Hi, Tom. You may put all of this in the category of "well, gee, I already knew that much" :) but here goes. Seems to me that you'd need to use LLFF to read a line and parse for commas (ALINES should work well for that). Then for each element, you could do additional parsing to come up with at least a default data type (e.g. - roughly, " indicates C or M, . indicates N, 2x / or - indicates date, unless also : which indicates datetime, none of the above but all digits indicates I). Of course, you really need to read the entire file, so that you can then determine the max-allowed length of each column.

Seems to me that you may be able to use the code for the VFP import wizard as a shortcut. You should find a file named xsource.zip in your HOME() + 'TOOLS\XSOURCE' folder. In the ZIP file, look for files in the path "VFPSource\Wizards\wzimport."

Kelly

>I need to write a generic program that can examine the structure of a comma seperated file. Once the structure of the has been determined, I will do an APPEND FROM into the cursor that will be created from the first step.
>
>The question is what is the best technique to determine the structure of any CSV file and create a cursor based on that structure for any file that is thrown at the generic import program?