Find missing sequences by category - sql

Question

I have to identify missing records from the example below.
Category BatchNo TransactionNo
+++++++++++++++++++++++++++++++++
CAT1 1 1
CAT1 1 2
CAT1 2 3
CAT1 2 4
CAT1 2 5
CAT1 3 6
CAT1 3 7
CAT1 3 8
CAT1 5 12
CAT1 5 13
CAT1 5 14
CAT1 5 15
CAT1 7 18
CAT2 1 1
CAT2 1 2
CAT2 3 6
CAT2 3 7
CAT2 3 8
CAT2 3 9
CAT2 4 10
CAT2 4 11
CAT2 4 12
CAT2 6 14
I need a script that will identify missing records as below
Category BatchNo
+++++++++++++++++++
CAT1 4
CAT1 6
CAT2 2
CAT2 5
I do not need to know that CAT1 8 and CAT2 7 are not there as they potentially have not been inserted yet.

Answer 1

You can create temporary result set with all possible batch no up to max batch number for each category than select batch no which are not available.
create table TEMP(
Category varchar(10),
BatchNo int,
TransactionNo int
)
insert into TEMP values
('CAT1', 1, 1),
('CAT1', 1, 2),
('CAT1', 2, 3),
('CAT1', 2, 4),
('CAT1', 2, 5),
('CAT1', 3, 6),
('CAT1', 3, 7),
('CAT1', 3, 8),
('CAT1', 5, 9),
('CAT1', 7, 10),
('CAT2', 1, 1),
('CAT2', 1, 2),
('CAT2', 3, 3),
('CAT2', 4, 4),
('CAT2', 4, 5),
('CAT2', 4, 6),
('CAT2', 6, 7);
WITH BatchNo (BatchID,Category,MaxBatch) AS (
SELECT 1, Category, MAX(BatchNo) AS MaxBatch FROM TEMP GROUP BY Category
UNION ALL
SELECT BatchID + 1, Category, MaxBatch FROM BatchNo
WHERE BatchID < MaxBatch
)
SELECT
BatchNo.Category,
BatchNo.BatchID
FROM
BatchNo
WHERE
BatchID NOT IN (SELECT BatchNo FROM TEMP WHERE Category = BatchNo.Category)
ORDER BY
BatchNo.Category,
BatchNo.BatchID
DROP TABLE TEMP

Answer 2

This one uses a Tally Table. For reference: http://www.sqlservercentral.com/articles/T-SQL/62867/
SAMPLE DATA
create table MyTable(
Category varchar(10),
BatchNo int,
TransactionNo int
)
insert into MyTable values
('CAT1', 1, 1),
('CAT1', 1, 2),
('CAT1', 2, 3),
('CAT1', 2, 4),
('CAT1', 2, 5),
('CAT1', 3, 6),
('CAT1', 3, 7),
('CAT1', 3, 8),
('CAT1', 5, 12),
('CAT1', 5, 13),
('CAT1', 5, 14),
('CAT1', 5, 15),
('CAT1', 7, 18),
('CAT2', 1, 1),
('CAT2', 1, 2),
('CAT2', 3, 6),
('CAT2', 3, 7),
('CAT2', 3, 8),
('CAT2', 3, 9),
('CAT2', 4, 10),
('CAT2', 4, 11),
('CAT2', 4, 12),
('CAT2', 6, 14);
SOLUTION
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
tally(n) as(
select
top (select top 1 BatchNo from MyTable order by BatchNo desc)
row_number() over(order by (select null))
from e4
)
select
c.Category,
t.n
from tally t
cross join(
select
Category,
max(BatchNo) as MaxBatchNo
from MyTable
group by Category
)c
left join MyTable m
on m.BatchNo = t.n
and m.Category = c.Category
where
m.Category is null
and t.n < c.MaxBatchNo
order by
c.Category,
t.n

Answer 3

It is better to create a projection table and use standard left join to find gaps:
declare #Sequencer table (
Id int primary key
);
insert into #Sequencer (Id)
select top (1000) row_number() over(order by (select null)) from master.dbo.spt_values;
select *
from #Sequencer s
inner join (
select Category, max(BatchNo) as [Size] from dbo.Table group by Category
) cat on cat.Size > s.Id
left join (
select distinct Category, BatchNo from dbo.Table
) t on t.Category = cat.Category and t.BatchNo = s.Id
where t.BatchNo is null;
Of course, in real life you might need more than 1000 rows, so adjust it accordingly.

Answer 4

WITH Numbers AS (
SELECT MAX(BatchNo) AS Number
FROM #MyTable
UNION ALL
SELECT Number - 1
FROM Numbers
WHERE Number > 1
)
,CategorySizes AS (
SELECT Category
,MIN(BatchNo) AS StartBatch
,MAX(BatchNo) AS EndBatch
FROM #MyTable
GROUP BY Category
)
,PossibleBatches AS (
SELECT Category
,Numbers.Number AS BatchNo
FROM CategorySizes
CROSS JOIN Numbers
WHERE Numbers.Number BETWEEN CategorySizes.StartBatch AND CategorySizes.EndBatch
)
,MissingBatches AS (
SELECT PossibleBatches.Category
,PossibleBatches.BatchNo
FROM PossibleBatches
LEFT JOIN #MyTable
ON #MyTable.Category = PossibleBatches.Category
AND #MyTable.BatchNo = PossibleBatches.BatchNo
WHERE #MyTable.BatchNo IS NULL
)
SELECT *
FROM MissingBatches

Answer 5

without use cycle or fetch you can use this one: (#Category is my eqvivalent of your table name). (Performance is perfect)
DECLARE #t TABLE (RN INT IDENTITY,Category VARCHAR(255), BatchNo INT)
INSERT INTO #t
SELECT DISTINCT Category, BatchNo
FROM #Category
SELECT a.Category,a.BatchNo+1 AS BatchNo
FROM #t a
CROSS APPLY (SELECT * FROM #t b
WHERE a.RN+1 = b.RN AND
a.Category = b.Category AND
a.BatchNo+1 != b.BatchNo) x

Answer 6

create table #cat(
Category varchar(10),
BatchNo int,
TransactionNo int
)
insert into #cat values
('CAT1', 1, 1),
('CAT1', 1, 2),
('CAT1', 2, 3),
('CAT1', 2, 4),
('CAT1', 2, 5),
('CAT1', 3, 6),
('CAT1', 3, 7),
('CAT1', 3, 8),
('CAT1', 5, 9),
('CAT1', 7, 10),
('CAT2', 1, 1),
('CAT2', 1, 2),
('CAT2', 3, 3),
('CAT2', 4, 4),
('CAT2', 4, 5),
('CAT2', 4, 6),
('CAT2', 6, 7);
SELECT DISTINCT C.Category, C.BatchNo + 1
FROM #cat c
OUTER APPLY
(
SELECT *
FROM #cat c1
WHERE C1.BatchNo = C.BatchNo + 1 AND C1.Category = C.Category
) C2
WHERE C2.BatchNo IS NULL
AND
C.BatchNo <> (SELECT MAX(BatchNo) FROM #cat C3 WHERE c3.Category = c.Category)