| 1 | #if defined IMPL_STL_NA_NA | 
|---|
| 2 | #define IMPL_STL | 
|---|
| 3 | #endif | 
|---|
| 4 |  | 
|---|
| 5 | #if defined IMPL_BUHR94_NA_NA | 
|---|
| 6 | #define IMPL_BUHR94 | 
|---|
| 7 | #endif | 
|---|
| 8 |  | 
|---|
| 9 | #if defined IMPL_STL | 
|---|
| 10 | #include <string> | 
|---|
| 11 | #include <iostream> | 
|---|
| 12 | #include <cstdio> | 
|---|
| 13 | using namespace std; | 
|---|
| 14 | #define IMPL_CXX | 
|---|
| 15 |  | 
|---|
| 16 | #elif defined IMPL_CFA_HL_SHARE | 
|---|
| 17 | #define IMPL_CFA_HL | 
|---|
| 18 | #define IMPL_CFA | 
|---|
| 19 |  | 
|---|
| 20 | #elif defined IMPL_CFA_LL_SHARE | 
|---|
| 21 | #define IMPL_CFA_LL | 
|---|
| 22 | #define IMPL_CFA | 
|---|
| 23 |  | 
|---|
| 24 | #elif defined IMPL_CFA_HL_NOSHARE | 
|---|
| 25 | #define IMPL_CFA_HL | 
|---|
| 26 | #define CFA_NOSHARE | 
|---|
| 27 | #define IMPL_CFA | 
|---|
| 28 |  | 
|---|
| 29 | #elif defined IMPL_CFA_LL_NOSHARE | 
|---|
| 30 | #define IMPL_CFA_LL | 
|---|
| 31 | #define CFA_NOSHARE | 
|---|
| 32 | #define IMPL_CFA | 
|---|
| 33 |  | 
|---|
| 34 | #elif defined IMPL_BUHR94 | 
|---|
| 35 | #include <iostream> | 
|---|
| 36 | #include <cstdio> | 
|---|
| 37 | #include "/u0/mlbrooks/usys1/sm/string/StringSharing/src/string.h" | 
|---|
| 38 | #define IMPL_CXX | 
|---|
| 39 |  | 
|---|
| 40 | #else | 
|---|
| 41 | #error Bad IMPL_ | 
|---|
| 42 | #endif | 
|---|
| 43 |  | 
|---|
| 44 |  | 
|---|
| 45 | #if defined IMPL_CFA_HL | 
|---|
| 46 | #include <string.hfa> | 
|---|
| 47 | #elif defined IMPL_CFA_LL | 
|---|
| 48 | #include <string_res.hfa> | 
|---|
| 49 | #endif | 
|---|
| 50 |  | 
|---|
| 51 | #if defined CFA_NOSHARE | 
|---|
| 52 | #include <string_sharectx.hfa> | 
|---|
| 53 | #define STRING_SHARING_CONTROL \ | 
|---|
| 54 | string_sharectx c = { NO_SHARING }; | 
|---|
| 55 | #else | 
|---|
| 56 | #define STRING_SHARING_CONTROL | 
|---|
| 57 | #endif | 
|---|
| 58 |  | 
|---|
| 59 | #if defined IMPL_CFA | 
|---|
| 60 | #include <math.hfa> | 
|---|
| 61 | extern "C" { | 
|---|
| 62 | void malloc_stats( void ); | 
|---|
| 63 | } | 
|---|
| 64 | #elif defined IMPL_CXX | 
|---|
| 65 | #include <algorithm> | 
|---|
| 66 | using std::min; | 
|---|
| 67 | #include <malloc.h> | 
|---|
| 68 | #endif | 
|---|
| 69 |  | 
|---|
| 70 | #include <time.h> | 
|---|
| 71 | #include <stdlib.h> // atoi | 
|---|
| 72 | #include <string.h> // strlen, only during setup | 
|---|
| 73 |  | 
|---|
| 74 | #if defined IMPL_STL || defined IMPL_BUHR94 | 
|---|
| 75 | #define PRINT(s) std::cout << s << std::endl | 
|---|
| 76 | #elif defined IMPL_CFA_HL || defined IMPL_CFA_LL | 
|---|
| 77 | #define PRINT(s) sout | s; | 
|---|
| 78 | #else | 
|---|
| 79 | #error Unhandled print case | 
|---|
| 80 | #endif | 
|---|
| 81 |  | 
|---|
| 82 | #if defined IMPL_CFA_LL | 
|---|
| 83 | #define STRING_T string_res | 
|---|
| 84 | #define ASSIGN_CHAR(str, idx, val) assignAt(str, idx, val) | 
|---|
| 85 | #else | 
|---|
| 86 | #define STRING_T string | 
|---|
| 87 | #define ASSIGN_CHAR(str, idx, val) str[idx] = val | 
|---|
| 88 | #endif | 
|---|
| 89 |  | 
|---|
| 90 | #if defined IMPL_CFA | 
|---|
| 91 | #define LEN(str) size(str) | 
|---|
| 92 | #elif defined IMPL_STL | 
|---|
| 93 | #define LEN(str) str.length() | 
|---|
| 94 | #else | 
|---|
| 95 | #error need LEN definition for this IMPL_ | 
|---|
| 96 | #endif | 
|---|
| 97 |  | 
|---|
| 98 | double meanLen(int N, char ** strings) { | 
|---|
| 99 | int totalLen = 0; | 
|---|
| 100 | for (int i = 0 ; i < N; i ++) { | 
|---|
| 101 | totalLen += strlen(strings[i]); | 
|---|
| 102 | } | 
|---|
| 103 | return (double)totalLen / (double)N; | 
|---|
| 104 | } | 
|---|
| 105 |  | 
|---|
| 106 | volatile int checkthis = 0; | 
|---|
| 107 | #define MAYBE( op ) if (checkthis) { op; } | 
|---|
| 108 |  | 
|---|
| 109 | // setup, not being timed, nor judged for aesthetics | 
|---|
| 110 | void makeSuperCorpus( STRING_T & target, int corpusLen, char ** corpus, int scSize, char scDelimiter ) { | 
|---|
| 111 | char delimiterStr[2] = { scDelimiter, '\0' }; | 
|---|
| 112 | for ( int i = 0;  i < scSize; i++ ) { | 
|---|
| 113 | target += corpus[ i % corpusLen ]; | 
|---|
| 114 | target += delimiterStr; | 
|---|
| 115 | } | 
|---|
| 116 | } | 
|---|
| 117 |  | 
|---|
| 118 | // the function at issue | 
|---|
| 119 |  | 
|---|
| 120 | #if defined IMPL_CFA_HL | 
|---|
| 121 | void processThing( string & thing, char magicChar ) { | 
|---|
| 122 | MAYBE(PRINT(thing)); | 
|---|
| 123 | size_t foundPos; | 
|---|
| 124 | for (;;) { | 
|---|
| 125 | size_t size_thing = size(thing); | 
|---|
| 126 | foundPos = find(thing, magicChar); | 
|---|
| 127 | if( foundPos == size_thing ) break; | 
|---|
| 128 | string mcOccur = thing(foundPos, foundPos+1)`shareEdits; | 
|---|
| 129 | mcOccur = ""; | 
|---|
| 130 | } | 
|---|
| 131 | return thing; | 
|---|
| 132 | } | 
|---|
| 133 |  | 
|---|
| 134 | #elif defined IMPL_CFA_LL | 
|---|
| 135 | void processThing( string_res & thing, char magicChar ) { | 
|---|
| 136 | MAYBE(PRINT(thing)); | 
|---|
| 137 | size_t foundPos; | 
|---|
| 138 | for (;;) { | 
|---|
| 139 | size_t size_thing = size(thing); | 
|---|
| 140 | foundPos = find(thing, magicChar); | 
|---|
| 141 | if( foundPos == size_thing ) break; | 
|---|
| 142 | string_res mcOccur = { thing, SHARE_EDITS, foundPos, foundPos+1 }; | 
|---|
| 143 | mcOccur = ""; | 
|---|
| 144 | } | 
|---|
| 145 | } | 
|---|
| 146 |  | 
|---|
| 147 | #elif defined IMPL_STL | 
|---|
| 148 | void processThing( string & thing, char magicChar ) { | 
|---|
| 149 | MAYBE(PRINT(thing)); | 
|---|
| 150 | size_t foundPos; | 
|---|
| 151 | for (;;) { | 
|---|
| 152 | size_t size_thing = thing.length(); | 
|---|
| 153 | foundPos = thing.find(magicChar); | 
|---|
| 154 | if (foundPos == string::npos) break; | 
|---|
| 155 | thing.erase(foundPos, 1); | 
|---|
| 156 | } | 
|---|
| 157 | } | 
|---|
| 158 |  | 
|---|
| 159 | #endif | 
|---|
| 160 |  | 
|---|
| 161 | /* | 
|---|
| 162 | void runCorrectnessDemo(char** corpus, int corpusLen) { | 
|---|
| 163 | STRING_T item = "asdf";    PRINT( processThing( item, '-' ) ); | 
|---|
| 164 | item = "as-df";            PRINT( processThing( item, '-' ) ); | 
|---|
| 165 | item = "-asdf-";           PRINT( processThing( item, '-' ) ); | 
|---|
| 166 | item = "-";                PRINT( processThing( item, '-' ) ); | 
|---|
| 167 | for ( int i = 0; i < corpusLen; i ++ ) { | 
|---|
| 168 | item = corpus[i];      PRINT( processThing( item, '-' ) ); | 
|---|
| 169 | } | 
|---|
| 170 |  | 
|---|
| 171 | STRING_T supercorpus; | 
|---|
| 172 | makeSuperCorpus( supercorpus, corpusLen, corpus, 30, ','); | 
|---|
| 173 | PRINT( supercorpus ); | 
|---|
| 174 | } | 
|---|
| 175 | */ | 
|---|
| 176 |  | 
|---|
| 177 | int main( int argc, char ** argv ) { | 
|---|
| 178 |  | 
|---|
| 179 | STRING_SHARING_CONTROL | 
|---|
| 180 |  | 
|---|
| 181 | int corpuslen = 0; | 
|---|
| 182 | char ** corpus = (char**) 0; | 
|---|
| 183 |  | 
|---|
| 184 | clock_t start, endTarget, end_actual; | 
|---|
| 185 |  | 
|---|
| 186 | const char * usage_args = "SCsize MagicChar ExecTimeSecs Corpus..."; | 
|---|
| 187 | const int static_arg_posns = 4; | 
|---|
| 188 |  | 
|---|
| 189 | int scSize = -1; | 
|---|
| 190 | char magicChar = '\0'; | 
|---|
| 191 | int execTimeSecs = -1; | 
|---|
| 192 |  | 
|---|
| 193 | switch (min(argc, static_arg_posns)) { | 
|---|
| 194 | case 4: execTimeSecs = atoi(argv[3]); | 
|---|
| 195 | case 3: magicChar = argv[2][0]; | 
|---|
| 196 | case 2: scSize = atoi(argv[1]); | 
|---|
| 197 | } | 
|---|
| 198 |  | 
|---|
| 199 | corpuslen = argc - static_arg_posns; | 
|---|
| 200 | corpus = argv + static_arg_posns; | 
|---|
| 201 |  | 
|---|
| 202 | if (scSize < 1 || magicChar == '\0' || execTimeSecs < 1 || corpuslen < 1) { | 
|---|
| 203 | printf("usage: %s %s\n", argv[0], usage_args); | 
|---|
| 204 | printf("output:\nxxx,corpusItemCount,corpusMeanLenChars,invcationCountActual,execTimeActualSec\n"); | 
|---|
| 205 | exit(1); | 
|---|
| 206 | } | 
|---|
| 207 |  | 
|---|
| 208 | double meanCorpusLen = meanLen(corpuslen, corpus); | 
|---|
| 209 |  | 
|---|
| 210 | //    runCorrectnessDemo(corpus, corpuslen); | 
|---|
| 211 |  | 
|---|
| 212 |  | 
|---|
| 213 | STRING_T supercorpus; | 
|---|
| 214 | makeSuperCorpus( supercorpus, corpuslen, corpus, scSize, ','); | 
|---|
| 215 |  | 
|---|
| 216 | size_t sc_charlen = LEN(supercorpus); | 
|---|
| 217 |  | 
|---|
| 218 | start = clock(); | 
|---|
| 219 | endTarget = start + CLOCKS_PER_SEC * execTimeSecs; | 
|---|
| 220 |  | 
|---|
| 221 | volatile unsigned int t = 0; | 
|---|
| 222 | size_t lastChunkEnd = -1; | 
|---|
| 223 | for ( ; t % 10000 != 0 || clock() < endTarget ; t += 1 ) { | 
|---|
| 224 | #if defined OP_PNO | 
|---|
| 225 |  | 
|---|
| 226 | lastChunkEnd++; | 
|---|
| 227 | if ( lastChunkEnd >= sc_charlen ) { | 
|---|
| 228 | lastChunkEnd = 0; | 
|---|
| 229 | } | 
|---|
| 230 |  | 
|---|
| 231 | #if defined IMPL_CFA_HL | 
|---|
| 232 | size_t thisChunkEnd = findFrom( supercorpus, lastChunkEnd, ',' ); | 
|---|
| 233 | //if (thisChunkEnd == string::npos) throw 1; | 
|---|
| 234 | string chunkProcessed = supercorpus(lastChunkEnd, thisChunkEnd); | 
|---|
| 235 | processThing(chunkProcessed, '-'); | 
|---|
| 236 | #elif defined IMPL_CFA_LL | 
|---|
| 237 | size_t thisChunkEnd = findFrom( supercorpus, lastChunkEnd, ',' ); | 
|---|
| 238 | //if (thisChunkEnd == string::npos) throw 1; | 
|---|
| 239 |  | 
|---|
| 240 | string_res chunkProcessed = { supercorpus, COPY_VALUE, lastChunkEnd, thisChunkEnd }; | 
|---|
| 241 | processThing( chunkProcessed, '-' ); | 
|---|
| 242 | #elif defined IMPL_STL | 
|---|
| 243 | size_t thisChunkEnd = supercorpus.find( ',', lastChunkEnd ); | 
|---|
| 244 | if (thisChunkEnd == string::npos) throw 1; | 
|---|
| 245 | string chunkProcessed = supercorpus.substr(lastChunkEnd, thisChunkEnd-lastChunkEnd); | 
|---|
| 246 | processThing(chunkProcessed, '-'); | 
|---|
| 247 | #else | 
|---|
| 248 | #error Bad IMPL_ | 
|---|
| 249 | #endif | 
|---|
| 250 | MAYBE( PRINT(chunkProcessed) ); | 
|---|
| 251 |  | 
|---|
| 252 | lastChunkEnd = thisChunkEnd; | 
|---|
| 253 | #else | 
|---|
| 254 | #error Bad OP_ | 
|---|
| 255 | #endif | 
|---|
| 256 | } | 
|---|
| 257 | end_actual = clock(); | 
|---|
| 258 | double elapsed = ((double) (end_actual - start)) / CLOCKS_PER_SEC; | 
|---|
| 259 | printf("xxx,%d,%f,%d,%f\n", corpuslen, meanCorpusLen, t, elapsed); | 
|---|
| 260 |  | 
|---|
| 261 | // malloc_stats(); | 
|---|
| 262 |  | 
|---|
| 263 | return 0; | 
|---|
| 264 | } | 
|---|