(Documentation work in progress)
The main routine is fdf_read, which opens the main file and starts
adding material to the fdf object. If an %include line is found,
the routine calls itself again with the new file.
Support for the Label ... Label < other_file idiom complicates the
code quite a bit, since an alternative hierarchy of possibly nested
file reads must be instantiated. This is done with the fdf_readlabel
routine, which is also recursive and may call fdf_read again.
Low-level support for all the parsing needed is provided by the
fdf_parse module, which is useful in itself and could potentially be
made into an independent library. The basic idiom is to digest an
input line into tokens, tagging them as integers, reals, names
(strings), etc. There are query functions for number of tokens of each
kind, plus extractors for them. Also, a quite powerful match
routine can be used to identify at once the structure of a line.