xdiff/xmerge.c

   1 /*
   2  *  LibXDiff by Davide Libenzi ( File Differential Library )
   3  *  Copyright (C) 2003-2006 Davide Libenzi, Johannes E. Schindelin
   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 typedef struct s_xdmerge {
  26         struct s_xdmerge *next;
  27         /*
  28          * 0 = conflict,
  29          * 1 = no conflict, take first,
  30          * 2 = no conflict, take second.
  31          */
  32         int mode;
  33         /*
  34          * These point at the respective postimages.  E.g. <i1,chg1> is
  35          * how side #1 wants to change the common ancestor; if there is no
  36          * overlap, lines before i1 in the postimage of side #1 appear
  37          * in the merge result as a region touched by neither side.
  38          */
  39         long i1, i2;
  40         long chg1, chg2;
  41         /*
  42          * These point at the preimage; of course there is just one
  43          * preimage, that is from the shared common ancestor.
  44          */
  45         long i0;
  46         long chg0;
  47 } xdmerge_t;
  48
  49 static int xdl_append_merge(xdmerge_t **merge, int mode,
  50                             long i0, long chg0,
  51                             long i1, long chg1,
  52                             long i2, long chg2)
  53 {
  54         xdmerge_t *m = *merge;
  55         if (m && (i1 <= m->i1 + m->chg1 || i2 <= m->i2 + m->chg2)) {
  56                 if (mode != m->mode)
  57                         m->mode = 0;
  58                 m->chg0 = i0 + chg0 - m->i0;
  59                 m->chg1 = i1 + chg1 - m->i1;
  60                 m->chg2 = i2 + chg2 - m->i2;
  61         } else {
  62                 m = xdl_malloc(sizeof(xdmerge_t));
  63                 if (!m)
  64                         return -1;
  65                 m->next = NULL;
  66                 m->mode = mode;
  67                 m->i0 = i0;
  68                 m->chg0 = chg0;
  69                 m->i1 = i1;
  70                 m->chg1 = chg1;
  71                 m->i2 = i2;
  72                 m->chg2 = chg2;
  73                 if (*merge)
  74                         (*merge)->next = m;
  75                 *merge = m;
  76         }
  77         return 0;
  78 }
  79
  80 static int xdl_cleanup_merge(xdmerge_t *c)
  81 {
  82         int count = 0;
  83         xdmerge_t *next_c;
  84
  85         /* were there conflicts? */
  86         for (; c; c = next_c) {
  87                 if (c->mode == 0)
  88                         count++;
  89                 next_c = c->next;
  90                 free(c);
  91         }
  92         return count;
  93 }
  94
  95 static int xdl_merge_cmp_lines(xdfenv_t *xe1, int i1, xdfenv_t *xe2, int i2,
  96                 int line_count, long flags)
  97 {
  98         int i;
  99         xrecord_t **rec1 = xe1->xdf2.recs + i1;
 100         xrecord_t **rec2 = xe2->xdf2.recs + i2;
 101
 102         for (i = 0; i < line_count; i++) {
 103                 int result = xdl_recmatch(rec1[i]->ptr, rec1[i]->size,
 104                         rec2[i]->ptr, rec2[i]->size, flags);
 105                 if (!result)
 106                         return -1;
 107         }
 108         return 0;
 109 }
 110
 111 static int xdl_recs_copy_0(int use_orig, xdfenv_t *xe, int i, int count, int add_nl, char *dest)
 112 {
 113         xrecord_t **recs;
 114         int size = 0;
 115
 116         recs = (use_orig ? xe->xdf1.recs : xe->xdf2.recs) + i;
 117
 118         if (count < 1)
 119                 return 0;
 120
 121         for (i = 0; i < count; size += recs[i++]->size)
 122                 if (dest)
 123                         memcpy(dest + size, recs[i]->ptr, recs[i]->size);
 124         if (add_nl) {
 125                 i = recs[count - 1]->size;
 126                 if (i == 0 || recs[count - 1]->ptr[i - 1] != '\n') {
 127                         if (dest)
 128                                 dest[size] = '\n';
 129                         size++;
 130                 }
 131         }
 132         return size;
 133 }
 134
 135 static int xdl_recs_copy(xdfenv_t *xe, int i, int count, int add_nl, char *dest)
 136 {
 137         return xdl_recs_copy_0(0, xe, i, count, add_nl, dest);
 138 }
 139
 140 static int xdl_orig_copy(xdfenv_t *xe, int i, int count, int add_nl, char *dest)
 141 {
 142         return xdl_recs_copy_0(1, xe, i, count, add_nl, dest);
 143 }
 144
 145 static int fill_conflict_hunk(xdfenv_t *xe1, const char *name1,
 146                               xdfenv_t *xe2, const char *name2,
 147                               int size, int i, int style,
 148                               xdmerge_t *m, char *dest, int marker_size)
 149 {
 150         int marker1_size = (name1 ? strlen(name1) + 1 : 0);
 151         int marker2_size = (name2 ? strlen(name2) + 1 : 0);
 152         int j;
 153
 154         if (marker_size <= 0)
 155                 marker_size = DEFAULT_CONFLICT_MARKER_SIZE;
 156
 157         /* Before conflicting part */
 158         size += xdl_recs_copy(xe1, i, m->i1 - i, 0,
 159                               dest ? dest + size : NULL);
 160
 161         if (!dest) {
 162                 size += marker_size + 1 + marker1_size;
 163         } else {
 164                 for (j = 0; j < marker_size; j++)
 165                         dest[size++] = '<';
 166                 if (marker1_size) {
 167                         dest[size] = ' ';
 168                         memcpy(dest + size + 1, name1, marker1_size - 1);
 169                         size += marker1_size;
 170                 }
 171                 dest[size++] = '\n';
 172         }
 173
 174         /* Postimage from side #1 */
 175         size += xdl_recs_copy(xe1, m->i1, m->chg1, 1,
 176                               dest ? dest + size : NULL);
 177
 178         if (style == XDL_MERGE_DIFF3) {
 179                 /* Shared preimage */
 180                 if (!dest) {
 181                         size += marker_size + 1;
 182                 } else {
 183                         for (j = 0; j < marker_size; j++)
 184                                 dest[size++] = '|';
 185                         dest[size++] = '\n';
 186                 }
 187                 size += xdl_orig_copy(xe1, m->i0, m->chg0, 1,
 188                                       dest ? dest + size : NULL);
 189         }
 190
 191         if (!dest) {
 192                 size += marker_size + 1;
 193         } else {
 194                 for (j = 0; j < marker_size; j++)
 195                         dest[size++] = '=';
 196                 dest[size++] = '\n';
 197         }
 198
 199         /* Postimage from side #2 */
 200         size += xdl_recs_copy(xe2, m->i2, m->chg2, 1,
 201                               dest ? dest + size : NULL);
 202         if (!dest) {
 203                 size += marker_size + 1 + marker2_size;
 204         } else {
 205                 for (j = 0; j < marker_size; j++)
 206                         dest[size++] = '>';
 207                 if (marker2_size) {
 208                         dest[size] = ' ';
 209                         memcpy(dest + size + 1, name2, marker2_size - 1);
 210                         size += marker2_size;
 211                 }
 212                 dest[size++] = '\n';
 213         }
 214         return size;
 215 }
 216
 217 static int xdl_fill_merge_buffer(xdfenv_t *xe1, const char *name1,
 218                                  xdfenv_t *xe2, const char *name2,
 219                                  xdmerge_t *m, char *dest, int style,
 220                                  int marker_size)
 221 {
 222         int size, i;
 223
 224         for (size = i = 0; m; m = m->next) {
 225                 if (m->mode == 0)
 226                         size = fill_conflict_hunk(xe1, name1, xe2, name2,
 227                                                   size, i, style, m, dest,
 228                                                   marker_size);
 229                 else if (m->mode == 1)
 230                         size += xdl_recs_copy(xe1, i, m->i1 + m->chg1 - i, 0,
 231                                               dest ? dest + size : NULL);
 232                 else if (m->mode == 2)
 233                         size += xdl_recs_copy(xe2, m->i2 - m->i1 + i,
 234                                               m->i1 + m->chg2 - i, 0,
 235                                               dest ? dest + size : NULL);
 236                 else
 237                         continue;
 238                 i = m->i1 + m->chg1;
 239         }
 240         size += xdl_recs_copy(xe1, i, xe1->xdf2.nrec - i, 0,
 241                               dest ? dest + size : NULL);
 242         return size;
 243 }
 244
 245 /*
 246  * Sometimes, changes are not quite identical, but differ in only a few
 247  * lines. Try hard to show only these few lines as conflicting.
 248  */
 249 static int xdl_refine_conflicts(xdfenv_t *xe1, xdfenv_t *xe2, xdmerge_t *m,
 250                 xpparam_t const *xpp)
 251 {
 252         for (; m; m = m->next) {
 253                 mmfile_t t1, t2;
 254                 xdfenv_t xe;
 255                 xdchange_t *xscr, *x;
 256                 int i1 = m->i1, i2 = m->i2;
 257
 258                 /* let's handle just the conflicts */
 259                 if (m->mode)
 260                         continue;
 261
 262                 /* no sense refining a conflict when one side is empty */
 263                 if (m->chg1 == 0 || m->chg2 == 0)
 264                         continue;
 265
 266                 /*
 267                  * This probably does not work outside git, since
 268                  * we have a very simple mmfile structure.
 269                  */
 270                 t1.ptr = (char *)xe1->xdf2.recs[m->i1]->ptr;
 271                 t1.size = xe1->xdf2.recs[m->i1 + m->chg1 - 1]->ptr
 272                         + xe1->xdf2.recs[m->i1 + m->chg1 - 1]->size - t1.ptr;
 273                 t2.ptr = (char *)xe2->xdf2.recs[m->i2]->ptr;
 274                 t2.size = xe2->xdf2.recs[m->i2 + m->chg2 - 1]->ptr
 275                         + xe2->xdf2.recs[m->i2 + m->chg2 - 1]->size - t2.ptr;
 276                 if (xdl_do_diff(&t1, &t2, xpp, &xe) < 0)
 277                         return -1;
 278                 if (xdl_change_compact(&xe.xdf1, &xe.xdf2, xpp->flags) < 0 ||
 279                     xdl_change_compact(&xe.xdf2, &xe.xdf1, xpp->flags) < 0 ||
 280                     xdl_build_script(&xe, &xscr) < 0) {
 281                         xdl_free_env(&xe);
 282                         return -1;
 283                 }
 284                 if (!xscr) {
 285                         /* If this happens, the changes are identical. */
 286                         xdl_free_env(&xe);
 287                         m->mode = 4;
 288                         continue;
 289                 }
 290                 x = xscr;
 291                 m->i1 = xscr->i1 + i1;
 292                 m->chg1 = xscr->chg1;
 293                 m->i2 = xscr->i2 + i2;
 294                 m->chg2 = xscr->chg2;
 295                 while (xscr->next) {
 296                         xdmerge_t *m2 = xdl_malloc(sizeof(xdmerge_t));
 297                         if (!m2) {
 298                                 xdl_free_env(&xe);
 299                                 xdl_free_script(x);
 300                                 return -1;
 301                         }
 302                         xscr = xscr->next;
 303                         m2->next = m->next;
 304                         m->next = m2;
 305                         m = m2;
 306                         m->mode = 0;
 307                         m->i1 = xscr->i1 + i1;
 308                         m->chg1 = xscr->chg1;
 309                         m->i2 = xscr->i2 + i2;
 310                         m->chg2 = xscr->chg2;
 311                 }
 312                 xdl_free_env(&xe);
 313                 xdl_free_script(x);
 314         }
 315         return 0;
 316 }
 317
 318 static int line_contains_alnum(const char *ptr, long size)
 319 {
 320         while (size--)
 321                 if (isalnum(*(ptr++)))
 322                         return 1;
 323         return 0;
 324 }
 325
 326 static int lines_contain_alnum(xdfenv_t *xe, int i, int chg)
 327 {
 328         for (; chg; chg--, i++)
 329                 if (line_contains_alnum(xe->xdf2.recs[i]->ptr,
 330                                 xe->xdf2.recs[i]->size))
 331                         return 1;
 332         return 0;
 333 }
 334
 335 /*
 336  * This function merges m and m->next, marking everything between those hunks
 337  * as conflicting, too.
 338  */
 339 static void xdl_merge_two_conflicts(xdmerge_t *m)
 340 {
 341         xdmerge_t *next_m = m->next;
 342         m->chg1 = next_m->i1 + next_m->chg1 - m->i1;
 343         m->chg2 = next_m->i2 + next_m->chg2 - m->i2;
 344         m->next = next_m->next;
 345         free(next_m);
 346 }
 347
 348 /*
 349  * If there are less than 3 non-conflicting lines between conflicts,
 350  * it appears simpler -- because it takes up less (or as many) lines --
 351  * if the lines are moved into the conflicts.
 352  */
 353 static int xdl_simplify_non_conflicts(xdfenv_t *xe1, xdmerge_t *m,
 354                                       int simplify_if_no_alnum)
 355 {
 356         int result = 0;
 357
 358         if (!m)
 359                 return result;
 360         for (;;) {
 361                 xdmerge_t *next_m = m->next;
 362                 int begin, end;
 363
 364                 if (!next_m)
 365                         return result;
 366
 367                 begin = m->i1 + m->chg1;
 368                 end = next_m->i1;
 369
 370                 if (m->mode != 0 || next_m->mode != 0 ||
 371                     (end - begin > 3 &&
 372                      (!simplify_if_no_alnum ||
 373                       lines_contain_alnum(xe1, begin, end - begin)))) {
 374                         m = next_m;
 375                 } else {
 376                         result++;
 377                         xdl_merge_two_conflicts(m);
 378                 }
 379         }
 380 }
 381
 382 /*
 383  * level == 0: mark all overlapping changes as conflict
 384  * level == 1: mark overlapping changes as conflict only if not identical
 385  * level == 2: analyze non-identical changes for minimal conflict set
 386  * level == 3: analyze non-identical changes for minimal conflict set, but
 387  *             treat hunks not containing any letter or number as conflicting
 388  *
 389  * returns < 0 on error, == 0 for no conflicts, else number of conflicts
 390  */
 391 static int xdl_do_merge(xdfenv_t *xe1, xdchange_t *xscr1, const char *name1,
 392                 xdfenv_t *xe2, xdchange_t *xscr2, const char *name2,
 393                 int flags, xmparam_t const *xmp, mmbuffer_t *result) {
 394         xdmerge_t *changes, *c;
 395         xpparam_t const *xpp = &xmp->xpp;
 396         int i0, i1, i2, chg0, chg1, chg2;
 397         int level = flags & XDL_MERGE_LEVEL_MASK;
 398         int style = flags & XDL_MERGE_STYLE_MASK;
 399
 400         if (style == XDL_MERGE_DIFF3) {
 401                 /*
 402                  * "diff3 -m" output does not make sense for anything
 403                  * more aggressive than XDL_MERGE_EAGER.
 404                  */
 405                 if (XDL_MERGE_EAGER < level)
 406                         level = XDL_MERGE_EAGER;
 407         }
 408
 409         c = changes = NULL;
 410
 411         while (xscr1 && xscr2) {
 412                 if (!changes)
 413                         changes = c;
 414                 if (xscr1->i1 + xscr1->chg1 < xscr2->i1) {
 415                         i0 = xscr1->i1;
 416                         i1 = xscr1->i2;
 417                         i2 = xscr2->i2 - xscr2->i1 + xscr1->i1;
 418                         chg0 = xscr1->chg1;
 419                         chg1 = xscr1->chg2;
 420                         chg2 = xscr1->chg1;
 421                         if (xdl_append_merge(&c, 1,
 422                                              i0, chg0, i1, chg1, i2, chg2)) {
 423                                 xdl_cleanup_merge(changes);
 424                                 return -1;
 425                         }
 426                         xscr1 = xscr1->next;
 427                         continue;
 428                 }
 429                 if (xscr2->i1 + xscr2->chg1 < xscr1->i1) {
 430                         i0 = xscr2->i1;
 431                         i1 = xscr1->i2 - xscr1->i1 + xscr2->i1;
 432                         i2 = xscr2->i2;
 433                         chg0 = xscr2->chg1;
 434                         chg1 = xscr2->chg1;
 435                         chg2 = xscr2->chg2;
 436                         if (xdl_append_merge(&c, 2,
 437                                              i0, chg0, i1, chg1, i2, chg2)) {
 438                                 xdl_cleanup_merge(changes);
 439                                 return -1;
 440                         }
 441                         xscr2 = xscr2->next;
 442                         continue;
 443                 }
 444                 if (level == XDL_MERGE_MINIMAL || xscr1->i1 != xscr2->i1 ||
 445                                 xscr1->chg1 != xscr2->chg1 ||
 446                                 xscr1->chg2 != xscr2->chg2 ||
 447                                 xdl_merge_cmp_lines(xe1, xscr1->i2,
 448                                         xe2, xscr2->i2,
 449                                         xscr1->chg2, xpp->flags)) {
 450                         /* conflict */
 451                         int off = xscr1->i1 - xscr2->i1;
 452                         int ffo = off + xscr1->chg1 - xscr2->chg1;
 453
 454                         i0 = xscr1->i1;
 455                         i1 = xscr1->i2;
 456                         i2 = xscr2->i2;
 457                         if (off > 0) {
 458                                 i0 -= off;
 459                                 i1 -= off;
 460                         }
 461                         else
 462                                 i2 += off;
 463                         chg0 = xscr1->i1 + xscr1->chg1 - i0;
 464                         chg1 = xscr1->i2 + xscr1->chg2 - i1;
 465                         chg2 = xscr2->i2 + xscr2->chg2 - i2;
 466                         if (ffo < 0) {
 467                                 chg0 -= ffo;
 468                                 chg1 -= ffo;
 469                         } else
 470                                 chg2 += ffo;
 471                         if (xdl_append_merge(&c, 0,
 472                                              i0, chg0, i1, chg1, i2, chg2)) {
 473                                 xdl_cleanup_merge(changes);
 474                                 return -1;
 475                         }
 476                 }
 477
 478                 i1 = xscr1->i1 + xscr1->chg1;
 479                 i2 = xscr2->i1 + xscr2->chg1;
 480
 481                 if (i1 >= i2)
 482                         xscr2 = xscr2->next;
 483                 if (i2 >= i1)
 484                         xscr1 = xscr1->next;
 485         }
 486         while (xscr1) {
 487                 if (!changes)
 488                         changes = c;
 489                 i0 = xscr1->i1;
 490                 i1 = xscr1->i2;
 491                 i2 = xscr1->i1 + xe2->xdf2.nrec - xe2->xdf1.nrec;
 492                 chg0 = xscr1->chg1;
 493                 chg1 = xscr1->chg2;
 494                 chg2 = xscr1->chg1;
 495                 if (xdl_append_merge(&c, 1,
 496                                      i0, chg0, i1, chg1, i2, chg2)) {
 497                         xdl_cleanup_merge(changes);
 498                         return -1;
 499                 }
 500                 xscr1 = xscr1->next;
 501         }
 502         while (xscr2) {
 503                 if (!changes)
 504                         changes = c;
 505                 i0 = xscr2->i1;
 506                 i1 = xscr2->i1 + xe1->xdf2.nrec - xe1->xdf1.nrec;
 507                 i2 = xscr2->i2;
 508                 chg0 = xscr2->chg1;
 509                 chg1 = xscr2->chg1;
 510                 chg2 = xscr2->chg2;
 511                 if (xdl_append_merge(&c, 2,
 512                                      i0, chg0, i1, chg1, i2, chg2)) {
 513                         xdl_cleanup_merge(changes);
 514                         return -1;
 515                 }
 516                 xscr2 = xscr2->next;
 517         }
 518         if (!changes)
 519                 changes = c;
 520         /* refine conflicts */
 521         if (XDL_MERGE_ZEALOUS <= level &&
 522             (xdl_refine_conflicts(xe1, xe2, changes, xpp) < 0 ||
 523              xdl_simplify_non_conflicts(xe1, changes,
 524                                         XDL_MERGE_ZEALOUS < level) < 0)) {
 525                 xdl_cleanup_merge(changes);
 526                 return -1;
 527         }
 528         /* output */
 529         if (result) {
 530                 int marker_size = xmp->marker_size;
 531                 int size = xdl_fill_merge_buffer(xe1, name1, xe2, name2,
 532                                                  changes, NULL, style,
 533                                                  marker_size);
 534                 result->ptr = xdl_malloc(size);
 535                 if (!result->ptr) {
 536                         xdl_cleanup_merge(changes);
 537                         return -1;
 538                 }
 539                 result->size = size;
 540                 xdl_fill_merge_buffer(xe1, name1, xe2, name2, changes,
 541                                       result->ptr, style, marker_size);
 542         }
 543         return xdl_cleanup_merge(changes);
 544 }
 545
 546 int xdl_merge(mmfile_t *orig, mmfile_t *mf1, const char *name1,
 547                 mmfile_t *mf2, const char *name2,
 548                 xmparam_t const *xmp, int flags, mmbuffer_t *result) {
 549         xdchange_t *xscr1, *xscr2;
 550         xdfenv_t xe1, xe2;
 551         int status;
 552         xpparam_t const *xpp = &xmp->xpp;
 553
 554         result->ptr = NULL;
 555         result->size = 0;
 556
 557         if (xdl_do_diff(orig, mf1, xpp, &xe1) < 0 ||
 558                         xdl_do_diff(orig, mf2, xpp, &xe2) < 0) {
 559                 return -1;
 560         }
 561         if (xdl_change_compact(&xe1.xdf1, &xe1.xdf2, xpp->flags) < 0 ||
 562             xdl_change_compact(&xe1.xdf2, &xe1.xdf1, xpp->flags) < 0 ||
 563             xdl_build_script(&xe1, &xscr1) < 0) {
 564                 xdl_free_env(&xe1);
 565                 return -1;
 566         }
 567         if (xdl_change_compact(&xe2.xdf1, &xe2.xdf2, xpp->flags) < 0 ||
 568             xdl_change_compact(&xe2.xdf2, &xe2.xdf1, xpp->flags) < 0 ||
 569             xdl_build_script(&xe2, &xscr2) < 0) {
 570                 xdl_free_env(&xe2);
 571                 return -1;
 572         }
 573         status = 0;
 574         if (!xscr1) {
 575                 result->ptr = xdl_malloc(mf2->size);
 576                 memcpy(result->ptr, mf2->ptr, mf2->size);
 577                 result->size = mf2->size;
 578         } else if (!xscr2) {
 579                 result->ptr = xdl_malloc(mf1->size);
 580                 memcpy(result->ptr, mf1->ptr, mf1->size);
 581                 result->size = mf1->size;
 582         } else {
 583                 status = xdl_do_merge(&xe1, xscr1, name1,
 584                                       &xe2, xscr2, name2,
 585                                       flags, xmp, result);
 586         }
 587         xdl_free_script(xscr1);
 588         xdl_free_script(xscr2);
 589
 590         xdl_free_env(&xe1);
 591         xdl_free_env(&xe2);
 592
 593         return status;
 594 }