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