merge2.c

   1 /*
   2  * wiggle - apply rejected patches
   3  *
   4  * Copyright (C) 2005 Neil Brown <neilb@cse.unsw.edu.au>
   5  *
   6  *
   7  *    This program is free software; you can redistribute it and/or modify
   8  *    it under the terms of the GNU General Public License as published by
   9  *    the Free Software Foundation; either version 2 of the License, or
  10  *    (at your option) any later version.
  11  *
  12  *    This program is distributed in the hope that it will be useful,
  13  *    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  *    GNU General Public License for more details.
  16  *
  17  *    You should have received a copy of the GNU General Public License
  18  *    along with this program; if not, write to the Free Software
  19  *    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  20  *
  21  *    Author: Neil Brown
  22  *    Email: <neilb@cse.unsw.edu.au>
  23  *    Paper: Neil Brown
  24  *           School of Computer Science and Engineering
  25  *           The University of New South Wales
  26  *           Sydney, 2052
  27  *           Australia
  28  */
  29
  30 #include "wiggle.h"
  31 #include <stdlib.h>
  32 #include <malloc.h>
  33
  34 /*
  35  * Second attempt at merging....
  36  *
  37  * We want to create a mergelist which identifies 'orig' and 'after'
  38  * sections (from a and c) and conflicts (which are ranges of a,b,c which
  39  * all don't match).
  40  * It is also helpful to differentiate 'orig' sections that aren't
  41  * matched in 'b' with orig sections that are.
  42  * To help with highlighting, it will be useful to know where
  43  * the conflicts match the csl lists.
  44  *
  45  * This can all be achieved with a list of (a,b,c,c1,c1) 5-tuples.
  46  * If two consecutive differ in more than one of a,b,c, it is a
  47  * conflict.
  48  * If only 'a' differ, it is un-matched original.
  49  * If only 'b' differ, it is matched, unchanged original
  50  * If only 'c' differ, it is 1
  51  */
  52
  53 static inline int min(int a, int b)
  54 {
  55         return a<b?a:b;
  56 }
  57 static inline void assert(int a)
  58 {
  59         if (!a) abort();
  60 }
  61
  62 int check_alreadyapplied(struct file af, struct file cf,
  63                           struct merge *m)
  64 {
  65         int i;
  66         if (m->al != m->cl)
  67                 return 0;
  68         for (i=0; i<m->al; i++) {
  69                 if (af.list[m->a+i].len != cf.list[m->c+i].len)
  70                         return 0;
  71                 if (strncmp(af.list[m->a+i].start,
  72                             cf.list[m->c+i].start,
  73                             af.list[m->a+i].len)!= 0)
  74                         return 0;
  75         }
  76 #if 0
  77         printf("already applied %d,%d,%d - %d,%d,%d\n",
  78                m->a,m->b,m->c,m->al,m->bl,m->cl);
  79         printf(" %.10s - %.10s\n", af.list[m->a].start,
  80                cf.list[m->c].start);
  81 #endif
  82         m->type = AlreadyApplied;
  83         return 1;
  84 }
  85
  86 inline int isolate_conflicts(struct file af, struct file bf, struct file cf,
  87                               struct csl *csl1, struct csl *csl2, int words,
  88                               struct merge *m)
  89 {
  90         /* A conflict indicates that something is definitely wrong
  91          * and so we need to be a bit suspicious of nearby apparent matches.
  92          * To display a conflict effectively we expands it's effect to
  93          * include any Extraneous, Unmatched or Changed text.
  94          * Also, unless 'words', we need to include any partial lines
  95          * in the Unchanged text that forms the border of a conflict.
  96          *
  97          * A Changed text may also border a conflict, but it can
  98          * only border one conflict (where as an Unchanged can border
  99          * a preceeding and a following conflict).
 100          * The 'new' section of a Changed text appears in the
 101          * conflict as does any part of the original before
 102          * a newline.
 103          *
 104          */
 105         int i,j,k;
 106         int cnt = 0;
 107
 108         for (i=0; m[i].type != End; i++)
 109                 if (m[i].type == Conflict) {
 110                         /* We have a conflict here.
 111                          * First search backwards for an Unchanged marking
 112                          * things as in_conflict.  Then find the
 113                          * cut-point in the Unchanged.  If there isn't one,
 114                          * keep looking.
 115                          *
 116                          * Then search forward doing the same thing.
 117                          */
 118                         cnt++;
 119                         m[i].in_conflict = 1;
 120                         j = i;
 121                         while (--j >= 0) {
 122                                 if (!m[j].in_conflict) {
 123                                         m[j].in_conflict = 1;
 124                                         m[j].lo = 0;
 125                                 } else if (m[j].type == Changed) {
 126                                         /* This can no longer form a border */
 127                                         m[j].lo = 0; m[j].hi = -1;
 128                                         /* We merge these conflicts and stop searching */
 129                                         cnt--;
 130                                         break;
 131                                 }
 132
 133                                 if (m[j].type == Unchanged || m[j].type == Changed) {
 134                                         if (words) {
 135                                                 m[j].hi = m[j].al;
 136                                                 break;
 137                                         }
 138                                         /* need to find the last line-break, which
 139                                          * might be after the last newline, if there
 140                                          * is one, or might be at the start
 141                                          */
 142                                         for (k=m[j].al; k>0; k--)
 143                                                 if (ends_line(af.list[m[j].a+k-1]))
 144                                                         break;
 145                                         if (k > 0)
 146                                                 m[j].hi = k;
 147                                         else if ((m[j].a == 0 || ends_line(af.list[m[j].a-1])) &&
 148                                                  (m[j].b == 0 || ends_line(bf.list[m[j].b-1])) &&
 149                                                  (m[j].c == 0 || ends_line(cf.list[m[j].c-1])))
 150                                                 m[j].hi = 0;
 151                                         else
 152                                                 /* no start-of-line found... */
 153                                                 m[j].hi = -1;
 154                                         if (m[j].hi > 0 && m[j].type == Changed) {
 155                                                 /* this can only work if start is also a linke break */
 156                                                 if ((m[j].a == 0 || ends_line(af.list[m[j].a-1])) &&
 157                                                     (m[j].b == 0 || ends_line(bf.list[m[j].b-1])) &&
 158                                                     (m[j].c == 0 || ends_line(cf.list[m[j].c-1])))
 159                                                         /* ok */;
 160                                                 else
 161                                                         m[j].hi = -1;
 162                                         }
 163                                         if (m[j].hi >= 0)
 164                                                 break;
 165                                 }
 166                         }
 167 #if 0
 168                         if (j>=0 && m[j].in_conflict && m[j].type == Unchanged &&
 169                             m[j].hi == m[j].lo) {
 170                                 /* nothing to separate conflicts, merge them */
 171                                 m[j].lo = 0;
 172                                 m[j].hi = -1;
 173                                 cnt--;
 174                         }
 175 #endif
 176                         /* now the forward search */
 177                         for (j=i+1; m[j].type != End; j++) {
 178                                 m[j].in_conflict = 1;
 179                                 if (m[j].type == Unchanged || m[j].type == Changed) {
 180                                         m[j].hi = m[j].al;
 181                                         if (words) {
 182                                                 m[j].lo = 0;
 183                                                 break;
 184                                         }
 185                                         /* need to find a line-break, which might be at
 186                                          * the very beginning, or might be after the
 187                                          * first newline - if there is one
 188                                          */
 189                                         if ((m[j].a == 0 || ends_line(af.list[m[j].a-1])) &&
 190                                             (m[j].b == 0 || ends_line(bf.list[m[j].b-1])) &&
 191                                             (m[j].c == 0 || ends_line(cf.list[m[j].c-1])))
 192                                                 m[j].lo = 0;
 193                                         else {
 194                                                 for (k = 0 ; k < m[j].al ; k++)
 195                                                         if (ends_line(af.list[m[j].a+k]))
 196                                                                 break;
 197                                                 if (k<m[j].al)
 198                                                         m[j].lo = k+1;
 199                                                 else
 200                                                         /* no start-of-line found */
 201                                                         m[j].lo = m[j].al+1;
 202                                         }
 203                                         if (m[j].lo <= m[j].al+1 && m[j].type == Changed) {
 204                                                 /* this can only work if the end is a line break */
 205                                                 if (ends_line(af.list[m[j].a+m[j].al-1]) &&
 206                                                     ends_line(bf.list[m[j].b+m[j].bl-1]) &&
 207                                                     ends_line(cf.list[m[j].c+m[j].cl-1]))
 208                                                         /* ok */;
 209                                                 else
 210                                                         m[j].lo = m[j].al+1;
 211                                         }
 212                                         if (m[j].lo < m[j].al+1)
 213                                                 break;
 214                                 }
 215                         }
 216                         i = j;
 217                 }
 218         return cnt;
 219 }
 220
 221
 222 struct ci make_merger(struct file af, struct file bf, struct file cf,
 223                       struct csl *csl1, struct csl *csl2, int words)
 224 {
 225         /* find the wiggles and conflicts between csl1 and csl2
 226          */
 227         struct ci rv;
 228         int i,l;
 229         int a,b,c,c1,c2;
 230         int wiggle_found = 0;
 231
 232         rv.conflicts = rv.wiggles = rv.ignored = 0;
 233
 234         for (i=0; csl1[i].len; i++);
 235         l = i;
 236         for (i=0; csl2[i].len; i++);
 237         l += i;
 238         /* maybe a bit of slack at each end */
 239         l = l* 4 + 10;
 240
 241         rv.merger = malloc(sizeof(struct merge)*l);
 242         if (!rv.merger)
 243                 return rv;
 244
 245         a=b=c=c1=c2 = 0;
 246         i = 0;
 247         while (1) {
 248                 int match1, match2;
 249                 match1 = (a>=csl1[c1].a && b >= csl1[c1].b); /* c1 doesn't match */
 250                 match2 = (b>=csl2[c2].a && c >= csl2[c2].b);
 251
 252                 rv.merger[i].a = a;
 253                 rv.merger[i].b = b;
 254                 rv.merger[i].c = c;
 255                 rv.merger[i].c1 = c1;
 256                 rv.merger[i].c2 = c2;
 257                 rv.merger[i].in_conflict = 0;
 258
 259                 if (!match1 && match2) {
 260                         if (a < csl1[c1].a) {
 261                                 /* some unmatched text */
 262                                 rv.merger[i].type = Unmatched;
 263                                 rv.merger[i].al = csl1[c1].a - a;
 264                                 rv.merger[i].bl = 0;
 265                                 rv.merger[i].cl = 0;
 266                                 wiggle_found ++;
 267                         } else {
 268                                 int newb;
 269                                 int j;
 270                                 assert(b<csl1[c1].b);
 271                                 /* some Extraneous text */
 272                                 /* length is min of unmatched on left
 273                                  * and matched on right
 274                                  */
 275                                 rv.merger[i].type = Extraneous;
 276                                 rv.merger[i].al = 0;
 277                                 rv.merger[i].cl =
 278                                         rv.merger[i].bl =
 279                                         min(csl1[c1].b - b,
 280                                             csl2[c2].len - (b-csl2[c2].a));
 281                                 newb = b+rv.merger[i].bl;
 282                                 for (j=b; j<newb; j++) {
 283                                         if (bf.list[j].start[0] == '\0') {
 284                                                 if (wiggle_found > 1)
 285                                                         rv.wiggles++;
 286                                                 wiggle_found = 0;
 287                                         } else
 288                                                 wiggle_found ++;
 289                                 }
 290                         }
 291                 } else if (match1 && !match2) {
 292                         /* some changed text
 293                          * if 'c' is currently at a suitable cut-point, then
 294                          * we can look for a triple-cut-point for start.
 295                          * Also, if csl2[c2].b isn't in a conflict, and is
 296                          * a suitable cut-point, then we could make a
 297                          * triple-cut-point for end of a conflict.
 298                          */
 299
 300                         rv.merger[i].type = Changed;
 301                         rv.merger[i].bl = min(csl1[c1].b+csl1[c1].len, csl2[c2].a) - b;
 302                         rv.merger[i].al = rv.merger[i].bl;
 303                         rv.merger[i].cl = csl2[c2].b - c;
 304                 } else if (match1 && match2) {
 305                         /* Some unchanged text
 306                          */
 307                         rv.merger[i].type = Unchanged;
 308                         rv.merger[i].bl =
 309                                 min(csl1[c1].len - (b-csl1[c1].b),
 310                                     csl2[c2].len - (b-csl2[c2].a));
 311                         rv.merger[i].al = rv.merger[i].cl =
 312                                 rv.merger[i].bl;
 313                 } else {
 314                         /* must be a conflict.
 315                          * Move a and c to next match, and b to closest of the two
 316                          */
 317                         rv.merger[i].type = Conflict;
 318                         rv.merger[i].al = csl1[c1].a - a;
 319                         rv.merger[i].cl = csl2[c2].b - c;
 320                         rv.merger[i].bl = min(csl1[c1].b, csl2[c2].a) - b;
 321                         if (check_alreadyapplied(af,cf,&rv.merger[i]))
 322                                 rv.ignored ++;
 323                 }
 324                 a += rv.merger[i].al;
 325                 b += rv.merger[i].bl;
 326                 c += rv.merger[i].cl;
 327                 i++;
 328
 329                 while (csl1[c1].a + csl1[c1].len <= a && csl1[c1].len) c1++;
 330                 assert(csl1[c1].b + csl1[c1].len >= b);
 331                 while (csl2[c2].b + csl2[c2].len <= c && csl2[c2].len) c2++;
 332                 assert(csl2[c2].a + csl2[c2].len >= b);
 333                 if (csl1[c1].len == 0 && csl2[c2].len == 0 &&
 334                     a == csl1[c1].a && b == csl1[c1].b &&
 335                     b == csl2[c2].a && c == csl2[c2].b)
 336                         break;
 337         }
 338         rv.merger[i].type = End;
 339         rv.merger[i].in_conflict = 0;
 340         rv.conflicts = isolate_conflicts(af,bf,cf,csl1,csl2,words,rv.merger);
 341         if (wiggle_found)
 342                 rv.wiggles++;
 343         return rv;
 344 }
 345
 346 void printrange(FILE *out, struct file *f, int start, int len)
 347 {
 348         while (len> 0) {
 349                 printword(out, f->list[start]);
 350                 start++;
 351                 len--;
 352         }
 353 }
 354
 355 struct ci print_merge2(FILE *out, struct file *a, struct file *b, struct file *c,
 356                  struct csl *c1, struct csl *c2,
 357                  int words)
 358 {
 359         struct ci rv = make_merger(*a, *b, *c, c1, c2, words);
 360         struct merge *m;
 361
 362         for (m=rv.merger; m->type != End ; m++) {
 363                 struct merge *cm;
 364 #if 1
 365                 static int v= -1;
 366                 if (v == -1) {
 367                         if (getenv("WiggleVerbose"))
 368                                 v = 1;
 369                         else
 370                                 v = 0;
 371                 }
 372                 if (v)
 373                 printf("[%s: %d-%d,%d-%d,%d-%d%s(%d,%d)]\n",
 374                        m->type==Unmatched?"Unmatched":
 375                        m->type==Unchanged?"Unchanged":
 376                        m->type==Extraneous?"Extraneous":
 377                        m->type==Changed?"Changed":
 378                        m->type==AlreadyApplied?"AlreadyApplied":
 379                        m->type==Conflict?"Conflict":"unknown",
 380                        m->a, m->a+m->al-1,
 381                        m->b, m->b+m->bl-1,
 382                        m->c, m->c+m->cl-1, m->in_conflict?" in_conflict":"",
 383                        m->lo, m->hi);
 384 #endif
 385                 while (m->in_conflict) {
 386                         /* need to print from 'hi' to 'lo' of next
 387                          * Unchanged which is < it's hi
 388                          */
 389                         int st = m->hi;
 390                         if (m->hi <= m->lo)
 391                                 st = 0;
 392
 393                         if (m->type == Unchanged)
 394                                 printrange(out, a, m->a+m->lo, m->hi - m->lo);
 395
 396 #if 1
 397                 if (v)
 398                         for (cm=m; cm->in_conflict; cm++) {
 399                                 printf("{%s: %d-%d,%d-%d,%d-%d%s(%d,%d)}\n",
 400                                        cm->type==Unmatched?"Unmatched":
 401                                        cm->type==Unchanged?"Unchanged":
 402                                        cm->type==Extraneous?"Extraneous":
 403                                        cm->type==Changed?"Changed":
 404                                        cm->type==AlreadyApplied?"AlreadyApplied":
 405                                        cm->type==Conflict?"Conflict":"unknown",
 406                                        cm->a, cm->a+cm->al-1,
 407                                        cm->b, cm->b+cm->bl-1,
 408                                        cm->c, cm->c+cm->cl-1, cm->in_conflict?" in_conflict":"",
 409                                        cm->lo, cm->hi);
 410                                 if ((cm->type == Unchanged ||cm->type == Changed) && cm != m && cm->lo < cm->hi)
 411                                         break;
 412                         }
 413 #endif
 414
 415                         fputs(words?"<<<---":"<<<<<<<\n", out);
 416                         for (cm=m; cm->in_conflict; cm++) {
 417                                 if ((cm->type == Unchanged || cm->type == Changed) && cm != m && cm->lo < cm->hi) {
 418                                         printrange(out, a, cm->a, cm->lo);
 419                                         break;
 420                                 }
 421                                 printrange(out, a, cm->a+st, cm->al-st);
 422                                 st = 0;
 423                         }
 424                         fputs(words?"|||":"|||||||\n", out);
 425                         st = m->hi;
 426                         for (cm=m; cm->in_conflict; cm++) {
 427                                 if ((cm->type == Unchanged || cm->type == Changed) && cm != m && cm->lo < cm->hi) {
 428                                         printrange(out, b, cm->b, cm->lo);
 429                                         break;
 430                                 }
 431                                 printrange(out, b, cm->b+st, cm->bl-st);
 432                                 st = 0;
 433                         }
 434                         fputs(words?"===":"=======\n", out);
 435                         st = m->hi;
 436                         for (cm=m; cm->in_conflict; cm++) {
 437                                 if (cm->type == Unchanged && cm != m && cm->lo < cm->hi) {
 438                                         printrange(out, c, cm->c, cm->lo);
 439                                         break;
 440                                 }
 441                                 if (cm->type == Changed)
 442                                         st = 0; /* All of result of change must be printed */
 443                                 printrange(out, c, cm->c+st, cm->cl-st);
 444                                 st = 0;
 445                         }
 446                         fputs(words?"--->>>":">>>>>>>\n", out);
 447                         m = cm;
 448                         if (m->in_conflict && m->hi >= m->al) {
 449                                 assert(m->type == Unchanged);
 450                                 printrange(out, a, m->a+m->lo, m->hi-m->lo);
 451                                 m++;
 452                         }
 453                 }
 454
 455                 /* there is always some non-conflict after a conflict,
 456                  * unless we hit the end
 457                  */
 458                 if (m->type == End)
 459                         break;
 460 #if 1
 461                 if (v) {
 462                         printf("<<%s: %d-%d,%d-%d,%d-%d%s(%d,%d)>>\n",
 463                                m->type==Unmatched?"Unmatched":
 464                                m->type==Unchanged?"Unchanged":
 465                                m->type==Extraneous?"Extraneous":
 466                                m->type==Changed?"Changed":
 467                                m->type==AlreadyApplied?"AlreadyApplied":
 468                                m->type==Conflict?"Conflict":"unknown",
 469                                m->a, m->a+m->al-1,
 470                                m->b, m->b+m->bl-1,
 471                                m->c, m->c+m->cl-1, m->in_conflict?" in_conflict":"",
 472                                m->lo, m->hi);
 473                 }
 474 #endif
 475                 switch(m->type) {
 476                 case Unchanged:
 477                 case AlreadyApplied:
 478                 case Unmatched:
 479                         printrange(out, a, m->a, m->al);
 480                         break;
 481                 case Extraneous:
 482                         break;
 483                 case Changed:
 484                         printrange(out, c, m->c, m->cl);
 485                         break;
 486                 case Conflict:
 487                 case End: assert(0);
 488                 }
 489         }
 490         return rv;
 491 }