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 "    --objects | --objects-edge\n"
  41 "    --unpacked\n"
  42 "    --header | --pretty\n"
  43 "    --abbrev=nr | --no-abbrev\n"
  44 "    --abbrev-commit\n"
  45 "    --left-right\n"
  46 "  special purpose:\n"
  47 "    --bisect\n"
  48 "    --bisect-vars\n"
  49 "    --bisect-all"
  50 ;
  51
  52 static struct rev_info revs;
  53
  54 static int bisect_list;
  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         if (commit->object.flags & BOUNDARY)
  69                 putchar('-');
  70         else if (commit->object.flags & UNINTERESTING)
  71                 putchar('^');
  72         else if (revs.left_right) {
  73                 if (commit->object.flags & SYMMETRIC_LEFT)
  74                         putchar('<');
  75                 else
  76                         putchar('>');
  77         }
  78         if (revs.abbrev_commit && revs.abbrev)
  79                 fputs(find_unique_abbrev(commit->object.sha1, revs.abbrev),
  80                       stdout);
  81         else
  82                 fputs(sha1_to_hex(commit->object.sha1), stdout);
  83         if (revs.print_parents) {
  84                 struct commit_list *parents = commit->parents;
  85                 while (parents) {
  86                         printf(" %s", sha1_to_hex(parents->item->object.sha1));
  87                         parents = parents->next;
  88                 }
  89         }
  90         show_decorations(commit);
  91         if (revs.commit_format == CMIT_FMT_ONELINE)
  92                 putchar(' ');
  93         else
  94                 putchar('\n');
  95
  96         if (revs.verbose_header && commit->buffer) {
  97                 struct strbuf buf;
  98                 strbuf_init(&buf, 0);
  99                 pretty_print_commit(revs.commit_format, commit,
 100                                     &buf, revs.abbrev, NULL, NULL,
 101                                     revs.date_mode, 0);
 102                 if (revs.graph) {
 103                         if (buf.len) {
 104                                 if (revs.commit_format != CMIT_FMT_ONELINE)
 105                                         graph_show_oneline(revs.graph);
 106
 107                                 graph_show_commit_msg(revs.graph, &buf);
 108
 109                                 /*
 110                                  * Add a newline after the commit message.
 111                                  *
 112                                  * Usually, this newline produces a blank
 113                                  * padding line between entries, in which case
 114                                  * we need to add graph padding on this line.
 115                                  *
 116                                  * However, the commit message may not end in a
 117                                  * newline.  In this case the newline simply
 118                                  * ends the last line of the commit message,
 119                                  * and we don't need any graph output.  (This
 120                                  * always happens with CMIT_FMT_ONELINE, and it
 121                                  * happens with CMIT_FMT_USERFORMAT when the
 122                                  * format doesn't explicitly end in a newline.)
 123                                  */
 124                                 if (buf.len && buf.buf[buf.len - 1] == '\n')
 125                                         graph_show_padding(revs.graph);
 126                                 putchar('\n');
 127                         } else {
 128                                 /*
 129                                  * If the message buffer is empty, just show
 130                                  * the rest of the graph output for this
 131                                  * commit.
 132                                  */
 133                                 if (graph_show_remainder(revs.graph))
 134                                         putchar('\n');
 135                         }
 136                 } else {
 137                         if (buf.len)
 138                                 printf("%s%c", buf.buf, hdr_termination);
 139                 }
 140                 strbuf_release(&buf);
 141         } else {
 142                 if (graph_show_remainder(revs.graph))
 143                         putchar('\n');
 144         }
 145         maybe_flush_or_die(stdout, "stdout");
 146         finish_commit(commit);
 147 }
 148
 149 static void finish_commit(struct commit *commit)
 150 {
 151         if (commit->parents) {
 152                 free_commit_list(commit->parents);
 153                 commit->parents = NULL;
 154         }
 155         free(commit->buffer);
 156         commit->buffer = NULL;
 157 }
 158
 159 static void finish_object(struct object_array_entry *p)
 160 {
 161         if (p->item->type == OBJ_BLOB && !has_sha1_file(p->item->sha1))
 162                 die("missing blob object '%s'", sha1_to_hex(p->item->sha1));
 163 }
 164
 165 static void show_object(struct object_array_entry *p)
 166 {
 167         /* An object with name "foo\n0000000..." can be used to
 168          * confuse downstream git-pack-objects very badly.
 169          */
 170         const char *ep = strchr(p->name, '\n');
 171
 172         finish_object(p);
 173         if (ep) {
 174                 printf("%s %.*s\n", sha1_to_hex(p->item->sha1),
 175                        (int) (ep - p->name),
 176                        p->name);
 177         }
 178         else
 179                 printf("%s %s\n", sha1_to_hex(p->item->sha1), p->name);
 180 }
 181
 182 static void show_edge(struct commit *commit)
 183 {
 184         printf("-%s\n", sha1_to_hex(commit->object.sha1));
 185 }
 186
 187 /*
 188  * This is a truly stupid algorithm, but it's only
 189  * used for bisection, and we just don't care enough.
 190  *
 191  * We care just barely enough to avoid recursing for
 192  * non-merge entries.
 193  */
 194 static int count_distance(struct commit_list *entry)
 195 {
 196         int nr = 0;
 197
 198         while (entry) {
 199                 struct commit *commit = entry->item;
 200                 struct commit_list *p;
 201
 202                 if (commit->object.flags & (UNINTERESTING | COUNTED))
 203                         break;
 204                 if (!(commit->object.flags & TREESAME))
 205                         nr++;
 206                 commit->object.flags |= COUNTED;
 207                 p = commit->parents;
 208                 entry = p;
 209                 if (p) {
 210                         p = p->next;
 211                         while (p) {
 212                                 nr += count_distance(p);
 213                                 p = p->next;
 214                         }
 215                 }
 216         }
 217
 218         return nr;
 219 }
 220
 221 static void clear_distance(struct commit_list *list)
 222 {
 223         while (list) {
 224                 struct commit *commit = list->item;
 225                 commit->object.flags &= ~COUNTED;
 226                 list = list->next;
 227         }
 228 }
 229
 230 #define DEBUG_BISECT 0
 231
 232 static inline int weight(struct commit_list *elem)
 233 {
 234         return *((int*)(elem->item->util));
 235 }
 236
 237 static inline void weight_set(struct commit_list *elem, int weight)
 238 {
 239         *((int*)(elem->item->util)) = weight;
 240 }
 241
 242 static int count_interesting_parents(struct commit *commit)
 243 {
 244         struct commit_list *p;
 245         int count;
 246
 247         for (count = 0, p = commit->parents; p; p = p->next) {
 248                 if (p->item->object.flags & UNINTERESTING)
 249                         continue;
 250                 count++;
 251         }
 252         return count;
 253 }
 254
 255 static inline int halfway(struct commit_list *p, int nr)
 256 {
 257         /*
 258          * Don't short-cut something we are not going to return!
 259          */
 260         if (p->item->object.flags & TREESAME)
 261                 return 0;
 262         if (DEBUG_BISECT)
 263                 return 0;
 264         /*
 265          * 2 and 3 are halfway of 5.
 266          * 3 is halfway of 6 but 2 and 4 are not.
 267          */
 268         switch (2 * weight(p) - nr) {
 269         case -1: case 0: case 1:
 270                 return 1;
 271         default:
 272                 return 0;
 273         }
 274 }
 275
 276 #if !DEBUG_BISECT
 277 #define show_list(a,b,c,d) do { ; } while (0)
 278 #else
 279 static void show_list(const char *debug, int counted, int nr,
 280                       struct commit_list *list)
 281 {
 282         struct commit_list *p;
 283
 284         fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
 285
 286         for (p = list; p; p = p->next) {
 287                 struct commit_list *pp;
 288                 struct commit *commit = p->item;
 289                 unsigned flags = commit->object.flags;
 290                 enum object_type type;
 291                 unsigned long size;
 292                 char *buf = read_sha1_file(commit->object.sha1, &type, &size);
 293                 char *ep, *sp;
 294
 295                 fprintf(stderr, "%c%c%c ",
 296                         (flags & TREESAME) ? ' ' : 'T',
 297                         (flags & UNINTERESTING) ? 'U' : ' ',
 298                         (flags & COUNTED) ? 'C' : ' ');
 299                 if (commit->util)
 300                         fprintf(stderr, "%3d", weight(p));
 301                 else
 302                         fprintf(stderr, "---");
 303                 fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.sha1));
 304                 for (pp = commit->parents; pp; pp = pp->next)
 305                         fprintf(stderr, " %.*s", 8,
 306                                 sha1_to_hex(pp->item->object.sha1));
 307
 308                 sp = strstr(buf, "\n\n");
 309                 if (sp) {
 310                         sp += 2;
 311                         for (ep = sp; *ep && *ep != '\n'; ep++)
 312                                 ;
 313                         fprintf(stderr, " %.*s", (int)(ep - sp), sp);
 314                 }
 315                 fprintf(stderr, "\n");
 316         }
 317 }
 318 #endif /* DEBUG_BISECT */
 319
 320 static struct commit_list *best_bisection(struct commit_list *list, int nr)
 321 {
 322         struct commit_list *p, *best;
 323         int best_distance = -1;
 324
 325         best = list;
 326         for (p = list; p; p = p->next) {
 327                 int distance;
 328                 unsigned flags = p->item->object.flags;
 329
 330                 if (flags & TREESAME)
 331                         continue;
 332                 distance = weight(p);
 333                 if (nr - distance < distance)
 334                         distance = nr - distance;
 335                 if (distance > best_distance) {
 336                         best = p;
 337                         best_distance = distance;
 338                 }
 339         }
 340
 341         return best;
 342 }
 343
 344 struct commit_dist {
 345         struct commit *commit;
 346         int distance;
 347 };
 348
 349 static int compare_commit_dist(const void *a_, const void *b_)
 350 {
 351         struct commit_dist *a, *b;
 352
 353         a = (struct commit_dist *)a_;
 354         b = (struct commit_dist *)b_;
 355         if (a->distance != b->distance)
 356                 return b->distance - a->distance; /* desc sort */
 357         return hashcmp(a->commit->object.sha1, b->commit->object.sha1);
 358 }
 359
 360 static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
 361 {
 362         struct commit_list *p;
 363         struct commit_dist *array = xcalloc(nr, sizeof(*array));
 364         int cnt, i;
 365
 366         for (p = list, cnt = 0; p; p = p->next) {
 367                 int distance;
 368                 unsigned flags = p->item->object.flags;
 369
 370                 if (flags & TREESAME)
 371                         continue;
 372                 distance = weight(p);
 373                 if (nr - distance < distance)
 374                         distance = nr - distance;
 375                 array[cnt].commit = p->item;
 376                 array[cnt].distance = distance;
 377                 cnt++;
 378         }
 379         qsort(array, cnt, sizeof(*array), compare_commit_dist);
 380         for (p = list, i = 0; i < cnt; i++) {
 381                 struct name_decoration *r = xmalloc(sizeof(*r) + 100);
 382                 struct object *obj = &(array[i].commit->object);
 383
 384                 sprintf(r->name, "dist=%d", array[i].distance);
 385                 r->next = add_decoration(&name_decoration, obj, r);
 386                 p->item = array[i].commit;
 387                 p = p->next;
 388         }
 389         if (p)
 390                 p->next = NULL;
 391         free(array);
 392         return list;
 393 }
 394
 395 /*
 396  * zero or positive weight is the number of interesting commits it can
 397  * reach, including itself.  Especially, weight = 0 means it does not
 398  * reach any tree-changing commits (e.g. just above uninteresting one
 399  * but traversal is with pathspec).
 400  *
 401  * weight = -1 means it has one parent and its distance is yet to
 402  * be computed.
 403  *
 404  * weight = -2 means it has more than one parent and its distance is
 405  * unknown.  After running count_distance() first, they will get zero
 406  * or positive distance.
 407  */
 408 static struct commit_list *do_find_bisection(struct commit_list *list,
 409                                              int nr, int *weights,
 410                                              int find_all)
 411 {
 412         int n, counted;
 413         struct commit_list *p;
 414
 415         counted = 0;
 416
 417         for (n = 0, p = list; p; p = p->next) {
 418                 struct commit *commit = p->item;
 419                 unsigned flags = commit->object.flags;
 420
 421                 p->item->util = &weights[n++];
 422                 switch (count_interesting_parents(commit)) {
 423                 case 0:
 424                         if (!(flags & TREESAME)) {
 425                                 weight_set(p, 1);
 426                                 counted++;
 427                                 show_list("bisection 2 count one",
 428                                           counted, nr, list);
 429                         }
 430                         /*
 431                          * otherwise, it is known not to reach any
 432                          * tree-changing commit and gets weight 0.
 433                          */
 434                         break;
 435                 case 1:
 436                         weight_set(p, -1);
 437                         break;
 438                 default:
 439                         weight_set(p, -2);
 440                         break;
 441                 }
 442         }
 443
 444         show_list("bisection 2 initialize", counted, nr, list);
 445
 446         /*
 447          * If you have only one parent in the resulting set
 448          * then you can reach one commit more than that parent
 449          * can reach.  So we do not have to run the expensive
 450          * count_distance() for single strand of pearls.
 451          *
 452          * However, if you have more than one parents, you cannot
 453          * just add their distance and one for yourself, since
 454          * they usually reach the same ancestor and you would
 455          * end up counting them twice that way.
 456          *
 457          * So we will first count distance of merges the usual
 458          * way, and then fill the blanks using cheaper algorithm.
 459          */
 460         for (p = list; p; p = p->next) {
 461                 if (p->item->object.flags & UNINTERESTING)
 462                         continue;
 463                 if (weight(p) != -2)
 464                         continue;
 465                 weight_set(p, count_distance(p));
 466                 clear_distance(list);
 467
 468                 /* Does it happen to be at exactly half-way? */
 469                 if (!find_all && halfway(p, nr))
 470                         return p;
 471                 counted++;
 472         }
 473
 474         show_list("bisection 2 count_distance", counted, nr, list);
 475
 476         while (counted < nr) {
 477                 for (p = list; p; p = p->next) {
 478                         struct commit_list *q;
 479                         unsigned flags = p->item->object.flags;
 480
 481                         if (0 <= weight(p))
 482                                 continue;
 483                         for (q = p->item->parents; q; q = q->next) {
 484                                 if (q->item->object.flags & UNINTERESTING)
 485                                         continue;
 486                                 if (0 <= weight(q))
 487                                         break;
 488                         }
 489                         if (!q)
 490                                 continue;
 491
 492                         /*
 493                          * weight for p is unknown but q is known.
 494                          * add one for p itself if p is to be counted,
 495                          * otherwise inherit it from q directly.
 496                          */
 497                         if (!(flags & TREESAME)) {
 498                                 weight_set(p, weight(q)+1);
 499                                 counted++;
 500                                 show_list("bisection 2 count one",
 501                                           counted, nr, list);
 502                         }
 503                         else
 504                                 weight_set(p, weight(q));
 505
 506                         /* Does it happen to be at exactly half-way? */
 507                         if (!find_all && halfway(p, nr))
 508                                 return p;
 509                 }
 510         }
 511
 512         show_list("bisection 2 counted all", counted, nr, list);
 513
 514         if (!find_all)
 515                 return best_bisection(list, nr);
 516         else
 517                 return best_bisection_sorted(list, nr);
 518 }
 519
 520 static struct commit_list *find_bisection(struct commit_list *list,
 521                                           int *reaches, int *all,
 522                                           int find_all)
 523 {
 524         int nr, on_list;
 525         struct commit_list *p, *best, *next, *last;
 526         int *weights;
 527
 528         show_list("bisection 2 entry", 0, 0, list);
 529
 530         /*
 531          * Count the number of total and tree-changing items on the
 532          * list, while reversing the list.
 533          */
 534         for (nr = on_list = 0, last = NULL, p = list;
 535              p;
 536              p = next) {
 537                 unsigned flags = p->item->object.flags;
 538
 539                 next = p->next;
 540                 if (flags & UNINTERESTING)
 541                         continue;
 542                 p->next = last;
 543                 last = p;
 544                 if (!(flags & TREESAME))
 545                         nr++;
 546                 on_list++;
 547         }
 548         list = last;
 549         show_list("bisection 2 sorted", 0, nr, list);
 550
 551         *all = nr;
 552         weights = xcalloc(on_list, sizeof(*weights));
 553
 554         /* Do the real work of finding bisection commit. */
 555         best = do_find_bisection(list, nr, weights, find_all);
 556         if (best) {
 557                 if (!find_all)
 558                         best->next = NULL;
 559                 *reaches = weight(best);
 560         }
 561         free(weights);
 562         return best;
 563 }
 564
 565 static void read_revisions_from_stdin(struct rev_info *revs)
 566 {
 567         char line[1000];
 568
 569         while (fgets(line, sizeof(line), stdin) != NULL) {
 570                 int len = strlen(line);
 571                 if (len && line[len - 1] == '\n')
 572                         line[--len] = 0;
 573                 if (!len)
 574                         break;
 575                 if (line[0] == '-')
 576                         die("options not supported in --stdin mode");
 577                 if (handle_revision_arg(line, revs, 0, 1))
 578                         die("bad revision '%s'", line);
 579         }
 580 }
 581
 582 int cmd_rev_list(int argc, const char **argv, const char *prefix)
 583 {
 584         struct commit_list *list;
 585         int i;
 586         int read_from_stdin = 0;
 587         int bisect_show_vars = 0;
 588         int bisect_find_all = 0;
 589         int quiet = 0;
 590
 591         git_config(git_default_config);
 592         init_revisions(&revs, prefix);
 593         revs.abbrev = 0;
 594         revs.commit_format = CMIT_FMT_UNSPECIFIED;
 595         argc = setup_revisions(argc, argv, &revs, NULL);
 596
 597         for (i = 1 ; i < argc; i++) {
 598                 const char *arg = argv[i];
 599
 600                 if (!strcmp(arg, "--header")) {
 601                         revs.verbose_header = 1;
 602                         continue;
 603                 }
 604                 if (!strcmp(arg, "--timestamp")) {
 605                         show_timestamp = 1;
 606                         continue;
 607                 }
 608                 if (!strcmp(arg, "--bisect")) {
 609                         bisect_list = 1;
 610                         continue;
 611                 }
 612                 if (!strcmp(arg, "--bisect-all")) {
 613                         bisect_list = 1;
 614                         bisect_find_all = 1;
 615                         continue;
 616                 }
 617                 if (!strcmp(arg, "--bisect-vars")) {
 618                         bisect_list = 1;
 619                         bisect_show_vars = 1;
 620                         continue;
 621                 }
 622                 if (!strcmp(arg, "--stdin")) {
 623                         if (read_from_stdin++)
 624                                 die("--stdin given twice?");
 625                         read_revisions_from_stdin(&revs);
 626                         continue;
 627                 }
 628                 if (!strcmp(arg, "--quiet")) {
 629                         quiet = 1;
 630                         continue;
 631                 }
 632                 usage(rev_list_usage);
 633
 634         }
 635         if (revs.commit_format != CMIT_FMT_UNSPECIFIED) {
 636                 /* The command line has a --pretty  */
 637                 hdr_termination = '\n';
 638                 if (revs.commit_format == CMIT_FMT_ONELINE)
 639                         header_prefix = "";
 640                 else
 641                         header_prefix = "commit ";
 642         }
 643         else if (revs.verbose_header)
 644                 /* Only --header was specified */
 645                 revs.commit_format = CMIT_FMT_RAW;
 646
 647         list = revs.commits;
 648
 649         if ((!list &&
 650              (!(revs.tag_objects||revs.tree_objects||revs.blob_objects) &&
 651               !revs.pending.nr)) ||
 652             revs.diff)
 653                 usage(rev_list_usage);
 654
 655         save_commit_buffer = revs.verbose_header || revs.grep_filter;
 656         if (bisect_list)
 657                 revs.limited = 1;
 658
 659         if (prepare_revision_walk(&revs))
 660                 die("revision walk setup failed");
 661         if (revs.tree_objects)
 662                 mark_edges_uninteresting(revs.commits, &revs, show_edge);
 663
 664         if (bisect_list) {
 665                 int reaches = reaches, all = all;
 666
 667                 revs.commits = find_bisection(revs.commits, &reaches, &all,
 668                                               bisect_find_all);
 669                 if (bisect_show_vars) {
 670                         int cnt;
 671                         char hex[41];
 672                         if (!revs.commits)
 673                                 return 1;
 674                         /*
 675                          * revs.commits can reach "reaches" commits among
 676                          * "all" commits.  If it is good, then there are
 677                          * (all-reaches) commits left to be bisected.
 678                          * On the other hand, if it is bad, then the set
 679                          * to bisect is "reaches".
 680                          * A bisect set of size N has (N-1) commits further
 681                          * to test, as we already know one bad one.
 682                          */
 683                         cnt = all - reaches;
 684                         if (cnt < reaches)
 685                                 cnt = reaches;
 686                         strcpy(hex, sha1_to_hex(revs.commits->item->object.sha1));
 687
 688                         if (bisect_find_all) {
 689                                 traverse_commit_list(&revs, show_commit, show_object);
 690                                 printf("------\n");
 691                         }
 692
 693                         printf("bisect_rev=%s\n"
 694                                "bisect_nr=%d\n"
 695                                "bisect_good=%d\n"
 696                                "bisect_bad=%d\n"
 697                                "bisect_all=%d\n",
 698                                hex,
 699                                cnt - 1,
 700                                all - reaches - 1,
 701                                reaches - 1,
 702                                all);
 703                         return 0;
 704                 }
 705         }
 706
 707         traverse_commit_list(&revs,
 708                 quiet ? finish_commit : show_commit,
 709                 quiet ? finish_object : show_object);
 710
 711         return 0;
 712 }