Is there a way I can specify output file format in SQLCMD mode so that I can read it back using BULK INSERT
? I want to do something like this:
:CONNECT SERVER1
:OUT 'E:\test'
SELECT TOP 100 ID, NAME
FROM DB1.dbo.TABLE1
GO
:CONNECT SERVER2
:OUT 'E:\test2'
CREATE TABLE #TEMP(ID INT, NAME VARCHAR(100))
BULK
INSERT #TEMP
FROM 'E:\test'
WITH
(
FIELDTERMINATOR = ',',
ROWTERMINATOR = '\n'
)
GO
SELECT *
FROM DB2.dbo.TABLE2
WHERE ID IN (SELECT ID FROM #TEMP)
-- Other queries for #TEMP
GO
Here's a good example for using sqlcmd to generate a CSV file:
http://www.siusic.com/wphchen/sqlcmd-example-to-output-in-csv-format-350.html
sqlcmd
-S “ServerName” -U “UID” -P “PWD” -d “DBName”
-i “c:\input.txt”
-o “c:\output.txt”
-s “,”
-h -1
You can use SQLCmd /? to display all applicable
parameters.
...
Parameter -h -1 instructs the utility not to output result
set header. Parameter -s “,” instructs the utility to use comma as
column separator.
One way to get the data bulk inserted in a somewhat automated fashion is to use an intermediate table.
(using your original scenario)
:CONNECT SERVER1
:OUT 'E:\test'
SELECT TOP 100 '^'+cast(ID as varchar(16))+'^^'+NAME
FROM DB1.dbo.TABLE1
GO
:CONNECT SERVER2
:OUT 'E:\test2'
CREATE TABLE #TEMP1(BOTHCOLUMNS VARCHAR(116))
BULK
INSERT #TEMP1
FROM 'E:\test'
GO
CREATE TABLE #TEMP2(ID INT, NAME VARCHAR(100))
INSERT #TEMP2
SELECT
Substring(BOTHCOLUMNS, 2, Patindex('%^^%', BOTHCOLUMNS) - 2),
Substring(BOTHCOLUMNS, Patindex('%^^%', BOTHCOLUMNS) + 2,
Len(BOTHCOLUMNS) - Patindex('%^^%', BOTHCOLUMNS) - 1)
FROM #TEMP1
WHERE Patindex('%^^%', BOTHCOLUMNS) > 1
SELECT t2.*
FROM DB2.dbo.TABLE2 t2
JOIN #TEMP2 tmp2 on tmp2.ID=t2.ID
-- Other queries for #TEMP
GO