umgeher/leveled
0
Fork 0
Code Issues Pull requests Releases Activity Actions

Revert to non-split tinybloom

Browse source
This commit is contained in:
Martin Sumner 2017-01-03 23:53:57 +00:00
parent be1d678d85
commit 85aaccfe31
1 changed files with 13 additions and 32 deletions
Download patch file Download diff file Expand all files Collapse all files

45
src/leveled_tinybloom.erl
View file

@ -36,32 +36,30 @@ empty(Width) when Width =< 256 ->
enter({hash, no_lookup}, Bloom) -> enter({hash, no_lookup}, Bloom) ->
Bloom; Bloom;
enter({hash, Hash}, Bloom) -> enter({hash, Hash}, Bloom) ->
{Slot0, Q, Bit1, Bit2} = split_hash(Hash), {Slot0, Bit1, Bit2} = split_hash(Hash),
Slot = Slot0 rem dict:size(Bloom), Slot = Slot0 rem dict:size(Bloom),
BitArray0 = dict:fetch(Slot, Bloom), BitArray0 = dict:fetch(Slot, Bloom),
{Pre, SplitArray0, Post} = split_array(BitArray0, Q),
FoldFun = FoldFun =
fun(Bit, Arr) -> add_to_array(Bit, Arr, 1024) end, fun(Bit, Arr) -> add_to_array(Bit, Arr, 4096) end,
SplitArray1 = lists:foldl(FoldFun, BitArray1 = lists:foldl(FoldFun,
SplitArray0, BitArray0,
[Bit1, Bit2]), [Bit1, Bit2]),
dict:store(Slot, <<Pre/binary, SplitArray1/binary, Post/binary>>, Bloom); dict:store(Slot, <<BitArray1/binary>>, Bloom);
enter(Key, Bloom) -> enter(Key, Bloom) ->
Hash = leveled_codec:magic_hash(Key), Hash = leveled_codec:magic_hash(Key),
enter({hash, Hash}, Bloom). enter({hash, Hash}, Bloom).
check({hash, Hash}, Bloom) -> check({hash, Hash}, Bloom) ->
{Slot0, Q, Bit1, Bit2} = split_hash(Hash), {Slot0, Bit1, Bit2} = split_hash(Hash),
Slot = Slot0 rem dict:size(Bloom), Slot = Slot0 rem dict:size(Bloom),
BitArray = dict:fetch(Slot, Bloom), BitArray = dict:fetch(Slot, Bloom),
{_Pre, SplitArray, _Post} = split_array(BitArray, Q),
case getbit(Bit1, SplitArray, 1024) of case getbit(Bit1, BitArray, 4096) of
<<0:1>> -> <<0:1>> ->
false; false;
<<1:1>> -> <<1:1>> ->
case getbit(Bit2, SplitArray, 1024) of case getbit(Bit2, BitArray, 4096) of
<<0:1>> -> <<0:1>> ->
false; false;
<<1:1>> -> <<1:1>> ->
@ -79,33 +77,16 @@ check(Key, Bloom) ->
split_hash(Hash) -> split_hash(Hash) ->
Slot = split_for_slot(Hash), Slot = split_for_slot(Hash),
{Q1, H1, H2} = split_for_bits(Hash), {H1, H2} = split_for_bits(Hash),
{Slot, Q1, H1, H2}. {Slot, H1, H2}.
split_for_slot(Hash) -> split_for_slot(Hash) ->
Hash band 255. Hash band 255.
split_for_bits(Hash) -> split_for_bits(Hash) ->
H1 = (Hash bsr 8) band 1023, H1 = (Hash bsr 8) band 4095,
H2 = (Hash bsr 18) band 1023, H2 = (Hash bsr 20) band 4095,
Q1 = (Hash bsr 28) band 3, {H1, H2}.
{Q1, H1, H2}.
split_array(Bin, Q) ->
case Q of
0 ->
<<ToUse:128/binary, Post/binary>> = Bin,
{<<>>, ToUse, Post};
1 ->
<<Pre:128/binary, ToUse:128/binary, Post/binary>> = Bin,
{Pre, ToUse, Post};
2 ->
<<Pre:256/binary, ToUse:128/binary, Post/binary>> = Bin,
{Pre, ToUse, Post};
3 ->
<<Pre:384/binary, ToUse:128/binary>> = Bin,
{Pre, ToUse, <<>>}
end.
add_to_array(Bit, BitArray, ArrayLength) -> add_to_array(Bit, BitArray, ArrayLength) ->
RestLen = ArrayLength - Bit - 1, RestLen = ArrayLength - Bit - 1,