From 0d0a8f8764d2d746361212c59fb4ca5553aabdb5 Mon Sep 17 00:00:00 2001
From: Simon Cruanes <simon.cruanes.2007@m4x.org>
Date: Sun, 2 Aug 2015 20:54:28 +0200
Subject: [PATCH] add `CCHashSet` into `containers.data`, a mutable set

---
 README.md              |   1 +
 _oasis                 |   2 +-
 doc/intro.txt          |   1 +
 src/data/CCHashSet.ml  | 200 +++++++++++++++++++++++++++++++++++++++++
 src/data/CCHashSet.mli |  98 ++++++++++++++++++++
 5 files changed, 301 insertions(+), 1 deletion(-)
 create mode 100644 src/data/CCHashSet.ml
 create mode 100644 src/data/CCHashSet.mli

diff --git a/README.md b/README.md
index e9365848..d04e3b04 100644
--- a/README.md
+++ b/README.md
@@ -121,6 +121,7 @@ Documentation [here](http://cedeela.fr/~simon/software/containers).
 - `CCMultimap` and `CCMultiset`, functors defining persistent structures
 - `CCFQueue`, a purely functional double-ended queue structure
 - `CCBV`, mutable bitvectors
+- `CCHashSet`, mutable set
 - `CCPersistentHashtbl` and `CCPersistentArray`, a semi-persistent array and hashtable
   (similar to [persistent arrays](https://www.lri.fr/~filliatr/ftp/ocaml/ds/parray.ml.html))
 - `CCMixmap`, `CCMixtbl`, `CCMixset`, containers of universal types (heterogenous containers)
diff --git a/_oasis b/_oasis
index 0afbb4d2..93411ced 100644
--- a/_oasis
+++ b/_oasis
@@ -84,7 +84,7 @@ Library "containers_data"
   Modules:          CCMultiMap, CCMultiSet, CCTrie, CCFlatHashtbl, CCCache,
                     CCPersistentHashtbl, CCDeque, CCFQueue, CCBV, CCMixtbl,
                     CCMixmap, CCRingBuffer, CCIntMap, CCPersistentArray,
-                    CCMixset, CCHashconsedSet, CCGraph
+                    CCMixset, CCHashconsedSet, CCGraph, CCHashSet
   BuildDepends:     bytes
   FindlibParent:    containers
   FindlibName:      data
diff --git a/doc/intro.txt b/doc/intro.txt
index 0b692889..fd5258bd 100644
--- a/doc/intro.txt
+++ b/doc/intro.txt
@@ -67,6 +67,7 @@ CCBV
 CCCache
 CCFQueue
 CCFlatHashtbl
+CCHashSet
 CCIntMap
 CCMixmap
 CCMixset
diff --git a/src/data/CCHashSet.ml b/src/data/CCHashSet.ml
new file mode 100644
index 00000000..20a350d6
--- /dev/null
+++ b/src/data/CCHashSet.ml
@@ -0,0 +1,200 @@
+(* This file is free softwarem part of containers. See file "license" for more details. *)
+
+(** {1 Mutable Set} *)
+
+type 'a sequence = ('a -> unit) -> unit
+type 'a printer = Format.formatter -> 'a -> unit
+
+module type S = sig
+  type t
+  type elt
+
+  val create : int -> t
+  (** [create n] makes a new set with the given capacity [n] *)
+
+  val clear : t -> unit
+  (** [clear s] removes all elements from [s] *)
+
+  val copy : t -> t
+  (** Fresh copy *)
+
+  val copy_into : into:t -> t -> unit
+  (** [copy_into ~into s] copies all elements of [s] into [into] *)
+
+  val insert : t -> elt -> unit
+  (** [insert s x] adds [x] into [s] *)
+
+  val remove : t -> elt -> unit
+  (** Remove the element, if it were in there *)
+
+  val cardinal : t -> int
+  (** [cardinal s] returns the number of elements in [s] *)
+
+  val mem : t -> elt -> bool
+  (** [mem s x] returns [true] iff [x] is in [s] *)
+
+  val find_exn : t -> elt -> elt
+  (** [find s x] returns [y] if [x] and [y] are equal, and [mem s y].
+      @raise Not_found if [x] not in [s] *)
+
+  val find : t -> elt -> elt option
+  (** Safe version of {!find_exn} *)
+
+  val inter : t -> t -> t
+  (** [inter a b] returns [a ∩ b] *)
+
+  val inter_mut : into:t -> t -> unit
+  (** [inter_mut ~into a] changes [into] into [a ∩ into] *)
+
+  val union : t -> t -> t
+  (** [union a b] returns [a ∪ b] *)
+
+  val union_mut : into:t -> t -> unit
+  (** [union_mut ~into a] changes [into] into [a ∪ into] *)
+
+  val diff : t -> t -> t
+  (** [diff a b] returns [a - b] *)
+
+  val subset : t -> t -> bool
+  (** [subset a b] returns [true] if all elements of [a] are in [b] *)
+
+  val for_all : (elt -> bool) -> t -> bool
+
+  val exists : (elt -> bool) -> t -> bool
+
+  val iter : (elt -> unit) -> t -> unit
+  (** Iterate on values *)
+
+  val fold : ('a -> elt -> 'a) -> 'a -> t -> 'a
+  (** Fold on values *)
+
+  val elements : t -> elt list
+  (** List of elements *)
+
+  val of_list : elt list -> t
+
+  val to_seq : t -> elt sequence
+
+  val of_seq : elt sequence -> t
+
+  val add_seq : t -> elt sequence -> unit
+
+  val pp : ?sep:string -> elt printer -> t printer
+  (** [pp pp_elt] returns a set printer, given a printer for
+      individual elements *)
+end
+
+module type ELEMENT = sig
+  type t
+  val equal : t -> t -> bool
+  val hash : t -> int (** Positive value *)
+end
+
+module Make(E : ELEMENT) : S with type elt = E.t = struct
+  module Tbl = Hashtbl.Make(E)
+
+  type elt = E.t
+
+  type t = elt Tbl.t  (* map [x -> x], for find *)
+
+  let create = Tbl.create
+
+  let clear = Tbl.clear
+
+  let copy = Tbl.copy
+
+  let copy_into ~into s =
+    Tbl.iter (fun x _ -> Tbl.replace into x x) s
+
+  let insert s x = Tbl.replace s x x
+
+  let remove = Tbl.remove
+
+  let cardinal = Tbl.length
+
+  let mem = Tbl.mem
+
+  let find_exn = Tbl.find
+
+  let find s x =
+    try Some (Tbl.find s x)
+    with Not_found -> None
+
+  let iter f s = Tbl.iter (fun x _ -> f x) s
+
+  let fold f acc s = Tbl.fold (fun x _ acc -> f acc x) s acc
+
+  let inter a b =
+    let res = create (min (cardinal a) (cardinal b)) in
+    iter (fun x -> if mem a x then insert res x) b;
+    res
+
+  let inter_mut ~into a =
+    iter
+      (fun x ->
+        if not (mem a x) then remove into x
+      ) into
+
+  let union a b =
+    let res = copy a in
+    copy_into ~into:res b;
+    res
+
+  let union_mut ~into a =
+    copy_into ~into a
+
+  let diff a b =
+    let res = copy a in
+    iter
+      (fun x -> remove res x) b;
+    res
+
+  exception FastExit
+
+  let for_all p s =
+    try
+      Tbl.iter (fun x _ -> if not (p x) then raise FastExit) s;
+      true
+    with FastExit -> false
+
+  let exists p s =
+    try
+      Tbl.iter (fun x _ -> if p x then raise FastExit) s;
+      false
+    with FastExit -> true
+
+  let subset a b =
+    for_all (fun x -> mem b x) a
+
+  let elements s =
+    Tbl.fold (fun x _ acc -> x::acc) s []
+
+  let of_list l =
+    let res = create (List.length l) in
+    List.iter (insert res) l;
+    res
+
+  let to_seq s yield = iter yield s
+
+  let add_seq s seq = seq (insert s)
+
+  let of_seq seq =
+    let s = create 32 in
+    seq (insert s);
+    s
+
+  let pp ?(sep=",") pp_x out s =
+    Format.pp_print_string out "{";
+    let first = ref true in
+    Tbl.iter
+      (fun x _ ->
+         if !first
+         then first := false
+         else (
+          Format.pp_print_string out sep;
+          Format.pp_print_cut out ();
+         );
+         pp_x out x
+      ) s;
+    Format.pp_print_string out "}"
+end
diff --git a/src/data/CCHashSet.mli b/src/data/CCHashSet.mli
new file mode 100644
index 00000000..ae60ed02
--- /dev/null
+++ b/src/data/CCHashSet.mli
@@ -0,0 +1,98 @@
+(* This file is free softwarem part of containers. See file "license" for more details. *)
+
+(** {1 Mutable Set}
+
+    {b status: unstable}
+
+    @since NEXT_RELEASE *)
+
+type 'a sequence = ('a -> unit) -> unit
+type 'a printer = Format.formatter -> 'a -> unit
+
+module type S = sig
+  type t
+  type elt
+
+  val create : int -> t
+  (** [create n] makes a new set with the given capacity [n] *)
+
+  val clear : t -> unit
+  (** [clear s] removes all elements from [s] *)
+
+  val copy : t -> t
+  (** Fresh copy *)
+
+  val copy_into : into:t -> t -> unit
+  (** [copy_into ~into s] copies all elements of [s] into [into] *)
+
+  val insert : t -> elt -> unit
+  (** [insert s x] adds [x] into [s] *)
+
+  val remove : t -> elt -> unit
+  (** Remove the element, if it were in there *)
+
+  val cardinal : t -> int
+  (** [cardinal s] returns the number of elements in [s] *)
+
+  val mem : t -> elt -> bool
+  (** [mem s x] returns [true] iff [x] is in [s] *)
+
+  val find_exn : t -> elt -> elt
+  (** [find s x] returns [y] if [x] and [y] are equal, and [mem s y].
+      @raise Not_found if [x] not in [s] *)
+
+  val find : t -> elt -> elt option
+  (** Safe version of {!find_exn} *)
+
+  val inter : t -> t -> t
+  (** [inter a b] returns [a ∩ b] *)
+
+  val inter_mut : into:t -> t -> unit
+  (** [inter_mut ~into a] changes [into] into [a ∩ into] *)
+
+  val union : t -> t -> t
+  (** [union a b] returns [a ∪ b] *)
+
+  val union_mut : into:t -> t -> unit
+  (** [union_mut ~into a] changes [into] into [a ∪ into] *)
+
+  val diff : t -> t -> t
+  (** [diff a b] returns [a - b] *)
+
+  val subset : t -> t -> bool
+  (** [subset a b] returns [true] if all elements of [a] are in [b] *)
+
+  val for_all : (elt -> bool) -> t -> bool
+
+  val exists : (elt -> bool) -> t -> bool
+
+  val iter : (elt -> unit) -> t -> unit
+  (** Iterate on values *)
+
+  val fold : ('a -> elt -> 'a) -> 'a -> t -> 'a
+  (** Fold on values *)
+
+  val elements : t -> elt list
+  (** List of elements *)
+
+  val of_list : elt list -> t
+
+  val to_seq : t -> elt sequence
+
+  val of_seq : elt sequence -> t
+
+  val add_seq : t -> elt sequence -> unit
+
+  val pp : ?sep:string -> elt printer -> t printer
+  (** [pp pp_elt] returns a set printer, given a printer for
+      individual elements *)
+end
+
+module type ELEMENT = sig
+  type t
+  val equal : t -> t -> bool
+  val hash : t -> int (** Positive value *)
+end
+
+module Make(E : ELEMENT) : S with type elt = E.t
+