xdiff/xdiffi.c

   1 /*
   2  *  LibXDiff by Davide Libenzi ( File Differential Library )
   3  *  Copyright (C) 2003  Davide Libenzi
   4  *
   5  *  This library is free software; you can redistribute it and/or
   6  *  modify it under the terms of the GNU Lesser General Public
   7  *  License as published by the Free Software Foundation; either
   8  *  version 2.1 of the License, or (at your option) any later version.
   9  *
  10  *  This library is distributed in the hope that it will be useful,
  11  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  *  Lesser General Public License for more details.
  14  *
  15  *  You should have received a copy of the GNU Lesser General Public
  16  *  License along with this library; if not, write to the Free Software
  17  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  18  *
  19  *  Davide Libenzi <davidel@xmailserver.org>
  20  *
  21  */
  22
  23 #include "xinclude.h"
  24
  25
  26
  27 #define XDL_MAX_COST_MIN 256
  28 #define XDL_HEUR_MIN_COST 256
  29 #define XDL_LINE_MAX (long)((1UL << (8 * sizeof(long) - 1)) - 1)
  30 #define XDL_SNAKE_CNT 20
  31 #define XDL_K_HEUR 4
  32
  33
  34
  35 typedef struct s_xdpsplit {
  36         long i1, i2;
  37         int min_lo, min_hi;
  38 } xdpsplit_t;
  39
  40
  41
  42
  43 static long xdl_split(unsigned long const *ha1, long off1, long lim1,
  44                       unsigned long const *ha2, long off2, long lim2,
  45                       long *kvdf, long *kvdb, int need_min, xdpsplit_t *spl,
  46                       xdalgoenv_t *xenv);
  47 static xdchange_t *xdl_add_change(xdchange_t *xscr, long i1, long i2, long chg1, long chg2);
  48
  49
  50
  51
  52 /*
  53  * See "An O(ND) Difference Algorithm and its Variations", by Eugene Myers.
  54  * Basically considers a "box" (off1, off2, lim1, lim2) and scan from both
  55  * the forward diagonal starting from (off1, off2) and the backward diagonal
  56  * starting from (lim1, lim2). If the K values on the same diagonal crosses
  57  * returns the furthest point of reach. We might end up having to expensive
  58  * cases using this algorithm is full, so a little bit of heuristic is needed
  59  * to cut the search and to return a suboptimal point.
  60  */
  61 static long xdl_split(unsigned long const *ha1, long off1, long lim1,
  62                       unsigned long const *ha2, long off2, long lim2,
  63                       long *kvdf, long *kvdb, int need_min, xdpsplit_t *spl,
  64                       xdalgoenv_t *xenv) {
  65         long dmin = off1 - lim2, dmax = lim1 - off2;
  66         long fmid = off1 - off2, bmid = lim1 - lim2;
  67         long odd = (fmid - bmid) & 1;
  68         long fmin = fmid, fmax = fmid;
  69         long bmin = bmid, bmax = bmid;
  70         long ec, d, i1, i2, prev1, best, dd, v, k;
  71
  72         /*
  73          * Set initial diagonal values for both forward and backward path.
  74          */
  75         kvdf[fmid] = off1;
  76         kvdb[bmid] = lim1;
  77
  78         for (ec = 1;; ec++) {
  79                 int got_snake = 0;
  80
  81                 /*
  82                  * We need to extent the diagonal "domain" by one. If the next
  83                  * values exits the box boundaries we need to change it in the
  84                  * opposite direction because (max - min) must be a power of two.
  85                  * Also we initialize the extenal K value to -1 so that we can
  86                  * avoid extra conditions check inside the core loop.
  87                  */
  88                 if (fmin > dmin)
  89                         kvdf[--fmin - 1] = -1;
  90                 else
  91                         ++fmin;
  92                 if (fmax < dmax)
  93                         kvdf[++fmax + 1] = -1;
  94                 else
  95                         --fmax;
  96
  97                 for (d = fmax; d >= fmin; d -= 2) {
  98                         if (kvdf[d - 1] >= kvdf[d + 1])
  99                                 i1 = kvdf[d - 1] + 1;
 100                         else
 101                                 i1 = kvdf[d + 1];
 102                         prev1 = i1;
 103                         i2 = i1 - d;
 104                         for (; i1 < lim1 && i2 < lim2 && ha1[i1] == ha2[i2]; i1++, i2++);
 105                         if (i1 - prev1 > xenv->snake_cnt)
 106                                 got_snake = 1;
 107                         kvdf[d] = i1;
 108                         if (odd && bmin <= d && d <= bmax && kvdb[d] <= i1) {
 109                                 spl->i1 = i1;
 110                                 spl->i2 = i2;
 111                                 spl->min_lo = spl->min_hi = 1;
 112                                 return ec;
 113                         }
 114                 }
 115
 116                 /*
 117                  * We need to extent the diagonal "domain" by one. If the next
 118                  * values exits the box boundaries we need to change it in the
 119                  * opposite direction because (max - min) must be a power of two.
 120                  * Also we initialize the extenal K value to -1 so that we can
 121                  * avoid extra conditions check inside the core loop.
 122                  */
 123                 if (bmin > dmin)
 124                         kvdb[--bmin - 1] = XDL_LINE_MAX;
 125                 else
 126                         ++bmin;
 127                 if (bmax < dmax)
 128                         kvdb[++bmax + 1] = XDL_LINE_MAX;
 129                 else
 130                         --bmax;
 131
 132                 for (d = bmax; d >= bmin; d -= 2) {
 133                         if (kvdb[d - 1] < kvdb[d + 1])
 134                                 i1 = kvdb[d - 1];
 135                         else
 136                                 i1 = kvdb[d + 1] - 1;
 137                         prev1 = i1;
 138                         i2 = i1 - d;
 139                         for (; i1 > off1 && i2 > off2 && ha1[i1 - 1] == ha2[i2 - 1]; i1--, i2--);
 140                         if (prev1 - i1 > xenv->snake_cnt)
 141                                 got_snake = 1;
 142                         kvdb[d] = i1;
 143                         if (!odd && fmin <= d && d <= fmax && i1 <= kvdf[d]) {
 144                                 spl->i1 = i1;
 145                                 spl->i2 = i2;
 146                                 spl->min_lo = spl->min_hi = 1;
 147                                 return ec;
 148                         }
 149                 }
 150
 151                 if (need_min)
 152                         continue;
 153
 154                 /*
 155                  * If the edit cost is above the heuristic trigger and if
 156                  * we got a good snake, we sample current diagonals to see
 157                  * if some of the, have reached an "interesting" path. Our
 158                  * measure is a function of the distance from the diagonal
 159                  * corner (i1 + i2) penalized with the distance from the
 160                  * mid diagonal itself. If this value is above the current
 161                  * edit cost times a magic factor (XDL_K_HEUR) we consider
 162                  * it interesting.
 163                  */
 164                 if (got_snake && ec > xenv->heur_min) {
 165                         for (best = 0, d = fmax; d >= fmin; d -= 2) {
 166                                 dd = d > fmid ? d - fmid: fmid - d;
 167                                 i1 = kvdf[d];
 168                                 i2 = i1 - d;
 169                                 v = (i1 - off1) + (i2 - off2) - dd;
 170
 171                                 if (v > XDL_K_HEUR * ec && v > best &&
 172                                     off1 + xenv->snake_cnt <= i1 && i1 < lim1 &&
 173                                     off2 + xenv->snake_cnt <= i2 && i2 < lim2) {
 174                                         for (k = 1; ha1[i1 - k] == ha2[i2 - k]; k++)
 175                                                 if (k == xenv->snake_cnt) {
 176                                                         best = v;
 177                                                         spl->i1 = i1;
 178                                                         spl->i2 = i2;
 179                                                         break;
 180                                                 }
 181                                 }
 182                         }
 183                         if (best > 0) {
 184                                 spl->min_lo = 1;
 185                                 spl->min_hi = 0;
 186                                 return ec;
 187                         }
 188
 189                         for (best = 0, d = bmax; d >= bmin; d -= 2) {
 190                                 dd = d > bmid ? d - bmid: bmid - d;
 191                                 i1 = kvdb[d];
 192                                 i2 = i1 - d;
 193                                 v = (lim1 - i1) + (lim2 - i2) - dd;
 194
 195                                 if (v > XDL_K_HEUR * ec && v > best &&
 196                                     off1 < i1 && i1 <= lim1 - xenv->snake_cnt &&
 197                                     off2 < i2 && i2 <= lim2 - xenv->snake_cnt) {
 198                                         for (k = 0; ha1[i1 + k] == ha2[i2 + k]; k++)
 199                                                 if (k == xenv->snake_cnt - 1) {
 200                                                         best = v;
 201                                                         spl->i1 = i1;
 202                                                         spl->i2 = i2;
 203                                                         break;
 204                                                 }
 205                                 }
 206                         }
 207                         if (best > 0) {
 208                                 spl->min_lo = 0;
 209                                 spl->min_hi = 1;
 210                                 return ec;
 211                         }
 212                 }
 213
 214                 /*
 215                  * Enough is enough. We spent too much time here and now we collect
 216                  * the furthest reaching path using the (i1 + i2) measure.
 217                  */
 218                 if (ec >= xenv->mxcost) {
 219                         long fbest, fbest1, bbest, bbest1;
 220
 221                         fbest = -1;
 222                         for (d = fmax; d >= fmin; d -= 2) {
 223                                 i1 = XDL_MIN(kvdf[d], lim1);
 224                                 i2 = i1 - d;
 225                                 if (lim2 < i2)
 226                                         i1 = lim2 + d, i2 = lim2;
 227                                 if (fbest < i1 + i2) {
 228                                         fbest = i1 + i2;
 229                                         fbest1 = i1;
 230                                 }
 231                         }
 232
 233                         bbest = XDL_LINE_MAX;
 234                         for (d = bmax; d >= bmin; d -= 2) {
 235                                 i1 = XDL_MAX(off1, kvdb[d]);
 236                                 i2 = i1 - d;
 237                                 if (i2 < off2)
 238                                         i1 = off2 + d, i2 = off2;
 239                                 if (i1 + i2 < bbest) {
 240                                         bbest = i1 + i2;
 241                                         bbest1 = i1;
 242                                 }
 243                         }
 244
 245                         if ((lim1 + lim2) - bbest < fbest - (off1 + off2)) {
 246                                 spl->i1 = fbest1;
 247                                 spl->i2 = fbest - fbest1;
 248                                 spl->min_lo = 1;
 249                                 spl->min_hi = 0;
 250                         } else {
 251                                 spl->i1 = bbest1;
 252                                 spl->i2 = bbest - bbest1;
 253                                 spl->min_lo = 0;
 254                                 spl->min_hi = 1;
 255                         }
 256                         return ec;
 257                 }
 258         }
 259
 260         return -1;
 261 }
 262
 263
 264 /*
 265  * Rule: "Divide et Impera". Recursively split the box in sub-boxes by calling
 266  * the box splitting function. Note that the real job (marking changed lines)
 267  * is done in the two boundary reaching checks.
 268  */
 269 int xdl_recs_cmp(diffdata_t *dd1, long off1, long lim1,
 270                  diffdata_t *dd2, long off2, long lim2,
 271                  long *kvdf, long *kvdb, int need_min, xdalgoenv_t *xenv) {
 272         unsigned long const *ha1 = dd1->ha, *ha2 = dd2->ha;
 273
 274         /*
 275          * Shrink the box by walking through each diagonal snake (SW and NE).
 276          */
 277         for (; off1 < lim1 && off2 < lim2 && ha1[off1] == ha2[off2]; off1++, off2++);
 278         for (; off1 < lim1 && off2 < lim2 && ha1[lim1 - 1] == ha2[lim2 - 1]; lim1--, lim2--);
 279
 280         /*
 281          * If one dimension is empty, then all records on the other one must
 282          * be obviously changed.
 283          */
 284         if (off1 == lim1) {
 285                 char *rchg2 = dd2->rchg;
 286                 long *rindex2 = dd2->rindex;
 287
 288                 for (; off2 < lim2; off2++)
 289                         rchg2[rindex2[off2]] = 1;
 290         } else if (off2 == lim2) {
 291                 char *rchg1 = dd1->rchg;
 292                 long *rindex1 = dd1->rindex;
 293
 294                 for (; off1 < lim1; off1++)
 295                         rchg1[rindex1[off1]] = 1;
 296         } else {
 297                 long ec;
 298                 xdpsplit_t spl;
 299
 300                 /*
 301                  * Divide ...
 302                  */
 303                 if ((ec = xdl_split(ha1, off1, lim1, ha2, off2, lim2, kvdf, kvdb,
 304                                     need_min, &spl, xenv)) < 0) {
 305
 306                         return -1;
 307                 }
 308
 309                 /*
 310                  * ... et Impera.
 311                  */
 312                 if (xdl_recs_cmp(dd1, off1, spl.i1, dd2, off2, spl.i2,
 313                                  kvdf, kvdb, spl.min_lo, xenv) < 0 ||
 314                     xdl_recs_cmp(dd1, spl.i1, lim1, dd2, spl.i2, lim2,
 315                                  kvdf, kvdb, spl.min_hi, xenv) < 0) {
 316
 317                         return -1;
 318                 }
 319         }
 320
 321         return 0;
 322 }
 323
 324
 325 int xdl_do_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
 326                 xdfenv_t *xe) {
 327         long ndiags;
 328         long *kvd, *kvdf, *kvdb;
 329         xdalgoenv_t xenv;
 330         diffdata_t dd1, dd2;
 331
 332         if (xdl_prepare_env(mf1, mf2, xpp, xe) < 0) {
 333
 334                 return -1;
 335         }
 336
 337         /*
 338          * Allocate and setup K vectors to be used by the differential algorithm.
 339          * One is to store the forward path and one to store the backward path.
 340          */
 341         ndiags = xe->xdf1.nreff + xe->xdf2.nreff + 3;
 342         if (!(kvd = (long *) xdl_malloc((2 * ndiags + 2) * sizeof(long)))) {
 343
 344                 xdl_free_env(xe);
 345                 return -1;
 346         }
 347         kvdf = kvd;
 348         kvdb = kvdf + ndiags;
 349         kvdf += xe->xdf2.nreff + 1;
 350         kvdb += xe->xdf2.nreff + 1;
 351
 352         xenv.mxcost = xdl_bogosqrt(ndiags);
 353         if (xenv.mxcost < XDL_MAX_COST_MIN)
 354                 xenv.mxcost = XDL_MAX_COST_MIN;
 355         xenv.snake_cnt = XDL_SNAKE_CNT;
 356         xenv.heur_min = XDL_HEUR_MIN_COST;
 357
 358         dd1.nrec = xe->xdf1.nreff;
 359         dd1.ha = xe->xdf1.ha;
 360         dd1.rchg = xe->xdf1.rchg;
 361         dd1.rindex = xe->xdf1.rindex;
 362         dd2.nrec = xe->xdf2.nreff;
 363         dd2.ha = xe->xdf2.ha;
 364         dd2.rchg = xe->xdf2.rchg;
 365         dd2.rindex = xe->xdf2.rindex;
 366
 367         if (xdl_recs_cmp(&dd1, 0, dd1.nrec, &dd2, 0, dd2.nrec,
 368                          kvdf, kvdb, (xpp->flags & XDF_NEED_MINIMAL) != 0, &xenv) < 0) {
 369
 370                 xdl_free(kvd);
 371                 xdl_free_env(xe);
 372                 return -1;
 373         }
 374
 375         xdl_free(kvd);
 376
 377         return 0;
 378 }
 379
 380
 381 static xdchange_t *xdl_add_change(xdchange_t *xscr, long i1, long i2, long chg1, long chg2) {
 382         xdchange_t *xch;
 383
 384         if (!(xch = (xdchange_t *) xdl_malloc(sizeof(xdchange_t))))
 385                 return NULL;
 386
 387         xch->next = xscr;
 388         xch->i1 = i1;
 389         xch->i2 = i2;
 390         xch->chg1 = chg1;
 391         xch->chg2 = chg2;
 392
 393         return xch;
 394 }
 395
 396
 397 int xdl_build_script(xdfenv_t *xe, xdchange_t **xscr) {
 398         xdchange_t *cscr = NULL, *xch;
 399         char *rchg1 = xe->xdf1.rchg, *rchg2 = xe->xdf2.rchg;
 400         long i1, i2, l1, l2;
 401
 402         /*
 403          * Trivial. Collects "groups" of changes and creates an edit script.
 404          */
 405         for (i1 = xe->xdf1.nrec, i2 = xe->xdf2.nrec; i1 >= 0 || i2 >= 0; i1--, i2--)
 406                 if (rchg1[i1 - 1] || rchg2[i2 - 1]) {
 407                         for (l1 = i1; rchg1[i1 - 1]; i1--);
 408                         for (l2 = i2; rchg2[i2 - 1]; i2--);
 409
 410                         if (!(xch = xdl_add_change(cscr, i1, i2, l1 - i1, l2 - i2))) {
 411                                 xdl_free_script(cscr);
 412                                 return -1;
 413                         }
 414                         cscr = xch;
 415                 }
 416
 417         *xscr = cscr;
 418
 419         return 0;
 420 }
 421
 422
 423 void xdl_free_script(xdchange_t *xscr) {
 424         xdchange_t *xch;
 425
 426         while ((xch = xscr) != NULL) {
 427                 xscr = xscr->next;
 428                 xdl_free(xch);
 429         }
 430 }
 431
 432
 433 int xdl_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
 434              xdemitconf_t const *xecfg, xdemitcb_t *ecb) {
 435         xdchange_t *xscr;
 436         xdfenv_t xe;
 437
 438         if (xdl_do_diff(mf1, mf2, xpp, &xe) < 0) {
 439
 440                 return -1;
 441         }
 442
 443         if (xdl_build_script(&xe, &xscr) < 0) {
 444
 445                 xdl_free_env(&xe);
 446                 return -1;
 447         }
 448
 449         if (xscr) {
 450                 if (xdl_emit_diff(&xe, xscr, ecb, xecfg) < 0) {
 451
 452                         xdl_free_script(xscr);
 453                         xdl_free_env(&xe);
 454                         return -1;
 455                 }
 456
 457                 xdl_free_script(xscr);
 458         }
 459
 460         xdl_free_env(&xe);
 461
 462         return 0;
 463 }
 464