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 adjust_hunk_tail(struct sline *sline,
 266                                       unsigned long all_mask,
 267                                       unsigned long hunk_begin,
 268                                       unsigned long i)
 269 {
 270         /* i points at the first uninteresting line.
 271          * If the last line of the hunk was interesting
 272          * only because it has some deletion, then
 273          * it is not all that interesting for the
 274          * purpose of giving trailing context lines.
 275          */
 276         if ((hunk_begin + 1 <= i) &&
 277             ((sline[i-1].flag & all_mask) == all_mask))
 278                 i--;
 279         return i;
 280 }
 281
 282 static unsigned long next_interesting(struct sline *sline,
 283                                       unsigned long mark,
 284                                       unsigned long i,
 285                                       unsigned long cnt,
 286                                       int uninteresting)
 287 {
 288         while (i < cnt)
 289                 if (uninteresting ?
 290                     !(sline[i].flag & mark) :
 291                     (sline[i].flag & mark))
 292                         return i;
 293                 else
 294                         i++;
 295         return cnt;
 296 }
 297
 298 static int give_context(struct sline *sline, unsigned long cnt, int num_parent)
 299 {
 300         unsigned long all_mask = (1UL<<num_parent) - 1;
 301         unsigned long mark = (1UL<<num_parent);
 302         unsigned long i;
 303
 304         i = next_interesting(sline, mark, 0, cnt, 0);
 305         if (cnt <= i)
 306                 return 0;
 307
 308         while (i < cnt) {
 309                 unsigned long j = (context < i) ? (i - context) : 0;
 310                 unsigned long k;
 311                 while (j < i)
 312                         sline[j++].flag |= mark;
 313
 314         again:
 315                 j = next_interesting(sline, mark, i, cnt, 1);
 316                 if (cnt <= j)
 317                         break; /* the rest are all interesting */
 318
 319                 /* lookahead context lines */
 320                 k = next_interesting(sline, mark, j, cnt, 0);
 321                 j = adjust_hunk_tail(sline, all_mask, i, j);
 322
 323                 if (k < j + context) {
 324                         /* k is interesting and [j,k) are not, but
 325                          * paint them interesting because the gap is small.
 326                          */
 327                         while (j < k)
 328                                 sline[j++].flag |= mark;
 329                         i = k;
 330                         goto again;
 331                 }
 332
 333                 /* j is the first uninteresting line and there is
 334                  * no overlap beyond it within context lines.
 335                  */
 336                 i = k;
 337                 k = (j + context < cnt) ? j + context : cnt;
 338                 while (j < k)
 339                         sline[j++].flag |= mark;
 340         }
 341         return 1;
 342 }
 343
 344 static int make_hunks(struct sline *sline, unsigned long cnt,
 345                        int num_parent, int dense)
 346 {
 347         unsigned long all_mask = (1UL<<num_parent) - 1;
 348         unsigned long mark = (1UL<<num_parent);
 349         unsigned long i;
 350         int has_interesting = 0;
 351
 352         for (i = 0; i < cnt; i++) {
 353                 if (interesting(&sline[i], all_mask))
 354                         sline[i].flag |= mark;
 355                 else
 356                         sline[i].flag &= ~mark;
 357         }
 358         if (!dense)
 359                 return give_context(sline, cnt, num_parent);
 360
 361         /* Look at each hunk, and if we have changes from only one
 362          * parent, or the changes are the same from all but one
 363          * parent, mark that uninteresting.
 364          */
 365         i = 0;
 366         while (i < cnt) {
 367                 unsigned long j, hunk_begin, hunk_end;
 368                 unsigned long same_diff;
 369                 while (i < cnt && !(sline[i].flag & mark))
 370                         i++;
 371                 if (cnt <= i)
 372                         break; /* No more interesting hunks */
 373                 hunk_begin = i;
 374                 for (j = i + 1; j < cnt; j++) {
 375                         if (!(sline[j].flag & mark)) {
 376                                 /* Look beyond the end to see if there
 377                                  * is an interesting line after this
 378                                  * hunk within context span.
 379                                  */
 380                                 unsigned long la; /* lookahead */
 381                                 int contin = 0;
 382                                 la = adjust_hunk_tail(sline, all_mask,
 383                                                      hunk_begin, j);
 384                                 la = (la + context < cnt) ?
 385                                         (la + context) : cnt;
 386                                 while (j <= --la) {
 387                                         if (sline[la].flag & mark) {
 388                                                 contin = 1;
 389                                                 break;
 390                                         }
 391                                 }
 392                                 if (!contin)
 393                                         break;
 394                                 j = la;
 395                         }
 396                 }
 397                 hunk_end = j;
 398
 399                 /* [i..hunk_end) are interesting.  Now is it really
 400                  * interesting?  We check if there are only two versions
 401                  * and the result matches one of them.  That is, we look
 402                  * at:
 403                  *   (+) line, which records lines added to which parents;
 404                  *       this line appears in the result.
 405                  *   (-) line, which records from what parents the line
 406                  *       was removed; this line does not appear in the result.
 407                  * then check the set of parents the result has difference
 408                  * from, from all lines.  If there are lines that has
 409                  * different set of parents that the result has differences
 410                  * from, that means we have more than two versions.
 411                  *
 412                  * Even when we have only two versions, if the result does
 413                  * not match any of the parents, the it should be considered
 414                  * interesting.  In such a case, we would have all '+' line.
 415                  * After passing the above "two versions" test, that would
 416                  * appear as "the same set of parents" to be "all parents".
 417                  */
 418                 same_diff = 0;
 419                 has_interesting = 0;
 420                 for (j = i; j < hunk_end && !has_interesting; j++) {
 421                         unsigned long this_diff = ~sline[j].flag & all_mask;
 422                         struct lline *ll = sline[j].lost_head;
 423                         if (this_diff) {
 424                                 /* This has some changes.  Is it the
 425                                  * same as others?
 426                                  */
 427                                 if (!same_diff)
 428                                         same_diff = this_diff;
 429                                 else if (same_diff != this_diff) {
 430                                         has_interesting = 1;
 431                                         break;
 432                                 }
 433                         }
 434                         while (ll && !has_interesting) {
 435                                 /* Lost this line from these parents;
 436                                  * who are they?  Are they the same?
 437                                  */
 438                                 this_diff = ll->parent_map;
 439                                 if (!same_diff)
 440                                         same_diff = this_diff;
 441                                 else if (same_diff != this_diff) {
 442                                         has_interesting = 1;
 443                                 }
 444                                 ll = ll->next;
 445                         }
 446                 }
 447
 448                 if (!has_interesting && same_diff != all_mask) {
 449                         /* This hunk is not that interesting after all */
 450                         for (j = hunk_begin; j < hunk_end; j++)
 451                                 sline[j].flag &= ~mark;
 452                 }
 453                 i = hunk_end;
 454         }
 455
 456         has_interesting = give_context(sline, cnt, num_parent);
 457         return has_interesting;
 458 }
 459
 460 static void dump_sline(struct sline *sline, int cnt, int num_parent)
 461 {
 462         unsigned long mark = (1UL<<num_parent);
 463         int i;
 464         int lno = 0;
 465
 466         while (1) {
 467                 struct sline *sl = &sline[lno];
 468                 int hunk_end;
 469                 while (lno < cnt && !(sline[lno].flag & mark))
 470                         lno++;
 471                 if (cnt <= lno)
 472                         break;
 473                 for (hunk_end = lno + 1; hunk_end < cnt; hunk_end++)
 474                         if (!(sline[hunk_end].flag & mark))
 475                                 break;
 476                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 477                 printf(" +%d,%d ", lno+1, hunk_end-lno);
 478                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 479                 putchar('\n');
 480                 while (lno < hunk_end) {
 481                         struct lline *ll;
 482                         int j;
 483                         sl = &sline[lno++];
 484                         ll = sl->lost_head;
 485                         while (ll) {
 486                                 for (j = 0; j < num_parent; j++) {
 487                                         if (ll->parent_map & (1UL<<j))
 488                                                 putchar('-');
 489                                         else
 490                                                 putchar(' ');
 491                                 }
 492                                 puts(ll->line);
 493                                 ll = ll->next;
 494                         }
 495                         for (j = 0; j < num_parent; j++) {
 496                                 if ((1UL<<j) & sl->flag)
 497                                         putchar(' ');
 498                                 else
 499                                         putchar('+');
 500                         }
 501                         printf("%.*s\n", sl->len, sl->bol);
 502                 }
 503         }
 504 }
 505
 506 static void reuse_combine_diff(struct sline *sline, unsigned long cnt,
 507                                int i, int j)
 508 {
 509         /* We have already examined parent j and we know parent i
 510          * and parent j are the same, so reuse the combined result
 511          * of parent j for parent i.
 512          */
 513         unsigned long lno, imask, jmask;
 514         imask = (1UL<<i);
 515         jmask = (1UL<<j);
 516
 517         for (lno = 0; lno < cnt; lno++) {
 518                 struct lline *ll = sline->lost_head;
 519                 while (ll) {
 520                         if (ll->parent_map & jmask)
 521                                 ll->parent_map |= imask;
 522                         ll = ll->next;
 523                 }
 524                 if (!(sline->flag & jmask))
 525                         sline->flag &= ~imask;
 526                 sline++;
 527         }
 528 }
 529
 530 int show_combined_diff(struct combine_diff_path *elem, int num_parent,
 531                        int dense, const char *header, int show_empty)
 532 {
 533         unsigned long size, cnt, lno;
 534         char *result, *cp, *ep;
 535         struct sline *sline; /* survived lines */
 536         int i, show_hunks, shown_header = 0;
 537         char ourtmp_buf[TMPPATHLEN];
 538         char *ourtmp = ourtmp_buf;
 539
 540         /* Read the result of merge first */
 541         if (memcmp(elem->sha1, null_sha1, 20)) {
 542                 result = grab_blob(elem->sha1, &size);
 543                 write_to_temp_file(ourtmp, result, size);
 544         }
 545         else {
 546                 struct stat st;
 547                 int fd;
 548                 ourtmp = elem->path;
 549                 if (0 <= (fd = open(ourtmp, O_RDONLY)) &&
 550                     !fstat(fd, &st)) {
 551                         int len = st.st_size;
 552                         int cnt = 0;
 553
 554                         size = len;
 555                         result = xmalloc(len + 1);
 556                         while (cnt < len) {
 557                                 int done = xread(fd, result+cnt, len-cnt);
 558                                 if (done == 0)
 559                                         break;
 560                                 if (done < 0)
 561                                         die("read error '%s'", ourtmp);
 562                                 cnt += done;
 563                         }
 564                         result[len] = 0;
 565                 }
 566                 else {
 567                         /* deleted file */
 568                         size = 0;
 569                         result = xmalloc(1);
 570                         result[0] = 0;
 571                         ourtmp = "/dev/null";
 572                 }
 573                 if (0 <= fd)
 574                         close(fd);
 575         }
 576
 577         for (cnt = 0, cp = result; cp - result < size; cp++) {
 578                 if (*cp == '\n')
 579                         cnt++;
 580         }
 581         if (result[size-1] != '\n')
 582                 cnt++; /* incomplete line */
 583
 584         sline = xcalloc(cnt, sizeof(*sline));
 585         ep = result;
 586         sline[0].bol = result;
 587         for (lno = 0, cp = result; cp - result < size; cp++) {
 588                 if (*cp == '\n') {
 589                         sline[lno].lost_tail = &sline[lno].lost_head;
 590                         sline[lno].len = cp - sline[lno].bol;
 591                         sline[lno].flag = (1UL<<num_parent) - 1;
 592                         lno++;
 593                         if (lno < cnt)
 594                                 sline[lno].bol = cp + 1;
 595                 }
 596         }
 597         if (result[size-1] != '\n') {
 598                 sline[cnt-1].lost_tail = &sline[cnt-1].lost_head;
 599                 sline[cnt-1].len = size - (sline[cnt-1].bol - result);
 600                 sline[cnt-1].flag = (1UL<<num_parent) - 1;
 601         }
 602
 603         for (i = 0; i < num_parent; i++) {
 604                 int j;
 605                 for (j = 0; j < i; j++) {
 606                         if (!memcmp(elem->parent_sha1[i],
 607                                     elem->parent_sha1[j], 20)) {
 608                                 reuse_combine_diff(sline, cnt, i, j);
 609                                 break;
 610                         }
 611                 }
 612                 if (i <= j)
 613                         combine_diff(elem->parent_sha1[i], ourtmp, sline,
 614                                      cnt, i);
 615         }
 616
 617         show_hunks = make_hunks(sline, cnt, num_parent, dense);
 618
 619         if (header && (show_hunks || show_empty)) {
 620                 shown_header++;
 621                 puts(header);
 622         }
 623         if (show_hunks) {
 624                 printf("diff --%s ", dense ? "cc" : "combined");
 625                 if (quote_c_style(elem->path, NULL, NULL, 0))
 626                         quote_c_style(elem->path, NULL, stdout, 0);
 627                 else
 628                         printf("%s", elem->path);
 629                 putchar('\n');
 630                 dump_sline(sline, cnt, num_parent);
 631         }
 632         if (ourtmp == ourtmp_buf)
 633                 unlink(ourtmp);
 634         free(result);
 635
 636         for (i = 0; i < cnt; i++) {
 637                 if (sline[i].lost_head) {
 638                         struct lline *ll = sline[i].lost_head;
 639                         while (ll) {
 640                                 struct lline *tmp = ll;
 641                                 ll = ll->next;
 642                                 free(tmp);
 643                         }
 644                 }
 645         }
 646         free(sline);
 647         return shown_header;
 648 }
 649
 650 int diff_tree_combined_merge(const unsigned char *sha1,
 651                              const char *header,
 652                              int show_empty_merge, int dense)
 653 {
 654         struct commit *commit = lookup_commit(sha1);
 655         struct diff_options diffopts;
 656         struct commit_list *parents;
 657         struct combine_diff_path *p, *paths = NULL;
 658         int num_parent, i, num_paths;
 659
 660         diff_setup(&diffopts);
 661         diffopts.output_format = DIFF_FORMAT_NO_OUTPUT;
 662         diffopts.recursive = 1;
 663
 664         /* count parents */
 665         for (parents = commit->parents, num_parent = 0;
 666              parents;
 667              parents = parents->next, num_parent++)
 668                 ; /* nothing */
 669
 670         /* find set of paths that everybody touches */
 671         for (parents = commit->parents, i = 0;
 672              parents;
 673              parents = parents->next, i++) {
 674                 struct commit *parent = parents->item;
 675                 diff_tree_sha1(parent->object.sha1, commit->object.sha1, "",
 676                                &diffopts);
 677                 paths = intersect_paths(paths, i, num_parent);
 678                 diff_flush(&diffopts);
 679         }
 680
 681         /* find out surviving paths */
 682         for (num_paths = 0, p = paths; p; p = p->next) {
 683                 if (p->len)
 684                         num_paths++;
 685         }
 686         if (num_paths || show_empty_merge) {
 687                 for (p = paths; p; p = p->next) {
 688                         if (!p->len)
 689                                 continue;
 690                         if (show_combined_diff(p, num_parent, dense, header,
 691                                                show_empty_merge))
 692                                 header = NULL;
 693                 }
 694         }
 695
 696         /* Clean things up */
 697         while (paths) {
 698                 struct combine_diff_path *tmp = paths;
 699                 paths = paths->next;
 700                 free(tmp);
 701         }
 702         return 0;
 703 }