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
  14 /* bits #0-15 in revision.h */
  15
  16 #define COUNTED         (1u<<16)
  17
  18 static const char rev_list_usage[] =
  19 "git-rev-list [OPTION] <commit-id>... [ -- paths... ]\n"
  20 "  limiting output:\n"
  21 "    --max-count=nr\n"
  22 "    --max-age=epoch\n"
  23 "    --min-age=epoch\n"
  24 "    --sparse\n"
  25 "    --no-merges\n"
  26 "    --remove-empty\n"
  27 "    --all\n"
  28 "    --branches\n"
  29 "    --tags\n"
  30 "    --remotes\n"
  31 "    --stdin\n"
  32 "    --quiet\n"
  33 "  ordering output:\n"
  34 "    --topo-order\n"
  35 "    --date-order\n"
  36 "  formatting output:\n"
  37 "    --parents\n"
  38 "    --objects | --objects-edge\n"
  39 "    --unpacked\n"
  40 "    --header | --pretty\n"
  41 "    --abbrev=nr | --no-abbrev\n"
  42 "    --abbrev-commit\n"
  43 "    --left-right\n"
  44 "  special purpose:\n"
  45 "    --bisect\n"
  46 "    --bisect-vars\n"
  47 "    --bisect-all"
  48 ;
  49
  50 static struct rev_info revs;
  51
  52 static int bisect_list;
  53 static int show_timestamp;
  54 static int hdr_termination;
  55 static const char *header_prefix;
  56
  57 static void finish_commit(struct commit *commit);
  58 static void show_commit(struct commit *commit)
  59 {
  60         if (show_timestamp)
  61                 printf("%lu ", commit->date);
  62         if (header_prefix)
  63                 fputs(header_prefix, stdout);
  64         if (commit->object.flags & BOUNDARY)
  65                 putchar('-');
  66         else if (commit->object.flags & UNINTERESTING)
  67                 putchar('^');
  68         else if (revs.left_right) {
  69                 if (commit->object.flags & SYMMETRIC_LEFT)
  70                         putchar('<');
  71                 else
  72                         putchar('>');
  73         }
  74         if (revs.abbrev_commit && revs.abbrev)
  75                 fputs(find_unique_abbrev(commit->object.sha1, revs.abbrev),
  76                       stdout);
  77         else
  78                 fputs(sha1_to_hex(commit->object.sha1), stdout);
  79         if (revs.parents) {
  80                 struct commit_list *parents = commit->parents;
  81                 while (parents) {
  82                         printf(" %s", sha1_to_hex(parents->item->object.sha1));
  83                         parents = parents->next;
  84                 }
  85         }
  86         show_decorations(commit);
  87         if (revs.commit_format == CMIT_FMT_ONELINE)
  88                 putchar(' ');
  89         else
  90                 putchar('\n');
  91
  92         if (revs.verbose_header && commit->buffer) {
  93                 struct strbuf buf;
  94                 strbuf_init(&buf, 0);
  95                 pretty_print_commit(revs.commit_format, commit,
  96                                     &buf, revs.abbrev, NULL, NULL,
  97                                     revs.date_mode, 0);
  98                 if (buf.len)
  99                         printf("%s%c", buf.buf, hdr_termination);
 100                 strbuf_release(&buf);
 101         }
 102         maybe_flush_or_die(stdout, "stdout");
 103         finish_commit(commit);
 104 }
 105
 106 static void finish_commit(struct commit *commit)
 107 {
 108         if (commit->parents) {
 109                 free_commit_list(commit->parents);
 110                 commit->parents = NULL;
 111         }
 112         free(commit->buffer);
 113         commit->buffer = NULL;
 114 }
 115
 116 static void finish_object(struct object_array_entry *p)
 117 {
 118         if (p->item->type == OBJ_BLOB && !has_sha1_file(p->item->sha1))
 119                 die("missing blob object '%s'", sha1_to_hex(p->item->sha1));
 120 }
 121
 122 static void show_object(struct object_array_entry *p)
 123 {
 124         /* An object with name "foo\n0000000..." can be used to
 125          * confuse downstream git-pack-objects very badly.
 126          */
 127         const char *ep = strchr(p->name, '\n');
 128
 129         finish_object(p);
 130         if (ep) {
 131                 printf("%s %.*s\n", sha1_to_hex(p->item->sha1),
 132                        (int) (ep - p->name),
 133                        p->name);
 134         }
 135         else
 136                 printf("%s %s\n", sha1_to_hex(p->item->sha1), p->name);
 137 }
 138
 139 static void show_edge(struct commit *commit)
 140 {
 141         printf("-%s\n", sha1_to_hex(commit->object.sha1));
 142 }
 143
 144 /*
 145  * This is a truly stupid algorithm, but it's only
 146  * used for bisection, and we just don't care enough.
 147  *
 148  * We care just barely enough to avoid recursing for
 149  * non-merge entries.
 150  */
 151 static int count_distance(struct commit_list *entry)
 152 {
 153         int nr = 0;
 154
 155         while (entry) {
 156                 struct commit *commit = entry->item;
 157                 struct commit_list *p;
 158
 159                 if (commit->object.flags & (UNINTERESTING | COUNTED))
 160                         break;
 161                 if (!(commit->object.flags & TREESAME))
 162                         nr++;
 163                 commit->object.flags |= COUNTED;
 164                 p = commit->parents;
 165                 entry = p;
 166                 if (p) {
 167                         p = p->next;
 168                         while (p) {
 169                                 nr += count_distance(p);
 170                                 p = p->next;
 171                         }
 172                 }
 173         }
 174
 175         return nr;
 176 }
 177
 178 static void clear_distance(struct commit_list *list)
 179 {
 180         while (list) {
 181                 struct commit *commit = list->item;
 182                 commit->object.flags &= ~COUNTED;
 183                 list = list->next;
 184         }
 185 }
 186
 187 #define DEBUG_BISECT 0
 188
 189 static inline int weight(struct commit_list *elem)
 190 {
 191         return *((int*)(elem->item->util));
 192 }
 193
 194 static inline void weight_set(struct commit_list *elem, int weight)
 195 {
 196         *((int*)(elem->item->util)) = weight;
 197 }
 198
 199 static int count_interesting_parents(struct commit *commit)
 200 {
 201         struct commit_list *p;
 202         int count;
 203
 204         for (count = 0, p = commit->parents; p; p = p->next) {
 205                 if (p->item->object.flags & UNINTERESTING)
 206                         continue;
 207                 count++;
 208         }
 209         return count;
 210 }
 211
 212 static inline int halfway(struct commit_list *p, int nr)
 213 {
 214         /*
 215          * Don't short-cut something we are not going to return!
 216          */
 217         if (p->item->object.flags & TREESAME)
 218                 return 0;
 219         if (DEBUG_BISECT)
 220                 return 0;
 221         /*
 222          * 2 and 3 are halfway of 5.
 223          * 3 is halfway of 6 but 2 and 4 are not.
 224          */
 225         switch (2 * weight(p) - nr) {
 226         case -1: case 0: case 1:
 227                 return 1;
 228         default:
 229                 return 0;
 230         }
 231 }
 232
 233 #if !DEBUG_BISECT
 234 #define show_list(a,b,c,d) do { ; } while (0)
 235 #else
 236 static void show_list(const char *debug, int counted, int nr,
 237                       struct commit_list *list)
 238 {
 239         struct commit_list *p;
 240
 241         fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
 242
 243         for (p = list; p; p = p->next) {
 244                 struct commit_list *pp;
 245                 struct commit *commit = p->item;
 246                 unsigned flags = commit->object.flags;
 247                 enum object_type type;
 248                 unsigned long size;
 249                 char *buf = read_sha1_file(commit->object.sha1, &type, &size);
 250                 char *ep, *sp;
 251
 252                 fprintf(stderr, "%c%c%c ",
 253                         (flags & TREESAME) ? ' ' : 'T',
 254                         (flags & UNINTERESTING) ? 'U' : ' ',
 255                         (flags & COUNTED) ? 'C' : ' ');
 256                 if (commit->util)
 257                         fprintf(stderr, "%3d", weight(p));
 258                 else
 259                         fprintf(stderr, "---");
 260                 fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.sha1));
 261                 for (pp = commit->parents; pp; pp = pp->next)
 262                         fprintf(stderr, " %.*s", 8,
 263                                 sha1_to_hex(pp->item->object.sha1));
 264
 265                 sp = strstr(buf, "\n\n");
 266                 if (sp) {
 267                         sp += 2;
 268                         for (ep = sp; *ep && *ep != '\n'; ep++)
 269                                 ;
 270                         fprintf(stderr, " %.*s", (int)(ep - sp), sp);
 271                 }
 272                 fprintf(stderr, "\n");
 273         }
 274 }
 275 #endif /* DEBUG_BISECT */
 276
 277 static struct commit_list *best_bisection(struct commit_list *list, int nr)
 278 {
 279         struct commit_list *p, *best;
 280         int best_distance = -1;
 281
 282         best = list;
 283         for (p = list; p; p = p->next) {
 284                 int distance;
 285                 unsigned flags = p->item->object.flags;
 286
 287                 if (flags & TREESAME)
 288                         continue;
 289                 distance = weight(p);
 290                 if (nr - distance < distance)
 291                         distance = nr - distance;
 292                 if (distance > best_distance) {
 293                         best = p;
 294                         best_distance = distance;
 295                 }
 296         }
 297
 298         return best;
 299 }
 300
 301 struct commit_dist {
 302         struct commit *commit;
 303         int distance;
 304 };
 305
 306 static int compare_commit_dist(const void *a_, const void *b_)
 307 {
 308         struct commit_dist *a, *b;
 309
 310         a = (struct commit_dist *)a_;
 311         b = (struct commit_dist *)b_;
 312         if (a->distance != b->distance)
 313                 return b->distance - a->distance; /* desc sort */
 314         return hashcmp(a->commit->object.sha1, b->commit->object.sha1);
 315 }
 316
 317 static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
 318 {
 319         struct commit_list *p;
 320         struct commit_dist *array = xcalloc(nr, sizeof(*array));
 321         int cnt, i;
 322
 323         for (p = list, cnt = 0; p; p = p->next) {
 324                 int distance;
 325                 unsigned flags = p->item->object.flags;
 326
 327                 if (flags & TREESAME)
 328                         continue;
 329                 distance = weight(p);
 330                 if (nr - distance < distance)
 331                         distance = nr - distance;
 332                 array[cnt].commit = p->item;
 333                 array[cnt].distance = distance;
 334                 cnt++;
 335         }
 336         qsort(array, cnt, sizeof(*array), compare_commit_dist);
 337         for (p = list, i = 0; i < cnt; i++) {
 338                 struct name_decoration *r = xmalloc(sizeof(*r) + 100);
 339                 struct object *obj = &(array[i].commit->object);
 340
 341                 sprintf(r->name, "dist=%d", array[i].distance);
 342                 r->next = add_decoration(&name_decoration, obj, r);
 343                 p->item = array[i].commit;
 344                 p = p->next;
 345         }
 346         if (p)
 347                 p->next = NULL;
 348         free(array);
 349         return list;
 350 }
 351
 352 /*
 353  * zero or positive weight is the number of interesting commits it can
 354  * reach, including itself.  Especially, weight = 0 means it does not
 355  * reach any tree-changing commits (e.g. just above uninteresting one
 356  * but traversal is with pathspec).
 357  *
 358  * weight = -1 means it has one parent and its distance is yet to
 359  * be computed.
 360  *
 361  * weight = -2 means it has more than one parent and its distance is
 362  * unknown.  After running count_distance() first, they will get zero
 363  * or positive distance.
 364  */
 365 static struct commit_list *do_find_bisection(struct commit_list *list,
 366                                              int nr, int *weights,
 367                                              int find_all)
 368 {
 369         int n, counted;
 370         struct commit_list *p;
 371
 372         counted = 0;
 373
 374         for (n = 0, p = list; p; p = p->next) {
 375                 struct commit *commit = p->item;
 376                 unsigned flags = commit->object.flags;
 377
 378                 p->item->util = &weights[n++];
 379                 switch (count_interesting_parents(commit)) {
 380                 case 0:
 381                         if (!(flags & TREESAME)) {
 382                                 weight_set(p, 1);
 383                                 counted++;
 384                                 show_list("bisection 2 count one",
 385                                           counted, nr, list);
 386                         }
 387                         /*
 388                          * otherwise, it is known not to reach any
 389                          * tree-changing commit and gets weight 0.
 390                          */
 391                         break;
 392                 case 1:
 393                         weight_set(p, -1);
 394                         break;
 395                 default:
 396                         weight_set(p, -2);
 397                         break;
 398                 }
 399         }
 400
 401         show_list("bisection 2 initialize", counted, nr, list);
 402
 403         /*
 404          * If you have only one parent in the resulting set
 405          * then you can reach one commit more than that parent
 406          * can reach.  So we do not have to run the expensive
 407          * count_distance() for single strand of pearls.
 408          *
 409          * However, if you have more than one parents, you cannot
 410          * just add their distance and one for yourself, since
 411          * they usually reach the same ancestor and you would
 412          * end up counting them twice that way.
 413          *
 414          * So we will first count distance of merges the usual
 415          * way, and then fill the blanks using cheaper algorithm.
 416          */
 417         for (p = list; p; p = p->next) {
 418                 if (p->item->object.flags & UNINTERESTING)
 419                         continue;
 420                 if (weight(p) != -2)
 421                         continue;
 422                 weight_set(p, count_distance(p));
 423                 clear_distance(list);
 424
 425                 /* Does it happen to be at exactly half-way? */
 426                 if (!find_all && halfway(p, nr))
 427                         return p;
 428                 counted++;
 429         }
 430
 431         show_list("bisection 2 count_distance", counted, nr, list);
 432
 433         while (counted < nr) {
 434                 for (p = list; p; p = p->next) {
 435                         struct commit_list *q;
 436                         unsigned flags = p->item->object.flags;
 437
 438                         if (0 <= weight(p))
 439                                 continue;
 440                         for (q = p->item->parents; q; q = q->next) {
 441                                 if (q->item->object.flags & UNINTERESTING)
 442                                         continue;
 443                                 if (0 <= weight(q))
 444                                         break;
 445                         }
 446                         if (!q)
 447                                 continue;
 448
 449                         /*
 450                          * weight for p is unknown but q is known.
 451                          * add one for p itself if p is to be counted,
 452                          * otherwise inherit it from q directly.
 453                          */
 454                         if (!(flags & TREESAME)) {
 455                                 weight_set(p, weight(q)+1);
 456                                 counted++;
 457                                 show_list("bisection 2 count one",
 458                                           counted, nr, list);
 459                         }
 460                         else
 461                                 weight_set(p, weight(q));
 462
 463                         /* Does it happen to be at exactly half-way? */
 464                         if (!find_all && halfway(p, nr))
 465                                 return p;
 466                 }
 467         }
 468
 469         show_list("bisection 2 counted all", counted, nr, list);
 470
 471         if (!find_all)
 472                 return best_bisection(list, nr);
 473         else
 474                 return best_bisection_sorted(list, nr);
 475 }
 476
 477 static struct commit_list *find_bisection(struct commit_list *list,
 478                                           int *reaches, int *all,
 479                                           int find_all)
 480 {
 481         int nr, on_list;
 482         struct commit_list *p, *best, *next, *last;
 483         int *weights;
 484
 485         show_list("bisection 2 entry", 0, 0, list);
 486
 487         /*
 488          * Count the number of total and tree-changing items on the
 489          * list, while reversing the list.
 490          */
 491         for (nr = on_list = 0, last = NULL, p = list;
 492              p;
 493              p = next) {
 494                 unsigned flags = p->item->object.flags;
 495
 496                 next = p->next;
 497                 if (flags & UNINTERESTING)
 498                         continue;
 499                 p->next = last;
 500                 last = p;
 501                 if (!(flags & TREESAME))
 502                         nr++;
 503                 on_list++;
 504         }
 505         list = last;
 506         show_list("bisection 2 sorted", 0, nr, list);
 507
 508         *all = nr;
 509         weights = xcalloc(on_list, sizeof(*weights));
 510
 511         /* Do the real work of finding bisection commit. */
 512         best = do_find_bisection(list, nr, weights, find_all);
 513         if (best) {
 514                 if (!find_all)
 515                         best->next = NULL;
 516                 *reaches = weight(best);
 517         }
 518         free(weights);
 519         return best;
 520 }
 521
 522 static void read_revisions_from_stdin(struct rev_info *revs)
 523 {
 524         char line[1000];
 525
 526         while (fgets(line, sizeof(line), stdin) != NULL) {
 527                 int len = strlen(line);
 528                 if (len && line[len - 1] == '\n')
 529                         line[--len] = 0;
 530                 if (!len)
 531                         break;
 532                 if (line[0] == '-')
 533                         die("options not supported in --stdin mode");
 534                 if (handle_revision_arg(line, revs, 0, 1))
 535                         die("bad revision '%s'", line);
 536         }
 537 }
 538
 539 int cmd_rev_list(int argc, const char **argv, const char *prefix)
 540 {
 541         struct commit_list *list;
 542         int i;
 543         int read_from_stdin = 0;
 544         int bisect_show_vars = 0;
 545         int bisect_find_all = 0;
 546         int quiet = 0;
 547
 548         git_config(git_default_config);
 549         init_revisions(&revs, prefix);
 550         revs.abbrev = 0;
 551         revs.commit_format = CMIT_FMT_UNSPECIFIED;
 552         argc = setup_revisions(argc, argv, &revs, NULL);
 553
 554         for (i = 1 ; i < argc; i++) {
 555                 const char *arg = argv[i];
 556
 557                 if (!strcmp(arg, "--header")) {
 558                         revs.verbose_header = 1;
 559                         continue;
 560                 }
 561                 if (!strcmp(arg, "--timestamp")) {
 562                         show_timestamp = 1;
 563                         continue;
 564                 }
 565                 if (!strcmp(arg, "--bisect")) {
 566                         bisect_list = 1;
 567                         continue;
 568                 }
 569                 if (!strcmp(arg, "--bisect-all")) {
 570                         bisect_list = 1;
 571                         bisect_find_all = 1;
 572                         continue;
 573                 }
 574                 if (!strcmp(arg, "--bisect-vars")) {
 575                         bisect_list = 1;
 576                         bisect_show_vars = 1;
 577                         continue;
 578                 }
 579                 if (!strcmp(arg, "--stdin")) {
 580                         if (read_from_stdin++)
 581                                 die("--stdin given twice?");
 582                         read_revisions_from_stdin(&revs);
 583                         continue;
 584                 }
 585                 if (!strcmp(arg, "--quiet")) {
 586                         quiet = 1;
 587                         continue;
 588                 }
 589                 usage(rev_list_usage);
 590
 591         }
 592         if (revs.commit_format != CMIT_FMT_UNSPECIFIED) {
 593                 /* The command line has a --pretty  */
 594                 hdr_termination = '\n';
 595                 if (revs.commit_format == CMIT_FMT_ONELINE)
 596                         header_prefix = "";
 597                 else
 598                         header_prefix = "commit ";
 599         }
 600         else if (revs.verbose_header)
 601                 /* Only --header was specified */
 602                 revs.commit_format = CMIT_FMT_RAW;
 603
 604         list = revs.commits;
 605
 606         if ((!list &&
 607              (!(revs.tag_objects||revs.tree_objects||revs.blob_objects) &&
 608               !revs.pending.nr)) ||
 609             revs.diff)
 610                 usage(rev_list_usage);
 611
 612         save_commit_buffer = revs.verbose_header || revs.grep_filter;
 613         if (bisect_list)
 614                 revs.limited = 1;
 615
 616         if (prepare_revision_walk(&revs))
 617                 die("revision walk setup failed");
 618         if (revs.tree_objects)
 619                 mark_edges_uninteresting(revs.commits, &revs, show_edge);
 620
 621         if (bisect_list) {
 622                 int reaches = reaches, all = all;
 623
 624                 revs.commits = find_bisection(revs.commits, &reaches, &all,
 625                                               bisect_find_all);
 626                 if (bisect_show_vars) {
 627                         int cnt;
 628                         char hex[41];
 629                         if (!revs.commits)
 630                                 return 1;
 631                         /*
 632                          * revs.commits can reach "reaches" commits among
 633                          * "all" commits.  If it is good, then there are
 634                          * (all-reaches) commits left to be bisected.
 635                          * On the other hand, if it is bad, then the set
 636                          * to bisect is "reaches".
 637                          * A bisect set of size N has (N-1) commits further
 638                          * to test, as we already know one bad one.
 639                          */
 640                         cnt = all - reaches;
 641                         if (cnt < reaches)
 642                                 cnt = reaches;
 643                         strcpy(hex, sha1_to_hex(revs.commits->item->object.sha1));
 644
 645                         if (bisect_find_all) {
 646                                 traverse_commit_list(&revs, show_commit, show_object);
 647                                 printf("------\n");
 648                         }
 649
 650                         printf("bisect_rev=%s\n"
 651                                "bisect_nr=%d\n"
 652                                "bisect_good=%d\n"
 653                                "bisect_bad=%d\n"
 654                                "bisect_all=%d\n",
 655                                hex,
 656                                cnt - 1,
 657                                all - reaches - 1,
 658                                reaches - 1,
 659                                all);
 660                         return 0;
 661                 }
 662         }
 663
 664         traverse_commit_list(&revs,
 665                 quiet ? finish_commit : show_commit,
 666                 quiet ? finish_object : show_object);
 667
 668         return 0;
 669 }