Bloom filter: Add faster, ETS-based build representation for dense bi…

…tmaps.

src/hanoidb_bloom.erl

@@ -187,16 +187,30 @@ set_bits(Mask, I1, I, [H|T], Acc) ->
 
 %%%========== Dispatch to appropriate representation:
 bitmask_new(LogN) ->
-    bitarray_new(1 bsl LogN).
+    if LogN >= 20 -> % Use sparse representation.
+            bitarray_new(1 bsl LogN);
+       true ->       % Use dense representation.
+            hanoidb_dense_bitmap:new(1 bsl LogN)
+    end.
 
 bitmask_set(I, BM) ->
     case element(1,BM) of
-        array -> bitarray_set(I, BM)
+        array -> bitarray_set(I, BM);
+        dense_bitmap_ets -> hanoidb_dense_bitmap:set(I, BM)
+    end.
+
+%%% Convert to external form.
+bitmask_build(BM) ->
+    case element(1,BM) of
+        array -> BM;
+        dense_bitmap_ets -> hanoidb_dense_bitmap:build(BM)
     end.
 
 bitmask_get(I, BM) ->
     case element(1,BM) of
-        array -> bitarray_get(I, BM)
+        array -> bitarray_get(I, BM);
+        dense_bitmap_ets -> hanoidb_dense_bitmap:member(I, BM);
+        dense_bitmap     -> hanoidb_dense_bitmap:member(I, BM)
     end.
 
 %%%========== Bitarray representation - suitable for sparse arrays ==========
@@ -216,11 +230,17 @@ bitarray_get(I, A) ->
 %%%^^^^^^^^^^ Bitarray representation - suitable for sparse arrays ^^^^^^^^^^
 
 encode(Bloom) ->
-    zlib:gzip(term_to_binary(Bloom)).
+    zlib:gzip(term_to_binary(bloom_build(Bloom))).
 
 decode(Bin) ->
     binary_to_term(zlib:gunzip(Bin)).
 
+%%% Convert to external form.
+bloom_build(Bloom=#bloom{a=Bitmasks}) ->
+    Bloom#bloom{a=[bitmask_build(X) || X <- Bitmasks]};
+bloom_build(Sbf=#sbf{b=Blooms}) ->
+    Sbf#sbf{b=[bloom_build(X) || X <- Blooms]}.
+
 %% UNIT TESTS
 
 -ifdef(TEST).

src/hanoidb_dense_bitmap.erl

@@ -0,0 +1,39 @@
+-module(hanoidb_dense_bitmap).
+
+-export([new/1, set/2, build/1, member/2]).
+-define(BITS_PER_CELL, 32).
+
+-define(REPR_NAME, dense_bitmap).
+
+new(N) ->
+    Tab = ets:new(dense_bitmap, [private, set]),
+    Width = 1 + (N-1) div ?BITS_PER_CELL,
+    Value = erlang:make_tuple(Width+1, 0, [{1,?REPR_NAME}]),
+    ets:insert(Tab, Value),
+    %io:format("DB| create(): ~p of width ~p\n", [Tab, Width]),
+    {dense_bitmap_ets, N, Width, Tab}.
+
+%% Set a bit.
+set(I, {dense_bitmap_ets, _,_, Tab}=DBM) ->
+    Cell = 2 + I div ?BITS_PER_CELL,
+    BitInCell = I rem ?BITS_PER_CELL,
+    Old = ets:lookup_element(Tab, ?REPR_NAME, Cell),
+    New = Old bor (1 bsl BitInCell),
+    ets:update_element(Tab, ?REPR_NAME, {Cell,New}),
+    DBM.
+
+build({dense_bitmap_ets, _, _, Tab}) ->
+    [Row] = ets:lookup(Tab, ?REPR_NAME),
+    ets:delete(Tab),
+    Row.
+
+member(I, Row) when element(1,Row)==?REPR_NAME ->
+    Cell = 2 + I div ?BITS_PER_CELL,
+    BitInCell = I rem ?BITS_PER_CELL,
+    CellValue = element(Cell, Row),
+    CellValue band (1 bsl BitInCell) =/= 0;
+member(I, {dense_bitmap_ets, _,_, Tab}) ->
+    Cell = 2 + I div ?BITS_PER_CELL,
+    BitInCell = I rem ?BITS_PER_CELL,
+    CellValue = ets:lookup_element(Tab, ?REPR_NAME, Cell),
+    CellValue band (1 bsl BitInCell) =/= 0.