Here is my data:
| ID | FIELD1 | FIELD2 | FIELD3 |
|-------------------------------|
| 1 | NULL | value1 | value2 |
|-------------------------------|
| 2 | NULL | value3 | NULL |
|-------------------------------|
| 3 | value4 | NULL | NULL |
|-------------------------------|
| 4 | value5 | value6 | value7 |
|-------------------------------|
| .. | ... | .... | .... |
Here is what I need to select:
| ID | ID2 | FIELDX |
|-------------------|
| 1 | 10 | value1 |
| 1 | 10 | value2 |
| 2 | 20 | value3 |
| 3 | 30 | value4 |
| 4 | 40 | value5 |
| 4 | 40 | value6 |
| 4 | 40 | value7 |
| .. | .. | .... |
The order of the data doesn't really matter. What matters is that each ID appears once for every associated FIELD1,2,3... value. Please note that there are many fields. I just chose to use these three as an example.
My attempt at the solution was this query:
SELECT x.ID, a.ID2, x.FIELDX
FROM (
SELECT t.ID, t.FIELD1
FROM SCHEMA1.TABLE1 t
UNION ALL
SELECT t.ID, t.FIELD2
FROM SCHEMA1.TABLE1 t
UNION ALL
SELECT t.ID, t.FIELD3
FROM SCHEMA1.TABLE1 t
) x
JOIN SCHEMA2.TABLE2 a ON x.ID = a.ID
WHERE x.FIELDX != NULL
WITH UR;
While this does do the job, I would rather not have to add a new inner select statement for each additional field. Moreover, I feel as though there is a more efficient way to do it.
Please advise.
DB2 doesn't have an explicit unpivot and your method is fine. A more efficient method is probably to do:
SELECT id, id2, fieldx
FROM (SELECT x.ID, a.ID2,
(case when col = 'field1' then field1
when col = 'field2' then field2
when col = 'field3' then field3
end) as FIELDX
FROM SCHEMA1.TABLE1 x join
SCHEMA2.TABLE2 a
on x.ID = a.ID cross join
(select 'field1' as col from sysibm.sysdummy1 union all
select 'field2' from sysibm.sysdummy1 union all
select 'field3' from sysibm.sysdummy1
) c
) x
WHERE x.FIELDX is not NULL;
This doesn't necessarily simplify the code. It does make it easier for DB2 to optimize the joins. And it only requires reading table1 once instead of one time for each column.
As a note: you should use fieldx is not null rather than fieldx != null.
Related
I've got one database with two columns (id and value). There are two types of values and each id has both of this values. How can I make a select to this database to have three columns in result (id, value1 and value2)
I've tried CASE and GROUP BY, but it shows only one result of each id
Example of a db:
| id | value |
| 0 | a |
| 0 | b |
| 1 | a |
| 1 | b |
Example of the result I am looking for is:
| id | value_a | value_b |
| 0 | a | b |
| 1 | a | b |
UPDATE:
As it was noted in comments, there is too simple data in the example.
The problem is more complicated
An example that would better describe it:
DB:
| id | value | value2 | value3 |
| 0 | a | a2 | a3 |
| 0 | b | b2 | b3 |
| 1 | a | c2 | c3 |
| 1 | b | d2 | d3 |
RESULT:
| id | value_a | value_b | value2_a | value2_b | value3_a | value3_b |
| 0 | a | b | a2 | b2 | a3 | b3 |
| 1 | a | b | c2 | d2 | c3 | d3 |
The output should be sorted by id an have all info from the both rows of each id.
If there are always two values per ID, you can try an aggregation using min() and max().
SELECT id,
min(value) value_a,
max(value) value_b
FROM elbat
GROUP BY id;
select t0.id,t0.Value as Value_A, t1.Value as Value_B
from test t0
inner join test t1 on t0.id = t1.id
where t0.Value = 'a' and t1.value = 'b';
I have used this method to turn "rows" into "columns". Depending on the number of unique values that exist in the table, you may or may not want to use this :)
SELECT id, SUM(CASE WHEN value = "a" then 1 else 0 END) value_a,
SUM(CASE WHEN value = "b" then 1 else 0 END) value_b,
SUM(CASE WHEN value = "c" then 1 else 0 END) value_c,
SUM(CASE WHEN value ="a2" then 1 else 0 END) value_a2,
.
.
.
FROM table
GROUP BY id;
Thanks all for the answers! This is the way how I did this:
WITH a_table AS
(
SELECT id, value, value2, value3 FROM table1 WHERE table1.value = 0
),
b_table AS
(
SELECT id, value, value2, value3 FROM table1 WHERE table1.value = 1
)
SELECT DISTINCT
a_table.id AS id,
a_table.value AS value_a,
a_table.value2 AS value2_a,
a_table.value3 AS value3_a,
b_table.value AS value_b,
b_table.value2 AS value2_b,
b_table.value3 AS value3_b
FROM a_table
JOIN b_table ON a_table.id = b_table.id
GROUP BY id;
I'm not quite sure how to properly phrase the question, but I am basically trying to develop an SQL query that SELECTs information from this table:
-------------------
| id | Val | Date |
|----|-----|------|
| 1 | A | 10/9 |
| 1 | B | 3/14 |
| 2 | A | 1/6 |
| 3 | A | 4/4 |
| 4 | B | 7/12 |
| 5 | A | 8/6 |
-------------------
And produces a table that looks like this:
------------------------------------------------
| id | Val_1 | Val_1_Date | Val_2 | Val_2_Date |
|----|-------|------------|-------|-------------
| 1 | A | 10/9 | B | 3/14 |
| 2 | A | 1/6 | | |
| 3 | A | 4/4 | | |
| 4 | | | B | 7/12 |
| 5 | A | 8/6 | | |
------------------------------------------------
I have already begun and developed the query to pull out the values in the Val fields into distinct columns:
SELECT * FROM
(
SELECT id, MAX(SWITCH( val='A', 'A')) as Val_1,
MAX(SWITCH( val='B', 'B')) as Val_2
FROM table1 GROUP BY id
)a
WHERE Val_1 IS NULL OR Val_2 IS NULL;
How would I expand on this to pull out their associated dates?
(I am using SWITCH() instead of CASE WHEN because I am using a driver similar to that of MS Access.)
Thanks!
I think following should work:
select id, SWITCH( val='A', 'A') as Val_1, SWITCH( val='A', Date) as Val_1_Date, SWITCH( val='B', 'B') as Val_2, SWITCH( val='B', Date) as Val_2_Date FROM table1 GROUP BY id
I do not prefer switches, so here is a query that does what you want without switches. This also answers your previous question.
Select distinct table1.ID, tableA.Val as Val_1, tableA.Date as Val_1_Date,
tableB.Val as Val_2, tableB.Date as Val_2_Date
FROM table1 left outer join
table1 as tableA on table1.id = tableA.id and tableA.Val = 'A' left outer join
table1 as tableB on table1.id = tableB.id and tableB.Val = 'B'
You can use ISNULL if that is preferred. This works because the first tables selects a distinct column of ID's, and the two joins get the A and B values. When creating selects using this method, make sure that you use tableA.Val = 'A' in the join conditions, and not in the where clause. Having tableA.Val = 'A' in the where clause will filter out all NULL's.
I'm trying to write a sql with a where clause, that checks if any element in a list is in another list. Is there a shorter way to accomplish this rather than check each member of the first list?
SELECT * from FOO
WHERE FOO.A IN ('2','3', '5', '7','11','13','17','19') OR
FOO.B IN ('2','3', '5', '7','11','13','17','19') OR
FOO.C IN ('2','3', '5', '7','11','13','17','19') OR
FOO.D IN ('2','3', '5', '7','11','13','17','19') OR
FOO.E IN ('2','3', '5', '7','11','13','17','19') OR
FOO.F IN ('2','3', '5', '7','11','13','17','19')
That is the simplified sql.
Was trying not to muddy waters too much, but since you ask:
Ultimately what I am trying to do here is, select rows from FOO, that has columns fulfilling various criteria. These criteria are stored in a second table (call it BAR), mainly db, name, type must match and flag must be 1. Was planning to build the IN list from BAR, comparing them with column names in INFORMATION_SCHEMA.COLUMNS containing FOO
FOO:
+--------+--------+---------+---------+--------+-------+
| DB | Name | Type | Col1 | Col2 | Col3 |
+--------+--------+---------+---------+--------+-------+
| 4 | AC1 | LO | 1 | 10 | 2 |
| 4 | AC1 | HI | 2 | 20 | 4 |
| 1 | DC2 | HI-HI | 11 | 5 | 2 |
| 1 | DC2 | HI | 22 | 10 | 4 |
| 1 | DC2 | LO | 33 | 15 | 6 |
+--------+--------+---------+---------+--------+-------+
BAR:
+--------+--------+---------+---------+--------+
| DB | Name | Type | Field | Flag |
+--------+--------+---------+---------+--------+
| 4 | AC1 | LO | Col1 | 1 |
| 4 | AC1 | HI | Col1 | 1 |
| 1 | DC2 | HI-HI | Col1 | 1 |
| 1 | DC2 | HI | Col1 | 1 |
| 1 | DC2 | LO | Col1 | 1 |
| 4 | AC1 | LO | Col2 | 0 |
| 4 | AC1 | HI | Col2 | 0 |
| 1 | DC2 | LO | Col2 | 0 |
| 1 | DC2 | HI-HI | Col2 | 0 |
| 1 | DC2 | HI | Col2 | 0 |
| 4 | AC1 | LO | Col3 | 0 |
| 4 | AC1 | HI | Col3 | 0 |
| 1 | DC2 | LO | Col3 | 0 |
| 1 | DC2 | HI-HI | Col3 | 0 |
| 1 | DC2 | HI | Col3 | 0 |
+--------+--------+---------+---------+--------+
On first examination, it would seem your schema is not appropriate for the type of query you're performing. It seems like you would want a FOOVAL table with a type and a value then you're query simply becomes:
CREATE TABLE FOOVAL
{
ID int, -- References FOO.ID
TYPE char, -- A, B, C, D, E, F
VAL int
}
SELECT * FROM FOO WHERE FOO.ID IN
(SELECT DISTINCT FOOVAL.ID WHERE FOOVAL.VAL IN ('2','3', '5', '7','11','13','17','19'))
Your method probably performs the best. Here is an alternative that only requires creating the list once. It uses a CTE to create a list of the values and then an exists clause to check whether any values match:
with vals as (
select '2' as p union all
select '3' union all
select '5' union all
select '7' union all
select '11' union all
select '13' union all
select '17' union all
select '19'
)
select *
from foo
where exists (select 1 from vals where vals.p in (foo.A, foo.B, foo.C, foo.D, foo.E, foo.F))
If you are using a database that doesn't support CTEs, you can just put the code in the where clause:
select 8
from foo
where exists (select 1
from (select '2' as p union all
select '3' union all
select '5' union all
select '7' union all
select '11' union all
select '13' union all
select '17' union all
select '19'
) t
where vals.p in (foo.A, foo.B, foo.C, foo.D, foo.E, foo.F)
)
If you are using Oracle, then you need to add from dual in the statements after the string constants. Otherwise, I think one or the other should work in any SQL database.
While it is not exactly clear what you want to do with the data, since you are using SQL Server my suggestion would be to use the UNPIVOT function to turn the col1, col2 and col3 columns into rows which will make it easier to filter the data:
select db, name, type, col, value
from foo
unpivot
(
value
for col in (Col1, Col2, Col3)
) unpiv;
See SQL Fiddle with Demo. This gives the data in the following format:
| DB | NAME | TYPE | COL | VALUE |
------------------------------------
| 4 | AC1 | LO | Col1 | 1 |
| 4 | AC1 | LO | Col2 | 10 |
| 4 | AC1 | LO | Col3 | 2 |
| 4 | AC1 | HI | Col1 | 2 |
Once the is in the row format, it should be significantly easier to apply any filters or even join to your BAR table.
suppose I have a table
-----------------------------------------------
| id | value1 | value2 | value3 |
-----------------------------------------------
| 102 | 10 | 1 | 3 |
-----------------------------------------------
| 102 | 2 | 11 | 0 |
-----------------------------------------------
| 102 | 0 | 9 | 13 |
-----------------------------------------------
| 102 | 3 | 5 | 7 |
-----------------------------------------------
and for each distinct id I want to return a row with max value in columns value1, value2 and value3, i.e.
-----------------------------------------------
| id | value1 | value2 | value3 |
-----------------------------------------------
| 102 | 10 | 11 | 13 |
-----------------------------------------------
(of course there are other ids than 102 in the table)
I managed to do it with "partition by" but the problem is that I have to use it in powerbuilder's datawindow, and as soon as I paste it there the whole IDE crashes and project gets corrupted.
I managed to create an sql that for each row does 3 inner joins with selects that return max of every column.
Is there any other easier way?
Thanks in advance for answering!
use GROUP BY and MAX()
SELECT id,
MAX(value1) val1,
MAX(value2) val2,
MAX(value3) val3
FROM tableName
GROUP BY ID
SQLFiddle Demo
SELECT id, MAX(value1) value1, MAX(value2) value2, MAX(value3) value3
FROM yourtable
GROUP BY id
I try to consolidate two rows of the same table whereas each row has a priority.
The value of interest is the value having priority 1 if it is not NULL; otherwise the value with priority 0.
An example data source could be:
| Id | GroupId | Priority | Col1 | Col2 | Col3 | ... | Coln |
-----------------------------------------------------------------
| 1 | 1 | 0 | NULL | 4711 | 3.41 | ... | f00 |
| 2 | 1 | 1 | NULL | NULL | 2.83 | ... | bar |
| 3 | 2 | 0 | NULL | 4711 | 3.41 | ... | f00 |
| 4 | 2 | 1 | 23 | NULL | 2.83 | ... | NULL |
and I want to have:
| GroupId | Col1 | Col2 | Col3 | ... | Coln |
-------------------------------------------------
| 1 | NULL | 4711 | 2.83 | ... | bar |
| 2 | 23 | 4711 | 2.83 | ... | f00 |
Is there a generic way in TSQL without the need to check each column explicitly?
SELECT
t1.GroupId,
ISNULL(t2.Col1, t1.Col1) as Col1,
ISNULL(t2.Col2, t1.Col2) as Col2,
ISNULL(t2.Col3, t1.Col3) as Col3,
...
ISNULL(t2.Coln, t1.Coln) as Coln
FROM mytable t1
JOIN mytable t2 ON t1.GroupId = t2.GroupId
WHERE
t1.Priority = 0 AND
t2.Priority = 1
Regards
I'll elaborate the ROW_NUMBER() solution that #KM suggested since IMO it's the best solution for this. (In CTE form for easier readability)
WITH cte AS (
SELECT
t1.GroupId,
t1.Col1,
t1.Col2,
ROW_NUMBER() OVER(PARTITION BY t1.GroupId ORDER BY ISNULL(GroupId ,-1) ) AS [row_id]
FROM
mytable t1
)
SELECT
*
FROM
cte
WHERE
row_id = 1
That will give you the row with the highest priority (according to your rules) for each GroupId in mytable.
ROW_NUMBER and RANK are two of my favorite TSQL tricks. http://msdn.microsoft.com/en-us/library/ms186734.aspx
edit: Another favorite of mine is PIVOT/UNPIVOT which you can use to transpose rows/columns which is another way of going about this type of problem. http://msdn.microsoft.com/en-us/library/ms177410.aspx
I think this would do what you are asking for without using isnull for every column
select
*
from
mytable t1
where
priority=(select max(priority) from mytable where groupid=t1.groupid group by groupid)