tanjent@gmail.com | f3b7897 | 2012-03-01 03:38:55 +0000 | [diff] [blame] | 1 | //----------------------------------------------------------------------------- |
| 2 | // Keyset tests generate various sorts of difficult-to-hash keysets and compare |
| 3 | // the distribution and collision frequency of the hash results against an |
| 4 | // ideal random distribution |
| 5 | |
| 6 | // The sanity checks are also in this cpp/h |
| 7 | |
| 8 | #pragma once |
| 9 | |
| 10 | #include "Types.h" |
| 11 | #include "Stats.h" |
| 12 | #include "Random.h" // for rand_p |
| 13 | |
| 14 | #include <algorithm> // for std::swap |
| 15 | #include <assert.h> |
| 16 | |
| 17 | //----------------------------------------------------------------------------- |
| 18 | // Sanity tests |
| 19 | |
| 20 | bool VerificationTest ( pfHash hash, const int hashbits, uint32_t expected, bool verbose ); |
| 21 | bool SanityTest ( pfHash hash, const int hashbits ); |
| 22 | void AppendedZeroesTest ( pfHash hash, const int hashbits ); |
| 23 | |
| 24 | //----------------------------------------------------------------------------- |
| 25 | // Keyset 'Combination' - all possible combinations of input blocks |
| 26 | |
| 27 | template< typename hashtype > |
| 28 | void CombinationKeygenRecurse ( uint32_t * key, int len, int maxlen, |
| 29 | uint32_t * blocks, int blockcount, |
| 30 | pfHash hash, std::vector<hashtype> & hashes ) |
| 31 | { |
| 32 | if(len == maxlen) return; |
| 33 | |
| 34 | for(int i = 0; i < blockcount; i++) |
| 35 | { |
| 36 | key[len] = blocks[i]; |
| 37 | |
| 38 | //if(len == maxlen-1) |
| 39 | { |
| 40 | hashtype h; |
| 41 | hash(key,(len+1) * sizeof(uint32_t),0,&h); |
| 42 | hashes.push_back(h); |
| 43 | } |
| 44 | |
| 45 | //else |
| 46 | { |
| 47 | CombinationKeygenRecurse(key,len+1,maxlen,blocks,blockcount,hash,hashes); |
| 48 | } |
| 49 | } |
| 50 | } |
| 51 | |
| 52 | template< typename hashtype > |
| 53 | bool CombinationKeyTest ( hashfunc<hashtype> hash, int maxlen, uint32_t * blocks, int blockcount, bool testColl, bool testDist, bool drawDiagram ) |
| 54 | { |
| 55 | printf("Keyset 'Combination' - up to %d blocks from a set of %d - ",maxlen,blockcount); |
| 56 | |
| 57 | //---------- |
| 58 | |
| 59 | std::vector<hashtype> hashes; |
| 60 | |
| 61 | uint32_t * key = new uint32_t[maxlen]; |
| 62 | |
| 63 | CombinationKeygenRecurse<hashtype>(key,0,maxlen,blocks,blockcount,hash,hashes); |
| 64 | |
| 65 | delete [] key; |
| 66 | |
| 67 | printf("%d keys\n",(int)hashes.size()); |
| 68 | |
| 69 | //---------- |
| 70 | |
| 71 | bool result = true; |
| 72 | |
| 73 | result &= TestHashList<hashtype>(hashes,testColl,testDist,drawDiagram); |
| 74 | |
| 75 | printf("\n"); |
| 76 | |
| 77 | return result; |
| 78 | } |
| 79 | |
| 80 | //---------------------------------------------------------------------------- |
| 81 | // Keyset 'Permutation' - given a set of 32-bit blocks, generate keys |
| 82 | // consisting of all possible permutations of those blocks |
| 83 | |
| 84 | template< typename hashtype > |
| 85 | void PermutationKeygenRecurse ( pfHash hash, uint32_t * blocks, int blockcount, int k, std::vector<hashtype> & hashes ) |
| 86 | { |
| 87 | if(k == blockcount-1) |
| 88 | { |
| 89 | hashtype h; |
| 90 | |
| 91 | hash(blocks,blockcount * sizeof(uint32_t),0,&h); |
| 92 | |
| 93 | hashes.push_back(h); |
| 94 | |
| 95 | return; |
| 96 | } |
| 97 | |
| 98 | for(int i = k; i < blockcount; i++) |
| 99 | { |
| 100 | std::swap(blocks[k],blocks[i]); |
| 101 | |
| 102 | PermutationKeygenRecurse(hash,blocks,blockcount,k+1,hashes); |
| 103 | |
| 104 | std::swap(blocks[k],blocks[i]); |
| 105 | } |
| 106 | } |
| 107 | |
| 108 | template< typename hashtype > |
| 109 | bool PermutationKeyTest ( hashfunc<hashtype> hash, uint32_t * blocks, int blockcount, bool testColl, bool testDist, bool drawDiagram ) |
| 110 | { |
| 111 | printf("Keyset 'Permutation' - %d blocks - ",blockcount); |
| 112 | |
| 113 | //---------- |
| 114 | |
| 115 | std::vector<hashtype> hashes; |
| 116 | |
| 117 | PermutationKeygenRecurse<hashtype>(hash,blocks,blockcount,0,hashes); |
| 118 | |
| 119 | printf("%d keys\n",(int)hashes.size()); |
| 120 | |
| 121 | //---------- |
| 122 | |
| 123 | bool result = true; |
| 124 | |
| 125 | result &= TestHashList<hashtype>(hashes,testColl,testDist,drawDiagram); |
| 126 | |
| 127 | printf("\n"); |
| 128 | |
| 129 | return result; |
| 130 | } |
| 131 | |
| 132 | //----------------------------------------------------------------------------- |
| 133 | // Keyset 'Sparse' - generate all possible N-bit keys with up to K bits set |
| 134 | |
| 135 | template < typename keytype, typename hashtype > |
| 136 | void SparseKeygenRecurse ( pfHash hash, int start, int bitsleft, bool inclusive, keytype & k, std::vector<hashtype> & hashes ) |
| 137 | { |
| 138 | const int nbytes = sizeof(keytype); |
| 139 | const int nbits = nbytes * 8; |
| 140 | |
| 141 | hashtype h; |
| 142 | |
| 143 | for(int i = start; i < nbits; i++) |
| 144 | { |
| 145 | flipbit(&k,nbytes,i); |
| 146 | |
| 147 | if(inclusive || (bitsleft == 1)) |
| 148 | { |
| 149 | hash(&k,sizeof(keytype),0,&h); |
| 150 | hashes.push_back(h); |
| 151 | } |
| 152 | |
| 153 | if(bitsleft > 1) |
| 154 | { |
| 155 | SparseKeygenRecurse(hash,i+1,bitsleft-1,inclusive,k,hashes); |
| 156 | } |
| 157 | |
| 158 | flipbit(&k,nbytes,i); |
| 159 | } |
| 160 | } |
| 161 | |
| 162 | //---------- |
| 163 | |
| 164 | template < int keybits, typename hashtype > |
| 165 | bool SparseKeyTest ( hashfunc<hashtype> hash, const int setbits, bool inclusive, bool testColl, bool testDist, bool drawDiagram ) |
| 166 | { |
| 167 | printf("Keyset 'Sparse' - %d-bit keys with %s %d bits set - ",keybits, inclusive ? "up to" : "exactly", setbits); |
| 168 | |
| 169 | typedef Blob<keybits> keytype; |
| 170 | |
| 171 | std::vector<hashtype> hashes; |
| 172 | |
| 173 | keytype k; |
| 174 | memset(&k,0,sizeof(k)); |
| 175 | |
| 176 | if(inclusive) |
| 177 | { |
| 178 | hashtype h; |
| 179 | |
| 180 | hash(&k,sizeof(keytype),0,&h); |
| 181 | |
| 182 | hashes.push_back(h); |
| 183 | } |
| 184 | |
| 185 | SparseKeygenRecurse(hash,0,setbits,inclusive,k,hashes); |
| 186 | |
| 187 | printf("%d keys\n",(int)hashes.size()); |
| 188 | |
| 189 | bool result = true; |
| 190 | |
| 191 | result &= TestHashList<hashtype>(hashes,testColl,testDist,drawDiagram); |
| 192 | |
| 193 | printf("\n"); |
| 194 | |
| 195 | return result; |
| 196 | } |
| 197 | |
| 198 | //----------------------------------------------------------------------------- |
| 199 | // Keyset 'Windows' - for all possible N-bit windows of a K-bit key, generate |
| 200 | // all possible keys with bits set in that window |
| 201 | |
| 202 | template < typename keytype, typename hashtype > |
| 203 | bool WindowedKeyTest ( hashfunc<hashtype> hash, const int windowbits, bool testCollision, bool testDistribution, bool drawDiagram ) |
| 204 | { |
| 205 | const int keybits = sizeof(keytype) * 8; |
| 206 | const int keycount = 1 << windowbits; |
| 207 | |
| 208 | std::vector<hashtype> hashes; |
| 209 | hashes.resize(keycount); |
| 210 | |
| 211 | bool result = true; |
| 212 | |
| 213 | int testcount = keybits; |
| 214 | |
| 215 | printf("Keyset 'Windowed' - %3d-bit key, %3d-bit window - %d tests, %d keys per test\n",keybits,windowbits,testcount,keycount); |
| 216 | |
| 217 | for(int j = 0; j <= testcount; j++) |
| 218 | { |
| 219 | int minbit = j; |
| 220 | |
| 221 | keytype key; |
| 222 | |
| 223 | for(int i = 0; i < keycount; i++) |
| 224 | { |
| 225 | key = i; |
| 226 | //key = key << minbit; |
| 227 | |
| 228 | lrot(&key,sizeof(keytype),minbit); |
| 229 | |
| 230 | hash(&key,sizeof(keytype),0,&hashes[i]); |
| 231 | } |
| 232 | |
| 233 | printf("Window at %3d - ",j); |
| 234 | |
| 235 | result &= TestHashList(hashes,testCollision,testDistribution,drawDiagram); |
| 236 | |
| 237 | //printf("\n"); |
| 238 | } |
| 239 | |
| 240 | return result; |
| 241 | } |
| 242 | |
| 243 | //----------------------------------------------------------------------------- |
| 244 | // Keyset 'Cyclic' - generate keys that consist solely of N repetitions of M |
| 245 | // bytes. |
| 246 | |
| 247 | // (This keyset type is designed to make MurmurHash2 fail) |
| 248 | |
| 249 | template < typename hashtype > |
| 250 | bool CyclicKeyTest ( pfHash hash, int cycleLen, int cycleReps, const int keycount, bool drawDiagram ) |
| 251 | { |
| 252 | printf("Keyset 'Cyclic' - %d cycles of %d bytes - %d keys\n",cycleReps,cycleLen,keycount); |
| 253 | |
| 254 | Rand r(483723); |
| 255 | |
| 256 | std::vector<hashtype> hashes; |
| 257 | hashes.resize(keycount); |
| 258 | |
| 259 | int keyLen = cycleLen * cycleReps; |
| 260 | |
| 261 | uint8_t * cycle = new uint8_t[cycleLen + 16]; |
| 262 | uint8_t * key = new uint8_t[keyLen]; |
| 263 | |
| 264 | //---------- |
| 265 | |
| 266 | for(int i = 0; i < keycount; i++) |
| 267 | { |
| 268 | r.rand_p(cycle,cycleLen); |
| 269 | |
| 270 | *(uint32_t*)cycle = f3mix(i ^ 0x746a94f1); |
| 271 | |
| 272 | for(int j = 0; j < keyLen; j++) |
| 273 | { |
| 274 | key[j] = cycle[j % cycleLen]; |
| 275 | } |
| 276 | |
| 277 | hash(key,keyLen,0,&hashes[i]); |
| 278 | } |
| 279 | |
| 280 | //---------- |
| 281 | |
| 282 | bool result = true; |
| 283 | |
| 284 | result &= TestHashList(hashes,true,true,drawDiagram); |
| 285 | printf("\n"); |
| 286 | |
| 287 | delete [] cycle; |
| 288 | delete [] key; |
| 289 | |
| 290 | return result; |
| 291 | } |
| 292 | |
| 293 | //----------------------------------------------------------------------------- |
| 294 | // Keyset 'TwoBytes' - generate all keys up to length N with two non-zero bytes |
| 295 | |
| 296 | void TwoBytesKeygen ( int maxlen, KeyCallback & c ); |
| 297 | |
| 298 | template < typename hashtype > |
| 299 | bool TwoBytesTest2 ( pfHash hash, int maxlen, bool drawDiagram ) |
| 300 | { |
| 301 | std::vector<hashtype> hashes; |
| 302 | |
| 303 | HashCallback<hashtype> c(hash,hashes); |
| 304 | |
| 305 | TwoBytesKeygen(maxlen,c); |
| 306 | |
| 307 | bool result = true; |
| 308 | |
| 309 | result &= TestHashList(hashes,true,true,drawDiagram); |
| 310 | printf("\n"); |
| 311 | |
| 312 | return result; |
| 313 | } |
| 314 | |
| 315 | //----------------------------------------------------------------------------- |
| 316 | // Keyset 'Text' - generate all keys of the form "prefix"+"core"+"suffix", |
| 317 | // where "core" consists of all possible combinations of the given character |
| 318 | // set of length N. |
| 319 | |
| 320 | template < typename hashtype > |
| 321 | bool TextKeyTest ( hashfunc<hashtype> hash, const char * prefix, const char * coreset, const int corelen, const char * suffix, bool drawDiagram ) |
| 322 | { |
| 323 | const int prefixlen = (int)strlen(prefix); |
| 324 | const int suffixlen = (int)strlen(suffix); |
| 325 | const int corecount = (int)strlen(coreset); |
| 326 | |
| 327 | const int keybytes = prefixlen + corelen + suffixlen; |
| 328 | const int keycount = (int)pow(double(corecount),double(corelen)); |
| 329 | |
| 330 | printf("Keyset 'Text' - keys of form \"%s[",prefix); |
| 331 | for(int i = 0; i < corelen; i++) printf("X"); |
| 332 | printf("]%s\" - %d keys\n",suffix,keycount); |
| 333 | |
| 334 | uint8_t * key = new uint8_t[keybytes+1]; |
| 335 | |
| 336 | key[keybytes] = 0; |
| 337 | |
| 338 | memcpy(key,prefix,prefixlen); |
| 339 | memcpy(key+prefixlen+corelen,suffix,suffixlen); |
| 340 | |
| 341 | //---------- |
| 342 | |
| 343 | std::vector<hashtype> hashes; |
| 344 | hashes.resize(keycount); |
| 345 | |
| 346 | for(int i = 0; i < keycount; i++) |
| 347 | { |
| 348 | int t = i; |
| 349 | |
| 350 | for(int j = 0; j < corelen; j++) |
| 351 | { |
| 352 | key[prefixlen+j] = coreset[t % corecount]; t /= corecount; |
| 353 | } |
| 354 | |
| 355 | hash(key,keybytes,0,&hashes[i]); |
| 356 | } |
| 357 | |
| 358 | //---------- |
| 359 | |
| 360 | bool result = true; |
| 361 | |
| 362 | result &= TestHashList(hashes,true,true,drawDiagram); |
| 363 | |
| 364 | printf("\n"); |
| 365 | |
| 366 | delete [] key; |
| 367 | |
| 368 | return result; |
| 369 | } |
| 370 | |
| 371 | //----------------------------------------------------------------------------- |
| 372 | // Keyset 'Zeroes' - keys consisting of all zeroes, differing only in length |
| 373 | |
| 374 | // We reuse one block of empty bytes, otherwise the RAM cost is enormous. |
| 375 | |
| 376 | template < typename hashtype > |
| 377 | bool ZeroKeyTest ( pfHash hash, bool drawDiagram ) |
| 378 | { |
| 379 | int keycount = 64*1024; |
| 380 | |
| 381 | printf("Keyset 'Zeroes' - %d keys\n",keycount); |
| 382 | |
| 383 | unsigned char * nullblock = new unsigned char[keycount]; |
| 384 | memset(nullblock,0,keycount); |
| 385 | |
| 386 | //---------- |
| 387 | |
| 388 | std::vector<hashtype> hashes; |
| 389 | |
| 390 | hashes.resize(keycount); |
| 391 | |
| 392 | for(int i = 0; i < keycount; i++) |
| 393 | { |
| 394 | hash(nullblock,i,0,&hashes[i]); |
| 395 | } |
| 396 | |
| 397 | bool result = true; |
| 398 | |
| 399 | result &= TestHashList(hashes,true,true,drawDiagram); |
| 400 | |
| 401 | printf("\n"); |
| 402 | |
| 403 | delete [] nullblock; |
| 404 | |
| 405 | return result; |
| 406 | } |
| 407 | |
| 408 | //----------------------------------------------------------------------------- |
| 409 | // Keyset 'Seed' - hash "the quick brown fox..." using different seeds |
| 410 | |
| 411 | template < typename hashtype > |
| 412 | bool SeedTest ( pfHash hash, int keycount, bool drawDiagram ) |
| 413 | { |
| 414 | printf("Keyset 'Seed' - %d keys\n",keycount); |
| 415 | |
| 416 | const char * text = "The quick brown fox jumps over the lazy dog"; |
| 417 | const int len = (int)strlen(text); |
| 418 | |
| 419 | //---------- |
| 420 | |
| 421 | std::vector<hashtype> hashes; |
| 422 | |
| 423 | hashes.resize(keycount); |
| 424 | |
| 425 | for(int i = 0; i < keycount; i++) |
| 426 | { |
| 427 | hash(text,len,i,&hashes[i]); |
| 428 | } |
| 429 | |
| 430 | bool result = true; |
| 431 | |
| 432 | result &= TestHashList(hashes,true,true,drawDiagram); |
| 433 | |
| 434 | printf("\n"); |
| 435 | |
| 436 | return result; |
| 437 | } |
| 438 | |
| 439 | //----------------------------------------------------------------------------- |