TSQL help

  • create table t1
    (key int,
    cnt int 
    )
    insert into t1
    select 2,5
    union all
    select 100,20
    union all
    select 4,8
    union all
    select 15,5

    create table t2
    (c1 int,
    c2 int
    )

    insert into t2
    select 1001,2
    union all
    select 82,100
    union all
    select 83,100
    union all
    select 84,100
    union all
    select 101,4
    union all
    select 102,4
    union all
    select 2000,15
    union all
    select 2001,15

    Expected output should be as below:
    First off, please execuse me if there is any issue with my language.

    I have two tables, "t1" and "t2".
    Based on the "cnt" column from t1 table, "key" value from t1 table we should repeat the Key value from t1 in table t2 based on the count "cnt" column. Basically, the "key" value has to be repeated "cnt" number of times and should be inserted into t2.
    Also, we need to take the "c1" column into account, while repeating the rows. if we need to check the rows in t2 and if the "key" is not getting repeated "cnt" number of times in table t2, then we need to generate the extra rows to match the count.
    for newly generated rows, we need make use of c1 column values repeatedly starting from 1st value for that particular "key" value/

    For now, i am not looking of insertion, but I am looking for a select statement which returns rows as follows. Can anybody help?

    c1  c2
    1001,2
    1001,2
    1001,2
    1001,2
    1001,2

    82,100
    83,100
    84,100
    82,100
    83,100
    84,100
    82,100
    83,100
    84,100
    82,100
    83,100
    84,100
    82,100
    83,100
    84,100
    82,100
    83,100
    84,100
    82,100
    83,100

    101,4
    102,4
    101,4
    102,4
    101,4
    102,4
    101,4
    102,4

    2000,15
    2001,15
    2000,15
    2001,15
    2000,15

    Thanks,

    Sam

  • -- Works for cnt <= 10
    -- Expand the inline tally for cnt > 10
    ;WITH Process AS (
     SELECT t2.*, t1.cnt,
      rn = ROW_NUMBER() OVER(PARTITION BY c2 ORDER BY c1)
     FROM (SELECT c1, c2, GroupSize = COUNT(*) OVER(PARTITION BY c2) FROM #t2) t2
     LEFT JOIN #t1 t1
      ON t1. =  t2.c2
     CROSS APPLY (
      SELECT TOP(CAST(ROUND(t1.cnt/(GroupSize*1.0), 0) AS INT)) n
      FROM (VALUES (0),(0),(0),(0),(0),(0),(0),(0),(0),(0)) d (n)
     ) x
    )
    SELECT c1, c2
    FROM Process
    WHERE cnt >= rn
    ORDER BY c1

    “Write the query the simplest way. If through testing it becomes clear that the performance is inadequate, consider alternative query forms.” - Gail Shaw

    For fast, accurate and documented assistance in answering your questions, please read this article.
    Understanding and using APPLY, (I) and (II) Paul White
    Hidden RBAR: Triangular Joins / The "Numbers" or "Tally" Table: What it is and how it replaces a loop Jeff Moden

Viewing 2 posts - 1 through 1 (of 1 total)

You must be logged in to reply to this topic. Login to reply