Identifying repeated fields in SQL query - sql

Question

I have an SQL query that returns a column like this:
foo
-----------
1200
1200
1201
1200
1200
1202
1202
1202
It has already been ordered in a specific way, and I would like to perform another query on this result set to ID the repeated data like this:
foo ID
---- ----
1200 1
1200 1
1201 2
1200 3
1200 3
1202 4
1202 4
1202 4
It's important that the second group of 1200 is identified as separate from the first. Every variation of OVER/PARTITION seems to want to lump both groups together. Is there a way to window the partition to only these repeated groups?
Edit:
This is for Microsoft SQL Server 2012

Answer 1

Not sure this will be the fastest results...
select main.num, main.id from
(select x.num,row_number()
over (order by (select 0)) as id
from (select distinct num from num) x) main
join
(select num, row_number() over(order by (select 0)) as ordering
from num) x2 on
x2.num=main.num
order by x2.ordering
Assuming the table "num" has a column "num" that contains your data, in the order-- of course num could be made into a view or a "with" for your original query.
Please see the following sqlfiddle

Answer 2

Here is one way to do this without a CURSOR
-- Create a temporay table
DECLARE #table TABLE
(
SeqID INT IDENTITY(1,1) NOT NULL PRIMARY KEY,
foo INT,
id int null
)
DECLARE #i INT
DECLARE #j INT
DECLARE #k INT
declare #tFoo INT
declare #oldFoo INT
SET #k = 0
set #oldFoo = 0
-- Insert data into the temporary table
INSERT INTO #table(foo)
SELECT 1200
INSERT INTO #table(foo)
SELECT 1200
INSERT INTO #table(foo)
SELECT 1201
INSERT INTO #table(foo)
SELECT 1200
INSERT INTO #table(foo)
SELECT 1200
INSERT INTO #table(foo)
SELECT 1202
INSERT INTO #table(foo)
SELECT 1202
INSERT INTO #table(foo)
SELECT 1202
-- Get the max and min SeqIDs to loop through
SELECT #i = MIN(SeqID) FROM #table
SELECT #j = MAX(SeqID) FROM #table
-- Loop through the temp table using the SeqID indentity column
WHILE (#i <= #j)
BEGIN
SELECT #tFoo = foo FROM #table WHERE SeqID = #i
if #oldFoo <> #tFoo
set #k = #k + 1
update #table set id = #k where SeqID = #i
SET #oldFoo = #tFoo
-- Increment the counter
SET #i = #i + 1
END
SELECT * from #table

Answer 3

You can do it without using cursors but it does not look nice (at least what I came up with). So 1) I assume you have PK column which orders your main values.
Then 2) I assume you have an ID column which you want to set.
create table tbl(foo int, pk int, id int);
insert into tbl(foo, pk) values (1100, 5);
insert into tbl(foo, pk) values (1200, 10);
insert into tbl(foo, pk) values (1200, 20);
insert into tbl(foo, pk) values (1201, 30);
insert into tbl(foo, pk) values (1200, 40);
insert into tbl(foo, pk) values (1200, 50);
insert into tbl(foo, pk) values (1202, 60);
insert into tbl(foo, pk) values (1202, 70);
insert into tbl(foo, pk) values (1202, 80);
insert into tbl(foo, pk) values (1202, 90);
SQL Fiddle here: http://sqlfiddle.com/#!6/fdaaa/2
update tbl
set
ID = 1
update t
set
t.ID = m.RN2
from
tbl t
join
(
select
y1.RN as RN1, y1.PK as PK1,
y2.RN as RN2, y2.PK as PK2
FROM
(
SELECT
ROW_NUMBER() OVER(ORDER BY x.pk1 ASC) AS rn,
x.pk1 AS pk
FROM
(
SELECT t1.pk AS pk1, t2.pk AS pk2
FROM
tbl t1
LEFT JOIN tbl t2 ON
(
(t1.pk < t2.pk AND t1.foo = t2.foo)
AND
(
NOT EXISTS
(
SELECT tMid.pk FROM
tbl tMid WHERE
tMid.pk < t2.pk
AND
tMid.pk > t1.pk
)
)
)
) x WHERE x.pk2 IS NULL
) y1
left join
(
SELECT
ROW_NUMBER() OVER(ORDER BY x.pk1 ASC) AS rn,
x.pk1 AS pk
FROM
(
SELECT t1.pk AS pk1, t2.pk AS pk2
FROM
tbl t1
LEFT JOIN tbl t2 ON
(
(t1.pk < t2.pk AND t1.foo = t2.foo)
AND
(
NOT EXISTS
(
SELECT tMid.pk FROM
tbl tMid WHERE
tMid.pk < t2.pk
AND
tMid.pk > t1.pk
)
)
)
) x WHERE x.pk2 IS NULL
) y2 on y1.RN = y2.RN - 1
) m on
(
(t.pk > m.pk1 and ((m.pk2 is not null ) and (t.pk <= m.pk2)))
-- or
-- (t.pk<=m.pk1)
)

Answer 4

This is my solution using a cursor and a temporary table to hold the results.
DECLARE #foo INT
DECLARE #previousfoo INT = -1
DECLARE #id INT = 0
DECLARE #getid CURSOR
DECLARE #resultstable TABLE
(
primaryId INT IDENTITY(1, 1) NOT NULL PRIMARY KEY,
foo INT,
id int null
)
SET #getid = CURSOR FOR
SELECT originaltable.foo
FROM originaltable
OPEN #getid
FETCH NEXT
FROM #getid INTO #foo
WHILE ##FETCH_STATUS = 0
BEGIN
IF (#foo <> #previousfoo)
BEGIN
SET #id = #id + 1
END
INSERT INTO #resultstable VALUES (#foo, #id)
SET #previousfoo = #foo
FETCH NEXT
FROM #getid INTO #foo
END
CLOSE #getid
DEALLOCATE #getid