My client drops for me many files like this on a shared drive M: daily
I would like you to advice on how to write a SQL code (that can include DTS if possible) that will take this files, merge them into one (since they have same columns) and send them to another client as one file. But it must only take the files of the same date and must not resend files i have send already. I need this to be an automated process since the files are dumped into M:drive midnite and I need this code so that I can schedule it as a job and run around 4h00am.
I would do all (and have...now where is that code) in a stored procedure.
1. us xp_cmdshell to interogate the directory and store all of the contents in a temp table
2. I would then parse it out and look for the files I want
3. I would then load these file to a temp table and audit them 1 at a time using bcp
4. When everything looked good, I would insert from the temp table to the final destination
5. I would then use xp_cmdshell again to vreate an datetime stamped sub archive folder, and would move EVERYTHING to that archive (i'm a neat freak...well...)
6. Thoughout the process I would echo out messages as to what;s going on to a file, and on the way out I would bcp that file to a log table.
I would also go the route that Brett suggested, there might be some "features" that you could skip, but on the other hand you get quite flexible code if you follow his suggestions!
The only thing that I would do differently is that I would create one folder for the archived files and store them there (with dates as filenames), but this is just a matter of personal preference.
Why would you like to include DTS to this?
...I know that DTS is good for a lot of things, but personally I find scritping much more portable (this might have something to do with that I haven't used DTS that much, so my opinion could change in the future).
I would do like Brett Kaiser and query the directory where these files are kept. And store all the contents in a #temp table. What I would do differently is that I would use a directory structure. I would for example keep the files in C:\CSV and once the file has been processed I would move the file to another directory say C:\CSV\sent by using a stored procedure.
In the database say TEST, I would create a table test_csv with the same fields as the .csv file.
I will then use a procedure like this one, I have written it specially for you. The input parameter will come from the #temp table well which means this procedure will be called from another procedure that will have loop for the filenames field and the procedure csv_test will be called from inside the loop.
-- SQL CODING BEGINS HERE
-- csv_test 'C:\CSV\Book2.csv'
create proc csv_test
declare @cmd as varchar(100)
declare @cmd_insert as varchar(300)
SET @filename = replace(@filename, '''', '''''')
set @cmd_insert = 'BULK INSERT TEST.dbo.test_csv FROM ''' + @filename + ''' WITH ( FIELDTERMINATOR = '','', ROWTERMINATOR = ''\n'' )'