Translating SQL Server Cursor to Azure Synapse - sql-server

I have the following code that loops through a table with unique model numbers and creates a new table that contains, for each model numbers, a row based on the year and week number. How can I translate this so it doesn't use a cursor?
DECLARE #current_model varchar(50);
--declare a cursor that iterates through model numbers in ItemInformation table
DECLARE model_cursor CURSOR FOR
SELECT model from ItemInformation
--start the cursor
OPEN model_cursor
--get the next (first value)
FETCH NEXT FROM model_cursor INTO #current_model;
DECLARE #year_counter SMALLINT;
DECLARE #week_counter TINYINT;
WHILE (##FETCH_STATUS = 0) --fetch status returns the status of the last cursor, if 0 then there is a next value (FETCH statement was successful)
BEGIN
SET #year_counter = 2019;
WHILE (#year_counter <= Datepart(year, Getdate() - 1) + 2)
BEGIN
SET #week_counter = 1;
WHILE (#week_counter <= 52)
BEGIN
INSERT INTO dbo.ModelCalendar(
model,
sales_year,
sales_week
)
VALUES(
#current_model,
#year_counter,
#week_counter
)
SET #week_counter = #week_counter + 1
END
SET #year_counter = #year_counter + 1
END
FETCH NEXT FROM model_cursor INTO #current_model
END;
CLOSE model_cursor;
DEALLOCATE model_cursor;
If ItemInformation contains the following table:
model,invoice
a,4.99
b,9.99
c,1.99
d,8.99
then the expected output is:
model,sales_year,sales_week
A,2019,1
A,2019,2
A,2019,3
...
A,2019,52
A,2020,1
A,2020,2
A,2020,3
...
A,2020,51
A,2020,52
A,2020,53 (this is 53 because 2020 is leap year and has 53 weeks)
A,2021,1
A,2021,2
...
A,2022,1
A,2022,2
...
A,2022,52
B,2019,1
B,2019,2
...
D, 2022,52

Using CTE's you can get all combinations of weeks and years within the range required. Then join your data table on.
declare #Test table (model varchar(1), invoice varchar(4));
insert into #Test (model, invoice)
values
('a', '4.99'),
('b', '9.99'),
('c', '1.99'),
('d', '8.99');
with Week_CTE as (
select 1 as WeekNo
union all
select 1 + WeekNo
from Week_CTE
where WeekNo < 53
), Year_CTE as (
select 2019 YearNo
union all
select 1 + YearNo
from Year_CTE
where YearNo <= datepart(year, current_timestamp)
)
select T.model, yr.YearNo, wk.WeekNo
from Week_CTE wk
cross join (
select YearNo
-- Find the last week of the year (52 or 53) -- might need to change the start day of the week for this to be correct
, datepart(week, dateadd(day, -1, dateadd(year, 1, '01 Jan ' + convert(varchar(4),YearNo)))) LastWeek
from Year_CTE yr
) yr
cross join (
-- Assuming only a single row per model is required, and the invoice column can be ignored
select model
from #Test
group by model
) T
where wk.WeekNo <= yr.LastWeek
order by yr.YearNo, wk.WeekNo;
As you have advised that using a recursive CTE is not an option, you can try using a CTE without recursion:
with T(N) as (
select X.N
from (values (0),(0),(0),(0),(0),(0),(0),(0)) X(N)
), W(N) as (
select top (53) row_number() over (order by ##version) as N
from T T1
cross join T T2
), Y(N) as (
-- Upper limit on number of years
select top (12) 2018 + row_number() over (order by ##version) AS N
from T T1
cross join T T2
)
select W.N as WeekNo, Y.N YearNo, T.model
from W
cross join (
select N
-- Find the last week of the year (52 or 53) -- might need to change the start day of the week for this to be correct
, datepart(week, dateadd(day, -1, dateadd(year, 1, '01 Jan ' + convert(varchar(4),N)))) LastWeek
from Y
) Y
cross join (
-- Assuming only a single row per model is required, and the invoice column can be ignored
select model
from #Test
group by model
) T
-- Filter to required number of years.
where Y.N <= datepart(year, current_timestamp) + 1
and W.N <= Y.LastWeek
order by Y.N, W.N, T.model;
Note: If you setup your sample data in future with the DDL/DML as shown here you will greatly assist people attempting to answer.

I don't like to see a loop solution where a set solution can be found. So here goes Take II with no CTE, no values and no row_number() (the table variable is just to simulate your data so not part of the actual solution):
declare #Test table (model varchar(1), invoice varchar(4));
insert into #Test (model, invoice)
values
('a', '4.99'),
('b', '9.99'),
('c', '1.99'),
('d', '8.99');
select Y.N + 2019 YearNumber, W.WeekNumber, T.Model
from (
-- Cross join 5 * 10, then filter to 52/53 as required
select W1.N * 10 + W2.N + 1 WeekNumber
from (
select 0 N
union all select 1
union all select 2
union all select 3
union all select 4
union all select 5
) W1
cross join (
select 0 N
union all select 1
union all select 2
union all select 3
union all select 4
union all select 5
union all select 6
union all select 7
union all select 8
union all select 9
) W2
) W
-- Cross join number of years required, just ensure its more than will ever be needed then filter back
cross join (
select 0 N
union all select 1
union all select 2
union all select 3
union all select 4
union all select 5
union all select 6
union all select 7
union all select 8
union all select 9
) Y
cross join (
-- Assuming only a single row per model is required, and the invoice column can be ignored
select model
from #Test
group by model
) T
-- Some filter to restrict the years
where Y.N <= 3
-- Some filter to restrict the weeks
and W.WeekNumber <= 53
order by YearNumber, WeekNumber;

I created a table to temporary calendar table containing all the weeks and years. To account for leap years, I took the last 7 days of a year and got the ISO week for each day. To know how many weeks are in a year, I put these values into another temp table and took the max value of it. Azure Synapse doesn't support multiple values in one insert so it looks a lot longer than it should be. I also have to declare each as variable since Synapse can only insert literal or variable. I then cross-joined with my ItemInformation table.
CREATE TABLE #temp_dates
(
year SMALLINT,
week TINYINT
);
CREATE TABLE #temp_weeks
(
week_num TINYINT
);
DECLARE #year_counter SMALLINT
SET #year_counter = 2019
DECLARE #week_counter TINYINT
WHILE ( #year_counter <= Datepart(year, Getdate() - 1) + 2 )
BEGIN
SET #week_counter = 1;
DECLARE #day_1 TINYINT
SET #day_1 = Datepart(isowk, Concat('12-25-', #year_counter))
DECLARE #day_2 TINYINT
SET #day_2 = Datepart(isowk, Concat('12-26-', #year_counter))
DECLARE #day_3 TINYINT
SET #day_3 = Datepart(isowk, Concat('12-27-', #year_counter))
DECLARE #day_4 TINYINT
SET #day_4 = Datepart(isowk, Concat('12-28-', #year_counter))
DECLARE #day_5 TINYINT
SET #day_5 = Datepart(isowk, Concat('12-29-', #year_counter))
DECLARE #day_6 TINYINT
SET #day_6 = Datepart(isowk, Concat('12-30-', #year_counter))
DECLARE #day_7 TINYINT
SET #day_7 = Datepart(isowk, Concat('12-31-', #year_counter))
TRUNCATE TABLE #temp_weeks
INSERT INTO #temp_weeks
(week_num)
VALUES (#day_1)
INSERT INTO #temp_weeks
(week_num)
VALUES (#day_2)
INSERT INTO #temp_weeks
(week_num)
VALUES (#day_3)
INSERT INTO #temp_weeks
(week_num)
VALUES (#day_4)
INSERT INTO #temp_weeks
(week_num)
VALUES (#day_5)
INSERT INTO #temp_weeks
(week_num)
VALUES (#day_6)
INSERT INTO #temp_weeks
(week_num)
VALUES (#day_7)
DECLARE #max_week TINYINT
SET #max_week = (SELECT Max(week_num)
FROM #temp_weeks)
WHILE ( #week_counter <= #max_week )
BEGIN
INSERT INTO #temp_dates
(year,
week)
VALUES ( #year_counter,
#week_counter )
SET #week_counter = #week_counter + 1
END
SET #year_counter = #year_counter + 1
END
DROP TABLE #temp_weeks;
SELECT i.model,
d.year,
d.week
FROM dbo.iteminformation i
CROSS JOIN #temp_dates d
ORDER BY model,
year,
week
DROP TABLE #temp_dates

Related

Can I "left join" days between 2 dates in sql server?

There is a table in SQL Server where data is entered day by day. In this table, data is not filled in some days.
Therefore, there are no records in the table.
Sample: dataTable
I need to generate a report like the one below from this table.
Create a table with all the days of the year. I know that I can output a report by "joining" the "dataTable" table.
But this solution seems a bit strange to me.
Is there another way?
the code i use for temp date table
CREATE TABLE tempDate (
calendarDate date,
PRIMARY KEY (calendarDate)
)
DECLARE
#start DATE= '2021-01-01',
#dateCount INT= 730,
#rowNumber INT=1
WHILE (#rowNumber < #dateCount)
BEGIN
INSERT INTO tempDate values (DATEADD(DAY, #rowNumber, #start))
set #rowNumber=#rowNumber+1
END
GO
select * from tempDate
This is how I join using this table
SELECT
*
FROM
tempDate td WITH (NOLOCK)
LEFT JOIN dataTable dt WITH (NOLOCK) ON dt.reportDate = td.calendarDate
WHERE
td.calendarDate BETWEEN '2021-09-05' AND '2021-09-15'
Create a table with all the days of the year. I know that I can output a report by "joining" the "dataTable" table.
This is the way. You can generate that "table" on the fly if you really want to, but normally the best way is to simply have a calendar table.
You can use common expression tables for dates. The code you need:
IF(OBJECT_ID('tempdb..#t') IS NOT NULL)
BEGIN
DROP TABLE #t
END
CREATE TABLE #t
(
id int,
dt date,
dsc varchar(100),
)
INSERT INTO #t
VALUES
(1, '2021.09.08', 'a'),
(1, '2021.09.09', 'b'),
(1, '2021.09.12', 'c')
DECLARE #minDate AS DATE
SET #minDate = (SELECT MIN(dt) FROM #t)
DECLARE #maxDate AS DATE
SET #maxDate = (SELECT MAX(dt) FROM #t)
;WITH cte
AS
(
SELECT #minDate AS [dt]
UNION ALL
SELECT DATEADD(DAY, 1, [dt])
FROM cte
WHERE DATEADD(DAY, 1, [dt])<=#maxDate
)
SELECT
ISNULL(CAST(t.id AS VARCHAR(10)), '') AS [id],
cte.dt AS [dt],
ISNULL(t.dsc, 'No record has been entered in the table.') AS [dsc]
FROM
cte
LEFT JOIN #t t on t.dt=cte.dt
The fastest method is to use a numbers table, you can get a date list between 2 dates with that:
DECLARE #Date1 DATE, #Date2 DATE
SET #Date1 = '20200528'
SET #Date2 = '20200625'
SELECT DATEADD(DAY,number+1,#Date1) [Date]
FROM master..spt_values
WHERE type = 'P'
AND DATEADD(DAY,number+1,#Date1) < #Date2
If you go go in LEFT JOIN this select, whit your table, you have the result that you want.
SELECT *
FROM (SELECT DATEADD(DAY,number+1,#Date1) [Date]
FROM master..spt_values WITH (NOLOCK)
WHERE type = 'P'
AND DATEADD(DAY,number+1,#Date1) < #Date2 ) as a
LEFT JOIN yourTable dt WITH (NOLOCK) ON a.date = dt.reportDate
WHERE td.[Date] BETWEEN '2021-09-05' AND '2021-09-15'

How To Get WeekDay Names Between Two WeekDay Name Fields in SQL Server

I have two columns with weekday names my question is how to get all weekday names between those two weekday names
Ex:
WeekStartDate WeekEndDate
monday friday
I want result like this
Monday, Tuesday, Wednesday, Thursday, Friday
Please help me
Thank you
Could you use the Between Operator in Where Clause and add a INT column as between won't work on nvarchar i guess.
Create table #temp(
data int,
dayN Nvarchar(100)
,dayc int
)
SELECT * FROM #TEMP
Insert into #temp values (10,'Monday',2)
Insert into #temp values (20,'tuesday',3)
Insert into #temp values (30,'wednesday',4)
Insert into #temp values (40,'thursday',5)
Insert into #temp values (50,'friday',6)
SELECT Data
FROM #TEMP
where dayN between 'Monday' and 'Wednesday'
--Above query gives wrong result
SELECT Data
FROM #TEMP
where dayC between 3 and 6
You can do It in following:
declare #tbl table (WeekStartDate nvarchar(60),WeekEndDate nvarchar(60) )
insert into #tbl values ('Monday', 'Friday')
declare #weeknums table (name nvarchar(60), num int)
insert into #weeknums values
('Monday' , 1 ),
('Tuesday' , 2 ),
('Wednesday' , 3 ),
('Thursday' , 4 ),
('Friday' , 5 ),
('Saturday' , 6 ),
('Sunday' , 7 )
declare #min nvarchar(max) = (select min(num)
from #tbl t
join #weeknums w on t.WeekStartDate = w.name or t.WeekEndDate = w.name)
declare #max nvarchar(max) = (select max(num)
from #tbl t
join #weeknums w on t.WeekStartDate = w.name or t.WeekEndDate = w.name)
select w.*
from #weeknums w
where w.num between #min and #max

Passing in Week Day name to get nearest date in SQL

I'm working on a query that deals with a frequency value (i.e. Mondays, Tuesdays, etc. - Think assignments).
So in my query I currently have a result of
jobId:1, personId:100, frequencyVal: 'Mondays'
jobId:2, personId:101, frequencyVal: 'Saturdays'
What I need is the next the 4 future(or current) dates for the frequencyVal.
So if today is 1/3/2015
I would need my result set to be
jobId:1, personId:100, frequencyVal: 'Mondays', futureDates: '1/5,1/12,1/19,1/26'
jobId:2, personId:102, frequencyVal: 'Saturdays', futureDates: '1/3,1/10,1/17,1/24'
I was looking at the following post:
How to find the Nearest (day of the week) for a given date
But that sets it for a specific date. And I'm looking at this being a web application and I want the dates for the current date. So if I try to run this query next Tuesday the future dates for jobId:1 would remove the 1/5 and add the 2/2.
Is there a way to pass in a weekday value to get the next nearest date?
I prefer a calendar table for this kind of query. Actually, I prefer a calendar table over date functions for most queries. Here's a minimal one. The one I use in production has more columns and more rows. (100 years of data is only 37k rows.)
create table calendar (
cal_date date not null primary key,
day_of_week varchar(15)
);
insert into calendar (cal_date) values
('2015-01-01'), ('2015-01-02'), ('2015-01-03'), ('2015-01-04'),
('2015-01-05'), ('2015-01-06'), ('2015-01-07'), ('2015-01-08'),
('2015-01-09'), ('2015-01-10'), ('2015-01-11'), ('2015-01-12'),
('2015-01-13'), ('2015-01-14'), ('2015-01-15'), ('2015-01-16'),
('2015-01-17'), ('2015-01-18'), ('2015-01-19'), ('2015-01-20'),
('2015-01-21'), ('2015-01-22'), ('2015-01-23'), ('2015-01-24'),
('2015-01-25'), ('2015-01-26'), ('2015-01-27'), ('2015-01-28'),
('2015-01-29'), ('2015-01-30'), ('2015-01-31'),
('2015-02-01'), ('2015-02-02'), ('2015-02-03'), ('2015-02-04'),
('2015-02-05'), ('2015-02-06'), ('2015-02-07'), ('2015-02-08'),
('2015-02-09'), ('2015-02-10'), ('2015-02-11'), ('2015-02-12'),
('2015-02-13'), ('2015-02-14'), ('2015-02-15'), ('2015-02-16'),
('2015-02-17'), ('2015-02-18'), ('2015-02-19'), ('2015-02-20'),
('2015-02-21'), ('2015-02-22'), ('2015-02-23'), ('2015-02-24'),
('2015-02-25'), ('2015-02-26'), ('2015-02-27'), ('2015-02-28')
;
update calendar
set day_of_week = datename(weekday, cal_date);
alter table calendar
alter column day_of_week varchar(15) not null;
alter table calendar
add constraint cal_date_matches_dow
check (datename(weekday, cal_date) = day_of_week);
create index day_of_week_ix on calendar (day_of_week);
Set the privileges so that
everyone can select, but
almost nobody can insert new rows, and
even fewer people can delete rows.
(Or write a constraint that can guarantee there are no gaps. I think you can do that in SQL Server.)
You can select the next four Mondays after today with a very simple SQL statement. (The current date is 2015-01-05, which is a Monday.)
select top 4 cal_date
from calendar
where cal_date > convert(date, getdate())
and day_of_week = 'Monday'
order by cal_date;
CAL_DATE
--
2015-01-12
2015-01-19
2015-01-26
2015-02-02
For me, this is a huge advantage. No procedural code. Simple SQL that is obviously right. Big win.
Your sample table
create table #t
(
jobId int,
personId int,
frequencyVal varchar(10)
);
insert into #t values (1,100,'Mondays'),(2,101,'Saturdays');
QUERY 1 : Select nearest 4 week of days in current month for particular week day
-- Gets first day of month
DECLARE #FIRSTDAY DATE=DATEADD(month, DATEDIFF(month, 0, GETDATE()), 0)
;WITH CTE as
(
-- Will find all dates in current month
SELECT CAST(#FIRSTDAY AS DATE) as DATES
UNION ALL
SELECT DATEADD(DAY,1,DATES)
FROM CTE
WHERE DATES < DATEADD(MONTH,1,#FIRSTDAY)
)
,CTE2 AS
(
-- Join the #t table with CTE on the datename+'s'
SELECT jobId,personId,frequencyVal,DATES,
-- Get week difference for each weekday
DATEDIFF(WEEK,DATES,GETDATE()) WEEKDIFF,
-- Count the number of weekdays in a month
COUNT(DATES) OVER(PARTITION BY DATENAME(WEEKDAY,CTE.DATES)) WEEKCOUNT
FROM CTE
JOIN #t ON DATENAME(WEEKDAY,CTE.DATES)+'s' = #t.frequencyVal
WHERE MONTH(DATES)= MONTH(GETDATE())
)
-- Converts to CSV and make sure that only nearest 4 week of days are generated for month
SELECT DISTINCT C2.jobId,C2.personId,frequencyVal,
SUBSTRING(
(SELECT ', ' + CAST(DATEPART(MONTH,DATES) AS VARCHAR(2)) + '/' +
CAST(DATEPART(DAY,DATES) AS VARCHAR(2))
FROM CTE2
WHERE C2.jobId=jobId AND C2.personId=personId AND C2.frequencyVal=frequencyVal AND
((WEEKDIFF<3 AND WEEKDIFF>-3 AND WEEKCOUNT = 5) OR WEEKCOUNT <= 4)
ORDER BY CTE2.DATES
FOR XML PATH('')),2,200000) futureDates
FROM CTE2 C2
SQL FIDDLE
For example, in Query2 the nearest date(here we take example as Saturday) of
2015-Jan-10 will be 01/03,01/10,01/17,01/24
2015-Jan-24 will be 01/10,01/17,01/24,01/31
QUERY 2 : Select next 4 week's dates for particular week day irrelevant of month
;WITH CTE as
(
-- Will find the next 4 week details
SELECT CAST(GETDATE() AS DATE) as DATES
UNION ALL
SELECT DATEADD(DAY,1,DATES)
FROM CTE
WHERE DATES < DATEADD(DAY,28,GETDATE())
)
,CTE2 AS
(
-- Join the #t table with CTE on the datename+'s'
SELECT jobId,personId,frequencyVal, DATES,
ROW_NUMBER() OVER(PARTITION BY DATENAME(WEEKDAY,CTE.DATES) ORDER BY CTE.DATES) DATECNT
FROM CTE
JOIN #t ON DATENAME(WEEKDAY,CTE.DATES)+'s' = #t.frequencyVal
)
-- Converts to CSV and make sure that only 4 days are generated for month
SELECT DISTINCT C2.jobId,C2.personId,frequencyVal,
SUBSTRING(
(SELECT ', ' + CAST(DATEPART(MONTH,DATES) AS VARCHAR(2)) + '/' +
CAST(DATEPART(DAY,DATES) AS VARCHAR(2))
FROM CTE2
WHERE C2.jobId=jobId AND C2.personId=personId AND C2.frequencyVal=frequencyVal
AND DATECNT < 5
ORDER BY CTE2.DATES
FOR XML PATH('')),2,200000) futureDates
FROM CTE2 C2
SQL FIDDLE
The following would be the output if the GETDATE() (if its Saturday) is
2015-01-05 - 1/10, 1/17, 1/24, 1/31
2015-01-24 - 1/24, 1/31, 2/7, 2/14
There's no built-in function to do it. But you can try this, you may place it inside a Scalar-Valued Function:
DECLARE #WeekDay VARCHAR(10) = 'Monday';
DECLARE #WeekDayInt INT;
SELECT #WeekDayInt = CASE #WeekDay
WHEN 'SUNDAY' THEN 1
WHEN 'MONDAY' THEN 2
WHEN 'TUESDAY' THEN 3
WHEN 'WEDNESDAY' THEN 4
WHEN 'THURSDAY' THEN 5
WHEN 'FRIDAY' THEN 6
WHEN 'SATURDAY' THEN 7 END
SELECT CONVERT(DATE, DATEADD(DAY, (DATEPART(WEEKDAY, GETDATE()) + #WeekDayInt) % 7, GETDATE())) AS NearestDate
UPDATE:
Looks like radar was right, here's the solution:
DECLARE #WeekDay VARCHAR(10) = 'Monday';
DECLARE #WeekDayInt INT;
DECLARE #Date DATETIME = GETDATE();
SELECT #WeekDayInt = CASE #WeekDay
WHEN 'SUNDAY' THEN 1
WHEN 'MONDAY' THEN 2
WHEN 'TUESDAY' THEN 3
WHEN 'WEDNESDAY' THEN 4
WHEN 'THURSDAY' THEN 5
WHEN 'FRIDAY' THEN 6
WHEN 'SATURDAY' THEN 7 END
DECLARE #Diff INT = DATEPART(WEEKDAY, #Date) - #WeekDayInt;
SELECT CONVERT(DATE, DATEADD(DAY, CASE WHEN #Diff >= 0 THEN 7 - #Diff ELSE ABS(#Diff) END, #Date)) AS NearestDate
Try this - based on king.code's answer to get the nearest date.
create table #t
(
jobId int,
personId int,
frequencyVal varchar(10)
);
insert into #t values (1,100,'Mondays'),(2,101,'Saturdays');
WITH cte(n) AS
(
SELECT 0
UNION ALL
SELECT n+1 FROM cte WHERE n < 3
)
select #t.jobId, #t.personId, #t.frequencyVal, STUFF(a.d, 1, 1, '') AS FutureDates
from #t
cross apply (SELECT CASE #t.frequencyVal
WHEN 'SUNDAYS' THEN 1
WHEN 'MONDAYS' THEN 2
WHEN 'TUESDAYS' THEN 3
WHEN 'WEDNESDAYS' THEN 4
WHEN 'THURSDAYS' THEN 5
WHEN 'FRIDAYS' THEN 6
WHEN 'SATURDAYS' THEN 7
END)tranlationWeekdays(n)
cross apply (select ',' + CONVERT(varchar(10), CONVERT(date,dateadd(WEEK, cte.n,CONVERT(DATE, DATEADD(DAY, (DATEPART(WEEKDAY, GETDATE()) + tranlationWeekdays.n) % 7, GETDATE()))))) from cte FOR XML PATH('')) a(d);
drop table #t;
Try this,
DECLARE #YEAR INT=2015
DECLARE #MONTH INT=1
DECLARE #DAY INT=1
DECLARE #DATE DATE = (SELECT DateFromParts(#Year, #Month, #Day))
DECLARE #TOTAL_DAYS INT =(SELECT DatePart(DY, #DATE));
WITH CTE1
AS (SELECT T_DAY=(SELECT DateName(DW, #DATE)),
#DATE AS T_DATE,
#DAY AS T_DDAY
UNION ALL
SELECT T_DAY=(SELECT DateName(DW, DateAdd(DAY, T_DDAY + 1, #DATE))),
DateAdd(DAY, T_DDAY + 1, #DATE) AS T_DATE,
T_DDAY + 1
FROM CTE1
WHERE T_DDAY + 1 <= 364)
SELECT DISTINCT T_DAY,
Stuff((SELECT ',' + CONVERT(VARCHAR(30), T_DATE)
FROM CTE1 A
WHERE A.T_DAY=CTE1.T_DAY AND A.T_DATE > GetDate() AND A.T_DATE<(DATEADD(WEEK,4,GETDATE()))
FOR XML PATH('')), 1, 1, '') AS FUTURE
FROM CTE1
ORDER BY T_DAY
OPTION (MAXRECURSION 365)
This is a simpler way I think, and I think it fits your requirements.
Note that I have changed your frequency_val column to an integer that represents the day of the week from SQL servers perspective and added a calculated column to illustrate how you can easily derive the day name from that.
declare #t table
(
jobId int,
personId int,
--frequencyVal varchar(10)
frequency_val int,
frequency_day as datename(weekday,frequency_val -1) + 's'
);
declare #num_occurances int = 4
declare #from_date date = dateadd(dd,3,getdate()) -- this will allow you to play with the date simply by changing the increment value
insert into #t
values
(1,100,1),--'Mondays'),
(2,101,6),--'Saturdays');
(3,101,7),--'Saturdays');
(4,100,2)--'Mondays'),
--select * from #t
;with r_cte (days_ahead, occurance_date)
as (select 0, convert(date,#from_date,121)
union all
select r_cte.days_ahead +1, convert(date,dateadd(DD, r_cte.days_ahead+1, #from_date),121)
from r_cte
where r_cte.days_ahead < 7 * #num_occurances
)
select t.*, r_cte.occurance_date
from
#t t
inner join r_cte
on DATEPART(WEEKDAY, dateadd(dd,##DATEFIRST - 1 ,r_cte.occurance_date)) = t.frequency_val
Having seen the use of DATENAME in some of the answers already given, I'd like to point out that return values of DATENAME might vary depending on your current language setting, but you can save the current language setting and ensure usage of us_english so you can be confident to use English weekday names.
Now here is my slightly different approach to get the 4 next dates that fall on a certain (known) weekday, using a user defined table valued function that allows to create a number sequence table (yes this is a pretty dull function, you have to pass MaxValue greater MinValue, but that could be easily enhanced, if needed, but hey, it does the job). Using that function span a table over 28 values (next 28 days should indeed include the next 4 relevant weekdays ;)), apply DATEADD on GETDATE and reduce the result set with WHERE to only those values that have the right weekday:
CREATE FUNCTION GetIntSequence(#MinValue INT, #MaxValue INT)
RETURNS #retSequence TABLE
(
IntValue INT NOT NULL
)
BEGIN
DECLARE #i INT = (SELECT #MinValue)
WHILE #i <= #MaxValue
BEGIN
INSERT INTO #retSequence (IntValue) SELECT #i
SELECT #i = #i + 1
END
RETURN
END
GO
DECLARE #weekDay NVARCHAR(MAX) = 'Monday' --(or Tuesday, wednesday, ...)
--save current language setting
DECLARE #languageBackup NVARCHAR(MAX) = (SELECT ##LANGUAGE)
--ensure us english language setting for reliable weekday names
SET LANGUAGE us_english;
SELECT FourWeeks.SomeDay FROM
(
SELECT
DATEADD(DAY, IntValue, GETDATE()) AS SomeDay
FROM dbo.GetIntSequence(1, 28)
) AS FourWeeks
WHERE DATENAME(WEEKDAY, SomeDay) = #weekDay
--restore old language setting
SET LANGUAGE #languageBackup;
GO
DROP FUNCTION dbo.GetIntSequence

How to make this sql query

I have 2 SQL Server tables with the following structure
Turns-time
cod_turn (PrimaryKey)
time (datetime)
Taken turns
cod_taken_turn (Primary Key)
cod_turn
...
and several other fields which are irrelevant to the problem. I cant alter the table structures because the app was made by someone else.
given a numeric variable parameter, which we will assume to be "3" for this example, and a given time, I need to create a query which looking from that time on, it looks the first 3 consecutive records by time which are not marked as "taken". For example:
For example, for these turns, starting by the time of "8:00" chosen by the user
8:00 (not taken)
9:00 (not taken)
10:00 (taken)
11:00 (not taken)
12:00 (not taken)
13:00 (not taken)
14:00 (taken)
The query it would have to list
11:00
12:00
13:00
I cant figure out how to make the query in pure sql, if possible.
with a cursor
declare #GivenTime datetime,
#GivenSequence int;
select #GivenTime = cast('08:00' as datetime),
#GivenSequence = 3;
declare #sequence int,
#code_turn int,
#time datetime,
#taked int,
#firstTimeInSequence datetime;
set #sequence = 0;
declare turnCursor cursor FAST_FORWARD for
select turn.cod_turn, turn.[time], taken.cod_taken_turn
from [Turns-time] as turn
left join [Taken turns] as taken on turn.cod_turn = taken.cod_turn
where turn.[time] >= #GivenTime
order by turn.[time] asc;
open turnCursor;
fetch next from turnCursor into #code_turn, #time, #taked;
while ##fetch_status = 0 AND #sequence < #GivenSequence
begin
if #taked IS NULL
select #firstTimeInSequence = coalesce(#firstTimeInSequence, #time)
,#sequence = #sequence + 1;
else
select #sequence = 0,
#firstTimeInSequence = null;
fetch next from turnCursor into #code_turn, #time, #taked;
end
close turnCursor;
deallocate turnCursor;
if #sequence = #GivenSequence
select top (#GivenSequence) * from [Turns-time] where [time] >= #firstTimeInSequence
order by [time] asc
WITH Base AS (
SELECT *,
CASE WHEN EXISTS(
SELECT *
FROM Taken_turns taken
WHERE taken.cod_turn = turns.cod_turn) THEN 1 ELSE 0 END AS taken
FROM [Turns-time] turns)
, RecursiveCTE As (
SELECT TOP 1 cod_turn, [time], taken AS run, 0 AS grp
FROM Base
WHERE [time] >= #start_time
ORDER BY [time]
UNION ALL
SELECT R.cod_turn, R.[time], R.run, R.grp
FROM (
SELECT T.*,
CASE WHEN T.taken = 0 THEN 0 ELSE run+1 END AS run,
CASE WHEN T.taken = 0 THEN grp + 1 ELSE grp END AS grp,
rn = ROW_NUMBER() OVER (ORDER BY T.[time])
FROM Base T
JOIN RecursiveCTE R
ON R.[time] < T.[time]
) R
WHERE R.rn = 1 AND run < #run_length
), T AS(
SELECT *,
MAX(grp) OVER () AS FinalGroup,
COUNT(*) OVER (PARTITION BY grp) AS group_size
FROM RecursiveCTE
)
SELECT cod_turn,time
FROM T
WHERE grp=FinalGroup AND group_size=#run_length
I think there is not a simple way to achieve this.
But probably there are many complex ways :). This is an approach that should work in Transact-SQL:
CREATE TABLE #CONSECUTIVE_TURNS (id int identity, time datetime, consecutive int)
INSERT INTO #CONSECUTIVE_TURNS (time, consecutive, 0)
SELECT cod_turn
, time
, 0
FROM Turns-time
ORDER BY time
DECLARE #i int
#n int
SET #i = 0
SET #n = 3 -- Number of consecutive not taken records
while (#i < #n) begin
UPDATE #CONSECUTIVE_TURNS
SET consecutive = consecutive + 1
WHERE not exists (SELECT 1
FROM Taken-turns
WHERE id = cod_turn + #i
)
SET #i = #i + 1
end
DECLARE #firstElement int
SELECT #firstElement = min(id)
FROM #CONSECUTIVE_TURNS
WHERE consecutive >= #n
SELECT *
FROM #CONSECUTIVE_TURNS
WHERE id between #firstElement
and #firstElement + #n - 1
This is untested but I think it will work.
Pure SQL
SELECT TOP 3 time FROM [turns-time] WHERE time >= (
-- get first result of the 3 consecutive results
SELECT TOP 1 time AS first_result
FROM [turns-time] tt
-- start from given time, which is 8:00 in this case
WHERE time >= '08:00'
-- turn is not taken
AND cod_turn NOT IN (SELECT cod_turn FROM taken_turns)
-- 3 consecutive turns from current turn are not taken
AND (
SELECT COUNT(*) FROM
(
SELECT TOP 3 cod_turn AS selected_turn FROM [turns-time] tt2 WHERE tt2.time >= tt.time
GROUP BY cod_turn ORDER BY tt2.time
) AS temp
WHERE selected_turn NOT IN (SELECT cod_turn FROM taken_turns)) = 3
) ORDER BY time
Note: I tested it on Postgresql (with some code modification), but not MS SQL Server. I'm not sure about performance compared to T-SQL.
Another set-based solution (tested):
DECLARE #Results TABLE
(
cod_turn INT NOT NULL
,[status] TINYINT NOT NULL
,RowNumber INT PRIMARY KEY
);
INSERT #Results (cod_turn, [status], RowNumber)
SELECT a.cod_turn
,CASE WHEN b.cod_turn IS NULL THEN 1 ELSE 0 END [status] --1=(not taken), 0=(taken)
,ROW_NUMBER() OVER(ORDER BY a.[time]) AS RowNumber
FROM [Turns-time] a
LEFT JOIN [Taken_turns] b ON a.cod_turn = b.cod_turn
WHERE a.[time] >= #Start;
--SELECT * FROM #Results r ORDER BY r.RowNumber;
SELECT *
FROM
(
SELECT TOP(1) ca.LastRowNumber
FROM #Results a
CROSS APPLY
(
SELECT SUM(c.status) CountNotTaken, MAX(c.RowNumber) LastRowNumber
FROM
(
SELECT TOP(#Len)
b.RowNumber, b.[status]
FROM #Results b
WHERE b.RowNumber <= a.RowNumber
ORDER BY b.RowNumber DESC
) c
) ca
WHERE ca.CountNotTaken = #Len
ORDER BY a.RowNumber ASC
) x INNER JOIN #Results y ON x.LastRowNumber - #Len + 1 <= y.RowNumber AND y.RowNumber <= x.LastRowNumber;

SQL Server related question

I have this thing that i need to do and some advices will be greatly appreciated.
I have a SQL server table with some phone calls.For each phone call i have the start and end time.
What i need to accomplish: a stored procedure which for a certain period of time, let's say 5 hours at a x interval, lets say 2 minutes returns the number of connected calls.
Something like:
Interval Nr of Calls Connected
01-01-2010 12:00:00 - 01-01-2010 12:05:00 30
01-01-2010 12:05:01 - 01-01-2010 12:10:00 10
.............
Which will be the fastest way to do that? Thank you for your help
This will work for intervals that have calls ...
Declare #datetimestart datetime
Declare #interval int
Set #datetimestart = '2009-01-01 12:00:00'
Set #interval = 5 --in minutes
Select
[start_interval], [end_interval] , count([start_interval]) as [calls]
From
(
Select
DateAdd( Minute,Floor(DateDiff(Minute,#datetimestart,[date])/#interval)*#interval
,#datetimestart) ,
DateAdd( Minute,#interval + Floor(DateDiff(Minute,#datetimestart,[date])/#interval)*#interval
,#datetimestart)
From yourTable
) As W([start_interval],[end_interval])
group by [start_interval], [end_interval]
This will work for all intervals regardless of number of calls..
Declare #datetimestart datetime, #datetimeend datetime, #datetimecurrent datetime
Declare #interval int
Set #datetimestart = '2009-01-01 12:00:00'
Set #interval = 10
Set #datetimeend = (Select max([date]) from yourtable)
SET #datetimecurrent = #datetimestart
declare #temp as table ([start_interval] datetime, [end_interval] datetime)
while #datetimecurrent < #datetimeend
BEGIN
insert into #temp select (#datetimecurrent), dateAdd( minute, #interval, #datetimecurrent)
set #datetimecurrent = dateAdd( minute, #interval, #datetimecurrent)
END
Select
*
From
(
Select
[start_interval],[end_interval], count(d.[start_time])
From #temp t left join yourtable d on d.[start_time] between t.[start_interval] and t.[end_interval]
) As W([start_interval],[end_interval], [calls])
I Altered Gaby's example a little to do What you expected
Declare #datetimeend datetime
,#datetimecurrent datetime
,#interval int
Set #interval = 10
Set #datetimeend = (Select max([end_time]) from Calls)
SET #datetimecurrent = '2010-04-17 14:20:00'
declare #temp as table ([start_interval] datetime, [end_interval] datetime)
while #datetimecurrent < #datetimeend
BEGIN
insert into #temp select (#datetimecurrent), dateAdd( minute, #interval, #datetimecurrent)
set #datetimecurrent = dateAdd( minute, #interval, #datetimecurrent)
END
Select
[start_interval],[end_interval], count(d.id) [COUNT]
From #temp t
left join Calls d on
d.end_time >= t.start_interval
AND d.start_time <= t.end_interval
GROUP BY [start_interval],[end_interval]
used this to create the table and fill it
CREATE TABLE dbo.Calls
(
id int NOT NULL IDENTITY (1, 1),
start_time datetime NOT NULL,
end_time datetime NULL,
caller nvarchar(50) NULL,
receiver nvarchar(50) NULL
) ON [PRIMARY]
GO
ALTER TABLE dbo.Calls ADD CONSTRAINT
PK_Calls PRIMARY KEY CLUSTERED
(
id
) ON [PRIMARY]
GO
DECLARE #I INT
SET #I = 0
WHILE #I < 100
BEGIN
INSERT INTO Calls
(start_time, end_time)
select
DATEADD(HOUR,-#I,DATEADD(MINUTE,-10,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-9,GETDATE()))
UNION
select
DATEADD(HOUR,-#I,DATEADD(MINUTE,-9,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-8,GETDATE()))
UNION
select
DATEADD(HOUR,-#I,DATEADD(MINUTE,-8,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-7,GETDATE()))
UNION
select
DATEADD(HOUR,-#I,DATEADD(MINUTE,-7,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-6,GETDATE()))
UNION
select
DATEADD(HOUR,-#I,DATEADD(MINUTE,-6,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-5,GETDATE()))
UNION
SELECT
DATEADD(HOUR,-#I,DATEADD(MINUTE,-5,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-4,GETDATE()))
UNION
select
DATEADD(HOUR,-#I,DATEADD(MINUTE,-4,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-3,GETDATE()))
UNION
select
DATEADD(HOUR,-#I,DATEADD(MINUTE,-3,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-2,GETDATE()))
UNION
select
DATEADD(HOUR,-#I,DATEADD(MINUTE,-2,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-1,GETDATE()))
UNION
select
DATEADD(HOUR,-#I,DATEADD(MINUTE,-1,GETDATE()))
,DATEADD(HOUR,-#I,DATEADD(MINUTE,-0,GETDATE()));
SET #I = #I + 1
END
Done in SQL Server 2008
but the script would work in other versions
I would use a Numbers pivot table to get the time intervals and then count all calls which overlapped the interval:
SELECT Intervals.IntervalStart
,Intervals.IntervalEnd
,COUNT(*)
FROM (
SELECT DATEADD(MINUTE, Numbers * 2, #StartTime) AS IntervalStart
,DATEADD(MINUTE, (Numbers + 1) * 2, #StartTime) AS IntervalEnd
FROM Numbers
WHERE Numbers BETWEEN 0 AND (5 * 60 / 2)
) AS Intervals
LEFT JOIN Calls
ON Calls.CallEnd >= Intervals.IntervalStart
AND Calls.CallStart < Intervals.IntervalEnd
GROUP BY Intervals.IntervalStart
,Intervals.IntervalEnd
To get the empty intervals, you would need to LEFT JOIN to this from another "Intervals" derived table.
How about this approach:
select Year(StartTime) as Year, Month(StartTime) as Month, Day(StartTime) as Day, datepart(hh, StartTime) as Hour, datepart(mm, StartTime) / 2 as TwoMinuteSegment, count(*)
from MyTable
where StartDate between '01-01-2010 12:00:00' and '01-01-2010 17:00:00'
group by Year(StartTime), Month(StartTime), Day(StartTime), datepart(hh, StartTime), datepart(mm, StartTime) / 2

Resources