exa/codecrypt
1
0
Fork 0
Code Issues Pull requests Actions Packages Projects Releases Wiki Activity

algos_enc: use faster unranking for decryption

Browse source
This commit is contained in:
Mirek Kratochvil 2013-05-17 15:16:42 +02:00
parent 93d6cccd4f
commit 3f73ba54c4
1 changed files with 6 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

15
src/algos_enc.cpp
View file

@ -371,18 +371,15 @@ static int fo_decrypt (const bvector&cipher, bvector&plain,
H = hf (M2);
/*
* prepare the error vector (again. Avoiding colex ranking which is
* little less deterministic than it could be (produces varying amounts
* of whitespace)
* Colex rank the vector to hash (it is faster than unranking)
*/
bvector ev_rank, ev2;
ev_rank.resize (ranksize);
bvector ev_rank;
ev.colex_rank (ev_rank);
ev_rank.resize (ranksize, 0);
for (i = 0; i < ranksize; ++i)
ev_rank[i] = 1 & (H[ (i >> 3) % H.size()] >> (i & 0x7) );
ev_rank.colex_unrank (ev2, ciphersize, errorcount);
if (ev_rank[i] != 1 & (H[ (i >> 3) % H.size()] >> (i & 0x7) ) )
return 8;
//now it should match, otherwise someone mangled the message.
if (ev != ev2) return 8;
//if the message seems okay, unpad and return it.
if (!message_unpad (M, plain) ) return 9;