Some time ago I discovered something really cool. It turns out Emacs has a package named range which can be used to operate on, well, ranges of integers. The truth is, it supports much more: arbitrary finite sets of integers.
For example, you can use the function range-compress-list to convert a (sorted) list of integers into a “compact” range notation:
(range-compress-list '(1 2 3 5 6 7 10 15 16)) ;; => ((1 . 3) (5 . 7) 10 (15 . 16))
Of course, you can also perform the opposite conversion:
(range-uncompress '((1 . 3) (5 . 7) 10 (15 . 16))) ;; => (1 2 3 5 6 7 10 15 16)
The cool thing is that there is a set of set-theoretic operations implemented there (pun intended
!), though one of them is named in a bit unexpected way:
(range-member-p 4 '((1 . 3) (5 . 7))) ;; => nil (range-concat '((1 . 3) (5 . 7)) '((2 . 6))) ;; => ((1 . 7)) (range-difference '((1 . 10)) '((4 . 6))) ;; => ((1 . 3) (7 . 10)) (range-intersection '((1 . 3) (5 . 7)) '((3 . 8))) ;; => (3 (5 . 7))
I have no idea if I will ever use this. The only use-case I can think of would be to say something like (range-uncompress '((1 . 10))) to get (1 2 3 4 5 6 7 8 9 10) – but this can be done without require’ing range, using (number-sequence 1 10). Still, range looks really cool! (If you are wondering, the only place in Emacs code where range is actually used seems to be Gnus, and I didn’t look how exactly it is used – Gnus is a huge beast, I wasn’t brave enough to dive into its code.)
I suppose the bottom line is that you can find real gems in Emacs sources if you look in the right place. Like I didn’t know that already…