I'm trying to get a distinct row using SQL from set of records that have matching key/id value, but NULLs in different columns. Hard to explain so please see screenshot. Any ideas?
create temporary table my_table (
id varchar(30), segmentdate1 date, converted1 varchar(10), segmentdate2 date, converted2 varchar(10)
);
insert into my_table (
id, segmentdate1, converted1, segmentdate2, converted2
)
values
('Michael','9/15/2020','No',NULL,NULL),
('Michael',NULL,NULL,'7/1/2019','Yes')
;
You seem to want aggregation:
select id, max(segmentdate1) as segmentdate1, max(converted1) as converted1,
max(segmentdate2) as segmentdate2, max(converted2) as converted2
from t
group by id;
Note: I made up names for the columns so they are unique.
This is probably a result set created from another query. That query probably has the wrong group by keys. You should probably fix that query.
declare #my_table table (
id varchar(30), segmentdate1 date, converted1 varchar(10), segmentdate2 date, converted2 varchar(10)
);
insert into #my_table (
id, segmentdate1, converted1, segmentdate2, converted2
)
values
('Michael','9/15/2020','No',NULL,NULL),
('Michael',NULL,NULL,'7/1/2019','Yes')
;
select id,max(isnull(segmentdate1,'1200-01-01')) segmentdate2
,max(isnull(converted1,'')) converted1, max(isnull(segmentdate2,'1200-01-01')) segmentdate2
,max(isnull(converted2,'')) converted2
from #my_table
group by id
Related
I have the following code and it give me an error when the table #ListaDeProducto has more than 1 row. Any idea?
insert into Solicitud_Plastico_Interna_Detalle(
IDSolicitud_Plastico_Interna
,IDTipo_Producto
,Cantidad_Solicitada
,Create_User
,Create_Date
,Contingencia
,Total
)
select
#IdSolicitud
,IDTipo_Producto
,Cantidad_Requerida
,#USUARIO
,getdate()
,Contingencia
,Total
from #ListaDeProducto
Table schema
CREATE TYPE [ListaProductoTableType2] AS TABLE
(
IDTipo_Producto int,
Tipo_Producto varchar(1000),
Cantidad_Requerida int,
Contingencia int ,
Total int,
IdSolicitud_batch varchar(100)
)
GO
I still will bet there is some trigger in the table.
So why you dont try create a new table to prove this query is ok with multiple rows
CREATE TABLE Solicitud_Plastico_Temporal AS (
select
#IdSolicitud as IDSolicitud_Plastico_Interna
,IDTipo_Producto
,Cantidad_Requerida
,#USUARIO as Create_User
,getdate() as Create_Date
,Contingencia
,Total
from #ListaDeProducto
)
I would like to take some data from a table from DB1 and insert some of that data to a table in DB2.
How would one proceed to do this?
This is what I've got so far:
CREATE VIEW old_study AS
SELECT *
FROM dblink('dbname=mydb', 'select name,begins,ends from study')
AS t1(name varchar(50), register_start date, register_end date);
/*old_study now contains the data I wanna transfer*/
INSERT INTO studies VALUES (nextval('studiesSequence'),name, '',3, 0, register_start, register_end)
SELECT name, register_start, register_end from old_study;
This is how my table in DB2 looks:
CREATE TABLE studies(
id int8 PRIMARY KEY NOT NULL,
name_string VARCHAR(255) NOT NULL,
description VARCHAR(255),
field int8 REFERENCES options_table(id) NOT NULL,
is_active INTEGER NOT NULL,
register_start DATE NOT NULL,
register_end DATE NOT NULL
);
You should include the column names in both the insert and select:
insert into vip_employees(name, age, occupation)
select name, age, occupation
from employees;
However, your data structure is suspect. Either you should use a flag in employees to identify the "VIP employees". Or you should have a primary key in employees and use this primary key in vip_employees to refer to employees. Copying over the data fields is rarely the right thing to do, especially for columns such as age which are going to change over time. Speaking of that, you normally derive age from the date of birth, rather than storing it directly in a table.
INSERT INTO studies
(
id
,name_string
,description
,field
,is_active
,register_start
,register_end
)
SELECT nextval('studiesSequence')
,NAME
,''
,3
,0
,register_start
,register_end
FROM dblink('dbname=mydb', 'select name,begins,ends from study')
AS t1(NAME VARCHAR(50), register_start DATE, register_end DATE);
You can directly insert values that retured by dblink()(that means no need to create a view)
Loop and cursor are weapons of last resort. Try to avoid them. You probably want INSERT INTO ... SELECT:
INSERT INTO x(x, y, z)
SELECT x, y, z
FROM t;
SqlFiddleDemo
EDIT:
INSERT INTO vip_employees(name, age, occupation) -- your column list may vary
SELECT name, age, occupation
FROM employees;
Your syntax is wrong. You cannot have both, a values clause for constant values and a select clause for a query in your INSERT statement.
You'd have to select constant values in your query:
insert into studies
(
id,
name_string,
description,
field,
is_active,
register_start,
register_end
)
select
studiesSequence.nextval,
name,
'Test',
null,
0,
register_start,
register_end
from old_study;
I have a view to maintain a componentid and componentname.
CREATE VIEW [dbo].[vDimNIS]
AS
SELECT
IndicatorID as TINYINT
IndicatorName as varchar(20)
FROM (
VALUES
(1,ant),
(2,bat),
(3,cat)
)AS Nis (IndicatorID , IndicatorName)
I would like to have an unique static IndicatorID for this table. What is the best to achieve this.
i assume you would like this view to eventually point to a table, e.g. DimNis, where all your entries have an indicatorid. the easiest way i know to achieve this would be to use an identitiy. this way each time you insert a new indicatorname into your table, a new indicatorid will be issued:
create table DimNis
( indicatorid tinyint identity(1,1) not null
, indicatorname varchar(20));
insert into DimNis
( indicatorname )
values
('ant');
insert into DimNis
( indicatorname )
values
('bat');
insert into DimNis
( indicatorname )
values
('cat');
select * from DimNis;
i hope this settles your issue.
If you add an identity column to your table (as suggested by Brett), the id will be autogenerated.
If you want to generate it on the fly, you can use ROW_NUMBER() that will give you a incremented number like identity, or you can use the hash function CHECKSUM that will compute an integer but may be with collisions.
The code with both Ids:
if object_id('vdimnis') is not null
drop view vDimNis
go
CREATE VIEW [dbo].[vDimNIS]
AS
SELECT
row_number() over (order by IndicatorName) as IndicatorId,
checksum(IndicatorName) as ChecksumId,
cast( IndicatorName as varchar(20) ) as IndicatorName
FROM ( VALUES
('ant'),
('bat'),
('cat')
) AS Nis ( IndicatorName)
go
select * from vDimNis
i have a table called table1 and it has following columns.
suppose there are records like localamount is 20,000, 30000,50000, 100000 then as per my condition i have to delete records from this table according to the group by set of site id, till id, transid,shift id where localamount exceeds 10,000... the rest of the records can be available?
my aim is to delete rows from this table where local amount is exceeds 10,0000 according to site id, till id, transid,shift id
SiteId varchar(10),
TillId tinyint,
ShiftId int,
TransId int,
TranDate datetime,
SettlementType varchar(5),
CreditCardNumber varchar(25),
ProductTypeCode varchar(10),
NewProductTypeCode varchar(10),
TransactionType int,
ForeignAmount money,
LocalAmount money,
ProductCode varchar(10)
Im not sure I understand what you are saying, but couldn't you do this without a group by?
delete from table1 where LocalAmount > 10,0000[sic] and SiteId = whatever and TillId = whatever...
obviously take the [sic] out...
Assuming you want to delete the whole group where the sum is > 10000
;with cte as
(
select sum(localamount) over
(partition by siteid, tillid, transid,shiftid) as l,
* from table1
)
delete from cte where l>10000
The following works, I'm just wondering if this is the correct approach to finding the latest value for each audit field.
USE tempdb
CREATE Table Tbl(
TblID Int,
AuditFieldID Int,
AuditValue Int,
AuditDate Date
)
GO
INSERT INTO Tbl(TblID,AuditFieldID,AuditValue,AuditDate) VALUES(1,10,101,'1/1/2001')
INSERT INTO Tbl(TblID,AuditFieldID,AuditValue,AuditDate) VALUES(2,10,102,'1/1/2002')
INSERT INTO Tbl(TblID,AuditFieldID,AuditValue,AuditDate) VALUES(3,20,201,'1/1/2001')
INSERT INTO Tbl(TblID,AuditFieldID,AuditValue,AuditDate) VALUES(4,20,202,'1/1/2009')
SELECT AuditFieldID,AuditValue,AuditDate
FROM Tbl A
WHERE TblID=
(SELECT TOP 1 TblID
FROM Tbl
WHERE AuditFieldID=A.AuditFieldID
ORDER BY AuditDate DESC
)
Aggregate/ranking to get key and latest date, join back to get value.
This assumes SQL Server 2005+
DECLARE #tbl Table (
TblID Int,
AuditFieldID Int,
AuditValue Int,
AuditDate Date
)
INSERT INTO #tbl(TblID,AuditFieldID,AuditValue,AuditDate) VALUES(1,10,101,'1/1/2001')
INSERT INTO #tbl(TblID,AuditFieldID,AuditValue,AuditDate) VALUES(2,10,102,'1/1/2002')
INSERT INTO #tbl(TblID,AuditFieldID,AuditValue,AuditDate) VALUES(3,20,201,'1/1/2001')
INSERT INTO #tbl(TblID,AuditFieldID,AuditValue,AuditDate) VALUES(4,20,202,'1/1/2009')
;WITH cLatest AS
(
SELECT
ROW_NUMBER() OVER (PARTITION BY AuditFieldID ORDER BY AuditDate DESC) AS Ranking,
AuditFieldID, AuditDate
FROM
#tbl
)
SELECT
A.AuditFieldID, A.AuditValue, A.AuditDate
FROM
#tbl A
JOIN
cLatest C ON A.AuditFieldID = C.AuditFieldID AND A.AuditDate = C.AuditDate
WHERE
C.Ranking = 1
Simpler:
SELECT top 1 AuditFieldID,AuditValue,AuditDate
FROM Tbl
order by AuditDate DES
There are various methods for doing this. Different methods perform differently. I encourage you to look at this blog which explains the various methods.
Including an Aggregated Column's Related Values
you don't need the where statement as you are already selecting from tbl A AND selecting on the same field.