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                 printf("%s%c", buf, hdr_termination);
  89                 free(buf);
  90         }
  91         maybe_flush_or_die(stdout, "stdout");
  92         if (commit->parents) {
  93                 free_commit_list(commit->parents);
  94                 commit->parents = NULL;
  95         }
  96         free(commit->buffer);
  97         commit->buffer = NULL;
  98 }
  99
 100 static void show_object(struct object_array_entry *p)
 101 {
 102         /* An object with name "foo\n0000000..." can be used to
 103          * confuse downstream git-pack-objects very badly.
 104          */
 105         const char *ep = strchr(p->name, '\n');
 106
 107         if (p->item->type == OBJ_BLOB && !has_sha1_file(p->item->sha1))
 108                 die("missing blob object '%s'", sha1_to_hex(p->item->sha1));
 109
 110         if (ep) {
 111                 printf("%s %.*s\n", sha1_to_hex(p->item->sha1),
 112                        (int) (ep - p->name),
 113                        p->name);
 114         }
 115         else
 116                 printf("%s %s\n", sha1_to_hex(p->item->sha1), p->name);
 117 }
 118
 119 static void show_edge(struct commit *commit)
 120 {
 121         printf("-%s\n", sha1_to_hex(commit->object.sha1));
 122 }
 123
 124 /*
 125  * This is a truly stupid algorithm, but it's only
 126  * used for bisection, and we just don't care enough.
 127  *
 128  * We care just barely enough to avoid recursing for
 129  * non-merge entries.
 130  */
 131 static int count_distance(struct commit_list *entry)
 132 {
 133         int nr = 0;
 134
 135         while (entry) {
 136                 struct commit *commit = entry->item;
 137                 struct commit_list *p;
 138
 139                 if (commit->object.flags & (UNINTERESTING | COUNTED))
 140                         break;
 141                 if (!revs.prune_fn || (commit->object.flags & TREECHANGE))
 142                         nr++;
 143                 commit->object.flags |= COUNTED;
 144                 p = commit->parents;
 145                 entry = p;
 146                 if (p) {
 147                         p = p->next;
 148                         while (p) {
 149                                 nr += count_distance(p);
 150                                 p = p->next;
 151                         }
 152                 }
 153         }
 154
 155         return nr;
 156 }
 157
 158 static void clear_distance(struct commit_list *list)
 159 {
 160         while (list) {
 161                 struct commit *commit = list->item;
 162                 commit->object.flags &= ~COUNTED;
 163                 list = list->next;
 164         }
 165 }
 166
 167 #define DEBUG_BISECT 0
 168
 169 static inline int weight(struct commit_list *elem)
 170 {
 171         return *((int*)(elem->item->util));
 172 }
 173
 174 static inline void weight_set(struct commit_list *elem, int weight)
 175 {
 176         *((int*)(elem->item->util)) = weight;
 177 }
 178
 179 static int count_interesting_parents(struct commit *commit)
 180 {
 181         struct commit_list *p;
 182         int count;
 183
 184         for (count = 0, p = commit->parents; p; p = p->next) {
 185                 if (p->item->object.flags & UNINTERESTING)
 186                         continue;
 187                 count++;
 188         }
 189         return count;
 190 }
 191
 192 static inline int halfway(struct commit_list *p, int distance, int nr)
 193 {
 194         /*
 195          * Don't short-cut something we are not going to return!
 196          */
 197         if (revs.prune_fn && !(p->item->object.flags & TREECHANGE))
 198                 return 0;
 199         if (DEBUG_BISECT)
 200                 return 0;
 201         /*
 202          * 2 and 3 are halfway of 5.
 203          * 3 is halfway of 6 but 2 and 4 are not.
 204          */
 205         distance *= 2;
 206         switch (distance - 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, distance;
 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                 n = weight(p);
 350                 if (n != -2)
 351                         continue;
 352                 distance = count_distance(p);
 353                 clear_distance(list);
 354                 weight_set(p, distance);
 355
 356                 /* Does it happen to be at exactly half-way? */
 357                 if (halfway(p, distance, nr))
 358                         return p;
 359                 counted++;
 360         }
 361
 362         show_list("bisection 2 count_distance", counted, nr, list);
 363
 364         while (counted < nr) {
 365                 for (p = list; p; p = p->next) {
 366                         struct commit_list *q;
 367                         unsigned flags = p->item->object.flags;
 368
 369                         if (0 <= weight(p))
 370                                 continue;
 371                         for (q = p->item->parents; q; q = q->next) {
 372                                 if (q->item->object.flags & UNINTERESTING)
 373                                         continue;
 374                                 if (0 <= weight(q))
 375                                         break;
 376                         }
 377                         if (!q)
 378                                 continue;
 379
 380                         /*
 381                          * weight for p is unknown but q is known.
 382                          * add one for p itself if p is to be counted,
 383                          * otherwise inherit it from q directly.
 384                          */
 385                         if (!revs.prune_fn || (flags & TREECHANGE)) {
 386                                 weight_set(p, weight(q)+1);
 387                                 counted++;
 388                                 show_list("bisection 2 count one",
 389                                           counted, nr, list);
 390                         }
 391                         else
 392                                 weight_set(p, weight(q));
 393
 394                         /* Does it happen to be at exactly half-way? */
 395                         distance = weight(p);
 396                         if (halfway(p, distance, nr))
 397                                 return p;
 398                 }
 399         }
 400
 401         show_list("bisection 2 counted all", counted, nr, list);
 402
 403         /* Then find the best one */
 404         return best_bisection(list, nr);
 405 }
 406
 407 static struct commit_list *find_bisection(struct commit_list *list,
 408                                           int *reaches, int *all)
 409 {
 410         int nr, on_list;
 411         struct commit_list *p, *best, *next, *last;
 412         int *weights;
 413
 414         show_list("bisection 2 entry", 0, 0, list);
 415
 416         /*
 417          * Count the number of total and tree-changing items on the
 418          * list, while reversing the list.
 419          */
 420         for (nr = on_list = 0, last = NULL, p = list;
 421              p;
 422              p = next) {
 423                 unsigned flags = p->item->object.flags;
 424
 425                 next = p->next;
 426                 if (flags & UNINTERESTING)
 427                         continue;
 428                 p->next = last;
 429                 last = p;
 430                 if (!revs.prune_fn || (flags & TREECHANGE))
 431                         nr++;
 432                 on_list++;
 433         }
 434         list = last;
 435         show_list("bisection 2 sorted", 0, nr, list);
 436
 437         *all = nr;
 438         weights = xcalloc(on_list, sizeof(*weights));
 439
 440         /* Do the real work of finding bisection commit. */
 441         best = do_find_bisection(list, nr, weights);
 442
 443         if (best)
 444                 best->next = NULL;
 445
 446         *reaches = weight(best);
 447         free(weights);
 448
 449         return best;
 450 }
 451
 452 static void read_revisions_from_stdin(struct rev_info *revs)
 453 {
 454         char line[1000];
 455
 456         while (fgets(line, sizeof(line), stdin) != NULL) {
 457                 int len = strlen(line);
 458                 if (line[len - 1] == '\n')
 459                         line[--len] = 0;
 460                 if (!len)
 461                         break;
 462                 if (line[0] == '-')
 463                         die("options not supported in --stdin mode");
 464                 if (handle_revision_arg(line, revs, 0, 1))
 465                         die("bad revision '%s'", line);
 466         }
 467 }
 468
 469 int cmd_rev_list(int argc, const char **argv, const char *prefix)
 470 {
 471         struct commit_list *list;
 472         int i;
 473         int read_from_stdin = 0;
 474         int bisect_show_vars = 0;
 475
 476         git_config(git_default_config);
 477         init_revisions(&revs, prefix);
 478         revs.abbrev = 0;
 479         revs.commit_format = CMIT_FMT_UNSPECIFIED;
 480         argc = setup_revisions(argc, argv, &revs, NULL);
 481
 482         for (i = 1 ; i < argc; i++) {
 483                 const char *arg = argv[i];
 484
 485                 if (!strcmp(arg, "--header")) {
 486                         revs.verbose_header = 1;
 487                         continue;
 488                 }
 489                 if (!strcmp(arg, "--timestamp")) {
 490                         show_timestamp = 1;
 491                         continue;
 492                 }
 493                 if (!strcmp(arg, "--bisect")) {
 494                         bisect_list = 1;
 495                         continue;
 496                 }
 497                 if (!strcmp(arg, "--bisect-vars")) {
 498                         bisect_list = 1;
 499                         bisect_show_vars = 1;
 500                         continue;
 501                 }
 502                 if (!strcmp(arg, "--stdin")) {
 503                         if (read_from_stdin++)
 504                                 die("--stdin given twice?");
 505                         read_revisions_from_stdin(&revs);
 506                         continue;
 507                 }
 508                 usage(rev_list_usage);
 509
 510         }
 511         if (revs.commit_format != CMIT_FMT_UNSPECIFIED) {
 512                 /* The command line has a --pretty  */
 513                 hdr_termination = '\n';
 514                 if (revs.commit_format == CMIT_FMT_ONELINE)
 515                         header_prefix = "";
 516                 else
 517                         header_prefix = "commit ";
 518         }
 519         else if (revs.verbose_header)
 520                 /* Only --header was specified */
 521                 revs.commit_format = CMIT_FMT_RAW;
 522
 523         list = revs.commits;
 524
 525         if ((!list &&
 526              (!(revs.tag_objects||revs.tree_objects||revs.blob_objects) &&
 527               !revs.pending.nr)) ||
 528             revs.diff)
 529                 usage(rev_list_usage);
 530
 531         save_commit_buffer = revs.verbose_header || revs.grep_filter;
 532         track_object_refs = 0;
 533         if (bisect_list)
 534                 revs.limited = 1;
 535
 536         prepare_revision_walk(&revs);
 537         if (revs.tree_objects)
 538                 mark_edges_uninteresting(revs.commits, &revs, show_edge);
 539
 540         if (bisect_list) {
 541                 int reaches = reaches, all = all;
 542
 543                 revs.commits = find_bisection(revs.commits, &reaches, &all);
 544                 if (bisect_show_vars) {
 545                         int cnt;
 546                         if (!revs.commits)
 547                                 return 1;
 548                         /*
 549                          * revs.commits can reach "reaches" commits among
 550                          * "all" commits.  If it is good, then there are
 551                          * (all-reaches) commits left to be bisected.
 552                          * On the other hand, if it is bad, then the set
 553                          * to bisect is "reaches".
 554                          * A bisect set of size N has (N-1) commits further
 555                          * to test, as we already know one bad one.
 556                          */
 557                         cnt = all-reaches;
 558                         if (cnt < reaches)
 559                                 cnt = reaches;
 560                         printf("bisect_rev=%s\n"
 561                                "bisect_nr=%d\n"
 562                                "bisect_good=%d\n"
 563                                "bisect_bad=%d\n"
 564                                "bisect_all=%d\n",
 565                                sha1_to_hex(revs.commits->item->object.sha1),
 566                                cnt - 1,
 567                                all - reaches - 1,
 568                                reaches - 1,
 569                                all);
 570                         return 0;
 571                 }
 572         }
 573
 574         traverse_commit_list(&revs, show_commit, show_object);
 575
 576         return 0;
 577 }