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                         if (nn == 0)
 239                                 lost_bucket = &sline[nb];
 240                         else
 241                                 lost_bucket = &sline[nb-1];
 242                         continue;
 243                 }
 244                 if (!lost_bucket)
 245                         continue;
 246                 switch (line[0]) {
 247                 case '-':
 248                         append_lost(lost_bucket, n, line+1);
 249                         break;
 250                 case '+':
 251                         sline[lno-1].flag &= pmask;
 252                         lno++;
 253                         break;
 254                 }
 255         }
 256         fclose(in);
 257         unlink(parent_tmp);
 258 }
 259
 260 static unsigned long context = 3;
 261 static char combine_marker = '@';
 262
 263 static int interesting(struct sline *sline, unsigned long all_mask)
 264 {
 265         return ((sline->flag & all_mask) != all_mask || sline->lost_head);
 266 }
 267
 268 static unsigned long adjust_hunk_tail(struct sline *sline,
 269                                       unsigned long all_mask,
 270                                       unsigned long hunk_begin,
 271                                       unsigned long i)
 272 {
 273         /* i points at the first uninteresting line.
 274          * If the last line of the hunk was interesting
 275          * only because it has some deletion, then
 276          * it is not all that interesting for the
 277          * purpose of giving trailing context lines.
 278          */
 279         if ((hunk_begin + 1 <= i) &&
 280             ((sline[i-1].flag & all_mask) == all_mask))
 281                 i--;
 282         return i;
 283 }
 284
 285 static unsigned long next_interesting(struct sline *sline,
 286                                       unsigned long mark,
 287                                       unsigned long i,
 288                                       unsigned long cnt,
 289                                       int uninteresting)
 290 {
 291         while (i < cnt)
 292                 if (uninteresting ?
 293                     !(sline[i].flag & mark) :
 294                     (sline[i].flag & mark))
 295                         return i;
 296                 else
 297                         i++;
 298         return cnt;
 299 }
 300
 301 static int give_context(struct sline *sline, unsigned long cnt, int num_parent)
 302 {
 303         unsigned long all_mask = (1UL<<num_parent) - 1;
 304         unsigned long mark = (1UL<<num_parent);
 305         unsigned long i;
 306
 307         i = next_interesting(sline, mark, 0, cnt, 0);
 308         if (cnt <= i)
 309                 return 0;
 310
 311         while (i < cnt) {
 312                 unsigned long j = (context < i) ? (i - context) : 0;
 313                 unsigned long k;
 314                 while (j < i)
 315                         sline[j++].flag |= mark;
 316
 317         again:
 318                 j = next_interesting(sline, mark, i, cnt, 1);
 319                 if (cnt <= j)
 320                         break; /* the rest are all interesting */
 321
 322                 /* lookahead context lines */
 323                 k = next_interesting(sline, mark, j, cnt, 0);
 324                 j = adjust_hunk_tail(sline, all_mask, i, j);
 325
 326                 if (k < j + context) {
 327                         /* k is interesting and [j,k) are not, but
 328                          * paint them interesting because the gap is small.
 329                          */
 330                         while (j < k)
 331                                 sline[j++].flag |= mark;
 332                         i = k;
 333                         goto again;
 334                 }
 335
 336                 /* j is the first uninteresting line and there is
 337                  * no overlap beyond it within context lines.
 338                  */
 339                 i = k;
 340                 k = (j + context < cnt) ? j + context : cnt;
 341                 while (j < k)
 342                         sline[j++].flag |= mark;
 343         }
 344         return 1;
 345 }
 346
 347 static int make_hunks(struct sline *sline, unsigned long cnt,
 348                        int num_parent, int dense)
 349 {
 350         unsigned long all_mask = (1UL<<num_parent) - 1;
 351         unsigned long mark = (1UL<<num_parent);
 352         unsigned long i;
 353         int has_interesting = 0;
 354
 355         for (i = 0; i < cnt; i++) {
 356                 if (interesting(&sline[i], all_mask))
 357                         sline[i].flag |= mark;
 358                 else
 359                         sline[i].flag &= ~mark;
 360         }
 361         if (!dense)
 362                 return give_context(sline, cnt, num_parent);
 363
 364         /* Look at each hunk, and if we have changes from only one
 365          * parent, or the changes are the same from all but one
 366          * parent, mark that uninteresting.
 367          */
 368         i = 0;
 369         while (i < cnt) {
 370                 unsigned long j, hunk_begin, hunk_end;
 371                 unsigned long same_diff;
 372                 while (i < cnt && !(sline[i].flag & mark))
 373                         i++;
 374                 if (cnt <= i)
 375                         break; /* No more interesting hunks */
 376                 hunk_begin = i;
 377                 for (j = i + 1; j < cnt; j++) {
 378                         if (!(sline[j].flag & mark)) {
 379                                 /* Look beyond the end to see if there
 380                                  * is an interesting line after this
 381                                  * hunk within context span.
 382                                  */
 383                                 unsigned long la; /* lookahead */
 384                                 int contin = 0;
 385                                 la = adjust_hunk_tail(sline, all_mask,
 386                                                      hunk_begin, j);
 387                                 la = (la + context < cnt) ?
 388                                         (la + context) : cnt;
 389                                 while (j <= --la) {
 390                                         if (sline[la].flag & mark) {
 391                                                 contin = 1;
 392                                                 break;
 393                                         }
 394                                 }
 395                                 if (!contin)
 396                                         break;
 397                                 j = la;
 398                         }
 399                 }
 400                 hunk_end = j;
 401
 402                 /* [i..hunk_end) are interesting.  Now is it really
 403                  * interesting?  We check if there are only two versions
 404                  * and the result matches one of them.  That is, we look
 405                  * at:
 406                  *   (+) line, which records lines added to which parents;
 407                  *       this line appears in the result.
 408                  *   (-) line, which records from what parents the line
 409                  *       was removed; this line does not appear in the result.
 410                  * then check the set of parents the result has difference
 411                  * from, from all lines.  If there are lines that has
 412                  * different set of parents that the result has differences
 413                  * from, that means we have more than two versions.
 414                  *
 415                  * Even when we have only two versions, if the result does
 416                  * not match any of the parents, the it should be considered
 417                  * interesting.  In such a case, we would have all '+' line.
 418                  * After passing the above "two versions" test, that would
 419                  * appear as "the same set of parents" to be "all parents".
 420                  */
 421                 same_diff = 0;
 422                 has_interesting = 0;
 423                 for (j = i; j < hunk_end && !has_interesting; j++) {
 424                         unsigned long this_diff = ~sline[j].flag & all_mask;
 425                         struct lline *ll = sline[j].lost_head;
 426                         if (this_diff) {
 427                                 /* This has some changes.  Is it the
 428                                  * same as others?
 429                                  */
 430                                 if (!same_diff)
 431                                         same_diff = this_diff;
 432                                 else if (same_diff != this_diff) {
 433                                         has_interesting = 1;
 434                                         break;
 435                                 }
 436                         }
 437                         while (ll && !has_interesting) {
 438                                 /* Lost this line from these parents;
 439                                  * who are they?  Are they the same?
 440                                  */
 441                                 this_diff = ll->parent_map;
 442                                 if (!same_diff)
 443                                         same_diff = this_diff;
 444                                 else if (same_diff != this_diff) {
 445                                         has_interesting = 1;
 446                                 }
 447                                 ll = ll->next;
 448                         }
 449                 }
 450
 451                 if (!has_interesting && same_diff != all_mask) {
 452                         /* This hunk is not that interesting after all */
 453                         for (j = hunk_begin; j < hunk_end; j++)
 454                                 sline[j].flag &= ~mark;
 455                 }
 456                 i = hunk_end;
 457         }
 458
 459         has_interesting = give_context(sline, cnt, num_parent);
 460         return has_interesting;
 461 }
 462
 463 static void dump_sline(struct sline *sline, int cnt, int num_parent)
 464 {
 465         unsigned long mark = (1UL<<num_parent);
 466         int i;
 467         int lno = 0;
 468
 469         while (1) {
 470                 struct sline *sl = &sline[lno];
 471                 int hunk_end;
 472                 while (lno < cnt && !(sline[lno].flag & mark))
 473                         lno++;
 474                 if (cnt <= lno)
 475                         break;
 476                 for (hunk_end = lno + 1; hunk_end < cnt; hunk_end++)
 477                         if (!(sline[hunk_end].flag & mark))
 478                                 break;
 479                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 480                 printf(" +%d,%d ", lno+1, hunk_end-lno);
 481                 for (i = 0; i <= num_parent; i++) putchar(combine_marker);
 482                 putchar('\n');
 483                 while (lno < hunk_end) {
 484                         struct lline *ll;
 485                         int j;
 486                         sl = &sline[lno++];
 487                         ll = sl->lost_head;
 488                         while (ll) {
 489                                 for (j = 0; j < num_parent; j++) {
 490                                         if (ll->parent_map & (1UL<<j))
 491                                                 putchar('-');
 492                                         else
 493                                                 putchar(' ');
 494                                 }
 495                                 puts(ll->line);
 496                                 ll = ll->next;
 497                         }
 498                         for (j = 0; j < num_parent; j++) {
 499                                 if ((1UL<<j) & sl->flag)
 500                                         putchar(' ');
 501                                 else
 502                                         putchar('+');
 503                         }
 504                         printf("%.*s\n", sl->len, sl->bol);
 505                 }
 506         }
 507 }
 508
 509 static void reuse_combine_diff(struct sline *sline, unsigned long cnt,
 510                                int i, int j)
 511 {
 512         /* We have already examined parent j and we know parent i
 513          * and parent j are the same, so reuse the combined result
 514          * of parent j for parent i.
 515          */
 516         unsigned long lno, imask, jmask;
 517         imask = (1UL<<i);
 518         jmask = (1UL<<j);
 519
 520         for (lno = 0; lno < cnt; lno++) {
 521                 struct lline *ll = sline->lost_head;
 522                 while (ll) {
 523                         if (ll->parent_map & jmask)
 524                                 ll->parent_map |= imask;
 525                         ll = ll->next;
 526                 }
 527                 if (!(sline->flag & jmask))
 528                         sline->flag &= ~imask;
 529                 sline++;
 530         }
 531 }
 532
 533 int show_combined_diff(struct combine_diff_path *elem, int num_parent,
 534                        int dense, const char *header, int show_empty)
 535 {
 536         unsigned long size, cnt, lno;
 537         char *result, *cp, *ep;
 538         struct sline *sline; /* survived lines */
 539         int i, show_hunks, shown_header = 0;
 540         char ourtmp_buf[TMPPATHLEN];
 541         char *ourtmp = ourtmp_buf;
 542
 543         /* Read the result of merge first */
 544         if (memcmp(elem->sha1, null_sha1, 20)) {
 545                 result = grab_blob(elem->sha1, &size);
 546                 write_to_temp_file(ourtmp, result, size);
 547         }
 548         else {
 549                 struct stat st;
 550                 int fd;
 551                 ourtmp = elem->path;
 552                 if (0 <= (fd = open(ourtmp, O_RDONLY)) &&
 553                     !fstat(fd, &st)) {
 554                         int len = st.st_size;
 555                         int cnt = 0;
 556
 557                         size = len;
 558                         result = xmalloc(len + 1);
 559                         while (cnt < len) {
 560                                 int done = xread(fd, result+cnt, len-cnt);
 561                                 if (done == 0)
 562                                         break;
 563                                 if (done < 0)
 564                                         die("read error '%s'", ourtmp);
 565                                 cnt += done;
 566                         }
 567                         result[len] = 0;
 568                 }
 569                 else {
 570                         /* deleted file */
 571                         size = 0;
 572                         result = xmalloc(1);
 573                         result[0] = 0;
 574                         ourtmp = "/dev/null";
 575                 }
 576                 if (0 <= fd)
 577                         close(fd);
 578         }
 579
 580         for (cnt = 0, cp = result; cp - result < size; cp++) {
 581                 if (*cp == '\n')
 582                         cnt++;
 583         }
 584         if (result[size-1] != '\n')
 585                 cnt++; /* incomplete line */
 586
 587         sline = xcalloc(cnt, sizeof(*sline));
 588         ep = result;
 589         sline[0].bol = result;
 590         for (lno = 0, cp = result; cp - result < size; cp++) {
 591                 if (*cp == '\n') {
 592                         sline[lno].lost_tail = &sline[lno].lost_head;
 593                         sline[lno].len = cp - sline[lno].bol;
 594                         sline[lno].flag = (1UL<<num_parent) - 1;
 595                         lno++;
 596                         if (lno < cnt)
 597                                 sline[lno].bol = cp + 1;
 598                 }
 599         }
 600         if (result[size-1] != '\n') {
 601                 sline[cnt-1].lost_tail = &sline[cnt-1].lost_head;
 602                 sline[cnt-1].len = size - (sline[cnt-1].bol - result);
 603                 sline[cnt-1].flag = (1UL<<num_parent) - 1;
 604         }
 605
 606         for (i = 0; i < num_parent; i++) {
 607                 int j;
 608                 for (j = 0; j < i; j++) {
 609                         if (!memcmp(elem->parent_sha1[i],
 610                                     elem->parent_sha1[j], 20)) {
 611                                 reuse_combine_diff(sline, cnt, i, j);
 612                                 break;
 613                         }
 614                 }
 615                 if (i <= j)
 616                         combine_diff(elem->parent_sha1[i], ourtmp, sline,
 617                                      cnt, i);
 618         }
 619
 620         show_hunks = make_hunks(sline, cnt, num_parent, dense);
 621
 622         if (header && (show_hunks || show_empty)) {
 623                 shown_header++;
 624                 puts(header);
 625         }
 626         if (show_hunks) {
 627                 printf("diff --%s ", dense ? "cc" : "combined");
 628                 if (quote_c_style(elem->path, NULL, NULL, 0))
 629                         quote_c_style(elem->path, NULL, stdout, 0);
 630                 else
 631                         printf("%s", elem->path);
 632                 putchar('\n');
 633                 printf("index ");
 634                 for (i = 0; i < num_parent; i++) {
 635                         printf("%s%s",
 636                                i ? ".." : "",
 637                                find_unique_abbrev(elem->parent_sha1[i],
 638                                                   DEFAULT_ABBREV));
 639                 }
 640                 printf("->%s\n",
 641                        find_unique_abbrev(elem->sha1, DEFAULT_ABBREV));
 642                 dump_sline(sline, cnt, num_parent);
 643         }
 644         if (ourtmp == ourtmp_buf)
 645                 unlink(ourtmp);
 646         free(result);
 647
 648         for (i = 0; i < cnt; i++) {
 649                 if (sline[i].lost_head) {
 650                         struct lline *ll = sline[i].lost_head;
 651                         while (ll) {
 652                                 struct lline *tmp = ll;
 653                                 ll = ll->next;
 654                                 free(tmp);
 655                         }
 656                 }
 657         }
 658         free(sline);
 659         return shown_header;
 660 }
 661
 662 int diff_tree_combined_merge(const unsigned char *sha1,
 663                              const char *header,
 664                              int show_empty_merge, int dense)
 665 {
 666         struct commit *commit = lookup_commit(sha1);
 667         struct diff_options diffopts;
 668         struct commit_list *parents;
 669         struct combine_diff_path *p, *paths = NULL;
 670         int num_parent, i, num_paths;
 671
 672         diff_setup(&diffopts);
 673         diffopts.output_format = DIFF_FORMAT_NO_OUTPUT;
 674         diffopts.recursive = 1;
 675
 676         /* count parents */
 677         for (parents = commit->parents, num_parent = 0;
 678              parents;
 679              parents = parents->next, num_parent++)
 680                 ; /* nothing */
 681
 682         /* find set of paths that everybody touches */
 683         for (parents = commit->parents, i = 0;
 684              parents;
 685              parents = parents->next, i++) {
 686                 struct commit *parent = parents->item;
 687                 diff_tree_sha1(parent->object.sha1, commit->object.sha1, "",
 688                                &diffopts);
 689                 paths = intersect_paths(paths, i, num_parent);
 690                 diff_flush(&diffopts);
 691         }
 692
 693         /* find out surviving paths */
 694         for (num_paths = 0, p = paths; p; p = p->next) {
 695                 if (p->len)
 696                         num_paths++;
 697         }
 698         if (num_paths || show_empty_merge) {
 699                 for (p = paths; p; p = p->next) {
 700                         if (!p->len)
 701                                 continue;
 702                         if (show_combined_diff(p, num_parent, dense, header,
 703                                                show_empty_merge))
 704                                 header = NULL;
 705                 }
 706         }
 707
 708         /* Clean things up */
 709         while (paths) {
 710                 struct combine_diff_path *tmp = paths;
 711                 paths = paths->next;
 712                 free(tmp);
 713         }
 714         return 0;
 715 }