View source with formatted comments or as raw
    1/*  Part of SWI-Prolog
    2
    3    Author:        Jan Wielemaker
    4    E-mail:        J.Wielemaker@vu.nl
    5    WWW:           http://www.swi-prolog.org
    6    Copyright (c)  2005-2024, VU University Amsterdam
    7                              CWI, Amsterdam
    8                              SWI-Prolog Solutions b.v.
    9    All rights reserved.
   10
   11    Redistribution and use in source and binary forms, with or without
   12    modification, are permitted provided that the following conditions
   13    are met:
   14
   15    1. Redistributions of source code must retain the above copyright
   16       notice, this list of conditions and the following disclaimer.
   17
   18    2. Redistributions in binary form must reproduce the above copyright
   19       notice, this list of conditions and the following disclaimer in
   20       the documentation and/or other materials provided with the
   21       distribution.
   22
   23    THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
   24    "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
   25    LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
   26    FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
   27    COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
   28    INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
   29    BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
   30    LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
   31    CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
   32    LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
   33    ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
   34    POSSIBILITY OF SUCH DAMAGE.
   35*/
   36
   37:- module(nb_set,
   38          [ empty_nb_set/1,		 % -EmptySet
   39            add_nb_set/2,		 % +Key, !Set
   40            add_nb_set/3,		 % +Key, !Set, ?New
   41            size_nb_set/2,		 % +Set, -Size
   42            nb_set_to_list/2,		 % +Set, -List
   43            gen_nb_set/2                 % +Set, -Key
   44          ]).   45:- autoload(library(terms), [term_factorized/3]).   46:- use_module(library(debug), [assertion/1]).   47
   48/** <module> Non-backtrackable sets
   49
   50This library provides  a  non-backtrackabe  _set_   of  terms  that  are
   51variants of each other. It is primarily intended to implement distinct/1
   52from library(solution_sequences). The set is implemented as a hash table
   53that is built using non-backtrackable primitives, notably nb_setarg/3.
   54
   55The original version of this library   used  binary trees which provides
   56immediate ordering. As the trees were   not  balanced, performance could
   57get   really   poor.   The   complexity   of   balancing   trees   using
   58non-backtrackable primitives is too high. The  next iteration used _open
   59hash tables_, while the current incarnation   uses _closed hash tables_,
   60providing better perfomance and less space usage.
   61
   62@author Jan Wielemaker
   63*/
   64
   65initial_capacity(4).                       % initial hash-table size
   66
   67%!  empty_nb_set(-Set)
   68%
   69%   Create an empty non-backtrackable set.
   70
   71empty_nb_set(NbSet) :-
   72    initial_capacity(Capacity),
   73    Empty = empty(1),
   74    '$filled_array'(Buckets, buckets, Capacity, Empty),
   75    NbSet = nb_set(Empty, Capacity, 0, Buckets).
   76
   77%!  add_nb_set(+Key, !Set) is det.
   78%!  add_nb_set(+Key, !Set, ?New) is semidet.
   79%
   80%   Insert Key into the set. If  a   variant  (see  =@=/2) of Key is
   81%   already in the set, the set is unchanged and New is unified with
   82%   `false`. Otherwise, New is unified with   `true` and a _copy of_
   83%   Key is added to the set.
   84%
   85%   @tbd    Computing the hash for cyclic terms is performed with
   86%           the help of term_factorized/3, which performs rather
   87%           poorly.
   88
   89add_nb_set(Key, Set) :-
   90    add_nb_set(Key, Set, _).
   91add_nb_set(Key, Set, New) :-
   92    Set = nb_set(Empty, Capacity, Size, Buckets),
   93    key_hash(Key, Hash),
   94    index(Hash, Capacity, KIndex),
   95        arg(KIndex, Buckets, StoredKey),
   96        (   same_term(StoredKey, Empty)
   97        ->  !,
   98            New = true,
   99            nb_setarg(KIndex, Buckets, Key),
  100            NSize is Size+1,
  101            nb_setarg(3, Set, NSize),
  102            (   NSize > Capacity//2
  103            ->  rehash(Set)
  104            ;   true
  105            )
  106        ;   Key =@= StoredKey
  107        ->  !,
  108            New = false
  109        ).
  110
  111%!  index(+Hash, +Capacity, -Index) is nondet.
  112%
  113%   Generate  candidate  values  for  Index,  starting  from  `Hash  mod
  114%   Capacity`, round tripping to 1 when Capacity is reached.
  115
  116index(Hash, Capacity, KIndex) :-
  117    KIndex0 is (Hash mod Capacity) + 1,
  118    next(KIndex0, Capacity, KIndex).
  119
  120next(KIndex, _, KIndex).
  121next(KIndex0, Capacity, KIndex) :-
  122    KIndex1 is 1+(KIndex0 mod Capacity),
  123    next(KIndex1, Capacity, KIndex).
  124
  125rehash(Set) :-
  126    Set = nb_set(Empty, Capacity, Size, Buckets),
  127    NCapacity is Capacity*2,
  128    '$filled_array'(NBuckets, buckets, NCapacity, Empty),
  129    nb_setarg(2, Set, NCapacity),
  130    nb_setarg(3, Set, 0),
  131    nb_linkarg(4, Set, NBuckets),
  132    forall(between(1, Capacity, I),
  133           reinsert(I, Empty, Buckets, Set)),
  134    arg(3, Set, NewSize),
  135    assertion(NewSize == Size).
  136
  137:- det(reinsert/4).  138reinsert(KIndex, Empty, Buckets, Set) :-
  139    arg(KIndex, Buckets, Key),
  140    (   same_term(Key, Empty)
  141    ->  true
  142    ;   add_nb_set(Key, Set, true)
  143    ).
  144
  145%!  hash_key(+Key, -Hash:integer) is det.
  146%
  147%   Compute a hash for Term. Note that variant_hash/2 currently does
  148%   not handle cyclic terms, so use  term_factorized/3 to get rid of
  149%   the cycles. This means that  this   library  is rather slow when
  150%   cyclic terms are involved.
  151
  152:- if(catch((A = f(A), variant_hash(A,_)), _, fail)).  153key_hash(Key, Hash) :-
  154    variant_hash(Key, Hash).
  155:- else.  156key_hash(Key, Hash) :-
  157    acyclic_term(Key),
  158    !,
  159    variant_hash(Key, Hash).
  160key_hash(Key, Hash) :-
  161    term_factorized(Key, Skeleton, Substiution),
  162    variant_hash(Skeleton+Substiution, Hash).
  163:- endif.  164
  165%!  nb_set_to_list(+NBSet, -OrdSet) is det.
  166%!  nb_set_to_list(-NBSet, +List) is det.
  167%
  168%   Get the elements of a an nb_set.   OrdSet  is sorted to the standard
  169%   order of terms, providing a set representation that is compatible to
  170%   library(ordsets).
  171
  172nb_set_to_list(NBSet, Set),
  173    NBSet = nb_set(Empty, Capacity, _Size, Buckets) =>
  174    buckets_to_list(1, Empty, Capacity, Buckets, List0),
  175    sort(List0, Set).
  176
  177buckets_to_list(KIndex, Empty, Capacity, Buckets, List) :-
  178    (   arg(KIndex, Buckets, Key)
  179    ->  (   same_term(Empty, Key)
  180        ->  KIndex1 is KIndex+1,
  181            buckets_to_list(KIndex1, Empty, Capacity, Buckets, List)
  182        ;   List = [Key|List1],
  183            KIndex1 is KIndex+2,
  184            buckets_to_list(KIndex1, Empty, Capacity, Buckets, List1)
  185        )
  186    ;   List = []
  187    ).
  188
  189%!  gen_nb_set(+Set, -Key) is nondet.
  190%
  191%   Enumerate the members of a set in the standard order of terms.
  192
  193gen_nb_set(nb_set(Empty, Capacity, _Size, Buckets), Key) =>
  194    between(1, Capacity, KIndex),
  195    arg(KIndex, Buckets, Key),
  196    \+ same_term(Empty, Key).
  197
  198%!  size_nb_set(+Set, -Size) is det.
  199%
  200%   Unify Size with the number of elements in the set
  201
  202size_nb_set(nb_set(_Empty, _Capacity, Sz, _Buckets), Size) =>
  203    Size = Sz