I suspect the parsing for CSV and the parsing for SYLK are completely different pieces of code, and the appropriate one gets called once a determination of the

I suspect the parsing for CSV and the parsing for SYLK are completely different pieces of code, and the appropriate one gets called once a determination of the file format is made. This would be in an upper level where you wouldn't want to put format-specific behavior. The problem then is that the code determining the format doesn't dig far enough into the file to make sure it's valid before sending it to the parser.