SqlBulkCopy from Spreadsheet

Question

I'm using SqlBulkCopy to import records from a spreadsheet into a SQL Server database.

The bulk copy operation works fine under most conditions. However it's doing something I can't get my head around.

I have a column defined in my column mappings:

bCopy.ColumnMappings.Add("Amount", "Amount");

In the spreadsheet this is a varchar, as it is in my work table (I import to a work table to do some validation before I move the data to my live table).

While I was testing, I went into a record and changed the value from a number to the word FOO. Then I ran my import.

My validation routine correctly identifies a problem with the column - but it's because the column is null, not because the value is not a number. When I checked the database, I found this field was imported with a null value. I thought the value should have been FOO.

This strikes me as kind of weird.

If it makes any difference, I'm using OLEDB to access the spreadsheet like a database table and passing the OLEDBDataReader in as the source in my BulkCopy object.

Is the OLEDB driver smart enough to realize this isn't a valid value in that column? I thought everything would be string/varchar by default. (We don't make any modifications to the column datatype in the spreadsheet)

EDIT

I've checked the dataReader as suggested - and strange enough, when I change the Amount field to have a string value, the value in the datareader is null. This is the first time I've seen this kind of behavior in an Excel datareader. Other string fields work just fine.

Answer 1

I fixed the issue by adding IMEX=1 to my OLEDB Connection string. Now the value is shown in the datareader. (I guess I need to do this to support mixed value columns)