combine-diff.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "blob.h"
   4 #include "diff.h"
   5 #include "diffcore.h"
   6 #include "quote.h"
   7 #include "xdiff-interface.h"
   8 #include "log-tree.h"
   9 #include "refs.h"
  10 #include "userdiff.h"
  11 #include "sha1-array.h"
  12
  13 static struct combine_diff_path *intersect_paths(struct combine_diff_path *curr, int n, int num_parent)
  14 {
  15         struct diff_queue_struct *q = &diff_queued_diff;
  16         struct combine_diff_path *p;
  17         int i;
  18
  19         if (!n) {
  20                 struct combine_diff_path *list = NULL, **tail = &list;
  21                 for (i = 0; i < q->nr; i++) {
  22                         int len;
  23                         const char *path;
  24                         if (diff_unmodified_pair(q->queue[i]))
  25                                 continue;
  26                         path = q->queue[i]->two->path;
  27                         len = strlen(path);
  28                         p = xmalloc(combine_diff_path_size(num_parent, len));
  29                         p->path = (char *) &(p->parent[num_parent]);
  30                         memcpy(p->path, path, len);
  31                         p->path[len] = 0;
  32                         p->len = len;
  33                         p->next = NULL;
  34                         memset(p->parent, 0,
  35                                sizeof(p->parent[0]) * num_parent);
  36
  37                         hashcpy(p->sha1, q->queue[i]->two->sha1);
  38                         p->mode = q->queue[i]->two->mode;
  39                         hashcpy(p->parent[n].sha1, q->queue[i]->one->sha1);
  40                         p->parent[n].mode = q->queue[i]->one->mode;
  41                         p->parent[n].status = q->queue[i]->status;
  42                         *tail = p;
  43                         tail = &p->next;
  44                 }
  45                 return list;
  46         }
  47
  48         for (p = curr; p; p = p->next) {
  49                 int found = 0;
  50                 if (!p->len)
  51                         continue;
  52                 for (i = 0; i < q->nr; i++) {
  53                         const char *path;
  54                         int len;
  55
  56                         if (diff_unmodified_pair(q->queue[i]))
  57                                 continue;
  58                         path = q->queue[i]->two->path;
  59                         len = strlen(path);
  60                         if (len == p->len && !memcmp(path, p->path, len)) {
  61                                 found = 1;
  62                                 hashcpy(p->parent[n].sha1, q->queue[i]->one->sha1);
  63                                 p->parent[n].mode = q->queue[i]->one->mode;
  64                                 p->parent[n].status = q->queue[i]->status;
  65                                 break;
  66                         }
  67                 }
  68                 if (!found)
  69                         p->len = 0;
  70         }
  71         return curr;
  72 }
  73
  74 /* Lines lost from parent */
  75 struct lline {
  76         struct lline *next;
  77         int len;
  78         unsigned long parent_map;
  79         char line[FLEX_ARRAY];
  80 };
  81
  82 /* Lines surviving in the merge result */
  83 struct sline {
  84         struct lline *lost_head, **lost_tail;
  85         struct lline *next_lost;
  86         char *bol;
  87         int len;
  88         /* bit 0 up to (N-1) are on if the parent has this line (i.e.
  89          * we did not change it).
  90          * bit N is used for "interesting" lines, including context.
  91          * bit (N+1) is used for "do not show deletion before this".
  92          */
  93         unsigned long flag;
  94         unsigned long *p_lno;
  95 };
  96
  97 static char *grab_blob(const unsigned char *sha1, unsigned int mode,
  98                        unsigned long *size, struct userdiff_driver *textconv,
  99                        const char *path)
 100 {
 101         char *blob;
 102         enum object_type type;
 103
 104         if (S_ISGITLINK(mode)) {
 105                 blob = xmalloc(100);
 106                 *size = snprintf(blob, 100,
 107                                  "Subproject commit %s\n", sha1_to_hex(sha1));
 108         } else if (is_null_sha1(sha1)) {
 109                 /* deleted blob */
 110                 *size = 0;
 111                 return xcalloc(1, 1);
 112         } else if (textconv) {
 113                 struct diff_filespec *df = alloc_filespec(path);
 114                 fill_filespec(df, sha1, 1, mode);
 115                 *size = fill_textconv(textconv, df, &blob);
 116                 free_filespec(df);
 117         } else {
 118                 blob = read_sha1_file(sha1, &type, size);
 119                 if (type != OBJ_BLOB)
 120                         die("object '%s' is not a blob!", sha1_to_hex(sha1));
 121         }
 122         return blob;
 123 }
 124
 125 static void append_lost(struct sline *sline, int n, const char *line, int len)
 126 {
 127         struct lline *lline;
 128         unsigned long this_mask = (1UL<<n);
 129         if (line[len-1] == '\n')
 130                 len--;
 131
 132         /* Check to see if we can squash things */
 133         if (sline->lost_head) {
 134                 lline = sline->next_lost;
 135                 while (lline) {
 136                         if (lline->len == len &&
 137                             !memcmp(lline->line, line, len)) {
 138                                 lline->parent_map |= this_mask;
 139                                 sline->next_lost = lline->next;
 140                                 return;
 141                         }
 142                         lline = lline->next;
 143                 }
 144         }
 145
 146         lline = xmalloc(sizeof(*lline) + len + 1);
 147         lline->len = len;
 148         lline->next = NULL;
 149         lline->parent_map = this_mask;
 150         memcpy(lline->line, line, len);
 151         lline->line[len] = 0;
 152         *sline->lost_tail = lline;
 153         sline->lost_tail = &lline->next;
 154         sline->next_lost = NULL;
 155 }
 156
 157 struct combine_diff_state {
 158         unsigned int lno;
 159         int ob, on, nb, nn;
 160         unsigned long nmask;
 161         int num_parent;
 162         int n;
 163         struct sline *sline;
 164         struct sline *lost_bucket;
 165 };
 166
 167 static void consume_line(void *state_, char *line, unsigned long len)
 168 {
 169         struct combine_diff_state *state = state_;
 170         if (5 < len && !memcmp("@@ -", line, 4)) {
 171                 if (parse_hunk_header(line, len,
 172                                       &state->ob, &state->on,
 173                                       &state->nb, &state->nn))
 174                         return;
 175                 state->lno = state->nb;
 176                 if (state->nn == 0) {
 177                         /* @@ -X,Y +N,0 @@ removed Y lines
 178                          * that would have come *after* line N
 179                          * in the result.  Our lost buckets hang
 180                          * to the line after the removed lines,
 181                          *
 182                          * Note that this is correct even when N == 0,
 183                          * in which case the hunk removes the first
 184                          * line in the file.
 185                          */
 186                         state->lost_bucket = &state->sline[state->nb];
 187                         if (!state->nb)
 188                                 state->nb = 1;
 189                 } else {
 190                         state->lost_bucket = &state->sline[state->nb-1];
 191                 }
 192                 if (!state->sline[state->nb-1].p_lno)
 193                         state->sline[state->nb-1].p_lno =
 194                                 xcalloc(state->num_parent,
 195                                         sizeof(unsigned long));
 196                 state->sline[state->nb-1].p_lno[state->n] = state->ob;
 197                 state->lost_bucket->next_lost = state->lost_bucket->lost_head;
 198                 return;
 199         }
 200         if (!state->lost_bucket)
 201                 return; /* not in any hunk yet */
 202         switch (line[0]) {
 203         case '-':
 204                 append_lost(state->lost_bucket, state->n, line+1, len-1);
 205                 break;
 206         case '+':
 207                 state->sline[state->lno-1].flag |= state->nmask;
 208                 state->lno++;
 209                 break;
 210         }
 211 }
 212
 213 static void combine_diff(const unsigned char *parent, unsigned int mode,
 214                          mmfile_t *result_file,
 215                          struct sline *sline, unsigned int cnt, int n,
 216                          int num_parent, int result_deleted,
 217                          struct userdiff_driver *textconv,
 218                          const char *path)
 219 {
 220         unsigned int p_lno, lno;
 221         unsigned long nmask = (1UL << n);
 222         xpparam_t xpp;
 223         xdemitconf_t xecfg;
 224         mmfile_t parent_file;
 225         struct combine_diff_state state;
 226         unsigned long sz;
 227
 228         if (result_deleted)
 229                 return; /* result deleted */
 230
 231         parent_file.ptr = grab_blob(parent, mode, &sz, textconv, path);
 232         parent_file.size = sz;
 233         memset(&xpp, 0, sizeof(xpp));
 234         xpp.flags = 0;
 235         memset(&xecfg, 0, sizeof(xecfg));
 236         memset(&state, 0, sizeof(state));
 237         state.nmask = nmask;
 238         state.sline = sline;
 239         state.lno = 1;
 240         state.num_parent = num_parent;
 241         state.n = n;
 242
 243         xdi_diff_outf(&parent_file, result_file, consume_line, &state,
 244                       &xpp, &xecfg);
 245         free(parent_file.ptr);
 246
 247         /* Assign line numbers for this parent.
 248          *
 249          * sline[lno].p_lno[n] records the first line number
 250          * (counting from 1) for parent N if the final hunk display
 251          * started by showing sline[lno] (possibly showing the lost
 252          * lines attached to it first).
 253          */
 254         for (lno = 0,  p_lno = 1; lno <= cnt; lno++) {
 255                 struct lline *ll;
 256                 sline[lno].p_lno[n] = p_lno;
 257
 258                 /* How many lines would this sline advance the p_lno? */
 259                 ll = sline[lno].lost_head;
 260                 while (ll) {
 261                         if (ll->parent_map & nmask)
 262                                 p_lno++; /* '-' means parent had it */
 263                         ll = ll->next;
 264                 }
 265                 if (lno < cnt && !(sline[lno].flag & nmask))
 266                         p_lno++; /* no '+' means parent had it */
 267         }
 268         sline[lno].p_lno[n] = p_lno; /* trailer */
 269 }
 270
 271 static unsigned long context = 3;
 272 static char combine_marker = '@';
 273
 274 static int interesting(struct sline *sline, unsigned long all_mask)
 275 {
 276         /* If some parents lost lines here, or if we have added to
 277          * some parent, it is interesting.
 278          */
 279         return ((sline->flag & all_mask) || sline->lost_head);
 280 }
 281
 282 static unsigned long adjust_hunk_tail(struct sline *sline,
 283                                       unsigned long all_mask,
 284                                       unsigned long hunk_begin,
 285                                       unsigned long i)
 286 {
 287         /* i points at the first uninteresting line.  If the last line
 288          * of the hunk was interesting only because it has some
 289          * deletion, then it is not all that interesting for the
 290          * purpose of giving trailing context lines.  This is because
 291          * we output '-' line and then unmodified sline[i-1] itself in
 292          * that case which gives us one extra context line.
 293          */
 294         if ((hunk_begin + 1 <= i) && !(sline[i-1].flag & all_mask))
 295                 i--;
 296         return i;
 297 }
 298
 299 static unsigned long find_next(struct sline *sline,
 300                                unsigned long mark,
 301                                unsigned long i,
 302                                unsigned long cnt,
 303                                int look_for_uninteresting)
 304 {
 305         /* We have examined up to i-1 and are about to look at i.
 306          * Find next interesting or uninteresting line.  Here,
 307          * "interesting" does not mean interesting(), but marked by
 308          * the give_context() function below (i.e. it includes context
 309          * lines that are not interesting to interesting() function
 310          * that are surrounded by interesting() ones.
 311          */
 312         while (i <= cnt)
 313                 if (look_for_uninteresting
 314                     ? !(sline[i].flag & mark)
 315                     : (sline[i].flag & mark))
 316                         return i;
 317                 else
 318                         i++;
 319         return i;
 320 }
 321
 322 static int give_context(struct sline *sline, unsigned long cnt, int num_parent)
 323 {
 324         unsigned long all_mask = (1UL<<num_parent) - 1;
 325         unsigned long mark = (1UL<<num_parent);
 326         unsigned long no_pre_delete = (2UL<<num_parent);
 327         unsigned long i;
 328
 329         /* Two groups of interesting lines may have a short gap of
 330          * uninteresting lines.  Connect such groups to give them a
 331          * bit of context.
 332          *
 333          * We first start from what the interesting() function says,
 334          * and mark them with "mark", and paint context lines with the
 335          * mark.  So interesting() would still say false for such context
 336          * lines but they are treated as "interesting" in the end.
 337          */
 338         i = find_next(sline, mark, 0, cnt, 0);
 339         if (cnt < i)
 340                 return 0;
 341
 342         while (i <= cnt) {
 343                 unsigned long j = (context < i) ? (i - context) : 0;
 344                 unsigned long k;
 345
 346                 /* Paint a few lines before the first interesting line. */
 347                 while (j < i)
 348                         sline[j++].flag |= mark | no_pre_delete;
 349
 350         again:
 351                 /* we know up to i is to be included.  where does the
 352                  * next uninteresting one start?
 353                  */
 354                 j = find_next(sline, mark, i, cnt, 1);
 355                 if (cnt < j)
 356                         break; /* the rest are all interesting */
 357
 358                 /* lookahead context lines */
 359                 k = find_next(sline, mark, j, cnt, 0);
 360                 j = adjust_hunk_tail(sline, all_mask, i, j);
 361
 362                 if (k < j + context) {
 363                         /* k is interesting and [j,k) are not, but
 364                          * paint them interesting because the gap is small.
 365                          */
 366                         while (j < k)
 367                                 sline[j++].flag |= mark;
 368                         i = k;
 369                         goto again;
 370                 }
 371
 372                 /* j is the first uninteresting line and there is
 373                  * no overlap beyond it within context lines.  Paint
 374                  * the trailing edge a bit.
 375                  */
 376                 i = k;
 377                 k = (j + context < cnt+1) ? j + context : cnt+1;
 378                 while (j < k)
 379                         sline[j++].flag |= mark;
 380         }
 381         return 1;
 382 }
 383
 384 static int make_hunks(struct sline *sline, unsigned long cnt,
 385                        int num_parent, int dense)
 386 {
 387         unsigned long all_mask = (1UL<<num_parent) - 1;
 388         unsigned long mark = (1UL<<num_parent);
 389         unsigned long i;
 390         int has_interesting = 0;
 391
 392         for (i = 0; i <= cnt; i++) {
 393                 if (interesting(&sline[i], all_mask))
 394                         sline[i].flag |= mark;
 395                 else
 396                         sline[i].flag &= ~mark;
 397         }
 398         if (!dense)
 399                 return give_context(sline, cnt, num_parent);
 400
 401         /* Look at each hunk, and if we have changes from only one
 402          * parent, or the changes are the same from all but one
 403          * parent, mark that uninteresting.
 404          */
 405         i = 0;
 406         while (i <= cnt) {
 407                 unsigned long j, hunk_begin, hunk_end;
 408                 unsigned long same_diff;
 409                 while (i <= cnt && !(sline[i].flag & mark))
 410                         i++;
 411                 if (cnt < i)
 412                         break; /* No more interesting hunks */
 413                 hunk_begin = i;
 414                 for (j = i + 1; j <= cnt; j++) {
 415                         if (!(sline[j].flag & mark)) {
 416                                 /* Look beyond the end to see if there
 417                                  * is an interesting line after this
 418                                  * hunk within context span.
 419                                  */
 420                                 unsigned long la; /* lookahead */
 421                                 int contin = 0;
 422                                 la = adjust_hunk_tail(sline, all_mask,
 423                                                      hunk_begin, j);
 424                                 la = (la + context < cnt + 1) ?
 425                                         (la + context) : cnt + 1;
 426                                 while (la && j <= --la) {
 427                                         if (sline[la].flag & mark) {
 428                                                 contin = 1;
 429                                                 break;
 430                                         }
 431                                 }
 432                                 if (!contin)
 433                                         break;
 434                                 j = la;
 435                         }
 436                 }
 437                 hunk_end = j;
 438
 439                 /* [i..hunk_end) are interesting.  Now is it really
 440                  * interesting?  We check if there are only two versions
 441                  * and the result matches one of them.  That is, we look
 442                  * at:
 443                  *   (+) line, which records lines added to which parents;
 444                  *       this line appears in the result.
 445                  *   (-) line, which records from what parents the line
 446                  *       was removed; this line does not appear in the result.
 447                  * then check the set of parents the result has difference
 448                  * from, from all lines.  If there are lines that has
 449                  * different set of parents that the result has differences
 450                  * from, that means we have more than two versions.
 451                  *
 452                  * Even when we have only two versions, if the result does
 453                  * not match any of the parents, the it should be considered
 454                  * interesting.  In such a case, we would have all '+' line.
 455                  * After passing the above "two versions" test, that would
 456                  * appear as "the same set of parents" to be "all parents".
 457                  */
 458                 same_diff = 0;
 459                 has_interesting = 0;
 460                 for (j = i; j < hunk_end && !has_interesting; j++) {
 461                         unsigned long this_diff = sline[j].flag & all_mask;
 462                         struct lline *ll = sline[j].lost_head;
 463                         if (this_diff) {
 464                                 /* This has some changes.  Is it the
 465                                  * same as others?
 466                                  */
 467                                 if (!same_diff)
 468                                         same_diff = this_diff;
 469                                 else if (same_diff != this_diff) {
 470                                         has_interesting = 1;
 471                                         break;
 472                                 }
 473                         }
 474                         while (ll && !has_interesting) {
 475                                 /* Lost this line from these parents;
 476                                  * who are they?  Are they the same?
 477                                  */
 478                                 this_diff = ll->parent_map;
 479                                 if (!same_diff)
 480                                         same_diff = this_diff;
 481                                 else if (same_diff != this_diff) {
 482                                         has_interesting = 1;
 483                                 }
 484                                 ll = ll->next;
 485                         }
 486                 }
 487
 488                 if (!has_interesting && same_diff != all_mask) {
 489                         /* This hunk is not that interesting after all */
 490                         for (j = hunk_begin; j < hunk_end; j++)
 491                                 sline[j].flag &= ~mark;
 492                 }
 493                 i = hunk_end;
 494         }
 495
 496         has_interesting = give_context(sline, cnt, num_parent);
 497         return has_interesting;
 498 }
 499
 500 static void show_parent_lno(struct sline *sline, unsigned long l0, unsigned long l1, int n, unsigned long null_context)
 501 {
 502         l0 = sline[l0].p_lno[n];
 503         l1 = sline[l1].p_lno[n];
 504         printf(" -%lu,%lu", l0, l1-l0-null_context);
 505 }
 506
 507 static int hunk_comment_line(const char *bol)
 508 {
 509         int ch;
 510
 511         if (!bol)
 512                 return 0;
 513         ch = *bol & 0xff;
 514         return (isalpha(ch) || ch == '_' || ch == '$');
 515 }
 516
 517 static void show_line_to_eol(const char *line, int len, const char *reset)
 518 {
 519         int saw_cr_at_eol = 0;
 520         if (len < 0)
 521                 len = strlen(line);
 522         saw_cr_at_eol = (len && line[len-1] == '\r');
 523
 524         printf("%.*s%s%s\n", len - saw_cr_at_eol, line,
 525                reset,
 526                saw_cr_at_eol ? "\r" : "");
 527 }
 528
 529 static void dump_sline(struct sline *sline, unsigned long cnt, int num_parent,
 530                        int use_color, int result_deleted)
 531 {
 532         unsigned long mark = (1UL<<num_parent);
 533         unsigned long no_pre_delete = (2UL<<num_parent);
 534         int i;
 535         unsigned long lno = 0;
 536         const char *c_frag = diff_get_color(use_color, DIFF_FRAGINFO);
 537         const char *c_func = diff_get_color(use_color, DIFF_FUNCINFO);
 538         const char *c_new = diff_get_color(use_color, DIFF_FILE_NEW);
 539         const char *c_old = diff_get_color(use_color, DIFF_FILE_OLD);
 540         const char *c_plain = diff_get_color(use_color, DIFF_PLAIN);
 541         const char *c_reset = diff_get_color(use_color, DIFF_RESET);
 542
 543         if (result_deleted)
 544                 return; /* result deleted */
 545
 546         while (1) {
 547                 unsigned long hunk_end;
 548                 unsigned long rlines;
 549                 const char *hunk_comment = NULL;
 550                 unsigned long null_context = 0;
 551
 552                 while (lno <= cnt && !(sline[lno].flag & mark)) {
 553                         if (hunk_comment_line(sline[lno].bol))
 554                                 hunk_comment = sline[lno].bol;
 555                         lno++;
 556                 }
 557                 if (cnt < lno)
 558                         break;
 559                 else {
 560                         for (hunk_end = lno + 1; hunk_end <= cnt; hunk_end++)
 561                                 if (!(sline[hunk_end].flag & mark))
 562                                         break;
 563                 }
 564                 rlines = hunk_end - lno;
 565                 if (cnt < hunk_end)
 566                         rlines--; /* pointing at the last delete hunk */
 567
 568                 if (!context) {
 569                         /*
 570                          * Even when running with --unified=0, all
 571                          * lines in the hunk needs to be processed in
 572                          * the loop below in order to show the
 573                          * deletion recorded in lost_head.  However,
 574                          * we do not want to show the resulting line
 575                          * with all blank context markers in such a
 576                          * case.  Compensate.
 577                          */
 578                         unsigned long j;
 579                         for (j = lno; j < hunk_end; j++)
 580                                 if (!(sline[j].flag & (mark-1)))
 581                                         null_context++;
 582                         rlines -= null_context;
 583                 }
 584
 585                 fputs(c_frag, stdout);
 586                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 587                 for (i = 0; i < num_parent; i++)
 588                         show_parent_lno(sline, lno, hunk_end, i, null_context);
 589                 printf(" +%lu,%lu ", lno+1, rlines);
 590                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 591
 592                 if (hunk_comment) {
 593                         int comment_end = 0;
 594                         for (i = 0; i < 40; i++) {
 595                                 int ch = hunk_comment[i] & 0xff;
 596                                 if (!ch || ch == '\n')
 597                                         break;
 598                                 if (!isspace(ch))
 599                                     comment_end = i;
 600                         }
 601                         if (comment_end)
 602                                 printf("%s%s %s%s", c_reset,
 603                                                     c_plain, c_reset,
 604                                                     c_func);
 605                         for (i = 0; i < comment_end; i++)
 606                                 putchar(hunk_comment[i]);
 607                 }
 608
 609                 printf("%s\n", c_reset);
 610                 while (lno < hunk_end) {
 611                         struct lline *ll;
 612                         int j;
 613                         unsigned long p_mask;
 614                         struct sline *sl = &sline[lno++];
 615                         ll = (sl->flag & no_pre_delete) ? NULL : sl->lost_head;
 616                         while (ll) {
 617                                 fputs(c_old, stdout);
 618                                 for (j = 0; j < num_parent; j++) {
 619                                         if (ll->parent_map & (1UL<<j))
 620                                                 putchar('-');
 621                                         else
 622                                                 putchar(' ');
 623                                 }
 624                                 show_line_to_eol(ll->line, -1, c_reset);
 625                                 ll = ll->next;
 626                         }
 627                         if (cnt < lno)
 628                                 break;
 629                         p_mask = 1;
 630                         if (!(sl->flag & (mark-1))) {
 631                                 /*
 632                                  * This sline was here to hang the
 633                                  * lost lines in front of it.
 634                                  */
 635                                 if (!context)
 636                                         continue;
 637                                 fputs(c_plain, stdout);
 638                         }
 639                         else
 640                                 fputs(c_new, stdout);
 641                         for (j = 0; j < num_parent; j++) {
 642                                 if (p_mask & sl->flag)
 643                                         putchar('+');
 644                                 else
 645                                         putchar(' ');
 646                                 p_mask <<= 1;
 647                         }
 648                         show_line_to_eol(sl->bol, sl->len, c_reset);
 649                 }
 650         }
 651 }
 652
 653 static void reuse_combine_diff(struct sline *sline, unsigned long cnt,
 654                                int i, int j)
 655 {
 656         /* We have already examined parent j and we know parent i
 657          * and parent j are the same, so reuse the combined result
 658          * of parent j for parent i.
 659          */
 660         unsigned long lno, imask, jmask;
 661         imask = (1UL<<i);
 662         jmask = (1UL<<j);
 663
 664         for (lno = 0; lno <= cnt; lno++) {
 665                 struct lline *ll = sline->lost_head;
 666                 sline->p_lno[i] = sline->p_lno[j];
 667                 while (ll) {
 668                         if (ll->parent_map & jmask)
 669                                 ll->parent_map |= imask;
 670                         ll = ll->next;
 671                 }
 672                 if (sline->flag & jmask)
 673                         sline->flag |= imask;
 674                 sline++;
 675         }
 676         /* the overall size of the file (sline[cnt]) */
 677         sline->p_lno[i] = sline->p_lno[j];
 678 }
 679
 680 static void dump_quoted_path(const char *head,
 681                              const char *prefix,
 682                              const char *path,
 683                              const char *c_meta, const char *c_reset)
 684 {
 685         static struct strbuf buf = STRBUF_INIT;
 686
 687         strbuf_reset(&buf);
 688         strbuf_addstr(&buf, c_meta);
 689         strbuf_addstr(&buf, head);
 690         quote_two_c_style(&buf, prefix, path, 0);
 691         strbuf_addstr(&buf, c_reset);
 692         puts(buf.buf);
 693 }
 694
 695 static void show_combined_header(struct combine_diff_path *elem,
 696                                  int num_parent,
 697                                  int dense,
 698                                  struct rev_info *rev,
 699                                  int mode_differs,
 700                                  int show_file_header)
 701 {
 702         struct diff_options *opt = &rev->diffopt;
 703         int abbrev = DIFF_OPT_TST(opt, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
 704         const char *a_prefix = opt->a_prefix ? opt->a_prefix : "a/";
 705         const char *b_prefix = opt->b_prefix ? opt->b_prefix : "b/";
 706         const char *c_meta = diff_get_color_opt(opt, DIFF_METAINFO);
 707         const char *c_reset = diff_get_color_opt(opt, DIFF_RESET);
 708         const char *abb;
 709         int added = 0;
 710         int deleted = 0;
 711         int i;
 712
 713         if (rev->loginfo && !rev->no_commit_id)
 714                 show_log(rev);
 715
 716         dump_quoted_path(dense ? "diff --cc " : "diff --combined ",
 717                          "", elem->path, c_meta, c_reset);
 718         printf("%sindex ", c_meta);
 719         for (i = 0; i < num_parent; i++) {
 720                 abb = find_unique_abbrev(elem->parent[i].sha1,
 721                                          abbrev);
 722                 printf("%s%s", i ? "," : "", abb);
 723         }
 724         abb = find_unique_abbrev(elem->sha1, abbrev);
 725         printf("..%s%s\n", abb, c_reset);
 726
 727         if (mode_differs) {
 728                 deleted = !elem->mode;
 729
 730                 /* We say it was added if nobody had it */
 731                 added = !deleted;
 732                 for (i = 0; added && i < num_parent; i++)
 733                         if (elem->parent[i].status !=
 734                             DIFF_STATUS_ADDED)
 735                                 added = 0;
 736                 if (added)
 737                         printf("%snew file mode %06o",
 738                                c_meta, elem->mode);
 739                 else {
 740                         if (deleted)
 741                                 printf("%sdeleted file ", c_meta);
 742                         printf("mode ");
 743                         for (i = 0; i < num_parent; i++) {
 744                                 printf("%s%06o", i ? "," : "",
 745                                        elem->parent[i].mode);
 746                         }
 747                         if (elem->mode)
 748                                 printf("..%06o", elem->mode);
 749                 }
 750                 printf("%s\n", c_reset);
 751         }
 752
 753         if (!show_file_header)
 754                 return;
 755
 756         if (added)
 757                 dump_quoted_path("--- ", "", "/dev/null",
 758                                  c_meta, c_reset);
 759         else
 760                 dump_quoted_path("--- ", a_prefix, elem->path,
 761                                  c_meta, c_reset);
 762         if (deleted)
 763                 dump_quoted_path("+++ ", "", "/dev/null",
 764                                  c_meta, c_reset);
 765         else
 766                 dump_quoted_path("+++ ", b_prefix, elem->path,
 767                                  c_meta, c_reset);
 768 }
 769
 770 static void show_patch_diff(struct combine_diff_path *elem, int num_parent,
 771                             int dense, int working_tree_file,
 772                             struct rev_info *rev)
 773 {
 774         struct diff_options *opt = &rev->diffopt;
 775         unsigned long result_size, cnt, lno;
 776         int result_deleted = 0;
 777         char *result, *cp;
 778         struct sline *sline; /* survived lines */
 779         int mode_differs = 0;
 780         int i, show_hunks;
 781         mmfile_t result_file;
 782         struct userdiff_driver *userdiff;
 783         struct userdiff_driver *textconv = NULL;
 784         int is_binary;
 785
 786         context = opt->context;
 787         userdiff = userdiff_find_by_path(elem->path);
 788         if (!userdiff)
 789                 userdiff = userdiff_find_by_name("default");
 790         if (DIFF_OPT_TST(opt, ALLOW_TEXTCONV))
 791                 textconv = userdiff_get_textconv(userdiff);
 792
 793         /* Read the result of merge first */
 794         if (!working_tree_file)
 795                 result = grab_blob(elem->sha1, elem->mode, &result_size,
 796                                    textconv, elem->path);
 797         else {
 798                 /* Used by diff-tree to read from the working tree */
 799                 struct stat st;
 800                 int fd = -1;
 801
 802                 if (lstat(elem->path, &st) < 0)
 803                         goto deleted_file;
 804
 805                 if (S_ISLNK(st.st_mode)) {
 806                         struct strbuf buf = STRBUF_INIT;
 807
 808                         if (strbuf_readlink(&buf, elem->path, st.st_size) < 0) {
 809                                 error("readlink(%s): %s", elem->path,
 810                                       strerror(errno));
 811                                 return;
 812                         }
 813                         result_size = buf.len;
 814                         result = strbuf_detach(&buf, NULL);
 815                         elem->mode = canon_mode(st.st_mode);
 816                 } else if (S_ISDIR(st.st_mode)) {
 817                         unsigned char sha1[20];
 818                         if (resolve_gitlink_ref(elem->path, "HEAD", sha1) < 0)
 819                                 result = grab_blob(elem->sha1, elem->mode,
 820                                                    &result_size, NULL, NULL);
 821                         else
 822                                 result = grab_blob(sha1, elem->mode,
 823                                                    &result_size, NULL, NULL);
 824                 } else if (textconv) {
 825                         struct diff_filespec *df = alloc_filespec(elem->path);
 826                         fill_filespec(df, null_sha1, 0, st.st_mode);
 827                         result_size = fill_textconv(textconv, df, &result);
 828                         free_filespec(df);
 829                 } else if (0 <= (fd = open(elem->path, O_RDONLY))) {
 830                         size_t len = xsize_t(st.st_size);
 831                         ssize_t done;
 832                         int is_file, i;
 833
 834                         elem->mode = canon_mode(st.st_mode);
 835                         /* if symlinks don't work, assume symlink if all parents
 836                          * are symlinks
 837                          */
 838                         is_file = has_symlinks;
 839                         for (i = 0; !is_file && i < num_parent; i++)
 840                                 is_file = !S_ISLNK(elem->parent[i].mode);
 841                         if (!is_file)
 842                                 elem->mode = canon_mode(S_IFLNK);
 843
 844                         result_size = len;
 845                         result = xmalloc(len + 1);
 846
 847                         done = read_in_full(fd, result, len);
 848                         if (done < 0)
 849                                 die_errno("read error '%s'", elem->path);
 850                         else if (done < len)
 851                                 die("early EOF '%s'", elem->path);
 852
 853                         result[len] = 0;
 854
 855                         /* If not a fake symlink, apply filters, e.g. autocrlf */
 856                         if (is_file) {
 857                                 struct strbuf buf = STRBUF_INIT;
 858
 859                                 if (convert_to_git(elem->path, result, len, &buf, safe_crlf)) {
 860                                         free(result);
 861                                         result = strbuf_detach(&buf, &len);
 862                                         result_size = len;
 863                                 }
 864                         }
 865                 }
 866                 else {
 867                 deleted_file:
 868                         result_deleted = 1;
 869                         result_size = 0;
 870                         elem->mode = 0;
 871                         result = xcalloc(1, 1);
 872                 }
 873
 874                 if (0 <= fd)
 875                         close(fd);
 876         }
 877
 878         for (i = 0; i < num_parent; i++) {
 879                 if (elem->parent[i].mode != elem->mode) {
 880                         mode_differs = 1;
 881                         break;
 882                 }
 883         }
 884
 885         if (textconv)
 886                 is_binary = 0;
 887         else if (userdiff->binary != -1)
 888                 is_binary = userdiff->binary;
 889         else {
 890                 is_binary = buffer_is_binary(result, result_size);
 891                 for (i = 0; !is_binary && i < num_parent; i++) {
 892                         char *buf;
 893                         unsigned long size;
 894                         buf = grab_blob(elem->parent[i].sha1,
 895                                         elem->parent[i].mode,
 896                                         &size, NULL, NULL);
 897                         if (buffer_is_binary(buf, size))
 898                                 is_binary = 1;
 899                         free(buf);
 900                 }
 901         }
 902         if (is_binary) {
 903                 show_combined_header(elem, num_parent, dense, rev,
 904                                      mode_differs, 0);
 905                 printf("Binary files differ\n");
 906                 free(result);
 907                 return;
 908         }
 909
 910         for (cnt = 0, cp = result; cp < result + result_size; cp++) {
 911                 if (*cp == '\n')
 912                         cnt++;
 913         }
 914         if (result_size && result[result_size-1] != '\n')
 915                 cnt++; /* incomplete line */
 916
 917         sline = xcalloc(cnt+2, sizeof(*sline));
 918         sline[0].bol = result;
 919         for (lno = 0; lno <= cnt + 1; lno++) {
 920                 sline[lno].lost_tail = &sline[lno].lost_head;
 921                 sline[lno].flag = 0;
 922         }
 923         for (lno = 0, cp = result; cp < result + result_size; cp++) {
 924                 if (*cp == '\n') {
 925                         sline[lno].len = cp - sline[lno].bol;
 926                         lno++;
 927                         if (lno < cnt)
 928                                 sline[lno].bol = cp + 1;
 929                 }
 930         }
 931         if (result_size && result[result_size-1] != '\n')
 932                 sline[cnt-1].len = result_size - (sline[cnt-1].bol - result);
 933
 934         result_file.ptr = result;
 935         result_file.size = result_size;
 936
 937         /* Even p_lno[cnt+1] is valid -- that is for the end line number
 938          * for deletion hunk at the end.
 939          */
 940         sline[0].p_lno = xcalloc((cnt+2) * num_parent, sizeof(unsigned long));
 941         for (lno = 0; lno <= cnt; lno++)
 942                 sline[lno+1].p_lno = sline[lno].p_lno + num_parent;
 943
 944         for (i = 0; i < num_parent; i++) {
 945                 int j;
 946                 for (j = 0; j < i; j++) {
 947                         if (!hashcmp(elem->parent[i].sha1,
 948                                      elem->parent[j].sha1)) {
 949                                 reuse_combine_diff(sline, cnt, i, j);
 950                                 break;
 951                         }
 952                 }
 953                 if (i <= j)
 954                         combine_diff(elem->parent[i].sha1,
 955                                      elem->parent[i].mode,
 956                                      &result_file, sline,
 957                                      cnt, i, num_parent, result_deleted,
 958                                      textconv, elem->path);
 959         }
 960
 961         show_hunks = make_hunks(sline, cnt, num_parent, dense);
 962
 963         if (show_hunks || mode_differs || working_tree_file) {
 964                 show_combined_header(elem, num_parent, dense, rev,
 965                                      mode_differs, 1);
 966                 dump_sline(sline, cnt, num_parent,
 967                            opt->use_color, result_deleted);
 968         }
 969         free(result);
 970
 971         for (lno = 0; lno < cnt; lno++) {
 972                 if (sline[lno].lost_head) {
 973                         struct lline *ll = sline[lno].lost_head;
 974                         while (ll) {
 975                                 struct lline *tmp = ll;
 976                                 ll = ll->next;
 977                                 free(tmp);
 978                         }
 979                 }
 980         }
 981         free(sline[0].p_lno);
 982         free(sline);
 983 }
 984
 985 static void show_raw_diff(struct combine_diff_path *p, int num_parent, struct rev_info *rev)
 986 {
 987         struct diff_options *opt = &rev->diffopt;
 988         int line_termination, inter_name_termination, i;
 989
 990         line_termination = opt->line_termination;
 991         inter_name_termination = '\t';
 992         if (!line_termination)
 993                 inter_name_termination = 0;
 994
 995         if (rev->loginfo && !rev->no_commit_id)
 996                 show_log(rev);
 997
 998         if (opt->output_format & DIFF_FORMAT_RAW) {
 999                 /* As many colons as there are parents */
1000                 for (i = 0; i < num_parent; i++)
1001                         putchar(':');
1002
1003                 /* Show the modes */
1004                 for (i = 0; i < num_parent; i++)
1005                         printf("%06o ", p->parent[i].mode);
1006                 printf("%06o", p->mode);
1007
1008                 /* Show sha1's */
1009                 for (i = 0; i < num_parent; i++)
1010                         printf(" %s", diff_unique_abbrev(p->parent[i].sha1,
1011                                                          opt->abbrev));
1012                 printf(" %s ", diff_unique_abbrev(p->sha1, opt->abbrev));
1013         }
1014
1015         if (opt->output_format & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS)) {
1016                 for (i = 0; i < num_parent; i++)
1017                         putchar(p->parent[i].status);
1018                 putchar(inter_name_termination);
1019         }
1020
1021         write_name_quoted(p->path, stdout, line_termination);
1022 }
1023
1024 /*
1025  * The result (p->elem) is from the working tree and their
1026  * parents are typically from multiple stages during a merge
1027  * (i.e. diff-files) or the state in HEAD and in the index
1028  * (i.e. diff-index).
1029  */
1030 void show_combined_diff(struct combine_diff_path *p,
1031                        int num_parent,
1032                        int dense,
1033                        struct rev_info *rev)
1034 {
1035         struct diff_options *opt = &rev->diffopt;
1036         if (!p->len)
1037                 return;
1038         if (opt->output_format & (DIFF_FORMAT_RAW |
1039                                   DIFF_FORMAT_NAME |
1040                                   DIFF_FORMAT_NAME_STATUS))
1041                 show_raw_diff(p, num_parent, rev);
1042         else if (opt->output_format & DIFF_FORMAT_PATCH)
1043                 show_patch_diff(p, num_parent, dense, 1, rev);
1044 }
1045
1046 static void free_combined_pair(struct diff_filepair *pair)
1047 {
1048         free(pair->two);
1049         free(pair);
1050 }
1051
1052 /*
1053  * A combine_diff_path expresses N parents on the LHS against 1 merge
1054  * result. Synthesize a diff_filepair that has N entries on the "one"
1055  * side and 1 entry on the "two" side.
1056  *
1057  * In the future, we might want to add more data to combine_diff_path
1058  * so that we can fill fields we are ignoring (most notably, size) here,
1059  * but currently nobody uses it, so this should suffice for now.
1060  */
1061 static struct diff_filepair *combined_pair(struct combine_diff_path *p,
1062                                            int num_parent)
1063 {
1064         int i;
1065         struct diff_filepair *pair;
1066         struct diff_filespec *pool;
1067
1068         pair = xmalloc(sizeof(*pair));
1069         pool = xcalloc(num_parent + 1, sizeof(struct diff_filespec));
1070         pair->one = pool + 1;
1071         pair->two = pool;
1072
1073         for (i = 0; i < num_parent; i++) {
1074                 pair->one[i].path = p->path;
1075                 pair->one[i].mode = p->parent[i].mode;
1076                 hashcpy(pair->one[i].sha1, p->parent[i].sha1);
1077                 pair->one[i].sha1_valid = !is_null_sha1(p->parent[i].sha1);
1078                 pair->one[i].has_more_entries = 1;
1079         }
1080         pair->one[num_parent - 1].has_more_entries = 0;
1081
1082         pair->two->path = p->path;
1083         pair->two->mode = p->mode;
1084         hashcpy(pair->two->sha1, p->sha1);
1085         pair->two->sha1_valid = !is_null_sha1(p->sha1);
1086         return pair;
1087 }
1088
1089 static void handle_combined_callback(struct diff_options *opt,
1090                                      struct combine_diff_path *paths,
1091                                      int num_parent,
1092                                      int num_paths)
1093 {
1094         struct combine_diff_path *p;
1095         struct diff_queue_struct q;
1096         int i;
1097
1098         q.queue = xcalloc(num_paths, sizeof(struct diff_filepair *));
1099         q.alloc = num_paths;
1100         q.nr = num_paths;
1101         for (i = 0, p = paths; p; p = p->next) {
1102                 if (!p->len)
1103                         continue;
1104                 q.queue[i++] = combined_pair(p, num_parent);
1105         }
1106         opt->format_callback(&q, opt, opt->format_callback_data);
1107         for (i = 0; i < num_paths; i++)
1108                 free_combined_pair(q.queue[i]);
1109         free(q.queue);
1110 }
1111
1112 void diff_tree_combined(const unsigned char *sha1,
1113                         const struct sha1_array *parents,
1114                         int dense,
1115                         struct rev_info *rev)
1116 {
1117         struct diff_options *opt = &rev->diffopt;
1118         struct diff_options diffopts;
1119         struct combine_diff_path *p, *paths = NULL;
1120         int i, num_paths, needsep, show_log_first, num_parent = parents->nr;
1121
1122         diffopts = *opt;
1123         diffopts.output_format = DIFF_FORMAT_NO_OUTPUT;
1124         DIFF_OPT_SET(&diffopts, RECURSIVE);
1125         DIFF_OPT_CLR(&diffopts, ALLOW_EXTERNAL);
1126
1127         show_log_first = !!rev->loginfo && !rev->no_commit_id;
1128         needsep = 0;
1129         /* find set of paths that everybody touches */
1130         for (i = 0; i < num_parent; i++) {
1131                 /* show stat against the first parent even
1132                  * when doing combined diff.
1133                  */
1134                 int stat_opt = (opt->output_format &
1135                                 (DIFF_FORMAT_NUMSTAT|DIFF_FORMAT_DIFFSTAT));
1136                 if (i == 0 && stat_opt)
1137                         diffopts.output_format = stat_opt;
1138                 else
1139                         diffopts.output_format = DIFF_FORMAT_NO_OUTPUT;
1140                 diff_tree_sha1(parents->sha1[i], sha1, "", &diffopts);
1141                 diffcore_std(&diffopts);
1142                 paths = intersect_paths(paths, i, num_parent);
1143
1144                 if (show_log_first && i == 0) {
1145                         show_log(rev);
1146                         if (rev->verbose_header && opt->output_format)
1147                                 putchar(opt->line_termination);
1148                 }
1149                 diff_flush(&diffopts);
1150         }
1151
1152         /* find out surviving paths */
1153         for (num_paths = 0, p = paths; p; p = p->next) {
1154                 if (p->len)
1155                         num_paths++;
1156         }
1157         if (num_paths) {
1158                 if (opt->output_format & (DIFF_FORMAT_RAW |
1159                                           DIFF_FORMAT_NAME |
1160                                           DIFF_FORMAT_NAME_STATUS)) {
1161                         for (p = paths; p; p = p->next) {
1162                                 if (p->len)
1163                                         show_raw_diff(p, num_parent, rev);
1164                         }
1165                         needsep = 1;
1166                 }
1167                 else if (opt->output_format &
1168                          (DIFF_FORMAT_NUMSTAT|DIFF_FORMAT_DIFFSTAT))
1169                         needsep = 1;
1170                 else if (opt->output_format & DIFF_FORMAT_CALLBACK)
1171                         handle_combined_callback(opt, paths, num_parent, num_paths);
1172
1173                 if (opt->output_format & DIFF_FORMAT_PATCH) {
1174                         if (needsep)
1175                                 putchar(opt->line_termination);
1176                         for (p = paths; p; p = p->next) {
1177                                 if (p->len)
1178                                         show_patch_diff(p, num_parent, dense,
1179                                                         0, rev);
1180                         }
1181                 }
1182         }
1183
1184         /* Clean things up */
1185         while (paths) {
1186                 struct combine_diff_path *tmp = paths;
1187                 paths = paths->next;
1188                 free(tmp);
1189         }
1190 }
1191
1192 void diff_tree_combined_merge(const struct commit *commit, int dense,
1193                               struct rev_info *rev)
1194 {
1195         struct commit_list *parent = commit->parents;
1196         struct sha1_array parents = SHA1_ARRAY_INIT;
1197
1198         while (parent) {
1199                 sha1_array_append(&parents, parent->item->object.sha1);
1200                 parent = parent->next;
1201         }
1202         diff_tree_combined(commit->object.sha1, &parents, dense, rev);
1203         sha1_array_clear(&parents);
1204 }