combine-diff.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "diff.h"
   4 #include "diffcore.h"
   5 #include "quote.h"
   6
   7 static int uninteresting(struct diff_filepair *p)
   8 {
   9         if (diff_unmodified_pair(p))
  10                 return 1;
  11         if (!S_ISREG(p->one->mode) || !S_ISREG(p->two->mode))
  12                 return 1;
  13         return 0;
  14 }
  15
  16 static struct combine_diff_path *intersect_paths(struct combine_diff_path *curr, int n, int num_parent)
  17 {
  18         struct diff_queue_struct *q = &diff_queued_diff;
  19         struct combine_diff_path *p;
  20         int i;
  21
  22         if (!n) {
  23                 struct combine_diff_path *list = NULL, **tail = &list;
  24                 for (i = 0; i < q->nr; i++) {
  25                         int len;
  26                         const char *path;
  27                         if (uninteresting(q->queue[i]))
  28                                 continue;
  29                         path = q->queue[i]->two->path;
  30                         len = strlen(path);
  31
  32                         p = xmalloc(sizeof(*p) + len + 1 + num_parent * 20);
  33                         p->path = (char*) &(p->parent_sha1[num_parent][0]);
  34                         memcpy(p->path, path, len);
  35                         p->path[len] = 0;
  36                         p->len = len;
  37                         p->next = NULL;
  38                         memcpy(p->sha1, q->queue[i]->two->sha1, 20);
  39                         memcpy(p->parent_sha1[n], q->queue[i]->one->sha1, 20);
  40                         *tail = p;
  41                         tail = &p->next;
  42                 }
  43                 return list;
  44         }
  45
  46         for (p = curr; p; p = p->next) {
  47                 int found = 0;
  48                 if (!p->len)
  49                         continue;
  50                 for (i = 0; i < q->nr; i++) {
  51                         const char *path;
  52                         int len;
  53
  54                         if (uninteresting(q->queue[i]))
  55                                 continue;
  56                         path = q->queue[i]->two->path;
  57                         len = strlen(path);
  58                         if (len == p->len && !memcmp(path, p->path, len)) {
  59                                 found = 1;
  60                                 memcpy(p->parent_sha1[n],
  61                                        q->queue[i]->one->sha1, 20);
  62                                 break;
  63                         }
  64                 }
  65                 if (!found)
  66                         p->len = 0;
  67         }
  68         return curr;
  69 }
  70
  71 struct lline {
  72         struct lline *next;
  73         int len;
  74         unsigned long parent_map;
  75         char line[FLEX_ARRAY];
  76 };
  77
  78 struct sline {
  79         struct lline *lost_head, **lost_tail;
  80         char *bol;
  81         int len;
  82         unsigned long flag;
  83 };
  84
  85 static char *grab_blob(const unsigned char *sha1, unsigned long *size)
  86 {
  87         char *blob;
  88         char type[20];
  89         if (!memcmp(sha1, null_sha1, 20)) {
  90                 /* deleted blob */
  91                 *size = 0;
  92                 return xcalloc(1, 1);
  93         }
  94         blob = read_sha1_file(sha1, type, size);
  95         if (strcmp(type, "blob"))
  96                 die("object '%s' is not a blob!", sha1_to_hex(sha1));
  97         return blob;
  98 }
  99
 100 #define TMPPATHLEN 50
 101 #define MAXLINELEN 10240
 102
 103 static void write_to_temp_file(char *tmpfile, void *blob, unsigned long size)
 104 {
 105         int fd = git_mkstemp(tmpfile, TMPPATHLEN, ".diff_XXXXXX");
 106         if (fd < 0)
 107                 die("unable to create temp-file");
 108         if (write(fd, blob, size) != size)
 109                 die("unable to write temp-file");
 110         close(fd);
 111 }
 112
 113 static void write_temp_blob(char *tmpfile, const unsigned char *sha1)
 114 {
 115         unsigned long size;
 116         void *blob;
 117         blob = grab_blob(sha1, &size);
 118         write_to_temp_file(tmpfile, blob, size);
 119         free(blob);
 120 }
 121
 122 static int parse_num(char **cp_p, unsigned int *num_p)
 123 {
 124         char *cp = *cp_p;
 125         unsigned int num = 0;
 126         int read_some;
 127
 128         while ('0' <= *cp && *cp <= '9')
 129                 num = num * 10 + *cp++ - '0';
 130         if (!(read_some = cp - *cp_p))
 131                 return -1;
 132         *cp_p = cp;
 133         *num_p = num;
 134         return 0;
 135 }
 136
 137 static int parse_hunk_header(char *line, int len,
 138                              unsigned int *ob, unsigned int *on,
 139                              unsigned int *nb, unsigned int *nn)
 140 {
 141         char *cp;
 142         cp = line + 4;
 143         if (parse_num(&cp, ob)) {
 144         bad_line:
 145                 return error("malformed diff output: %s", line);
 146         }
 147         if (*cp == ',') {
 148                 cp++;
 149                 if (parse_num(&cp, on))
 150                         goto bad_line;
 151         }
 152         else
 153                 *on = 1;
 154         if (*cp++ != ' ' || *cp++ != '+')
 155                 goto bad_line;
 156         if (parse_num(&cp, nb))
 157                 goto bad_line;
 158         if (*cp == ',') {
 159                 cp++;
 160                 if (parse_num(&cp, nn))
 161                         goto bad_line;
 162         }
 163         else
 164                 *nn = 1;
 165         return -!!memcmp(cp, " @@", 3);
 166 }
 167
 168 static void append_lost(struct sline *sline, int n, const char *line)
 169 {
 170         struct lline *lline;
 171         int len = strlen(line);
 172         unsigned long this_mask = (1UL<<n);
 173         if (line[len-1] == '\n')
 174                 len--;
 175
 176         /* Check to see if we can squash things */
 177         if (sline->lost_head) {
 178                 struct lline *last_one = NULL;
 179                 /* We cannot squash it with earlier one */
 180                 for (lline = sline->lost_head;
 181                      lline;
 182                      lline = lline->next)
 183                         if (lline->parent_map & this_mask)
 184                                 last_one = lline;
 185                 lline = last_one ? last_one->next : sline->lost_head;
 186                 while (lline) {
 187                         if (lline->len == len &&
 188                             !memcmp(lline->line, line, len)) {
 189                                 lline->parent_map |= this_mask;
 190                                 return;
 191                         }
 192                         lline = lline->next;
 193                 }
 194         }
 195
 196         lline = xmalloc(sizeof(*lline) + len + 1);
 197         lline->len = len;
 198         lline->next = NULL;
 199         lline->parent_map = this_mask;
 200         memcpy(lline->line, line, len);
 201         lline->line[len] = 0;
 202         *sline->lost_tail = lline;
 203         sline->lost_tail = &lline->next;
 204 }
 205
 206 static void combine_diff(const unsigned char *parent, const char *ourtmp,
 207                          struct sline *sline, int cnt, int n)
 208 {
 209         FILE *in;
 210         char parent_tmp[TMPPATHLEN];
 211         char cmd[TMPPATHLEN * 2 + 1024];
 212         char line[MAXLINELEN];
 213         unsigned int lno, ob, on, nb, nn;
 214         unsigned long pmask = ~(1UL << n);
 215         struct sline *lost_bucket = NULL;
 216
 217         write_temp_blob(parent_tmp, parent);
 218         sprintf(cmd, "diff --unified=0 -La/x -Lb/x '%s' '%s'",
 219                 parent_tmp, ourtmp);
 220         in = popen(cmd, "r");
 221         if (!in)
 222                 return;
 223
 224         lno = 1;
 225         while (fgets(line, sizeof(line), in) != NULL) {
 226                 int len = strlen(line);
 227                 if (5 < len && !memcmp("@@ -", line, 4)) {
 228                         if (parse_hunk_header(line, len,
 229                                               &ob, &on, &nb, &nn))
 230                                 break;
 231                         lno = nb;
 232                         if (!nb) {
 233                                 /* @@ -1,2 +0,0 @@ to remove the
 234                                  * first two lines...
 235                                  */
 236                                 nb = 1;
 237                         }
 238                         lost_bucket = &sline[nb-1]; /* sline is 0 based */
 239                         continue;
 240                 }
 241                 if (!lost_bucket)
 242                         continue;
 243                 switch (line[0]) {
 244                 case '-':
 245                         append_lost(lost_bucket, n, line+1);
 246                         break;
 247                 case '+':
 248                         sline[lno-1].flag &= pmask;
 249                         lno++;
 250                         break;
 251                 }
 252         }
 253         fclose(in);
 254         unlink(parent_tmp);
 255 }
 256
 257 static unsigned long context = 3;
 258 static char combine_marker = '@';
 259
 260 static int interesting(struct sline *sline, unsigned long all_mask)
 261 {
 262         return ((sline->flag & all_mask) != all_mask || sline->lost_head);
 263 }
 264
 265 static unsigned long line_common_diff(struct sline *sline, unsigned long all_mask)
 266 {
 267         /*
 268          * Look at the line and see from which parents we have the
 269          * same difference.
 270          */
 271
 272         /* Lower bits of sline->flag records if the parent had this
 273          * line, so XOR with all_mask gives us on-bits for parents we
 274          * have differences with.
 275          */
 276         unsigned long common_adds = (sline->flag ^ all_mask) & all_mask;
 277         unsigned long common_removes = all_mask;
 278
 279         /* If all the parents have this line, that also counts as
 280          * having the same difference.
 281          */
 282         if (!common_adds)
 283                 common_adds = all_mask;
 284
 285         if (sline->lost_head) {
 286                 /* Lost head list records the lines removed from
 287                  * the parents, and parent_map records from which
 288                  * parent the line was removed.
 289                  */
 290                 struct lline *ll;
 291                 for (ll = sline->lost_head; ll; ll = ll->next) {
 292                         common_removes &= ll->parent_map;
 293                 }
 294         }
 295         return common_adds & common_removes;
 296 }
 297
 298 static unsigned long line_all_diff(struct sline *sline, unsigned long all_mask)
 299 {
 300         /*
 301          * Look at the line and see from which parents we have some difference.
 302          */
 303         unsigned long different = (sline->flag ^ all_mask) & all_mask;
 304         if (sline->lost_head) {
 305                 /* Lost head list records the lines removed from
 306                  * the parents, and parent_map records from which
 307                  * parent the line was removed.
 308                  */
 309                 struct lline *ll;
 310                 for (ll = sline->lost_head; ll; ll = ll->next) {
 311                         different |= ll->parent_map;
 312                 }
 313         }
 314         return different;
 315 }
 316
 317 static unsigned long adjust_hunk_tail(struct sline *sline,
 318                                       unsigned long all_mask,
 319                                       unsigned long hunk_begin,
 320                                       unsigned long i)
 321 {
 322         /* i points at the first uninteresting line.
 323          * If the last line of the hunk was interesting
 324          * only because it has some deletion, then
 325          * it is not all that interesting for the
 326          * purpose of giving trailing context lines.
 327          */
 328         if ((hunk_begin + 1 <= i) &&
 329             ((sline[i-1].flag & all_mask) == all_mask))
 330                 i--;
 331         return i;
 332 }
 333
 334 static unsigned long next_interesting(struct sline *sline,
 335                                       unsigned long mark,
 336                                       unsigned long i,
 337                                       unsigned long cnt,
 338                                       int uninteresting)
 339 {
 340         while (i < cnt)
 341                 if (uninteresting ?
 342                     !(sline[i].flag & mark) :
 343                     (sline[i].flag & mark))
 344                         return i;
 345                 else
 346                         i++;
 347         return cnt;
 348 }
 349
 350 static int give_context(struct sline *sline, unsigned long cnt, int num_parent)
 351 {
 352         unsigned long all_mask = (1UL<<num_parent) - 1;
 353         unsigned long mark = (1UL<<num_parent);
 354         unsigned long i;
 355
 356         i = next_interesting(sline, mark, 0, cnt, 0);
 357         if (cnt <= i)
 358                 return 0;
 359
 360         while (i < cnt) {
 361                 unsigned long j = (context < i) ? (i - context) : 0;
 362                 unsigned long k;
 363                 while (j < i)
 364                         sline[j++].flag |= mark;
 365
 366         again:
 367                 j = next_interesting(sline, mark, i, cnt, 1);
 368                 if (cnt <= j)
 369                         break; /* the rest are all interesting */
 370
 371                 /* lookahead context lines */
 372                 k = next_interesting(sline, mark, j, cnt, 0);
 373                 j = adjust_hunk_tail(sline, all_mask, i, j);
 374
 375                 if (k < j + context) {
 376                         /* k is interesting and [j,k) are not, but
 377                          * paint them interesting because the gap is small.
 378                          */
 379                         while (j < k)
 380                                 sline[j++].flag |= mark;
 381                         i = k;
 382                         goto again;
 383                 }
 384
 385                 /* j is the first uninteresting line and there is
 386                  * no overlap beyond it within context lines.
 387                  */
 388                 i = k;
 389                 k = (j + context < cnt) ? j + context : cnt;
 390                 while (j < k)
 391                         sline[j++].flag |= mark;
 392         }
 393         return 1;
 394 }
 395
 396 static int make_hunks(struct sline *sline, unsigned long cnt,
 397                        int num_parent, int dense)
 398 {
 399         unsigned long all_mask = (1UL<<num_parent) - 1;
 400         unsigned long mark = (1UL<<num_parent);
 401         unsigned long i;
 402         int has_interesting = 0;
 403
 404         for (i = 0; i < cnt; i++) {
 405                 if (interesting(&sline[i], all_mask))
 406                         sline[i].flag |= mark;
 407                 else
 408                         sline[i].flag &= ~mark;
 409         }
 410         if (!dense)
 411                 return give_context(sline, cnt, num_parent);
 412
 413         /* Look at each hunk, and if we have changes from only one
 414          * parent, or the changes are the same from all but one
 415          * parent, mark that uninteresting.
 416          */
 417         i = 0;
 418         while (i < cnt) {
 419                 unsigned long j, hunk_begin, hunk_end;
 420                 int same, diff;
 421                 unsigned long same_diff, all_diff;
 422                 while (i < cnt && !(sline[i].flag & mark))
 423                         i++;
 424                 if (cnt <= i)
 425                         break; /* No more interesting hunks */
 426                 hunk_begin = i;
 427                 for (j = i + 1; j < cnt; j++) {
 428                         if (!(sline[j].flag & mark)) {
 429                                 /* Look beyond the end to see if there
 430                                  * is an interesting line after this
 431                                  * hunk within context span.
 432                                  */
 433                                 unsigned long la; /* lookahead */
 434                                 int contin = 0;
 435                                 la = adjust_hunk_tail(sline, all_mask,
 436                                                      hunk_begin, j);
 437                                 la = (la + context < cnt) ?
 438                                         (la + context) : cnt;
 439                                 while (j <= --la) {
 440                                         if (sline[la].flag & mark) {
 441                                                 contin = 1;
 442                                                 break;
 443                                         }
 444                                 }
 445                                 if (!contin)
 446                                         break;
 447                                 j = la;
 448                         }
 449                 }
 450                 hunk_end = j;
 451
 452                 /* [i..hunk_end) are interesting.  Now does it have
 453                  * the same change with all but one parent?
 454                  */
 455                 same_diff = all_mask;
 456                 all_diff = 0;
 457                 for (j = i; j < hunk_end; j++) {
 458                         same_diff &= line_common_diff(sline + j, all_mask);
 459                         all_diff |= line_all_diff(sline + j, all_mask);
 460                 }
 461                 diff = same = 0;
 462                 for (j = 0; j < num_parent; j++) {
 463                         if (same_diff & (1UL<<j))
 464                                 same++;
 465                         if (all_diff & (1UL<<j))
 466                                 diff++;
 467                 }
 468                 if ((num_parent - 1 <= same) || (diff == 1)) {
 469                         /* This hunk is not that interesting after all */
 470                         for (j = hunk_begin; j < hunk_end; j++)
 471                                 sline[j].flag &= ~mark;
 472                 }
 473                 i = hunk_end;
 474         }
 475
 476         has_interesting = give_context(sline, cnt, num_parent);
 477         return has_interesting;
 478 }
 479
 480 static void dump_sline(struct sline *sline, int cnt, int num_parent)
 481 {
 482         unsigned long mark = (1UL<<num_parent);
 483         int i;
 484         int lno = 0;
 485
 486         while (1) {
 487                 struct sline *sl = &sline[lno];
 488                 int hunk_end;
 489                 while (lno < cnt && !(sline[lno].flag & mark))
 490                         lno++;
 491                 if (cnt <= lno)
 492                         break;
 493                 for (hunk_end = lno + 1; hunk_end < cnt; hunk_end++)
 494                         if (!(sline[hunk_end].flag & mark))
 495                                 break;
 496                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 497                 printf(" +%d,%d ", lno+1, hunk_end-lno);
 498                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 499                 putchar('\n');
 500                 while (lno < hunk_end) {
 501                         struct lline *ll;
 502                         int j;
 503                         sl = &sline[lno++];
 504                         ll = sl->lost_head;
 505                         while (ll) {
 506                                 for (j = 0; j < num_parent; j++) {
 507                                         if (ll->parent_map & (1UL<<j))
 508                                                 putchar('-');
 509                                         else
 510                                                 putchar(' ');
 511                                 }
 512                                 puts(ll->line);
 513                                 ll = ll->next;
 514                         }
 515                         for (j = 0; j < num_parent; j++) {
 516                                 if ((1UL<<j) & sl->flag)
 517                                         putchar(' ');
 518                                 else
 519                                         putchar('+');
 520                         }
 521                         printf("%.*s\n", sl->len, sl->bol);
 522                 }
 523         }
 524 }
 525
 526 int show_combined_diff(struct combine_diff_path *elem, int num_parent,
 527                        int dense, const char *header, int show_empty)
 528 {
 529         unsigned long size, cnt, lno;
 530         char *result, *cp, *ep;
 531         struct sline *sline; /* survived lines */
 532         int i, show_hunks, shown_header = 0;
 533         char ourtmp_buf[TMPPATHLEN];
 534         char *ourtmp = ourtmp_buf;
 535
 536         /* Read the result of merge first */
 537         if (memcmp(elem->sha1, null_sha1, 20)) {
 538                 result = grab_blob(elem->sha1, &size);
 539                 write_to_temp_file(ourtmp, result, size);
 540         }
 541         else {
 542                 struct stat st;
 543                 int fd;
 544                 ourtmp = elem->path;
 545                 if (0 <= (fd = open(ourtmp, O_RDONLY)) &&
 546                     !fstat(fd, &st)) {
 547                         int len = st.st_size;
 548                         int cnt = 0;
 549
 550                         size = len;
 551                         result = xmalloc(len + 1);
 552                         while (cnt < len) {
 553                                 int done = xread(fd, result+cnt, len-cnt);
 554                                 if (done == 0)
 555                                         break;
 556                                 if (done < 0)
 557                                         die("read error '%s'", ourtmp);
 558                                 cnt += done;
 559                         }
 560                         result[len] = 0;
 561                 }
 562                 else {
 563                         /* deleted file */
 564                         size = 0;
 565                         result = xmalloc(1);
 566                         result[0] = 0;
 567                         ourtmp = "/dev/null";
 568                 }
 569                 if (0 <= fd)
 570                         close(fd);
 571         }
 572
 573         for (cnt = 0, cp = result; cp - result < size; cp++) {
 574                 if (*cp == '\n')
 575                         cnt++;
 576         }
 577         if (result[size-1] != '\n')
 578                 cnt++; /* incomplete line */
 579
 580         sline = xcalloc(cnt, sizeof(*sline));
 581         ep = result;
 582         sline[0].bol = result;
 583         for (lno = 0, cp = result; cp - result < size; cp++) {
 584                 if (*cp == '\n') {
 585                         sline[lno].lost_tail = &sline[lno].lost_head;
 586                         sline[lno].len = cp - sline[lno].bol;
 587                         sline[lno].flag = (1UL<<num_parent) - 1;
 588                         lno++;
 589                         if (lno < cnt)
 590                                 sline[lno].bol = cp + 1;
 591                 }
 592         }
 593         if (result[size-1] != '\n') {
 594                 sline[cnt-1].lost_tail = &sline[cnt-1].lost_head;
 595                 sline[cnt-1].len = size - (sline[cnt-1].bol - result);
 596                 sline[cnt-1].flag = (1UL<<num_parent) - 1;
 597         }
 598
 599         for (i = 0; i < num_parent; i++)
 600                 combine_diff(elem->parent_sha1[i], ourtmp, sline, cnt, i);
 601
 602         show_hunks = make_hunks(sline, cnt, num_parent, dense);
 603
 604         if (header && (show_hunks || show_empty)) {
 605                 shown_header++;
 606                 puts(header);
 607         }
 608         if (show_hunks) {
 609                 printf("diff --%s ", dense ? "cc" : "combined");
 610                 if (quote_c_style(elem->path, NULL, NULL, 0))
 611                         quote_c_style(elem->path, NULL, stdout, 0);
 612                 else
 613                         printf("%s", elem->path);
 614                 putchar('\n');
 615                 dump_sline(sline, cnt, num_parent);
 616         }
 617         if (ourtmp == ourtmp_buf)
 618                 unlink(ourtmp);
 619         free(result);
 620
 621         for (i = 0; i < cnt; i++) {
 622                 if (sline[i].lost_head) {
 623                         struct lline *ll = sline[i].lost_head;
 624                         while (ll) {
 625                                 struct lline *tmp = ll;
 626                                 ll = ll->next;
 627                                 free(tmp);
 628                         }
 629                 }
 630         }
 631         free(sline);
 632         return shown_header;
 633 }
 634
 635 int diff_tree_combined_merge(const unsigned char *sha1,
 636                              const char *header,
 637                              int show_empty_merge, int dense)
 638 {
 639         struct commit *commit = lookup_commit(sha1);
 640         struct diff_options diffopts;
 641         struct commit_list *parents;
 642         struct combine_diff_path *p, *paths = NULL;
 643         int num_parent, i, num_paths;
 644
 645         diff_setup(&diffopts);
 646         diffopts.output_format = DIFF_FORMAT_NO_OUTPUT;
 647         diffopts.recursive = 1;
 648
 649         /* count parents */
 650         for (parents = commit->parents, num_parent = 0;
 651              parents;
 652              parents = parents->next, num_parent++)
 653                 ; /* nothing */
 654
 655         /* find set of paths that everybody touches */
 656         for (parents = commit->parents, i = 0;
 657              parents;
 658              parents = parents->next, i++) {
 659                 struct commit *parent = parents->item;
 660                 diff_tree_sha1(parent->object.sha1, commit->object.sha1, "",
 661                                &diffopts);
 662                 paths = intersect_paths(paths, i, num_parent);
 663                 diff_flush(&diffopts);
 664         }
 665
 666         /* find out surviving paths */
 667         for (num_paths = 0, p = paths; p; p = p->next) {
 668                 if (p->len)
 669                         num_paths++;
 670         }
 671         if (num_paths || show_empty_merge) {
 672                 for (p = paths; p; p = p->next) {
 673                         if (!p->len)
 674                                 continue;
 675                         if (show_combined_diff(p, num_parent, dense, header,
 676                                                show_empty_merge))
 677                                 header = NULL;
 678                 }
 679         }
 680
 681         /* Clean things up */
 682         while (paths) {
 683                 struct combine_diff_path *tmp = paths;
 684                 paths = paths->next;
 685                 free(tmp);
 686         }
 687         return 0;
 688 }