mirror of
https://github.com/c-cube/iter.git
synced 2025-12-05 19:00:31 -05:00
API change for Sequence.persistent: no need to
provide a block size, the unrolled list manages it (doubling block size); new implementation for persistent is more efficient.
This commit is contained in:
parent
a3ff8ee854
commit
a5385ce649
3 changed files with 75 additions and 87 deletions
153
sequence.ml
153
sequence.ml
|
|
@ -144,97 +144,94 @@ let intersperse elem seq =
|
|||
|
||||
(** Mutable unrolled list to serve as intermediate storage *)
|
||||
module MList = struct
|
||||
type 'a t = {
|
||||
content : 'a array; (* elements of the node *)
|
||||
mutable len : int; (* number of elements in content *)
|
||||
mutable tl : 'a t; (* tail *)
|
||||
} (** A list that contains some elements, and may point to another list *)
|
||||
type 'a node =
|
||||
| Nil
|
||||
| Cons of 'a array * int ref * 'a node ref
|
||||
|
||||
let _empty () : 'a t = Obj.magic 0
|
||||
(** Empty list, for the tl field *)
|
||||
let of_seq seq =
|
||||
let start = ref Nil in
|
||||
let chunk_size = ref 8 in
|
||||
(* fill the list. prev: tail-reference from previous node *)
|
||||
let prev, cur = ref start, ref Nil in
|
||||
seq
|
||||
(fun x -> match !cur with
|
||||
| Nil ->
|
||||
let n = !chunk_size in
|
||||
if n < 4096 then chunk_size := 2 * !chunk_size;
|
||||
cur := Cons (Array.make n x, ref 1, ref Nil)
|
||||
| Cons (a,n,next) ->
|
||||
assert (!n < Array.length a);
|
||||
a.(!n) <- x;
|
||||
incr n;
|
||||
if !n = Array.length a then begin
|
||||
!prev := !cur;
|
||||
prev := next;
|
||||
cur := Nil
|
||||
end
|
||||
);
|
||||
!prev := !cur;
|
||||
!start
|
||||
|
||||
let make n =
|
||||
assert (n > 0);
|
||||
{ content = Array.make n (Obj.magic 0);
|
||||
len = 0;
|
||||
tl = _empty ();
|
||||
}
|
||||
let is_empty = function
|
||||
| Nil -> true
|
||||
| Cons _ -> false
|
||||
|
||||
let rec is_empty l =
|
||||
l.len = 0 && (l.tl == _empty () || is_empty l.tl)
|
||||
|
||||
let rec iter f l =
|
||||
for i = 0 to l.len - 1 do f l.content.(i); done;
|
||||
if l.tl != _empty () then iter f l.tl
|
||||
let rec iter f l = match l with
|
||||
| Nil -> ()
|
||||
| Cons (a, n, tl) ->
|
||||
for i=0 to !n - 1 do f a.(i) done;
|
||||
iter f !tl
|
||||
|
||||
let iteri f l =
|
||||
let rec iteri i f l =
|
||||
for j = 0 to l.len - 1 do f (i+j) l.content.(j); done;
|
||||
if l.tl != _empty () then iteri (i+l.len) f l.tl
|
||||
let rec iteri i f l = match l with
|
||||
| Nil -> ()
|
||||
| Cons (a, n, tl) ->
|
||||
for j=0 to !n - 1 do f (i+j) a.(j) done;
|
||||
iteri (i+ !n) f !tl
|
||||
in iteri 0 f l
|
||||
|
||||
let rec iter_rev f l =
|
||||
(if l.tl != _empty () then iter_rev f l.tl);
|
||||
for i = l.len - 1 downto 0 do f l.content.(i); done
|
||||
let rec iter_rev f l = match l with
|
||||
| Nil -> ()
|
||||
| Cons (a, n, tl) ->
|
||||
iter_rev f !tl;
|
||||
for i = !n-1 downto 0 do f a.(i) done
|
||||
|
||||
let length l =
|
||||
let rec len acc l =
|
||||
if l.tl == _empty () then acc+l.len else len (acc+l.len) l.tl
|
||||
let rec len acc l = match l with
|
||||
| Nil -> acc
|
||||
| Cons (_, n, tl) -> len (acc+ !n) !tl
|
||||
in len 0 l
|
||||
|
||||
(** Get element by index *)
|
||||
let rec get l i =
|
||||
if i < l.len then l.content.(i)
|
||||
else if i >= l.len && l.tl == _empty () then raise (Invalid_argument "MList.get")
|
||||
else get l.tl (i - l.len)
|
||||
let rec get l i = match l with
|
||||
| Nil -> raise (Invalid_argument "MList.get")
|
||||
| Cons (a, n, _) when i < !n -> a.(i)
|
||||
| Cons (_, n, tl) -> get !tl (i- !n)
|
||||
|
||||
(** Push [x] at the end of the list. It returns the block in which the
|
||||
element is inserted. *)
|
||||
let rec push x l =
|
||||
if l.len = Array.length l.content
|
||||
then begin (* insert in the next block *)
|
||||
(if l.tl == _empty () then
|
||||
let n = Array.length l.content in
|
||||
l.tl <- make (n + n lsr 1));
|
||||
push x l.tl
|
||||
end else begin (* insert in l *)
|
||||
l.content.(l.len) <- x;
|
||||
l.len <- l.len + 1;
|
||||
l
|
||||
end
|
||||
let to_seq l k = iter k l
|
||||
|
||||
(** Reverse list (in place), and returns the new head *)
|
||||
let rev l =
|
||||
let rec rev prev l =
|
||||
(* reverse array *)
|
||||
for i = 0 to (l.len-1) / 2 do
|
||||
let x = l.content.(i) in
|
||||
l.content.(i) <- l.content.(l.len - i - 1);
|
||||
l.content.(l.len - i - 1) <- x;
|
||||
done;
|
||||
(* reverse next block *)
|
||||
let l' = l.tl in
|
||||
l.tl <- prev;
|
||||
if l' == _empty () then l else rev l l'
|
||||
let to_stream l =
|
||||
let cur = ref l in
|
||||
let i = ref 0 in (* offset in cons *)
|
||||
let rec get_next _ = match !cur with
|
||||
| Nil -> None
|
||||
| Cons (_, n, tl) when !i = !n ->
|
||||
cur := !tl;
|
||||
i := 0;
|
||||
get_next 42 (* any value would do *)
|
||||
| Cons (a, n, _) ->
|
||||
let x = a.(!i) in
|
||||
incr i;
|
||||
Some x
|
||||
in
|
||||
rev (_empty ()) l
|
||||
|
||||
(** Build a MList of elements of the Seq. The optional argument indicates
|
||||
the size of the blocks *)
|
||||
let of_seq ?(size=8) seq =
|
||||
(* read sequence into a MList.t *)
|
||||
let start = make size in
|
||||
let l = ref start in
|
||||
seq (fun x -> l := push x !l);
|
||||
start
|
||||
Stream.from get_next
|
||||
end
|
||||
|
||||
(** Iterate on the sequence, storing elements in a data structure.
|
||||
The resulting sequence can be iterated on as many times as needed. *)
|
||||
let persistent ?(blocksize=64) seq =
|
||||
if blocksize < 2 then failwith "Sequence.persistent: blocksize too small";
|
||||
let l = MList.of_seq ~size:blocksize seq in
|
||||
from_iter (fun k -> MList.iter k l)
|
||||
let persistent seq =
|
||||
let l = MList.of_seq seq in
|
||||
MList.to_seq l
|
||||
|
||||
(** Sort the sequence. Eager, O(n) ram and O(n ln(n)) time. *)
|
||||
let sort ?(cmp=Pervasives.compare) seq =
|
||||
|
|
@ -337,7 +334,7 @@ let take n seq =
|
|||
incr count;
|
||||
k x;
|
||||
if !count = n then raise ExitSequence)
|
||||
with ExitSequence -> ()
|
||||
with ExitSequence -> ()
|
||||
|
||||
(** Drop the [n] first elements of the sequence *)
|
||||
let drop n seq =
|
||||
|
|
@ -470,14 +467,8 @@ let of_stream s =
|
|||
|
||||
(** Convert to a stream. The sequence is made persistent. *)
|
||||
let to_stream seq =
|
||||
let l = ref (MList.of_seq seq) in
|
||||
let i = ref 0 in
|
||||
let rec get_next () =
|
||||
if !l == MList._empty () then None
|
||||
else if (!l).MList.len = !i then (l := (!l).MList.tl; i := 0; get_next ())
|
||||
else let x = (!l).MList.content.(!i) in (incr i; Some x)
|
||||
in
|
||||
Stream.from (fun _ -> get_next ())
|
||||
let l = MList.of_seq seq in
|
||||
MList.to_stream l
|
||||
|
||||
(** Push elements of the sequence on the stack *)
|
||||
let to_stack s seq = iter (fun x -> Stack.push x s) seq
|
||||
|
|
|
|||
|
|
@ -151,15 +151,11 @@ val fmap : ('a -> 'b option) -> 'a t -> 'b t
|
|||
val intersperse : 'a -> 'a t -> 'a t
|
||||
(** Insert the single element between every element of the sequence *)
|
||||
|
||||
val persistent : ?blocksize:int -> 'a t -> 'a t
|
||||
val persistent : 'a t -> 'a t
|
||||
(** Iterate on the sequence, storing elements in a data structure.
|
||||
The resulting sequence can be iterated on as many times as needed.
|
||||
{b Note}: calling persistent on an already persistent sequence
|
||||
will still make a new copy of the sequence!
|
||||
|
||||
@param blocksize the size of chunks in the unrolled list
|
||||
used to store elements. Use bigger values for bigger sequences.
|
||||
Default: 64 *)
|
||||
will still make a new copy of the sequence! *)
|
||||
|
||||
val sort : ?cmp:('a -> 'a -> int) -> 'a t -> 'a t
|
||||
(** Sort the sequence. Eager, O(n) ram and O(n ln(n)) time.
|
||||
|
|
|
|||
|
|
@ -93,6 +93,7 @@ let test_persistent () =
|
|||
OUnit.assert_equal ~printer [] (seq |> S.to_list);
|
||||
OUnit.assert_equal ~printer [0;1;2;3;4] (seq' |> S.to_list);
|
||||
OUnit.assert_equal ~printer [0;1;2;3;4] (seq' |> S.to_list);
|
||||
OUnit.assert_equal ~printer [0;1;2;3;4] (seq' |> S.to_stream |> S.of_stream |> S.to_list);
|
||||
()
|
||||
|
||||
let test_big_persistent () =
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue