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