Data output can be divided roughly into formatting and buffering. Data input can be divided just as roughly into buffering and parsing; although, the parsing step gets easier if you can put a lexing step in front. (We'll talk more about lexing in the very next chapter!)
The classical iostreams API is built on top of <stdio.h>, which in turn is built on top of the POSIX file-descriptor API. You can't understand the higher levels without a good understanding of the levels beneath it. In particular, the mess of fopen mode strings and fstream constructor flags can be understood only by reference to lookup tables mapping them onto the actual underlying POSIX open flags.
The POSIX API is concerned only with moving chunks of data to and from file descriptors; it does not "buffer" data in the naive sense. The <stdio.h> API adds a layer of buffering...
