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