On a table I have a bigint column that stores a timestamp with a microsecond precision like:
636453251396217655
636453251398405201
636453251592389899
636453251668326820
I have to build a script that, if that date is older than a week, the row must moved to another table.
I tried to convert to date using:
CREATE FUNCTION [dbo].[UNIXToDateTime] (#timestamp bigint)
RETURNS datetime
AS
BEGIN
DECLARE #ret datetime
SELECT #ret = DATEADD(second, #timestamp, '1970/01/01 00:00:00')
RETURN #ret
END
and used like:
select dbo.UNIXToDateTime(636453251396217655)
but because of the bigint my script crash because:
Arithmetic overflow error during expression conversion in int data
type
I can lose precision, the important is the date part that is the main part of the sql filter.
Demo: http://sqlfiddle.com/#!6/24f05/1
There's an answer here for converting with epoch values:
CREATE FUNCTION [dbo].[fn_EpochToDatetime] (#Epoch BIGINT)
RETURNS DATETIME
AS
BEGIN
DECLARE #Days AS INT, #MilliSeconds AS INT
SET #Days = #Epoch / (1000*60*60*24)
SET #MilliSeconds = #Epoch % (1000*60*60*24)
RETURN (SELECT DATEADD(MILLISECOND, #MilliSeconds, DATEADD(DAY, #Days, '1/1/1970')))
END;
You can use that function but simply divide your epoch values. As you are fine with the loss of fidelity, this will suit your needs perfectly. For example:
DECLARE #epoch BIGINT = 636453251396217655
SELECT dbo.[fn_EpochToDatetime](#epoch/100000)
Related
I have the following query:
DECLARE #Month int
DECLARE #Year int
set #Month = 2
set #Year = 2004
Declare #MonthStartDate datetime
declare #MonthEndDate datetime
set #MonthStartDate = 'select DATEADD(month,#Month-1,DATEADD(year,#Year-1900,0))'
set #MonthEndDate = 'select DATEADD(day,-1,DATEADD(month,#Month,DATEADD(year,#Year-1900,0)))'
return #MonthStartDate , #MonthEndDate
But it returns:
"Conversion failed when converting date and/or time from character string."
What's wrong here?
Alternatively, you can also use as follow..
select #MonthStartDate = DATEADD(month,#Month-1,DATEADD(year,#Year-1900,0))
select #MonthEndDate = DATEADD(day,-1,DATEADD(month,#Month,DATEADD(year,#Year-1900,0)))
You should use DateTime expression instead of string literal. Just remove quotes:
DECLARE #Month int
DECLARE #Year int
set #Month = 2
set #Year = 2004
Declare #MonthStartDate datetime
declare #MonthEndDate datetime
set #MonthStartDate = DATEADD(month,#Month-1,DATEADD(year,#Year-1900,0))
set #MonthEndDate = DATEADD(day,-1,DATEADD(month,#Month,DATEADD(year,#Year-1900,0)))
Looking at your Query (Since you don't have enough Description on the Question ) What I understood is that you are trying to get the First and Last day of a Given month. If you are using a SQL Server Version 2012 or Above, then you have an Inbuild Function called EOMONTH() which can be used to calculate the End of any given month. Otherwise, you may try the below method on any Version on SQL Server
Declare #MonthStartDate datetime,
#MonthEndDate datetime,
#Year int,
#Month int --It's Better to Declare all the variables in the same space for easy handling
SELECT
#Month = 2,
#Year = 2004 -- Common Assignments can be done together
;WITH MNT
AS
(
SELECT
MonthStartDate = CAST(#Month AS VARCHAR(20))+'/01/'+CAST(#Year AS VARCHAR(20)) -- Hardcoded Day as 1 since Month Start always at 1
)
SELECT
#MonthStartDate = MonthStartDate,
#MonthEndDate = DATEADD(DAY,-1,DATEADD(MONTH,1,MonthStartDate))
FROM MNT
How to convert Epoch to DateTime SQL Server if epoch exceeds the year 2038?
Answer in Convert Epoch to DateTime SQL Server will not work.
Example:
SELECT DATEADD(ss, 2713795200000 / 1000, '19700101')
Thu, 30 Dec 2055 16:00:00 GMT
DATEADD function assumes an INT as an increment to your date, to bypass the limitation of INT you can either reduce the precision of your epoch, or do a slightly complex code to retain the precision of your epoch.
This reduces the precision to minutes:
SELECT DATEADD(MINUTE,#YourEpoch/60/1000, '1/1/1970')
This one splits your epoch to days and milliseconds and then combines them in a datetime
CREATE FUNCTION [dbo].[fn_EpochToDatetime] (#Epoch BIGINT)
RETURNS DATETIME
AS
BEGIN
DECLARE #Days AS INT, #MilliSeconds AS INT
SET #Days = #Epoch / (1000*60*60*24)
SET #MilliSeconds = #Epoch % (1000*60*60*24)
RETURN (SELECT DATEADD(MILLISECOND, #MilliSeconds, DATEADD(DAY, #Days, '1/1/1970')))
END;
However, I'm not quite sure why the 2nd solution is not as precise as I expect it to be.
Building on the response above, the solution provided works but does not protect from trying to convert to a date that is out of bounds for SQL server.
create function dbo.unixTimestampConversion (
#unixTime bigInt
)
returns dateTime2(7)
as
begin
declare
#output dateTime2(7)
, #days int
, #ms int
, #x int = (1000 * 60 * 60 * 24)
;
set #days = #unixTime / #x
;
set #ms = #unixTime % #x
;
if (#unixTime < 32503593600000 and #unixTime > -2208988800000)
begin
set #output = dateAdd (millisecond, #ms, dateAdd (day, #days, '1/1/1970'))
;
end
;
else if (#unixTime <= -2208988800000)
begin
set #output = '1/1/1900'
;
end
;
else if (#unixTime >= 32503593600000)
begin
set #output = '12/31/2999'
;
end
;
return #output
;
end
;
You can assign the epoch time to your datetime directly (I tried this on SQL Server 15.0). Although it considers the number as the number of days since 1900-1-1 00:00:00 so you have to add 2208988800 (the number of seconds in 70 years) and then divide by 86400(number of seconds in a day).
DECLARE #time DATETIME = (2208988800.0 + [your epoch time in seconds])/86400;
However, it seems to be 0.007s or 0.003s behind the given epoch. Also, I'm not sure if this is faster than the DATEADD() function.
create a function to convert epoch to datetime and use them in your query like
below
create FUNCTION [dbo].[from_unixtime] (#Datetime BIGINT)
RETURNS DATETIME
AS
BEGIN
DECLARE #LocalTimeOffset BIGINT
,#AdjustedLocalDatetime BIGINT;
SET #LocalTimeOffset = DATEDIFF(second,GETDATE(),GETUTCDATE())
SET #AdjustedLocalDatetime = #Datetime - #LocalTimeOffset
RETURN (SELECT DATEADD(second,#AdjustedLocalDatetime, CAST('1970-01-01 00:00:00' AS datetime)))
END;
and then use this function in your query
I am trying to create dynamic code for date calculation in SQL stored procedures but am having problems executing string expressions and parameters as date expressions.
I want to hold generic string expressions in a table that will create the dates according to the value of the parameters.
for example this is a generic expression :
DATEADD(#TimeResolution, -#IterationN, #CurrentCalc)
as you can see these generic expressions are composed out of parameters to.
in the stored procedures I intend to declare the variables that are in the expression and assign values to them using a select statement from a different table.
the problem is that after deriving these string values and writing the expression it does not give me the date I want but fails.
so for example if I write the following code
declare #Today date
declare #LastYear date
set #Today = getdate()
set #LastYear = DATEADD(year, -1, #Today)
select #Lastyear
it works fine and I will get last year's date.
but when I try something like this :
declare #Today date
declare #LastYear date
declare #Timeresolution varchar(5)
select #Timeresolution = [Timeresolution] from dbo.mytable where rule_id=1//Timeresolution is a varchar column in my table holding the values 'year' or 'month'
declare Iteration int
select #Iteration = [Iteration] from dbo.mytable where rule_id=1 //Iteration is a int column in my table holding the values 1 or 2, or 3
set #Today = getdate()
set #LastYear = DATEADD(Timeresolution , -Iteration , #Today)
select #Lastyear
this gives me a conversion error.
Is there a way to create such dynamic date expressions?
It isn't possible to use a variable for the interval in DATEADD, but you can do something like this:
IF #Timeresolution = 'year'
BEGIN
SET #LastYear = DATEADD(year, -#Iteration , #Today)
END
IF #Timeresolution = 'month'
BEGIN
SET #LastYear = DATEADD(month, -#Iteration , #Today)
END
The code select DATEPART(month,getutcdate()) returns an integer representing the numeric month value of the current date.
The value MONTH is one of a number of predefined options for the interval parameter.
What is the underlying data type of that interval value?
declare #date datetime = getutcdate()
, #interval notNVarChar(16) = 'month' --what type should this be?
--Would it need quotes for assignment?
select DATEPART(#interval, #date)
You can make it parameterize by making a dynamic SQL as below:
declare #date datetime
set #date = GETDATE()
declare #option varchar(50)
declare #sql varchar(max)
SET #option = 'MONTH' --Here you can set other options like hour, second, milisecond etc..
set #sql = 'SELECT DATEPART('+ #option + ',''' + CONVERT(varchar,#date,21)+''')'
print #sql
EXEC( #sql)
There isn't one - you cannot parameterize it.
From the documentation:
DATEPART ( datepart , date )
datepart
Is the part of date (a date or time value) for which an integer will be returned. The following table lists all valid datepart arguments. User-defined variable equivalents are not valid.
(my emphasis)
There are lots of options available like :
day
week
hour
minute
second
millisecond
etc..
PLease go thru thisLINK
I created the following sfunction
CREATE FUNCTION sf_GetTrainingDate (#ID_ITEM int,#EMPLOYEE VARCHAR)
RETURNS DATETIME AS
BEGIN
--(1)
DECLARE #RESULT DATETIME
DECLARE #DATE DATETIME
DECLARE #REPEAT_INTERVAL INT
SET #DATE = (/*subquery returning a date in datetime format*/)
SET #REPEAT_INTERVAL = (/*subquery returning a datetime */)
WHILE #DATE <= getdate()
BEGIN
SET #DATE=dateadd(month,#REPEAT_INTERVAL,#DATE)
IF #DATE > GETDATE()
BREAK
ELSE
CONTINUE
END
SET #RESULT = #DATE
RETURN #RESULT
--(2)
END
The code works correctly, I mean if I run only the part from (1) to (2) using some PRINT commands to read the value of #DATE all is fine, but as i use the stored function it returns NULL instead of a date.
Could you give me an hand?