"int 2or4bytes`depending`on`system"

In theory, an int should have the "natural" word size of the underlying architecture:
on an 8-bit machine, an int would be 8 bits;
on a 16-bit machine, an int would be 16 bits;
on a 32-bit machine, an int would be 32 bits;
etc.

However, it is common (usual?) these days to assume that an int will always be at least 16 bits - thus Keil C51, for example, has 16-bit ints even on the 8-bit 8051.

Note that the ANSI/ISO standards say nothing about the actual sizes of the types; just that an int must be no smaller than a char, a long must be no smaller than an int, etc, etc

"enum 2or4bytes`depending [on the system]"

Most compilers - especially for embedded targets - have at least the option to put "small" enums into a single byte.
Keil C51 certainly does, also Borland C++ Builder and (I think) MSVC.

Note that all these things are totally implementation-defined. Therefore, the only safe place to obtain these details is from the Manual for your specific compiler.
You will also need the Manual for other data-representation details, such as byte ordering, packing/padding, floating-point format, etc, etc...