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 static int xdl_change_compact(xdfile_t *xdf, xdfile_t *xdfo);
  49
  50
  51
  52
  53
  54 /*
  55  * See "An O(ND) Difference Algorithm and its Variations", by Eugene Myers.
  56  * Basically considers a "box" (off1, off2, lim1, lim2) and scan from both
  57  * the forward diagonal starting from (off1, off2) and the backward diagonal
  58  * starting from (lim1, lim2). If the K values on the same diagonal crosses
  59  * returns the furthest point of reach. We might end up having to expensive
  60  * cases using this algorithm is full, so a little bit of heuristic is needed
  61  * to cut the search and to return a suboptimal point.
  62  */
  63 static long xdl_split(unsigned long const *ha1, long off1, long lim1,
  64                       unsigned long const *ha2, long off2, long lim2,
  65                       long *kvdf, long *kvdb, int need_min, xdpsplit_t *spl,
  66                       xdalgoenv_t *xenv) {
  67         long dmin = off1 - lim2, dmax = lim1 - off2;
  68         long fmid = off1 - off2, bmid = lim1 - lim2;
  69         long odd = (fmid - bmid) & 1;
  70         long fmin = fmid, fmax = fmid;
  71         long bmin = bmid, bmax = bmid;
  72         long ec, d, i1, i2, prev1, best, dd, v, k;
  73
  74         /*
  75          * Set initial diagonal values for both forward and backward path.
  76          */
  77         kvdf[fmid] = off1;
  78         kvdb[bmid] = lim1;
  79
  80         for (ec = 1;; ec++) {
  81                 int got_snake = 0;
  82
  83                 /*
  84                  * We need to extent the diagonal "domain" by one. If the next
  85                  * values exits the box boundaries we need to change it in the
  86                  * opposite direction because (max - min) must be a power of two.
  87                  * Also we initialize the extenal K value to -1 so that we can
  88                  * avoid extra conditions check inside the core loop.
  89                  */
  90                 if (fmin > dmin)
  91                         kvdf[--fmin - 1] = -1;
  92                 else
  93                         ++fmin;
  94                 if (fmax < dmax)
  95                         kvdf[++fmax + 1] = -1;
  96                 else
  97                         --fmax;
  98
  99                 for (d = fmax; d >= fmin; d -= 2) {
 100                         if (kvdf[d - 1] >= kvdf[d + 1])
 101                                 i1 = kvdf[d - 1] + 1;
 102                         else
 103                                 i1 = kvdf[d + 1];
 104                         prev1 = i1;
 105                         i2 = i1 - d;
 106                         for (; i1 < lim1 && i2 < lim2 && ha1[i1] == ha2[i2]; i1++, i2++);
 107                         if (i1 - prev1 > xenv->snake_cnt)
 108                                 got_snake = 1;
 109                         kvdf[d] = i1;
 110                         if (odd && bmin <= d && d <= bmax && kvdb[d] <= i1) {
 111                                 spl->i1 = i1;
 112                                 spl->i2 = i2;
 113                                 spl->min_lo = spl->min_hi = 1;
 114                                 return ec;
 115                         }
 116                 }
 117
 118                 /*
 119                  * We need to extent the diagonal "domain" by one. If the next
 120                  * values exits the box boundaries we need to change it in the
 121                  * opposite direction because (max - min) must be a power of two.
 122                  * Also we initialize the extenal K value to -1 so that we can
 123                  * avoid extra conditions check inside the core loop.
 124                  */
 125                 if (bmin > dmin)
 126                         kvdb[--bmin - 1] = XDL_LINE_MAX;
 127                 else
 128                         ++bmin;
 129                 if (bmax < dmax)
 130                         kvdb[++bmax + 1] = XDL_LINE_MAX;
 131                 else
 132                         --bmax;
 133
 134                 for (d = bmax; d >= bmin; d -= 2) {
 135                         if (kvdb[d - 1] < kvdb[d + 1])
 136                                 i1 = kvdb[d - 1];
 137                         else
 138                                 i1 = kvdb[d + 1] - 1;
 139                         prev1 = i1;
 140                         i2 = i1 - d;
 141                         for (; i1 > off1 && i2 > off2 && ha1[i1 - 1] == ha2[i2 - 1]; i1--, i2--);
 142                         if (prev1 - i1 > xenv->snake_cnt)
 143                                 got_snake = 1;
 144                         kvdb[d] = i1;
 145                         if (!odd && fmin <= d && d <= fmax && i1 <= kvdf[d]) {
 146                                 spl->i1 = i1;
 147                                 spl->i2 = i2;
 148                                 spl->min_lo = spl->min_hi = 1;
 149                                 return ec;
 150                         }
 151                 }
 152
 153                 if (need_min)
 154                         continue;
 155
 156                 /*
 157                  * If the edit cost is above the heuristic trigger and if
 158                  * we got a good snake, we sample current diagonals to see
 159                  * if some of the, have reached an "interesting" path. Our
 160                  * measure is a function of the distance from the diagonal
 161                  * corner (i1 + i2) penalized with the distance from the
 162                  * mid diagonal itself. If this value is above the current
 163                  * edit cost times a magic factor (XDL_K_HEUR) we consider
 164                  * it interesting.
 165                  */
 166                 if (got_snake && ec > xenv->heur_min) {
 167                         for (best = 0, d = fmax; d >= fmin; d -= 2) {
 168                                 dd = d > fmid ? d - fmid: fmid - d;
 169                                 i1 = kvdf[d];
 170                                 i2 = i1 - d;
 171                                 v = (i1 - off1) + (i2 - off2) - dd;
 172
 173                                 if (v > XDL_K_HEUR * ec && v > best &&
 174                                     off1 + xenv->snake_cnt <= i1 && i1 < lim1 &&
 175                                     off2 + xenv->snake_cnt <= i2 && i2 < lim2) {
 176                                         for (k = 1; ha1[i1 - k] == ha2[i2 - k]; k++)
 177                                                 if (k == xenv->snake_cnt) {
 178                                                         best = v;
 179                                                         spl->i1 = i1;
 180                                                         spl->i2 = i2;
 181                                                         break;
 182                                                 }
 183                                 }
 184                         }
 185                         if (best > 0) {
 186                                 spl->min_lo = 1;
 187                                 spl->min_hi = 0;
 188                                 return ec;
 189                         }
 190
 191                         for (best = 0, d = bmax; d >= bmin; d -= 2) {
 192                                 dd = d > bmid ? d - bmid: bmid - d;
 193                                 i1 = kvdb[d];
 194                                 i2 = i1 - d;
 195                                 v = (lim1 - i1) + (lim2 - i2) - dd;
 196
 197                                 if (v > XDL_K_HEUR * ec && v > best &&
 198                                     off1 < i1 && i1 <= lim1 - xenv->snake_cnt &&
 199                                     off2 < i2 && i2 <= lim2 - xenv->snake_cnt) {
 200                                         for (k = 0; ha1[i1 + k] == ha2[i2 + k]; k++)
 201                                                 if (k == xenv->snake_cnt - 1) {
 202                                                         best = v;
 203                                                         spl->i1 = i1;
 204                                                         spl->i2 = i2;
 205                                                         break;
 206                                                 }
 207                                 }
 208                         }
 209                         if (best > 0) {
 210                                 spl->min_lo = 0;
 211                                 spl->min_hi = 1;
 212                                 return ec;
 213                         }
 214                 }
 215
 216                 /*
 217                  * Enough is enough. We spent too much time here and now we collect
 218                  * the furthest reaching path using the (i1 + i2) measure.
 219                  */
 220                 if (ec >= xenv->mxcost) {
 221                         long fbest, fbest1, bbest, bbest1;
 222
 223                         fbest = fbest1 = -1;
 224                         for (d = fmax; d >= fmin; d -= 2) {
 225                                 i1 = XDL_MIN(kvdf[d], lim1);
 226                                 i2 = i1 - d;
 227                                 if (lim2 < i2)
 228                                         i1 = lim2 + d, i2 = lim2;
 229                                 if (fbest < i1 + i2) {
 230                                         fbest = i1 + i2;
 231                                         fbest1 = i1;
 232                                 }
 233                         }
 234
 235                         bbest = bbest1 = XDL_LINE_MAX;
 236                         for (d = bmax; d >= bmin; d -= 2) {
 237                                 i1 = XDL_MAX(off1, kvdb[d]);
 238                                 i2 = i1 - d;
 239                                 if (i2 < off2)
 240                                         i1 = off2 + d, i2 = off2;
 241                                 if (i1 + i2 < bbest) {
 242                                         bbest = i1 + i2;
 243                                         bbest1 = i1;
 244                                 }
 245                         }
 246
 247                         if ((lim1 + lim2) - bbest < fbest - (off1 + off2)) {
 248                                 spl->i1 = fbest1;
 249                                 spl->i2 = fbest - fbest1;
 250                                 spl->min_lo = 1;
 251                                 spl->min_hi = 0;
 252                         } else {
 253                                 spl->i1 = bbest1;
 254                                 spl->i2 = bbest - bbest1;
 255                                 spl->min_lo = 0;
 256                                 spl->min_hi = 1;
 257                         }
 258                         return ec;
 259                 }
 260         }
 261
 262         return -1;
 263 }
 264
 265
 266 /*
 267  * Rule: "Divide et Impera". Recursively split the box in sub-boxes by calling
 268  * the box splitting function. Note that the real job (marking changed lines)
 269  * is done in the two boundary reaching checks.
 270  */
 271 int xdl_recs_cmp(diffdata_t *dd1, long off1, long lim1,
 272                  diffdata_t *dd2, long off2, long lim2,
 273                  long *kvdf, long *kvdb, int need_min, xdalgoenv_t *xenv) {
 274         unsigned long const *ha1 = dd1->ha, *ha2 = dd2->ha;
 275
 276         /*
 277          * Shrink the box by walking through each diagonal snake (SW and NE).
 278          */
 279         for (; off1 < lim1 && off2 < lim2 && ha1[off1] == ha2[off2]; off1++, off2++);
 280         for (; off1 < lim1 && off2 < lim2 && ha1[lim1 - 1] == ha2[lim2 - 1]; lim1--, lim2--);
 281
 282         /*
 283          * If one dimension is empty, then all records on the other one must
 284          * be obviously changed.
 285          */
 286         if (off1 == lim1) {
 287                 char *rchg2 = dd2->rchg;
 288                 long *rindex2 = dd2->rindex;
 289
 290                 for (; off2 < lim2; off2++)
 291                         rchg2[rindex2[off2]] = 1;
 292         } else if (off2 == lim2) {
 293                 char *rchg1 = dd1->rchg;
 294                 long *rindex1 = dd1->rindex;
 295
 296                 for (; off1 < lim1; off1++)
 297                         rchg1[rindex1[off1]] = 1;
 298         } else {
 299                 long ec;
 300                 xdpsplit_t spl;
 301                 spl.i1 = spl.i2 = 0;
 302
 303                 /*
 304                  * Divide ...
 305                  */
 306                 if ((ec = xdl_split(ha1, off1, lim1, ha2, off2, lim2, kvdf, kvdb,
 307                                     need_min, &spl, xenv)) < 0) {
 308
 309                         return -1;
 310                 }
 311
 312                 /*
 313                  * ... et Impera.
 314                  */
 315                 if (xdl_recs_cmp(dd1, off1, spl.i1, dd2, off2, spl.i2,
 316                                  kvdf, kvdb, spl.min_lo, xenv) < 0 ||
 317                     xdl_recs_cmp(dd1, spl.i1, lim1, dd2, spl.i2, lim2,
 318                                  kvdf, kvdb, spl.min_hi, xenv) < 0) {
 319
 320                         return -1;
 321                 }
 322         }
 323
 324         return 0;
 325 }
 326
 327
 328 int xdl_do_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
 329                 xdfenv_t *xe) {
 330         long ndiags;
 331         long *kvd, *kvdf, *kvdb;
 332         xdalgoenv_t xenv;
 333         diffdata_t dd1, dd2;
 334
 335         if (xdl_prepare_env(mf1, mf2, xpp, xe) < 0) {
 336
 337                 return -1;
 338         }
 339
 340         /*
 341          * Allocate and setup K vectors to be used by the differential algorithm.
 342          * One is to store the forward path and one to store the backward path.
 343          */
 344         ndiags = xe->xdf1.nreff + xe->xdf2.nreff + 3;
 345         if (!(kvd = (long *) xdl_malloc((2 * ndiags + 2) * sizeof(long)))) {
 346
 347                 xdl_free_env(xe);
 348                 return -1;
 349         }
 350         kvdf = kvd;
 351         kvdb = kvdf + ndiags;
 352         kvdf += xe->xdf2.nreff + 1;
 353         kvdb += xe->xdf2.nreff + 1;
 354
 355         xenv.mxcost = xdl_bogosqrt(ndiags);
 356         if (xenv.mxcost < XDL_MAX_COST_MIN)
 357                 xenv.mxcost = XDL_MAX_COST_MIN;
 358         xenv.snake_cnt = XDL_SNAKE_CNT;
 359         xenv.heur_min = XDL_HEUR_MIN_COST;
 360
 361         dd1.nrec = xe->xdf1.nreff;
 362         dd1.ha = xe->xdf1.ha;
 363         dd1.rchg = xe->xdf1.rchg;
 364         dd1.rindex = xe->xdf1.rindex;
 365         dd2.nrec = xe->xdf2.nreff;
 366         dd2.ha = xe->xdf2.ha;
 367         dd2.rchg = xe->xdf2.rchg;
 368         dd2.rindex = xe->xdf2.rindex;
 369
 370         if (xdl_recs_cmp(&dd1, 0, dd1.nrec, &dd2, 0, dd2.nrec,
 371                          kvdf, kvdb, (xpp->flags & XDF_NEED_MINIMAL) != 0, &xenv) < 0) {
 372
 373                 xdl_free(kvd);
 374                 xdl_free_env(xe);
 375                 return -1;
 376         }
 377
 378         xdl_free(kvd);
 379
 380         return 0;
 381 }
 382
 383
 384 static xdchange_t *xdl_add_change(xdchange_t *xscr, long i1, long i2, long chg1, long chg2) {
 385         xdchange_t *xch;
 386
 387         if (!(xch = (xdchange_t *) xdl_malloc(sizeof(xdchange_t))))
 388                 return NULL;
 389
 390         xch->next = xscr;
 391         xch->i1 = i1;
 392         xch->i2 = i2;
 393         xch->chg1 = chg1;
 394         xch->chg2 = chg2;
 395
 396         return xch;
 397 }
 398
 399
 400 static int xdl_change_compact(xdfile_t *xdf, xdfile_t *xdfo) {
 401         long ix, ixo, ixs, ixref, grpsiz, nrec = xdf->nrec;
 402         char *rchg = xdf->rchg, *rchgo = xdfo->rchg;
 403         xrecord_t **recs = xdf->recs;
 404
 405         /*
 406          * This is the same of what GNU diff does. Move back and forward
 407          * change groups for a consistent and pretty diff output. This also
 408          * helps in finding joineable change groups and reduce the diff size.
 409          */
 410         for (ix = ixo = 0;;) {
 411                 /*
 412                  * Find the first changed line in the to-be-compacted file.
 413                  * We need to keep track of both indexes, so if we find a
 414                  * changed lines group on the other file, while scanning the
 415                  * to-be-compacted file, we need to skip it properly. Note
 416                  * that loops that are testing for changed lines on rchg* do
 417                  * not need index bounding since the array is prepared with
 418                  * a zero at position -1 and N.
 419                  */
 420                 for (; ix < nrec && !rchg[ix]; ix++)
 421                         while (rchgo[ixo++]);
 422                 if (ix == nrec)
 423                         break;
 424
 425                 /*
 426                  * Record the start of a changed-group in the to-be-compacted file
 427                  * and find the end of it, on both to-be-compacted and other file
 428                  * indexes (ix and ixo).
 429                  */
 430                 ixs = ix;
 431                 for (ix++; rchg[ix]; ix++);
 432                 for (; rchgo[ixo]; ixo++);
 433
 434                 do {
 435                         grpsiz = ix - ixs;
 436
 437                         /*
 438                          * If the line before the current change group, is equal to
 439                          * the last line of the current change group, shift backward
 440                          * the group.
 441                          */
 442                         while (ixs > 0 && recs[ixs - 1]->ha == recs[ix - 1]->ha &&
 443                                XDL_RECMATCH(recs[ixs - 1], recs[ix - 1])) {
 444                                 rchg[--ixs] = 1;
 445                                 rchg[--ix] = 0;
 446
 447                                 /*
 448                                  * This change might have joined two change groups,
 449                                  * so we try to take this scenario in account by moving
 450                                  * the start index accordingly (and so the other-file
 451                                  * end-of-group index).
 452                                  */
 453                                 for (; rchg[ixs - 1]; ixs--);
 454                                 while (rchgo[--ixo]);
 455                         }
 456
 457                         /*
 458                          * Record the end-of-group position in case we are matched
 459                          * with a group of changes in the other file (that is, the
 460                          * change record before the enf-of-group index in the other
 461                          * file is set).
 462                          */
 463                         ixref = rchgo[ixo - 1] ? ix: nrec;
 464
 465                         /*
 466                          * If the first line of the current change group, is equal to
 467                          * the line next of the current change group, shift forward
 468                          * the group.
 469                          */
 470                         while (ix < nrec && recs[ixs]->ha == recs[ix]->ha &&
 471                                XDL_RECMATCH(recs[ixs], recs[ix])) {
 472                                 rchg[ixs++] = 0;
 473                                 rchg[ix++] = 1;
 474
 475                                 /*
 476                                  * This change might have joined two change groups,
 477                                  * so we try to take this scenario in account by moving
 478                                  * the start index accordingly (and so the other-file
 479                                  * end-of-group index). Keep tracking the reference
 480                                  * index in case we are shifting together with a
 481                                  * corresponding group of changes in the other file.
 482                                  */
 483                                 for (; rchg[ix]; ix++);
 484                                 while (rchgo[++ixo])
 485                                         ixref = ix;
 486                         }
 487                 } while (grpsiz != ix - ixs);
 488
 489                 /*
 490                  * Try to move back the possibly merged group of changes, to match
 491                  * the recorded postion in the other file.
 492                  */
 493                 while (ixref < ix) {
 494                         rchg[--ixs] = 1;
 495                         rchg[--ix] = 0;
 496                         while (rchgo[--ixo]);
 497                 }
 498         }
 499
 500         return 0;
 501 }
 502
 503
 504 int xdl_build_script(xdfenv_t *xe, xdchange_t **xscr) {
 505         xdchange_t *cscr = NULL, *xch;
 506         char *rchg1 = xe->xdf1.rchg, *rchg2 = xe->xdf2.rchg;
 507         long i1, i2, l1, l2;
 508
 509         /*
 510          * Trivial. Collects "groups" of changes and creates an edit script.
 511          */
 512         for (i1 = xe->xdf1.nrec, i2 = xe->xdf2.nrec; i1 >= 0 || i2 >= 0; i1--, i2--)
 513                 if (rchg1[i1 - 1] || rchg2[i2 - 1]) {
 514                         for (l1 = i1; rchg1[i1 - 1]; i1--);
 515                         for (l2 = i2; rchg2[i2 - 1]; i2--);
 516
 517                         if (!(xch = xdl_add_change(cscr, i1, i2, l1 - i1, l2 - i2))) {
 518                                 xdl_free_script(cscr);
 519                                 return -1;
 520                         }
 521                         cscr = xch;
 522                 }
 523
 524         *xscr = cscr;
 525
 526         return 0;
 527 }
 528
 529
 530 void xdl_free_script(xdchange_t *xscr) {
 531         xdchange_t *xch;
 532
 533         while ((xch = xscr) != NULL) {
 534                 xscr = xscr->next;
 535                 xdl_free(xch);
 536         }
 537 }
 538
 539
 540 int xdl_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
 541              xdemitconf_t const *xecfg, xdemitcb_t *ecb) {
 542         xdchange_t *xscr;
 543         xdfenv_t xe;
 544
 545         if (xdl_do_diff(mf1, mf2, xpp, &xe) < 0) {
 546
 547                 return -1;
 548         }
 549         if (xdl_change_compact(&xe.xdf1, &xe.xdf2) < 0 ||
 550             xdl_change_compact(&xe.xdf2, &xe.xdf1) < 0 ||
 551             xdl_build_script(&xe, &xscr) < 0) {
 552
 553                 xdl_free_env(&xe);
 554                 return -1;
 555         }
 556         if (xscr) {
 557                 if (xdl_emit_diff(&xe, xscr, ecb, xecfg) < 0) {
 558
 559                         xdl_free_script(xscr);
 560                         xdl_free_env(&xe);
 561                         return -1;
 562                 }
 563                 xdl_free_script(xscr);
 564         }
 565         xdl_free_env(&xe);
 566
 567         return 0;
 568 }
 569