I would guess that they used some sort of stream writer. You are going to have to write an application for this. There is no t-sql that can do this that I am aware of. Basically you read the contents into a byte array to "reassemble" the gibberish into a file. You could probably convert this into a CLR if you need to use it repeatedly.

If you are doing this in .NET the basic syntax would be very close to this. I am assuming that dt is a DataTable that has the text column as a column in the DataTable named "PDF".

System.IO.File.WriteAllBytes(@"C:\YourFolder\YourFile.pdf", dt.Rows[0]["PDF"] as byte[]);