asked on

Load CSV file with Commas within its columns into SQL server using SSIS

Hello,

I have a CSV file in the following format

Name, Tournament , Code , Cash
"a", "bit $1,000, express", "f" , $1000
"n", "Tour",hold $10", "y" , $10

so here "bit $1,000express" and "Tour",hold $10" are actually in the Tournament coloumn as you can see from my illustration. This is exactly how most of the data in my file is represented.

So when I use SSIS to try to load this CSV file into my SQL 2008 database, the middle column is truncated at the commas that occur with the text of the column into separate columns .

so here "000", and "express" will be moved to next two columns and "hold $10" will be in the next columns after its rightful column.

Does anyone know how to solve this kind of problem in SSIS .
By the way its a large CSV file with many columns .

Thanks.

ASKER CERTIFIED SOLUTION

Paul MacDonald

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

marvo2010

ASKER

Yeah Thats the main issue really. There are commas within the columns , and the column delimiter is also comma, so where ever the SSIS see a comma, it basically separates the column into a separate column. I am wondering if anyone has econutered this kind of problem and how they were able to solve it.

marvo2010

ASKER

If I export it as a Tab separated CSV file , will it not still insert Tabs where ever it sees commas just like SSIS is separating data into columns based on the columns.

Alex_W

If you can replace the , separator with another character, say ; leaving the internal commas in each column intact when perhaps a small .NET app could perform the necessary formatting for you. You could:

1) read the rows into arrays specifying the separation character (;)
2) find and replace the remaining commas (inside each column) to a more friendly character or even remove them
3) then write out the line as comma separated so its ready to import as CSV while having replaced the internal commas so they dont cause these issues.

If this sounds like a solution to you I can post some test code in VB.NET...

marvo2010

ASKER

yes this certainly sounds like a solution . Lets test the VB code using a Script task in SSIS.

Thanks

Paul MacDonald

Yes, that was what I was driving at. If you can replace the actual column delimiters at the export source, that would be ideal. I was illustrating how you could change the commas to tabs if you don't have the original source data (can't export it again). And no, when you import the data into SSIS, you should be allowed to indicate the data is tab-delimited rather than comma-delimited and SQL Server should consume it just fine.

Alex_W

If you can replace the column delimiters to a tab then it sounds as though paulmacd's solution would be easiest because you'll have no middlware and be able to import directly to SQL Server. If this for some reason isnt possible and your after some VB code to do some re-formatting, then please post and I'll supply some.

marvo2010

ASKER

yeah we have the files already and needs to load them rightly now. so if you can supply the vb code and how to apply it , that will be great
Thanks

Alex_W

This is made up from some bits and peices and untested for your purpose but should get you started with the functionality, maybe best in a console app to quickly run the routine against the file:

Dim m_source As String()   ' create an array to read the rows
m_source = System.IO.File.ReadAllLines("C:\Input.csv")   ' array of source file lines
Dim Output As New System.IO.StreamWriter("C:\Output.csv")   ' file to output

For x = 1 To m_source.Length - 1   ' iterate down source file rows (starting from 1 - row after headers)
  Dim DataToWrite() As String = m_source(x).Split(";"c)   ' data array or current row split with a ;
  DataToWrite = DataToWrite.Replace(",", "_")   ' replace the internal comma with something
  DataToWrite = DataToWrite.Replace(";", ",")   ' now swap the ; separator back to a , for CSV compatibility
  Output.WriteLine(DataToWrite)   ' write the output back out
Next

Output.Close()

Open in new window

seenall

The way I have preferred getting around this issue is by loading the entire record into a single column staging table and then performing the delimited extraction using T-SQL CHARINDEX and SUBSTRING functions to reach the target table

This will at least avoid maintaining an external .net module and keep the process integrated into your DTSX package.

Just a thought

marvo2010

ASKER

Hello Seenall,

Pls do have a step by step solution of how to implmement your suggested solution without an external .Net module like u said using T-SQL Charindex and SUBSTRING. I will appreciate to have a look.

Thanks

marvo2010

ASKER

Thanks Guys,

I have been able to potentially solve this problem.Converting the files to Tab delimited worked perfectly. I ve got a little solution that can actually convert the files to Tab delimited from folder to folder and then I need to Iterate through the files using SSIS and load to my staging table .
Microsoft have not got a solution for this from what I have learnt during this process but the "NEVER SAY NEVER ATTITUDE" always gets the job done no matter what. I love never to quit.

Thanks guys

Guy Hengel [angelIII / a3]

This question has been classified as abandoned and is closed as part of the Cleanup Program. See the recommendation for more details.