Sql-server – SQL recursion and cte – dates timeseries

cterecursivesql serversql-server-2016

I need to build vector(s) of time series dates (say 10) for a subset of start dates.

The following statement returns 10 consecutive dates starting from '2010-01-01'.

;with cte as
(select 1 i union all
 select i+1 i from cte where i < 10)
select '2010-01-01' as start_date, dateadd(d, i-1, '2010-01-01') AS trial_date
INTO #tmp_trialdates
FROM cte

Does anyone have any suggestions as to how I could incorporate querying a table (or other dataset) of multiple start dates (not necessarily sequential), generating time series dates for each start date, then populating the data in the one output table?

Best Answer

For this you need to change the anchor set of the recursive CTE, so each row starts with the dates from the other table before starting the recursive portion.

DECLARE @AmountDays INT = 10

;WITH GeneratedDays AS
(
    -- Anchor
    SELECT
        DateSeriesIdentifier = T.DateSeriesIdentifier,
        Date = T.Date, -- Assuming it's DATE data type
        RecursionLevel = 1
    FROM
        OtherTable AS T

    UNION ALL

    -- Recursion
    SELECT
        DateSeriesIdentifier = G.DateSeriesIdentifier,
        Date = DATEADD(DAY, 1, G.Date),
        RecursionLevel = G.RecursionLevel + 1
    FROM
        GeneratedDays AS G
    WHERE
        G.RecursionLevel + 1 <= @AmountDays
)
SELECT
    G.DateSeriesIdentifier,
    G.Date,
    G.RecursionLevel
FROM
    GeneratedDays AS G
ORDER BY
    G.DateSeriesIdentifier,
    G.RecursionLevel

This is a result from a set I used to test which had dates 2016-02-04, 2018-01-19 and 2018-08-30:

DateSeriesIdentifier    Date            RecursionLevel
21                      2016-02-04      1
21                      2016-02-05      2
21                      2016-02-06      3
21                      2016-02-07      4
21                      2016-02-08      5
21                      2016-02-09      6
21                      2016-02-10      7
21                      2016-02-11      8
21                      2016-02-12      9
21                      2016-02-13      10
620646                  2018-01-19      1
620646                  2018-01-20      2
620646                  2018-01-21      3
620646                  2018-01-22      4
620646                  2018-01-23      5
620646                  2018-01-24      6
620646                  2018-01-25      7
620646                  2018-01-26      8
620646                  2018-01-27      9
620646                  2018-01-28      10
639701                  2018-08-30      1
639701                  2018-08-31      2
639701                  2018-09-01      3
639701                  2018-09-02      4
639701                  2018-09-03      5
639701                  2018-09-04      6
639701                  2018-09-05      7
639701                  2018-09-06      8
639701                  2018-09-07      9
639701                  2018-09-08      10

Related Solutions

Sql-server – Is it possible to change recursion level of a CTE in a function

This is what Mikael was hinting at, I believe.

You will find things work a lot quicker without that MAX parameter, are you really going to need that big a string?

CREATE FUNCTION [dbo].[StringSplit]
(
    @String NVARCHAR(MAX)    
)
RETURNS TABLE
WITH SCHEMABINDING
AS
RETURN
(   
  WITH E1(N) AS (
                 SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL
                 SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL
                 SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1
                ),                          --10E+1 or 10 rows
       E2(N) AS (SELECT 1 FROM E1 a, E1 b), --10E+2 or 100 rows
       E4(N) AS (SELECT 1 FROM E2 a, E2 b), --10E+4 or 10,000 rows max
 cteTally(N) AS (SELECT TOP (ISNULL(DATALENGTH(@String),0)) ROW_NUMBER() OVER (ORDER BY (SELECT NULL)) FROM E4)
        SELECT N AS Pos, substring(@String, N, 1) AS Digit
        from cteTally
);

Sql-server – Getting Unavailable dates for renting a product that has stocks

I would use a Calendar table. This table simply has a list of dates for several decades.

CREATE TABLE [dbo].[Calendar](
    [dt] [date] NOT NULL,
 CONSTRAINT [PK_Calendar] PRIMARY KEY CLUSTERED 
(
    [dt] ASC
))

In my system it has few extra columns, such as [IsLastDayOfMonth], [IsLastDayOfQuarter], which are useful in some reports, but in your case you need just the date column. There are many ways to populate such table.

For example, 100K rows (~270 years) from 1900-01-01:

INSERT INTO dbo.Calendar (dt)
SELECT TOP (100000) 
    DATEADD(day, ROW_NUMBER() OVER (ORDER BY s1.[object_id])-1, '19000101') AS dt
FROM sys.all_objects AS s1 CROSS JOIN sys.all_objects AS s2
OPTION (MAXDOP 1);

Sample data

DECLARE @Stocks TABLE (
    StockId int
    , ProductId int
    , AvailableFromDate date);

INSERT INTO @Stocks(StockId, ProductId, AvailableFromDate) VALUES
(1, 1, '2016-01-01'),
(2, 1, '2016-01-01');

DECLARE @RentStockOrders TABLE (
    RentStockOrderId int
    , StockId int
    , BeginRentDate date
    , EndRentDate date);

INSERT INTO @RentStockOrders (RentStockOrderId, StockId, BeginRentDate, EndRentDate) VALUES
(1, 1, '2016-01-15', '2016-02-14'),
(2, 2, '2016-01-30', '2016-02-20'),
(3, 2, '2016-02-26', '2016-03-07'),
(4, 1, '2016-02-29', '2016-03-14');

Parameters

DECLARE @ParamProductID int = 1;
DECLARE @ParamDays int = 14;
DECLARE @ParamStartDate date = '2015-12-01';
DECLARE @ParamEndDate date = '2017-01-01';
-- these dates define some reasonable limit

First variant

It turns out that window functions accept only literal constant as a size of a window, not a variable. Alas. Still, I'll show this query, because it illustrates the approach and shows how simple it would be if SQL Server supported variables as the size of the window. It also gives the correct answer that we can use to verify the second variant.

WITH
CTE_AllDays
-- list all days and stock IDs between @ParamStartDate and @ParamEndDate
-- for each day and stock indicate whether it is available based on AvailableFromDate
AS
(
    SELECT
        S.StockId
        ,dbo.Calendar.dt
        ,CASE WHEN dbo.Calendar.dt >= S.AvailableFromDate 
            THEN 1 ELSE 0 END AS AvailableStockDay
        -- 1 - available
        -- 0 - not available
    FROM
        @Stocks AS S
        INNER JOIN dbo.Calendar ON
            dbo.Calendar.dt >= @ParamStartDate
            AND dbo.Calendar.dt <= @ParamEndDate
    WHERE
        S.ProductId = @ParamProductID
)
,CTE_BookedDays
-- list all booked (unavailable) days for each stock ID
AS
(
    SELECT
        S.StockId
        ,CA.dt
        ,0 AS AvailableStockDay
        -- 0 - not available
    FROM
        @RentStockOrders AS R
        INNER JOIN @Stocks AS S ON S.StockId = R.StockId
        CROSS APPLY
        (
            SELECT dbo.Calendar.dt
            FROM dbo.Calendar
            WHERE
                dbo.Calendar.dt >= @ParamStartDate
                AND dbo.Calendar.dt <= @ParamEndDate
                AND dbo.Calendar.dt >= R.BeginRentDate
                AND dbo.Calendar.dt <= R.EndRentDate
        ) AS CA
    WHERE
        S.ProductId = @ParamProductID
)
,CTE_Daily
-- combine individual availability flags
-- first: multiply to get the final availability for a stock ID and day
-- second: group further by day and SUM flags
AS
(
    SELECT
        CTE_AllDays.dt
        ,CASE WHEN 
            SUM(
            CTE_AllDays.AvailableStockDay * ISNULL(CTE_BookedDays.AvailableStockDay, 1)
            ) = 0 
        THEN 0 ELSE 1 END AS AvailableDay
        -- day is available, if any stock is available
        -- SUM=0 - not available
        -- SUM>0 - available
    FROM
        CTE_AllDays
        LEFT JOIN CTE_BookedDays ON
            CTE_BookedDays.StockId = CTE_AllDays.StockId AND
            CTE_BookedDays.dt = CTE_AllDays.dt
    GROUP BY
        CTE_AllDays.dt
)
,CTE_Sum
-- rolling sum of flags with 14 days window
AS
(
    SELECT
        dt
        ,SUM(CTE_Daily.AvailableDay) OVER (ORDER BY CTE_Daily.dt
            ROWS BETWEEN CURRENT ROW AND 13 FOLLOWING) AS AvailableConsecutive
    -- we can't put @ParamDays here instead of constant
    FROM CTE_Daily
)
-- If a rolling sum =  14, 
-- it means that all 14 consecutive days are available

-- If a rolling sum <> 14, 
-- it means that at least one of the 14 consecutive days is not available
SELECT dt
FROM CTE_Sum
WHERE AvailableConsecutive <> 14
-- we can put @ParamDays here instead of 14, but not above
ORDER BY dt;

Result

Note, I started the range of dates from 2015-12-01, before the general availability.

+------------+
|     dt     |
+------------+
| 2015-12-01 |
| 2015-12-02 |
| 2015-12-03 |
| 2015-12-04 |
| 2015-12-05 |
| 2015-12-06 |
| 2015-12-07 |
| 2015-12-08 |
| 2015-12-09 |
| 2015-12-10 |
| 2015-12-11 |
| 2015-12-12 |
| 2015-12-13 |
| 2015-12-14 |
| 2015-12-15 |
| 2015-12-16 |
| 2015-12-17 |
| 2015-12-18 |
| 2015-12-19 |
| 2015-12-20 |
| 2015-12-21 |
| 2015-12-22 |
| 2015-12-23 |
| 2015-12-24 |
| 2015-12-25 |
| 2015-12-26 |
| 2015-12-27 |
| 2015-12-28 |
| 2015-12-29 |
| 2015-12-30 |
| 2015-12-31 |
| 2016-01-17 |
| 2016-01-18 |
| 2016-01-19 |
| 2016-01-20 |
| 2016-01-21 |
| 2016-01-22 |
| 2016-01-23 |
| 2016-01-24 |
| 2016-01-25 |
| 2016-01-26 |
| 2016-01-27 |
| 2016-01-28 |
| 2016-01-29 |
| 2016-01-30 |
| 2016-01-31 |
| 2016-02-01 |
| 2016-02-02 |
| 2016-02-03 |
| 2016-02-04 |
| 2016-02-05 |
| 2016-02-06 |
| 2016-02-07 |
| 2016-02-08 |
| 2016-02-09 |
| 2016-02-10 |
| 2016-02-11 |
| 2016-02-12 |
| 2016-02-13 |
| 2016-02-14 |
| 2016-02-16 |
| 2016-02-17 |
| 2016-02-18 |
| 2016-02-19 |
| 2016-02-20 |
| 2016-02-21 |
| 2016-02-22 |
| 2016-02-23 |
| 2016-02-24 |
| 2016-02-25 |
| 2016-02-26 |
| 2016-02-27 |
| 2016-02-28 |
| 2016-02-29 |
| 2016-03-01 |
| 2016-03-02 |
| 2016-03-03 |
| 2016-03-04 |
| 2016-03-05 |
| 2016-03-06 |
| 2016-03-07 |
| 2016-12-20 |
| 2016-12-21 |
| 2016-12-22 |
| 2016-12-23 |
| 2016-12-24 |
| 2016-12-25 |
| 2016-12-26 |
| 2016-12-27 |
| 2016-12-28 |
| 2016-12-29 |
| 2016-12-30 |
| 2016-12-31 |
| 2017-01-01 |
+------------+

Second variant

The first part of the query until CTE_Daily is the same. Then, I'll use a gaps-and-islands approach to find the islands of available dates and calculate their sizes.

WITH
CTE_AllDays
-- list all days and stock IDs between @ParamStartDate and @ParamEndDate
-- for each day and stock indicate whether it is available based on AvailableFromDate
AS
(
    SELECT
        S.StockId
        ,dbo.Calendar.dt
        ,CASE WHEN dbo.Calendar.dt >= S.AvailableFromDate 
            THEN 1 ELSE 0 END AS AvailableStockDay
        -- 1 - available
        -- 0 - not available
    FROM
        @Stocks AS S
        INNER JOIN dbo.Calendar ON
            dbo.Calendar.dt >= @ParamStartDate
            AND dbo.Calendar.dt <= @ParamEndDate
    WHERE
        S.ProductId = @ParamProductID
)
,CTE_BookedDays
-- list all booked (unavailable) days for each stock ID
AS
(
    SELECT
        S.StockId
        ,CA.dt
        ,0 AS AvailableStockDay
        -- 0 - not available
    FROM
        @RentStockOrders AS R
        INNER JOIN @Stocks AS S ON S.StockId = R.StockId
        CROSS APPLY
        (
            SELECT dbo.Calendar.dt
            FROM dbo.Calendar
            WHERE
                dbo.Calendar.dt >= @ParamStartDate
                AND dbo.Calendar.dt <= @ParamEndDate
                AND dbo.Calendar.dt >= R.BeginRentDate
                AND dbo.Calendar.dt <= R.EndRentDate
        ) AS CA
    WHERE
        S.ProductId = @ParamProductID
)
,CTE_Daily
-- combine individual availability flags
-- first: multiply to get the final availability for a stock ID and day
-- second: group further by day and SUM flags
AS
(
    SELECT
        CTE_AllDays.dt
        ,CASE WHEN 
            SUM(
            CTE_AllDays.AvailableStockDay * ISNULL(CTE_BookedDays.AvailableStockDay, 1)
            ) = 0 
        THEN 0 ELSE 1 END AS AvailableDay
        -- day is available, if any stock is available
        -- SUM=0 - not available
        -- SUM>0 - available
    FROM
        CTE_AllDays
        LEFT JOIN CTE_BookedDays ON
            CTE_BookedDays.StockId = CTE_AllDays.StockId AND
            CTE_BookedDays.dt = CTE_AllDays.dt
    GROUP BY
        CTE_AllDays.dt
)
,CTE_RowNumbers
-- calculate two sets of row numbers to isolate consecutive rows with 0s and 1s 
-- (gaps and islands)
AS
(
    SELECT
        dt
        ,AvailableDay
        ,ROW_NUMBER() OVER (ORDER BY dt) AS rn1
        ,ROW_NUMBER() OVER (PARTITION BY AvailableDay ORDER BY dt) AS rn2
    FROM CTE_Daily
)
,CTE_Groups
-- each gaps and island will have the same GroupNumber
-- count the size of the Group
-- number the rows within each Group to find if @ParamDays rows fit into the Group
AS
(
    SELECT
        dt
        ,AvailableDay
        ,GroupNumber
        ,COUNT(*) OVER (PARTITION BY GroupNumber) AS GroupSize
        ,ROW_NUMBER() OVER (PARTITION BY GroupNumber ORDER BY dt) AS GroupRN
    FROM
        CTE_RowNumbers
        CROSS APPLY (SELECT rn1 - rn2 AS GroupNumber) AS CA
)
SELECT
    dt
    --,CASE WHEN AvailableDay = 1 AND GroupSize - GroupRN + 1 >= @ParamDays
    --THEN 1 ELSE 0 END AS AvailableConsecutive
FROM CTE_Groups
WHERE
    CASE WHEN AvailableDay = 1 AND GroupSize - GroupRN + 1 >= @ParamDays
    THEN 1 ELSE 0 END = 0
    -- AvailableConsecutive = 0 to list all unavailable days
ORDER BY dt;

The result is the same as in the first variant, but this variant uses parameter @ParamDays.

To understand how it works, run the query starting with the first CTE, examine the results, then add the next CTE, examine the results and so on.

Best Answer

Related Solutions

Sql-server – Is it possible to change recursion level of a CTE in a function

Sql-server – Getting Unavailable dates for renting a product that has stocks

Related Question