I agree. Most of the time, just accessing the file's extension should get you the right file type, and since the extension is how OnBase identifies a file, that's also the access the API gives you.
In cases where the file extension is insufficient to capture the file type (custom extensions, wrong extensions, etc.), I have had to build separate custom file identification classes.
You can take advantage of urlmon.dll, which contains classes and methods for looking at a file's MIME type. However, you will find that this has some significant disadvantages. For example, all MS Office XML file types (docx, xlsx, etc.) will be identified as pkZip files, since that's what they really are. So in cases like that, you would need to interrogate the XML files inside the zipped Office file to determine which type was in use.
You can open the file stream and read the initial bytes. For types that contain a signature (magic number), you should be able to identify them this way.
If you're uploading from a web page, consider taking advantage of the HTML5 file APIs. These do a great job identifying file types, and you can pass the type information along with the file stream.
Alternately, you could use a third-party utility or DLL that has already taken these steps and save yourself some programming time. The disadvantage here is that some of them have costs associated and some are slow.
The best course is to ensure that your files have correct extensions and then use them to determine the file type to use in OnBase.
