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