builtin-rev-list.c

   1 #include "cache.h"
   2 #include "refs.h"
   3 #include "tag.h"
   4 #include "commit.h"
   5 #include "tree.h"
   6 #include "blob.h"
   7 #include "tree-walk.h"
   8 #include "diff.h"
   9 #include "revision.h"
  10 #include "list-objects.h"
  11 #include "builtin.h"
  12 #include "log-tree.h"
  13 #include "graph.h"
  14
  15 /* bits #0-15 in revision.h */
  16
  17 #define COUNTED         (1u<<16)
  18
  19 static const char rev_list_usage[] =
  20 "git rev-list [OPTION] <commit-id>... [ -- paths... ]\n"
  21 "  limiting output:\n"
  22 "    --max-count=nr\n"
  23 "    --max-age=epoch\n"
  24 "    --min-age=epoch\n"
  25 "    --sparse\n"
  26 "    --no-merges\n"
  27 "    --remove-empty\n"
  28 "    --all\n"
  29 "    --branches\n"
  30 "    --tags\n"
  31 "    --remotes\n"
  32 "    --stdin\n"
  33 "    --quiet\n"
  34 "  ordering output:\n"
  35 "    --topo-order\n"
  36 "    --date-order\n"
  37 "    --reverse\n"
  38 "  formatting output:\n"
  39 "    --parents\n"
  40 "    --children\n"
  41 "    --objects | --objects-edge\n"
  42 "    --unpacked\n"
  43 "    --header | --pretty\n"
  44 "    --abbrev=nr | --no-abbrev\n"
  45 "    --abbrev-commit\n"
  46 "    --left-right\n"
  47 "  special purpose:\n"
  48 "    --bisect\n"
  49 "    --bisect-vars\n"
  50 "    --bisect-all"
  51 ;
  52
  53 static struct rev_info revs;
  54
  55 static int show_timestamp;
  56 static int hdr_termination;
  57 static const char *header_prefix;
  58
  59 static void finish_commit(struct commit *commit);
  60 static void show_commit(struct commit *commit)
  61 {
  62         graph_show_commit(revs.graph);
  63
  64         if (show_timestamp)
  65                 printf("%lu ", commit->date);
  66         if (header_prefix)
  67                 fputs(header_prefix, stdout);
  68
  69         if (!revs.graph) {
  70                 if (commit->object.flags & BOUNDARY)
  71                         putchar('-');
  72                 else if (commit->object.flags & UNINTERESTING)
  73                         putchar('^');
  74                 else if (revs.left_right) {
  75                         if (commit->object.flags & SYMMETRIC_LEFT)
  76                                 putchar('<');
  77                         else
  78                                 putchar('>');
  79                 }
  80         }
  81         if (revs.abbrev_commit && revs.abbrev)
  82                 fputs(find_unique_abbrev(commit->object.sha1, revs.abbrev),
  83                       stdout);
  84         else
  85                 fputs(sha1_to_hex(commit->object.sha1), stdout);
  86         if (revs.print_parents) {
  87                 struct commit_list *parents = commit->parents;
  88                 while (parents) {
  89                         printf(" %s", sha1_to_hex(parents->item->object.sha1));
  90                         parents = parents->next;
  91                 }
  92         }
  93         if (revs.children.name) {
  94                 struct commit_list *children;
  95
  96                 children = lookup_decoration(&revs.children, &commit->object);
  97                 while (children) {
  98                         printf(" %s", sha1_to_hex(children->item->object.sha1));
  99                         children = children->next;
 100                 }
 101         }
 102         show_decorations(&revs, commit);
 103         if (revs.commit_format == CMIT_FMT_ONELINE)
 104                 putchar(' ');
 105         else
 106                 putchar('\n');
 107
 108         if (revs.verbose_header && commit->buffer) {
 109                 struct strbuf buf = STRBUF_INIT;
 110                 pretty_print_commit(revs.commit_format, commit,
 111                                     &buf, revs.abbrev, NULL, NULL,
 112                                     revs.date_mode, 0);
 113                 if (revs.graph) {
 114                         if (buf.len) {
 115                                 if (revs.commit_format != CMIT_FMT_ONELINE)
 116                                         graph_show_oneline(revs.graph);
 117
 118                                 graph_show_commit_msg(revs.graph, &buf);
 119
 120                                 /*
 121                                  * Add a newline after the commit message.
 122                                  *
 123                                  * Usually, this newline produces a blank
 124                                  * padding line between entries, in which case
 125                                  * we need to add graph padding on this line.
 126                                  *
 127                                  * However, the commit message may not end in a
 128                                  * newline.  In this case the newline simply
 129                                  * ends the last line of the commit message,
 130                                  * and we don't need any graph output.  (This
 131                                  * always happens with CMIT_FMT_ONELINE, and it
 132                                  * happens with CMIT_FMT_USERFORMAT when the
 133                                  * format doesn't explicitly end in a newline.)
 134                                  */
 135                                 if (buf.len && buf.buf[buf.len - 1] == '\n')
 136                                         graph_show_padding(revs.graph);
 137                                 putchar('\n');
 138                         } else {
 139                                 /*
 140                                  * If the message buffer is empty, just show
 141                                  * the rest of the graph output for this
 142                                  * commit.
 143                                  */
 144                                 if (graph_show_remainder(revs.graph))
 145                                         putchar('\n');
 146                         }
 147                 } else {
 148                         if (buf.len)
 149                                 printf("%s%c", buf.buf, hdr_termination);
 150                 }
 151                 strbuf_release(&buf);
 152         } else {
 153                 if (graph_show_remainder(revs.graph))
 154                         putchar('\n');
 155         }
 156         maybe_flush_or_die(stdout, "stdout");
 157         finish_commit(commit);
 158 }
 159
 160 static void finish_commit(struct commit *commit)
 161 {
 162         if (commit->parents) {
 163                 free_commit_list(commit->parents);
 164                 commit->parents = NULL;
 165         }
 166         free(commit->buffer);
 167         commit->buffer = NULL;
 168 }
 169
 170 static void finish_object(struct object_array_entry *p)
 171 {
 172         if (p->item->type == OBJ_BLOB && !has_sha1_file(p->item->sha1))
 173                 die("missing blob object '%s'", sha1_to_hex(p->item->sha1));
 174 }
 175
 176 static void show_object(struct object_array_entry *p)
 177 {
 178         /* An object with name "foo\n0000000..." can be used to
 179          * confuse downstream "git pack-objects" very badly.
 180          */
 181         const char *ep = strchr(p->name, '\n');
 182
 183         finish_object(p);
 184         if (ep) {
 185                 printf("%s %.*s\n", sha1_to_hex(p->item->sha1),
 186                        (int) (ep - p->name),
 187                        p->name);
 188         }
 189         else
 190                 printf("%s %s\n", sha1_to_hex(p->item->sha1), p->name);
 191 }
 192
 193 static void show_edge(struct commit *commit)
 194 {
 195         printf("-%s\n", sha1_to_hex(commit->object.sha1));
 196 }
 197
 198 /*
 199  * This is a truly stupid algorithm, but it's only
 200  * used for bisection, and we just don't care enough.
 201  *
 202  * We care just barely enough to avoid recursing for
 203  * non-merge entries.
 204  */
 205 static int count_distance(struct commit_list *entry)
 206 {
 207         int nr = 0;
 208
 209         while (entry) {
 210                 struct commit *commit = entry->item;
 211                 struct commit_list *p;
 212
 213                 if (commit->object.flags & (UNINTERESTING | COUNTED))
 214                         break;
 215                 if (!(commit->object.flags & TREESAME))
 216                         nr++;
 217                 commit->object.flags |= COUNTED;
 218                 p = commit->parents;
 219                 entry = p;
 220                 if (p) {
 221                         p = p->next;
 222                         while (p) {
 223                                 nr += count_distance(p);
 224                                 p = p->next;
 225                         }
 226                 }
 227         }
 228
 229         return nr;
 230 }
 231
 232 static void clear_distance(struct commit_list *list)
 233 {
 234         while (list) {
 235                 struct commit *commit = list->item;
 236                 commit->object.flags &= ~COUNTED;
 237                 list = list->next;
 238         }
 239 }
 240
 241 #define DEBUG_BISECT 0
 242
 243 static inline int weight(struct commit_list *elem)
 244 {
 245         return *((int*)(elem->item->util));
 246 }
 247
 248 static inline void weight_set(struct commit_list *elem, int weight)
 249 {
 250         *((int*)(elem->item->util)) = weight;
 251 }
 252
 253 static int count_interesting_parents(struct commit *commit)
 254 {
 255         struct commit_list *p;
 256         int count;
 257
 258         for (count = 0, p = commit->parents; p; p = p->next) {
 259                 if (p->item->object.flags & UNINTERESTING)
 260                         continue;
 261                 count++;
 262         }
 263         return count;
 264 }
 265
 266 static inline int halfway(struct commit_list *p, int nr)
 267 {
 268         /*
 269          * Don't short-cut something we are not going to return!
 270          */
 271         if (p->item->object.flags & TREESAME)
 272                 return 0;
 273         if (DEBUG_BISECT)
 274                 return 0;
 275         /*
 276          * 2 and 3 are halfway of 5.
 277          * 3 is halfway of 6 but 2 and 4 are not.
 278          */
 279         switch (2 * weight(p) - nr) {
 280         case -1: case 0: case 1:
 281                 return 1;
 282         default:
 283                 return 0;
 284         }
 285 }
 286
 287 #if !DEBUG_BISECT
 288 #define show_list(a,b,c,d) do { ; } while (0)
 289 #else
 290 static void show_list(const char *debug, int counted, int nr,
 291                       struct commit_list *list)
 292 {
 293         struct commit_list *p;
 294
 295         fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
 296
 297         for (p = list; p; p = p->next) {
 298                 struct commit_list *pp;
 299                 struct commit *commit = p->item;
 300                 unsigned flags = commit->object.flags;
 301                 enum object_type type;
 302                 unsigned long size;
 303                 char *buf = read_sha1_file(commit->object.sha1, &type, &size);
 304                 char *ep, *sp;
 305
 306                 fprintf(stderr, "%c%c%c ",
 307                         (flags & TREESAME) ? ' ' : 'T',
 308                         (flags & UNINTERESTING) ? 'U' : ' ',
 309                         (flags & COUNTED) ? 'C' : ' ');
 310                 if (commit->util)
 311                         fprintf(stderr, "%3d", weight(p));
 312                 else
 313                         fprintf(stderr, "---");
 314                 fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.sha1));
 315                 for (pp = commit->parents; pp; pp = pp->next)
 316                         fprintf(stderr, " %.*s", 8,
 317                                 sha1_to_hex(pp->item->object.sha1));
 318
 319                 sp = strstr(buf, "\n\n");
 320                 if (sp) {
 321                         sp += 2;
 322                         for (ep = sp; *ep && *ep != '\n'; ep++)
 323                                 ;
 324                         fprintf(stderr, " %.*s", (int)(ep - sp), sp);
 325                 }
 326                 fprintf(stderr, "\n");
 327         }
 328 }
 329 #endif /* DEBUG_BISECT */
 330
 331 static struct commit_list *best_bisection(struct commit_list *list, int nr)
 332 {
 333         struct commit_list *p, *best;
 334         int best_distance = -1;
 335
 336         best = list;
 337         for (p = list; p; p = p->next) {
 338                 int distance;
 339                 unsigned flags = p->item->object.flags;
 340
 341                 if (flags & TREESAME)
 342                         continue;
 343                 distance = weight(p);
 344                 if (nr - distance < distance)
 345                         distance = nr - distance;
 346                 if (distance > best_distance) {
 347                         best = p;
 348                         best_distance = distance;
 349                 }
 350         }
 351
 352         return best;
 353 }
 354
 355 struct commit_dist {
 356         struct commit *commit;
 357         int distance;
 358 };
 359
 360 static int compare_commit_dist(const void *a_, const void *b_)
 361 {
 362         struct commit_dist *a, *b;
 363
 364         a = (struct commit_dist *)a_;
 365         b = (struct commit_dist *)b_;
 366         if (a->distance != b->distance)
 367                 return b->distance - a->distance; /* desc sort */
 368         return hashcmp(a->commit->object.sha1, b->commit->object.sha1);
 369 }
 370
 371 static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
 372 {
 373         struct commit_list *p;
 374         struct commit_dist *array = xcalloc(nr, sizeof(*array));
 375         int cnt, i;
 376
 377         for (p = list, cnt = 0; p; p = p->next) {
 378                 int distance;
 379                 unsigned flags = p->item->object.flags;
 380
 381                 if (flags & TREESAME)
 382                         continue;
 383                 distance = weight(p);
 384                 if (nr - distance < distance)
 385                         distance = nr - distance;
 386                 array[cnt].commit = p->item;
 387                 array[cnt].distance = distance;
 388                 cnt++;
 389         }
 390         qsort(array, cnt, sizeof(*array), compare_commit_dist);
 391         for (p = list, i = 0; i < cnt; i++) {
 392                 struct name_decoration *r = xmalloc(sizeof(*r) + 100);
 393                 struct object *obj = &(array[i].commit->object);
 394
 395                 sprintf(r->name, "dist=%d", array[i].distance);
 396                 r->next = add_decoration(&name_decoration, obj, r);
 397                 p->item = array[i].commit;
 398                 p = p->next;
 399         }
 400         if (p)
 401                 p->next = NULL;
 402         free(array);
 403         return list;
 404 }
 405
 406 /*
 407  * zero or positive weight is the number of interesting commits it can
 408  * reach, including itself.  Especially, weight = 0 means it does not
 409  * reach any tree-changing commits (e.g. just above uninteresting one
 410  * but traversal is with pathspec).
 411  *
 412  * weight = -1 means it has one parent and its distance is yet to
 413  * be computed.
 414  *
 415  * weight = -2 means it has more than one parent and its distance is
 416  * unknown.  After running count_distance() first, they will get zero
 417  * or positive distance.
 418  */
 419 static struct commit_list *do_find_bisection(struct commit_list *list,
 420                                              int nr, int *weights,
 421                                              int find_all)
 422 {
 423         int n, counted;
 424         struct commit_list *p;
 425
 426         counted = 0;
 427
 428         for (n = 0, p = list; p; p = p->next) {
 429                 struct commit *commit = p->item;
 430                 unsigned flags = commit->object.flags;
 431
 432                 p->item->util = &weights[n++];
 433                 switch (count_interesting_parents(commit)) {
 434                 case 0:
 435                         if (!(flags & TREESAME)) {
 436                                 weight_set(p, 1);
 437                                 counted++;
 438                                 show_list("bisection 2 count one",
 439                                           counted, nr, list);
 440                         }
 441                         /*
 442                          * otherwise, it is known not to reach any
 443                          * tree-changing commit and gets weight 0.
 444                          */
 445                         break;
 446                 case 1:
 447                         weight_set(p, -1);
 448                         break;
 449                 default:
 450                         weight_set(p, -2);
 451                         break;
 452                 }
 453         }
 454
 455         show_list("bisection 2 initialize", counted, nr, list);
 456
 457         /*
 458          * If you have only one parent in the resulting set
 459          * then you can reach one commit more than that parent
 460          * can reach.  So we do not have to run the expensive
 461          * count_distance() for single strand of pearls.
 462          *
 463          * However, if you have more than one parents, you cannot
 464          * just add their distance and one for yourself, since
 465          * they usually reach the same ancestor and you would
 466          * end up counting them twice that way.
 467          *
 468          * So we will first count distance of merges the usual
 469          * way, and then fill the blanks using cheaper algorithm.
 470          */
 471         for (p = list; p; p = p->next) {
 472                 if (p->item->object.flags & UNINTERESTING)
 473                         continue;
 474                 if (weight(p) != -2)
 475                         continue;
 476                 weight_set(p, count_distance(p));
 477                 clear_distance(list);
 478
 479                 /* Does it happen to be at exactly half-way? */
 480                 if (!find_all && halfway(p, nr))
 481                         return p;
 482                 counted++;
 483         }
 484
 485         show_list("bisection 2 count_distance", counted, nr, list);
 486
 487         while (counted < nr) {
 488                 for (p = list; p; p = p->next) {
 489                         struct commit_list *q;
 490                         unsigned flags = p->item->object.flags;
 491
 492                         if (0 <= weight(p))
 493                                 continue;
 494                         for (q = p->item->parents; q; q = q->next) {
 495                                 if (q->item->object.flags & UNINTERESTING)
 496                                         continue;
 497                                 if (0 <= weight(q))
 498                                         break;
 499                         }
 500                         if (!q)
 501                                 continue;
 502
 503                         /*
 504                          * weight for p is unknown but q is known.
 505                          * add one for p itself if p is to be counted,
 506                          * otherwise inherit it from q directly.
 507                          */
 508                         if (!(flags & TREESAME)) {
 509                                 weight_set(p, weight(q)+1);
 510                                 counted++;
 511                                 show_list("bisection 2 count one",
 512                                           counted, nr, list);
 513                         }
 514                         else
 515                                 weight_set(p, weight(q));
 516
 517                         /* Does it happen to be at exactly half-way? */
 518                         if (!find_all && halfway(p, nr))
 519                                 return p;
 520                 }
 521         }
 522
 523         show_list("bisection 2 counted all", counted, nr, list);
 524
 525         if (!find_all)
 526                 return best_bisection(list, nr);
 527         else
 528                 return best_bisection_sorted(list, nr);
 529 }
 530
 531 static struct commit_list *find_bisection(struct commit_list *list,
 532                                           int *reaches, int *all,
 533                                           int find_all)
 534 {
 535         int nr, on_list;
 536         struct commit_list *p, *best, *next, *last;
 537         int *weights;
 538
 539         show_list("bisection 2 entry", 0, 0, list);
 540
 541         /*
 542          * Count the number of total and tree-changing items on the
 543          * list, while reversing the list.
 544          */
 545         for (nr = on_list = 0, last = NULL, p = list;
 546              p;
 547              p = next) {
 548                 unsigned flags = p->item->object.flags;
 549
 550                 next = p->next;
 551                 if (flags & UNINTERESTING)
 552                         continue;
 553                 p->next = last;
 554                 last = p;
 555                 if (!(flags & TREESAME))
 556                         nr++;
 557                 on_list++;
 558         }
 559         list = last;
 560         show_list("bisection 2 sorted", 0, nr, list);
 561
 562         *all = nr;
 563         weights = xcalloc(on_list, sizeof(*weights));
 564
 565         /* Do the real work of finding bisection commit. */
 566         best = do_find_bisection(list, nr, weights, find_all);
 567         if (best) {
 568                 if (!find_all)
 569                         best->next = NULL;
 570                 *reaches = weight(best);
 571         }
 572         free(weights);
 573         return best;
 574 }
 575
 576 static inline int log2i(int n)
 577 {
 578         int log2 = 0;
 579
 580         for (; n > 1; n >>= 1)
 581                 log2++;
 582
 583         return log2;
 584 }
 585
 586 static inline int exp2i(int n)
 587 {
 588         return 1 << n;
 589 }
 590
 591 /*
 592  * Estimate the number of bisect steps left (after the current step)
 593  *
 594  * For any x between 0 included and 2^n excluded, the probability for
 595  * n - 1 steps left looks like:
 596  *
 597  * P(2^n + x) == (2^n - x) / (2^n + x)
 598  *
 599  * and P(2^n + x) < 0.5 means 2^n < 3x
 600  */
 601 static int estimate_bisect_steps(int all)
 602 {
 603         int n, x, e;
 604
 605         if (all < 3)
 606                 return 0;
 607
 608         n = log2i(all);
 609         e = exp2i(n);
 610         x = all - e;
 611
 612         return (e < 3 * x) ? n : n - 1;
 613 }
 614
 615 int cmd_rev_list(int argc, const char **argv, const char *prefix)
 616 {
 617         struct commit_list *list;
 618         int i;
 619         int read_from_stdin = 0;
 620         int bisect_list = 0;
 621         int bisect_show_vars = 0;
 622         int bisect_find_all = 0;
 623         int quiet = 0;
 624
 625         git_config(git_default_config, NULL);
 626         init_revisions(&revs, prefix);
 627         revs.abbrev = 0;
 628         revs.commit_format = CMIT_FMT_UNSPECIFIED;
 629         argc = setup_revisions(argc, argv, &revs, NULL);
 630
 631         quiet = DIFF_OPT_TST(&revs.diffopt, QUIET);
 632         for (i = 1 ; i < argc; i++) {
 633                 const char *arg = argv[i];
 634
 635                 if (!strcmp(arg, "--header")) {
 636                         revs.verbose_header = 1;
 637                         continue;
 638                 }
 639                 if (!strcmp(arg, "--timestamp")) {
 640                         show_timestamp = 1;
 641                         continue;
 642                 }
 643                 if (!strcmp(arg, "--bisect")) {
 644                         bisect_list = 1;
 645                         continue;
 646                 }
 647                 if (!strcmp(arg, "--bisect-all")) {
 648                         bisect_list = 1;
 649                         bisect_find_all = 1;
 650                         revs.show_decorations = 1;
 651                         continue;
 652                 }
 653                 if (!strcmp(arg, "--bisect-vars")) {
 654                         bisect_list = 1;
 655                         bisect_show_vars = 1;
 656                         continue;
 657                 }
 658                 if (!strcmp(arg, "--stdin")) {
 659                         if (read_from_stdin++)
 660                                 die("--stdin given twice?");
 661                         read_revisions_from_stdin(&revs);
 662                         continue;
 663                 }
 664                 usage(rev_list_usage);
 665
 666         }
 667         if (revs.commit_format != CMIT_FMT_UNSPECIFIED) {
 668                 /* The command line has a --pretty  */
 669                 hdr_termination = '\n';
 670                 if (revs.commit_format == CMIT_FMT_ONELINE)
 671                         header_prefix = "";
 672                 else
 673                         header_prefix = "commit ";
 674         }
 675         else if (revs.verbose_header)
 676                 /* Only --header was specified */
 677                 revs.commit_format = CMIT_FMT_RAW;
 678
 679         list = revs.commits;
 680
 681         if ((!list &&
 682              (!(revs.tag_objects||revs.tree_objects||revs.blob_objects) &&
 683               !revs.pending.nr)) ||
 684             revs.diff)
 685                 usage(rev_list_usage);
 686
 687         save_commit_buffer = revs.verbose_header ||
 688                 revs.grep_filter.pattern_list;
 689         if (bisect_list)
 690                 revs.limited = 1;
 691
 692         if (prepare_revision_walk(&revs))
 693                 die("revision walk setup failed");
 694         if (revs.tree_objects)
 695                 mark_edges_uninteresting(revs.commits, &revs, show_edge);
 696
 697         if (bisect_list) {
 698                 int reaches = reaches, all = all;
 699
 700                 revs.commits = find_bisection(revs.commits, &reaches, &all,
 701                                               bisect_find_all);
 702                 if (bisect_show_vars) {
 703                         int cnt;
 704                         char hex[41];
 705                         if (!revs.commits)
 706                                 return 1;
 707                         /*
 708                          * revs.commits can reach "reaches" commits among
 709                          * "all" commits.  If it is good, then there are
 710                          * (all-reaches) commits left to be bisected.
 711                          * On the other hand, if it is bad, then the set
 712                          * to bisect is "reaches".
 713                          * A bisect set of size N has (N-1) commits further
 714                          * to test, as we already know one bad one.
 715                          */
 716                         cnt = all - reaches;
 717                         if (cnt < reaches)
 718                                 cnt = reaches;
 719                         strcpy(hex, sha1_to_hex(revs.commits->item->object.sha1));
 720
 721                         if (bisect_find_all) {
 722                                 traverse_commit_list(&revs, show_commit, show_object);
 723                                 printf("------\n");
 724                         }
 725
 726                         printf("bisect_rev=%s\n"
 727                                "bisect_nr=%d\n"
 728                                "bisect_good=%d\n"
 729                                "bisect_bad=%d\n"
 730                                "bisect_all=%d\n"
 731                                "bisect_steps=%d\n",
 732                                hex,
 733                                cnt - 1,
 734                                all - reaches - 1,
 735                                reaches - 1,
 736                                all,
 737                                estimate_bisect_steps(all));
 738                         return 0;
 739                 }
 740         }
 741
 742         traverse_commit_list(&revs,
 743                 quiet ? finish_commit : show_commit,
 744                 quiet ? finish_object : show_object);
 745
 746         return 0;
 747 }