet VARCHAR(100) = '0123456789'
DECLARE #Length INT = LEN(#Alphabet)
DECLARE #Top INT = SQRT(#Quantity) + 1
;WITH CTE AS (
SELECT TOP (#Top) *
FROM sys.objects
)
SELECT TOP (#Quantity)
SUBSTRING(#Alphabet, ABS(CHECKSUM(NEWID())) % #Length + 1, 1)
+ SUBSTRING(#Alphabet, ABS(CHECKSUM(NEWID())) % #Length + 1, 1)
+ SUBSTRING(#Alphabet, ABS(CHECKSUM(NEWID())) % #Length + 1, 1)
+ SUBSTRING(#Alphabet, ABS(CHECKSUM(NEWID())) % #Length + 1, 1)
+ SUBSTRING(#Alphabet, ABS(CHECKSUM(NEWID())) % #Length + 1, 1)
AS [Code]
FROM CTE X
CROSS JOIN CTE Y
I am observing a strange behaviour. The following code that is related to conversion of a substring of nvarchar to tinyint works fine:
DECLARE #s nvarchar(50) = N'Line 1'
DECLARE #n tinyint = SUBSTRING(#s, 6, 10)
SELECT
N'Line 1' AS explicit_nvarchar,
#s AS nvarchar_variable,
SUBSTRING(#s, 6, 10) AS nvarchar_substring
, #n AS tinyint_var_converted_earlier
, CAST(SUBSTRING(#s, 6, 10) AS tinyint) AS cast_sub_var
, CAST(SUBSTRING(N'Line 1', 6, 10) AS tinyint) AS cast_nvarchar_substr
I can observe the following output in the Microsoft SQL Server Management Studio:
.
However, the code that tries to do with the SELECT from the tables fails if the commented line is uncommented:
Select
[order].[identifier] As [order_identifier],
[plan_data].[starts_on] As [from],
[plan_data].[ends_on] As [to],
[workplace].[identifier] AS line_identifier,
SUBSTRING([workplace].[identifier], 6, 10) AS line_no_str
--, CAST(SUBSTRING([workplace].[identifier], 6, 10) AS int) AS line_no
From
...
With commented line...
but with uncommented line (both for conversion to tinyint or int)...
Update: Strange. I have added the last line to the WHERE to check whether all lines
contain the 'Line...
Where
[plan].[identifier] = #lPlanIdentifier And
(
[plan_data].[starts_on] Between #fromUTC And #toUTC Or
[plan_data].[ends_on] Between #fromUTC And #toUTC)
AND LEFT([workplace].[identifier], 4) = N'Line'
... and all of a sudden it works also with the uncommented line. How can I discover what causes the problem?
Update2: I should have followed the Hamlet Hakobyan's answer exactly as it was published. When looking inside the table, I can see:
You have value in identifier column for which SUBSTRING([workplace].[identifier], 6, 10) returns ined.
Try this query:
SELECT * FROM [workplace]
WHERE SUBSTRING([identifier], 6, 10) = N'ined'
As indicated by the comments and other answer(s), you have a value in [workplace].[identifier] for one or more rows that fails to cast to an INT. When you added the following to the WHERE clause, you eliminated the row(s) that has this data before it was converted in the SELECT clause:
LEFT([workplace].[identifier], 4) = N'Line'
Run the following to get a distinct list of values in [workplace].[identifier]
SELECT DISTINCT [workplace].[identifier]
FROM [workplace]
Whenever you have an error converting a varchar to a number the easiest way to find the data is using IsNumeric. However IsNumeric doesn't mean IsInteger so its worth removing scientific or decimal formats as described by G Mastros in this answer
SELECT
[order].[identifier] As [order_identifier],
[plan_data].[starts_on] As [from],
[plan_data].[ends_on] As [to],
[workplace].[identifier] AS line_identifier,
SUBSTRING([workplace].[identifier], 6, 10) AS line_no_str
--, CAST(SUBSTRING([workplace].[identifier], 6, 10) AS int) AS line_no
FROM
...
WHERE
IsNumeric(SUBSTRING([workplace].[identifier], 6, 10) + '.0e0') = 0
If you can't fix it but you can use a default value than you could use a case statement to do the conversion only when its legal to do so.
SELECT
[order].[identifier] As [order_identifier],
[plan_data].[starts_on] As [from],
[plan_data].[ends_on] As [to],
[workplace].[identifier] AS line_identifier,
CASE WHEN IsNumeric(SUBSTRING([workplace].[identifier], 6, 10) + '.0e0') = 1
THEN CAST(SUBSTRING([workplace].[identifier], 6, 10) AS int)
ELSE Null --Default to null in this example
END as line_no