The problem of discriminating between csv and SYLK files is unsolvable, as every file with only ASCII characters and no commas or quotes is a valid csv file (with one column) and there are valid SYLK files in that set.
In the general sense, yeah, but you would think it shouldn't be too hard to make a reasonable guess about which one it is and handle failures gracefully instead of assuming that it's SYLK based on a tiny amount of data and then blowing up when it turns out not to be.
