tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 1 | #include "Bitvec.h"
|
| 2 | #include <vector>
|
| 3 | #include <assert.h>
|
| 4 |
|
| 5 | // handle xnor
|
| 6 |
|
| 7 | typedef std::vector<uint32_t> slice;
|
| 8 | typedef std::vector<slice> slice_vec;
|
| 9 |
|
| 10 | int countbits ( slice & v )
|
| 11 | {
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 12 | int c = 0;
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 13 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 14 | for(size_t i = 0; i < v.size(); i++)
|
| 15 | {
|
| 16 | int d = countbits(v[i]);
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 17 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 18 | c += d;
|
| 19 | }
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 20 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 21 | return c;
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 22 | }
|
| 23 |
|
| 24 | int countxor ( slice & a, slice & b )
|
| 25 | {
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 26 | assert(a.size() == b.size());
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 27 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 28 | int c = 0;
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 29 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 30 | for(size_t i = 0; i < a.size(); i++)
|
| 31 | {
|
| 32 | int d = countbits(a[i] ^ b[i]);
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 33 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 34 | c += d;
|
| 35 | }
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 36 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 37 | return c;
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 38 | }
|
| 39 |
|
| 40 | void xoreq ( slice & a, slice & b )
|
| 41 | {
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 42 | assert(a.size() == b.size());
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 43 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 44 | for(size_t i = 0; i < a.size(); i++)
|
| 45 | {
|
| 46 | a[i] ^= b[i];
|
| 47 | }
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 48 | }
|
| 49 |
|
| 50 | //-----------------------------------------------------------------------------
|
| 51 | // Bitslice a hash set
|
| 52 |
|
| 53 | template< typename hashtype >
|
| 54 | void Bitslice ( std::vector<hashtype> & hashes, slice_vec & slices )
|
| 55 | {
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 56 | const int hashbytes = sizeof(hashtype);
|
| 57 | const int hashbits = hashbytes * 8;
|
| 58 | const int slicelen = ((int)hashes.size() + 31) / 32;
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 59 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 60 | slices.clear();
|
| 61 | slices.resize(hashbits);
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 62 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 63 | for(int i = 0; i < (int)slices.size(); i++)
|
| 64 | {
|
| 65 | slices[i].resize(slicelen,0);
|
| 66 | }
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 67 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 68 | for(int j = 0; j < hashbits; j++)
|
| 69 | {
|
| 70 | void * sliceblob = &(slices[j][0]);
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 71 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 72 | for(int i = 0; i < (int)hashes.size(); i++)
|
| 73 | {
|
| 74 | int b = getbit(hashes[i],j);
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 75 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 76 | setbit(sliceblob,slicelen*4,i,b);
|
| 77 | }
|
| 78 | }
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 79 | }
|
| 80 |
|
| 81 | void FactorSlices ( slice_vec & slices )
|
| 82 | {
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 83 | std::vector<int> counts(slices.size(),0);
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 84 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 85 | for(size_t i = 0; i < slices.size(); i++)
|
| 86 | {
|
| 87 | counts[i] = countbits(slices[i]);
|
| 88 | }
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 89 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 90 | bool changed = true;
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 91 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 92 | while(changed)
|
| 93 | {
|
| 94 | int bestA = -1;
|
| 95 | int bestB = -1;
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 96 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 97 | for(int j = 0; j < (int)slices.size()-1; j++)
|
| 98 | {
|
| 99 | for(int i = j+1; i < (int)slices.size(); i++)
|
| 100 | {
|
| 101 | int d = countxor(slices[i],slices[j]);
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 102 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 103 | if((d < counts[i]) && (d < counts[j]))
|
| 104 | {
|
| 105 | if(counts[i] < counts[j])
|
| 106 | {
|
| 107 | bestA = j;
|
| 108 | bestB = i;
|
| 109 | }
|
| 110 | }
|
| 111 | else if(d < counts[i])
|
| 112 | {
|
| 113 | //bestA =
|
| 114 | }
|
| 115 | }
|
| 116 | }
|
| 117 | }
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 118 | }
|
| 119 |
|
| 120 |
|
| 121 | void foo ( void )
|
| 122 | {
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 123 | slice a;
|
| 124 | slice_vec b;
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 125 |
|
tanjent@gmail.com | 6ffe010 | 2011-03-19 21:28:26 +0000 | [diff] [blame] | 126 | Bitslice(a,b);
|
tanjent@gmail.com | f67ce94 | 2011-03-14 09:11:18 +0000 | [diff] [blame] | 127 | } |