I need a function that for any elisp mode buffer receives a position as
argument and returns non-nil if inside a defun (i.e. when there is a
top-level expression covering POS). If this is the case, the return
value is a cons (BEG . END) where BEG and END are the beginning and end
positions of this toplevel sexp.
This seems surprisingly hard, I can only think of solutions that are
really complicated. The most complicated part is actually to find out
whether the value is non-nil or nil. Not every toplevel expression
needs to be a list! If POS is not inside a level of parens, I think I
could use the second element ("character address of start of last
complete sexp terminated") of `syntax-ppss' but that's still a bit of
way from a result. There are really strange cases like the positions of
space characters in symbol\ with\ spaces.
I want to know if anyone knows some elegant solution I fail to see.
> I need a function that for any elisp mode
> buffer receives a position as argument and
> returns non-nil if inside a defun (i.e.
> when there is a top-level expression covering
> POS). If this is the case, the return value
> is a cons (BEG . END) where BEG and END are
> the beginning and end positions of this
> toplevel sexp.
If it only would be so simple. But that doesn't work for non-list
top-level expressions - try with point over a number, for example - your
function returns nil. That is because AFAIK `beginning-of-defun' and
`end-of-defun' only work for listp toplevel expressions.
Hmm, but I guess when paren depth at POS is 0 (syntax-ppss can tell you
that), `scan-sexps' could be used in the way you use
`beginning-of-defun' and `end-of-defun' above to find whether we are
between an expression beginning and the according expression end.