bundle.c

   1 #include "cache.h"
   2 #include "lockfile.h"
   3 #include "bundle.h"
   4 #include "object-store.h"
   5 #include "repository.h"
   6 #include "object.h"
   7 #include "commit.h"
   8 #include "diff.h"
   9 #include "revision.h"
  10 #include "list-objects.h"
  11 #include "run-command.h"
  12 #include "refs.h"
  13 #include "strvec.h"
  14 #include "list-objects-filter-options.h"
  15
  16 static const char v2_bundle_signature[] = "# v2 git bundle\n";
  17 static const char v3_bundle_signature[] = "# v3 git bundle\n";
  18 static struct {
  19         int version;
  20         const char *signature;
  21 } bundle_sigs[] = {
  22         { 2, v2_bundle_signature },
  23         { 3, v3_bundle_signature },
  24 };
  25
  26 void bundle_header_init(struct bundle_header *header)
  27 {
  28         struct bundle_header blank = BUNDLE_HEADER_INIT;
  29         memcpy(header, &blank, sizeof(*header));
  30 }
  31
  32 void bundle_header_release(struct bundle_header *header)
  33 {
  34         string_list_clear(&header->prerequisites, 1);
  35         string_list_clear(&header->references, 1);
  36         list_objects_filter_release(&header->filter);
  37 }
  38
  39 static int parse_capability(struct bundle_header *header, const char *capability)
  40 {
  41         const char *arg;
  42         if (skip_prefix(capability, "object-format=", &arg)) {
  43                 int algo = hash_algo_by_name(arg);
  44                 if (algo == GIT_HASH_UNKNOWN)
  45                         return error(_("unrecognized bundle hash algorithm: %s"), arg);
  46                 header->hash_algo = &hash_algos[algo];
  47                 return 0;
  48         }
  49         if (skip_prefix(capability, "filter=", &arg)) {
  50                 parse_list_objects_filter(&header->filter, arg);
  51                 return 0;
  52         }
  53         return error(_("unknown capability '%s'"), capability);
  54 }
  55
  56 static int parse_bundle_signature(struct bundle_header *header, const char *line)
  57 {
  58         int i;
  59
  60         for (i = 0; i < ARRAY_SIZE(bundle_sigs); i++) {
  61                 if (!strcmp(line, bundle_sigs[i].signature)) {
  62                         header->version = bundle_sigs[i].version;
  63                         return 0;
  64                 }
  65         }
  66         return -1;
  67 }
  68
  69 int read_bundle_header_fd(int fd, struct bundle_header *header,
  70                           const char *report_path)
  71 {
  72         struct strbuf buf = STRBUF_INIT;
  73         int status = 0;
  74
  75         /* The bundle header begins with the signature */
  76         if (strbuf_getwholeline_fd(&buf, fd, '\n') ||
  77             parse_bundle_signature(header, buf.buf)) {
  78                 if (report_path)
  79                         error(_("'%s' does not look like a v2 or v3 bundle file"),
  80                               report_path);
  81                 status = -1;
  82                 goto abort;
  83         }
  84
  85         header->hash_algo = the_hash_algo;
  86
  87         /* The bundle header ends with an empty line */
  88         while (!strbuf_getwholeline_fd(&buf, fd, '\n') &&
  89                buf.len && buf.buf[0] != '\n') {
  90                 struct object_id oid;
  91                 int is_prereq = 0;
  92                 const char *p;
  93
  94                 strbuf_rtrim(&buf);
  95
  96                 if (header->version == 3 && *buf.buf == '@') {
  97                         if (parse_capability(header, buf.buf + 1)) {
  98                                 status = -1;
  99                                 break;
 100                         }
 101                         continue;
 102                 }
 103
 104                 if (*buf.buf == '-') {
 105                         is_prereq = 1;
 106                         strbuf_remove(&buf, 0, 1);
 107                 }
 108
 109                 /*
 110                  * Tip lines have object name, SP, and refname.
 111                  * Prerequisites have object name that is optionally
 112                  * followed by SP and subject line.
 113                  */
 114                 if (parse_oid_hex_algop(buf.buf, &oid, &p, header->hash_algo) ||
 115                     (*p && !isspace(*p)) ||
 116                     (!is_prereq && !*p)) {
 117                         if (report_path)
 118                                 error(_("unrecognized header: %s%s (%d)"),
 119                                       (is_prereq ? "-" : ""), buf.buf, (int)buf.len);
 120                         status = -1;
 121                         break;
 122                 } else {
 123                         struct object_id *dup = oiddup(&oid);
 124                         if (is_prereq)
 125                                 string_list_append(&header->prerequisites, "")->util = dup;
 126                         else
 127                                 string_list_append(&header->references, p + 1)->util = dup;
 128                 }
 129         }
 130
 131  abort:
 132         if (status) {
 133                 close(fd);
 134                 fd = -1;
 135         }
 136         strbuf_release(&buf);
 137         return fd;
 138 }
 139
 140 int read_bundle_header(const char *path, struct bundle_header *header)
 141 {
 142         int fd = open(path, O_RDONLY);
 143
 144         if (fd < 0)
 145                 return error(_("could not open '%s'"), path);
 146         return read_bundle_header_fd(fd, header, path);
 147 }
 148
 149 int is_bundle(const char *path, int quiet)
 150 {
 151         struct bundle_header header = BUNDLE_HEADER_INIT;
 152         int fd = open(path, O_RDONLY);
 153
 154         if (fd < 0)
 155                 return 0;
 156         fd = read_bundle_header_fd(fd, &header, quiet ? NULL : path);
 157         if (fd >= 0)
 158                 close(fd);
 159         bundle_header_release(&header);
 160         return (fd >= 0);
 161 }
 162
 163 static int list_refs(struct string_list *r, int argc, const char **argv)
 164 {
 165         int i;
 166
 167         for (i = 0; i < r->nr; i++) {
 168                 struct object_id *oid;
 169                 const char *name;
 170
 171                 if (argc > 1) {
 172                         int j;
 173                         for (j = 1; j < argc; j++)
 174                                 if (!strcmp(r->items[i].string, argv[j]))
 175                                         break;
 176                         if (j == argc)
 177                                 continue;
 178                 }
 179
 180                 oid = r->items[i].util;
 181                 name = r->items[i].string;
 182                 printf("%s %s\n", oid_to_hex(oid), name);
 183         }
 184         return 0;
 185 }
 186
 187 /* Remember to update object flag allocation in object.h */
 188 #define PREREQ_MARK (1u<<16)
 189
 190 int verify_bundle(struct repository *r,
 191                   struct bundle_header *header,
 192                   int verbose)
 193 {
 194         /*
 195          * Do fast check, then if any prereqs are missing then go line by line
 196          * to be verbose about the errors
 197          */
 198         struct string_list *p = &header->prerequisites;
 199         struct rev_info revs = REV_INFO_INIT;
 200         const char *argv[] = {NULL, "--all", NULL};
 201         struct commit *commit;
 202         int i, ret = 0, req_nr;
 203         const char *message = _("Repository lacks these prerequisite commits:");
 204
 205         if (!r || !r->objects || !r->objects->odb) {
 206                 ret = error(_("need a repository to verify a bundle"));
 207                 goto cleanup;
 208         }
 209
 210         repo_init_revisions(r, &revs, NULL);
 211         for (i = 0; i < p->nr; i++) {
 212                 struct string_list_item *e = p->items + i;
 213                 const char *name = e->string;
 214                 struct object_id *oid = e->util;
 215                 struct object *o = parse_object(r, oid);
 216                 if (o) {
 217                         o->flags |= PREREQ_MARK;
 218                         add_pending_object(&revs, o, name);
 219                         continue;
 220                 }
 221                 if (++ret == 1)
 222                         error("%s", message);
 223                 error("%s %s", oid_to_hex(oid), name);
 224         }
 225         if (revs.pending.nr != p->nr)
 226                 goto cleanup;
 227         req_nr = revs.pending.nr;
 228         setup_revisions(2, argv, &revs, NULL);
 229
 230         list_objects_filter_copy(&revs.filter, &header->filter);
 231
 232         if (prepare_revision_walk(&revs))
 233                 die(_("revision walk setup failed"));
 234
 235         i = req_nr;
 236         while (i && (commit = get_revision(&revs)))
 237                 if (commit->object.flags & PREREQ_MARK)
 238                         i--;
 239
 240         for (i = 0; i < p->nr; i++) {
 241                 struct string_list_item *e = p->items + i;
 242                 const char *name = e->string;
 243                 const struct object_id *oid = e->util;
 244                 struct object *o = parse_object(r, oid);
 245                 assert(o); /* otherwise we'd have returned early */
 246                 if (o->flags & SHOWN)
 247                         continue;
 248                 if (++ret == 1)
 249                         error("%s", message);
 250                 error("%s %s", oid_to_hex(oid), name);
 251         }
 252
 253         /* Clean up objects used, as they will be reused. */
 254         for (i = 0; i < p->nr; i++) {
 255                 struct string_list_item *e = p->items + i;
 256                 struct object_id *oid = e->util;
 257                 commit = lookup_commit_reference_gently(r, oid, 1);
 258                 if (commit)
 259                         clear_commit_marks(commit, ALL_REV_FLAGS);
 260         }
 261
 262         if (verbose) {
 263                 struct string_list *r;
 264
 265                 r = &header->references;
 266                 printf_ln(Q_("The bundle contains this ref:",
 267                              "The bundle contains these %"PRIuMAX" refs:",
 268                              r->nr),
 269                           (uintmax_t)r->nr);
 270                 list_refs(r, 0, NULL);
 271
 272                 r = &header->prerequisites;
 273                 if (!r->nr) {
 274                         printf_ln(_("The bundle records a complete history."));
 275                 } else {
 276                         printf_ln(Q_("The bundle requires this ref:",
 277                                      "The bundle requires these %"PRIuMAX" refs:",
 278                                      r->nr),
 279                                   (uintmax_t)r->nr);
 280                         list_refs(r, 0, NULL);
 281                 }
 282
 283                 printf_ln("The bundle uses this hash algorithm: %s",
 284                           header->hash_algo->name);
 285                 if (header->filter.choice)
 286                         printf_ln("The bundle uses this filter: %s",
 287                                   list_objects_filter_spec(&header->filter));
 288         }
 289 cleanup:
 290         release_revisions(&revs);
 291         return ret;
 292 }
 293
 294 int list_bundle_refs(struct bundle_header *header, int argc, const char **argv)
 295 {
 296         return list_refs(&header->references, argc, argv);
 297 }
 298
 299 static int is_tag_in_date_range(struct object *tag, struct rev_info *revs)
 300 {
 301         unsigned long size;
 302         enum object_type type;
 303         char *buf = NULL, *line, *lineend;
 304         timestamp_t date;
 305         int result = 1;
 306
 307         if (revs->max_age == -1 && revs->min_age == -1)
 308                 goto out;
 309
 310         buf = read_object_file(&tag->oid, &type, &size);
 311         if (!buf)
 312                 goto out;
 313         line = memmem(buf, size, "\ntagger ", 8);
 314         if (!line++)
 315                 goto out;
 316         lineend = memchr(line, '\n', buf + size - line);
 317         line = memchr(line, '>', lineend ? lineend - line : buf + size - line);
 318         if (!line++)
 319                 goto out;
 320         date = parse_timestamp(line, NULL, 10);
 321         result = (revs->max_age == -1 || revs->max_age < date) &&
 322                 (revs->min_age == -1 || revs->min_age > date);
 323 out:
 324         free(buf);
 325         return result;
 326 }
 327
 328
 329 /* Write the pack data to bundle_fd */
 330 static int write_pack_data(int bundle_fd, struct rev_info *revs, struct strvec *pack_options)
 331 {
 332         struct child_process pack_objects = CHILD_PROCESS_INIT;
 333         int i;
 334
 335         strvec_pushl(&pack_objects.args,
 336                      "pack-objects",
 337                      "--stdout", "--thin", "--delta-base-offset",
 338                      NULL);
 339         strvec_pushv(&pack_objects.args, pack_options->v);
 340         if (revs->filter.choice)
 341                 strvec_pushf(&pack_objects.args, "--filter=%s",
 342                              list_objects_filter_spec(&revs->filter));
 343         pack_objects.in = -1;
 344         pack_objects.out = bundle_fd;
 345         pack_objects.git_cmd = 1;
 346
 347         /*
 348          * start_command() will close our descriptor if it's >1. Duplicate it
 349          * to avoid surprising the caller.
 350          */
 351         if (pack_objects.out > 1) {
 352                 pack_objects.out = dup(pack_objects.out);
 353                 if (pack_objects.out < 0) {
 354                         error_errno(_("unable to dup bundle descriptor"));
 355                         child_process_clear(&pack_objects);
 356                         return -1;
 357                 }
 358         }
 359
 360         if (start_command(&pack_objects))
 361                 return error(_("Could not spawn pack-objects"));
 362
 363         for (i = 0; i < revs->pending.nr; i++) {
 364                 struct object *object = revs->pending.objects[i].item;
 365                 if (object->flags & UNINTERESTING)
 366                         write_or_die(pack_objects.in, "^", 1);
 367                 write_or_die(pack_objects.in, oid_to_hex(&object->oid), the_hash_algo->hexsz);
 368                 write_or_die(pack_objects.in, "\n", 1);
 369         }
 370         close(pack_objects.in);
 371         if (finish_command(&pack_objects))
 372                 return error(_("pack-objects died"));
 373         return 0;
 374 }
 375
 376 /*
 377  * Write out bundle refs based on the tips already
 378  * parsed into revs.pending. As a side effect, may
 379  * manipulate revs.pending to include additional
 380  * necessary objects (like tags).
 381  *
 382  * Returns the number of refs written, or negative
 383  * on error.
 384  */
 385 static int write_bundle_refs(int bundle_fd, struct rev_info *revs)
 386 {
 387         int i;
 388         int ref_count = 0;
 389
 390         for (i = 0; i < revs->pending.nr; i++) {
 391                 struct object_array_entry *e = revs->pending.objects + i;
 392                 struct object_id oid;
 393                 char *ref;
 394                 const char *display_ref;
 395                 int flag;
 396
 397                 if (e->item->flags & UNINTERESTING)
 398                         continue;
 399                 if (dwim_ref(e->name, strlen(e->name), &oid, &ref, 0) != 1)
 400                         goto skip_write_ref;
 401                 if (read_ref_full(e->name, RESOLVE_REF_READING, &oid, &flag))
 402                         flag = 0;
 403                 display_ref = (flag & REF_ISSYMREF) ? e->name : ref;
 404
 405                 if (e->item->type == OBJ_TAG &&
 406                                 !is_tag_in_date_range(e->item, revs)) {
 407                         e->item->flags |= UNINTERESTING;
 408                         goto skip_write_ref;
 409                 }
 410
 411                 /*
 412                  * Make sure the refs we wrote out is correct; --max-count and
 413                  * other limiting options could have prevented all the tips
 414                  * from getting output.
 415                  *
 416                  * Non commit objects such as tags and blobs do not have
 417                  * this issue as they are not affected by those extra
 418                  * constraints.
 419                  */
 420                 if (!(e->item->flags & SHOWN) && e->item->type == OBJ_COMMIT) {
 421                         warning(_("ref '%s' is excluded by the rev-list options"),
 422                                 e->name);
 423                         goto skip_write_ref;
 424                 }
 425                 /*
 426                  * If you run "git bundle create bndl v1.0..v2.0", the
 427                  * name of the positive ref is "v2.0" but that is the
 428                  * commit that is referenced by the tag, and not the tag
 429                  * itself.
 430                  */
 431                 if (!oideq(&oid, &e->item->oid)) {
 432                         /*
 433                          * Is this the positive end of a range expressed
 434                          * in terms of a tag (e.g. v2.0 from the range
 435                          * "v1.0..v2.0")?
 436                          */
 437                         struct commit *one = lookup_commit_reference(revs->repo, &oid);
 438                         struct object *obj;
 439
 440                         if (e->item == &(one->object)) {
 441                                 /*
 442                                  * Need to include e->name as an
 443                                  * independent ref to the pack-objects
 444                                  * input, so that the tag is included
 445                                  * in the output; otherwise we would
 446                                  * end up triggering "empty bundle"
 447                                  * error.
 448                                  */
 449                                 obj = parse_object_or_die(&oid, e->name);
 450                                 obj->flags |= SHOWN;
 451                                 add_pending_object(revs, obj, e->name);
 452                         }
 453                         goto skip_write_ref;
 454                 }
 455
 456                 ref_count++;
 457                 write_or_die(bundle_fd, oid_to_hex(&e->item->oid), the_hash_algo->hexsz);
 458                 write_or_die(bundle_fd, " ", 1);
 459                 write_or_die(bundle_fd, display_ref, strlen(display_ref));
 460                 write_or_die(bundle_fd, "\n", 1);
 461  skip_write_ref:
 462                 free(ref);
 463         }
 464
 465         /* end header */
 466         write_or_die(bundle_fd, "\n", 1);
 467         return ref_count;
 468 }
 469
 470 struct bundle_prerequisites_info {
 471         struct object_array *pending;
 472         int fd;
 473 };
 474
 475 static void write_bundle_prerequisites(struct commit *commit, void *data)
 476 {
 477         struct bundle_prerequisites_info *bpi = data;
 478         struct object *object;
 479         struct pretty_print_context ctx = { 0 };
 480         struct strbuf buf = STRBUF_INIT;
 481
 482         if (!(commit->object.flags & BOUNDARY))
 483                 return;
 484         strbuf_addf(&buf, "-%s ", oid_to_hex(&commit->object.oid));
 485         write_or_die(bpi->fd, buf.buf, buf.len);
 486
 487         ctx.fmt = CMIT_FMT_ONELINE;
 488         ctx.output_encoding = get_log_output_encoding();
 489         strbuf_reset(&buf);
 490         pretty_print_commit(&ctx, commit, &buf);
 491         strbuf_trim(&buf);
 492
 493         object = (struct object *)commit;
 494         object->flags |= UNINTERESTING;
 495         add_object_array_with_path(object, buf.buf, bpi->pending, S_IFINVALID,
 496                                    NULL);
 497         strbuf_addch(&buf, '\n');
 498         write_or_die(bpi->fd, buf.buf, buf.len);
 499         strbuf_release(&buf);
 500 }
 501
 502 int create_bundle(struct repository *r, const char *path,
 503                   int argc, const char **argv, struct strvec *pack_options, int version)
 504 {
 505         struct lock_file lock = LOCK_INIT;
 506         int bundle_fd = -1;
 507         int bundle_to_stdout;
 508         int ref_count = 0;
 509         struct rev_info revs, revs_copy;
 510         int min_version = 2;
 511         struct bundle_prerequisites_info bpi;
 512         int i;
 513
 514         /* init revs to list objects for pack-objects later */
 515         save_commit_buffer = 0;
 516         repo_init_revisions(r, &revs, NULL);
 517
 518         /*
 519          * Pre-initialize the '--objects' flag so we can parse a
 520          * --filter option successfully.
 521          */
 522         revs.tree_objects = revs.blob_objects = 1;
 523
 524         argc = setup_revisions(argc, argv, &revs, NULL);
 525
 526         /*
 527          * Reasons to require version 3:
 528          *
 529          * 1. @object-format is required because our hash algorithm is not
 530          *    SHA1.
 531          * 2. @filter is required because we parsed an object filter.
 532          */
 533         if (the_hash_algo != &hash_algos[GIT_HASH_SHA1] || revs.filter.choice)
 534                 min_version = 3;
 535
 536         if (argc > 1) {
 537                 error(_("unrecognized argument: %s"), argv[1]);
 538                 goto err;
 539         }
 540
 541         bundle_to_stdout = !strcmp(path, "-");
 542         if (bundle_to_stdout)
 543                 bundle_fd = 1;
 544         else
 545                 bundle_fd = hold_lock_file_for_update(&lock, path,
 546                                                       LOCK_DIE_ON_ERROR);
 547
 548         if (version == -1)
 549                 version = min_version;
 550
 551         if (version < 2 || version > 3) {
 552                 die(_("unsupported bundle version %d"), version);
 553         } else if (version < min_version) {
 554                 die(_("cannot write bundle version %d with algorithm %s"), version, the_hash_algo->name);
 555         } else if (version == 2) {
 556                 write_or_die(bundle_fd, v2_bundle_signature, strlen(v2_bundle_signature));
 557         } else {
 558                 const char *capability = "@object-format=";
 559                 write_or_die(bundle_fd, v3_bundle_signature, strlen(v3_bundle_signature));
 560                 write_or_die(bundle_fd, capability, strlen(capability));
 561                 write_or_die(bundle_fd, the_hash_algo->name, strlen(the_hash_algo->name));
 562                 write_or_die(bundle_fd, "\n", 1);
 563
 564                 if (revs.filter.choice) {
 565                         const char *value = expand_list_objects_filter_spec(&revs.filter);
 566                         capability = "@filter=";
 567                         write_or_die(bundle_fd, capability, strlen(capability));
 568                         write_or_die(bundle_fd, value, strlen(value));
 569                         write_or_die(bundle_fd, "\n", 1);
 570                 }
 571         }
 572
 573         /* save revs.pending in revs_copy for later use */
 574         memcpy(&revs_copy, &revs, sizeof(revs));
 575         revs_copy.pending.nr = 0;
 576         revs_copy.pending.alloc = 0;
 577         revs_copy.pending.objects = NULL;
 578         for (i = 0; i < revs.pending.nr; i++) {
 579                 struct object_array_entry *e = revs.pending.objects + i;
 580                 if (e)
 581                         add_object_array_with_path(e->item, e->name,
 582                                                    &revs_copy.pending,
 583                                                    e->mode, e->path);
 584         }
 585
 586         /* write prerequisites */
 587         revs.boundary = 1;
 588         if (prepare_revision_walk(&revs))
 589                 die("revision walk setup failed");
 590         bpi.fd = bundle_fd;
 591         bpi.pending = &revs_copy.pending;
 592
 593         /*
 594          * Remove any object walking here. We only care about commits and
 595          * tags here. The revs_copy has the right instances of these values.
 596          */
 597         revs.blob_objects = revs.tree_objects = 0;
 598         traverse_commit_list(&revs, write_bundle_prerequisites, NULL, &bpi);
 599         object_array_remove_duplicates(&revs_copy.pending);
 600
 601         /* write bundle refs */
 602         ref_count = write_bundle_refs(bundle_fd, &revs_copy);
 603         if (!ref_count)
 604                 die(_("Refusing to create empty bundle."));
 605         else if (ref_count < 0)
 606                 goto err;
 607
 608         /* write pack */
 609         if (write_pack_data(bundle_fd, &revs_copy, pack_options))
 610                 goto err;
 611
 612         if (!bundle_to_stdout) {
 613                 if (commit_lock_file(&lock))
 614                         die_errno(_("cannot create '%s'"), path);
 615         }
 616         return 0;
 617 err:
 618         rollback_lock_file(&lock);
 619         return -1;
 620 }
 621
 622 int unbundle(struct repository *r, struct bundle_header *header,
 623              int bundle_fd, struct strvec *extra_index_pack_args)
 624 {
 625         struct child_process ip = CHILD_PROCESS_INIT;
 626         strvec_pushl(&ip.args, "index-pack", "--fix-thin", "--stdin", NULL);
 627
 628         /* If there is a filter, then we need to create the promisor pack. */
 629         if (header->filter.choice)
 630                 strvec_push(&ip.args, "--promisor=from-bundle");
 631
 632         if (extra_index_pack_args) {
 633                 strvec_pushv(&ip.args, extra_index_pack_args->v);
 634                 strvec_clear(extra_index_pack_args);
 635         }
 636
 637         if (verify_bundle(r, header, 0))
 638                 return -1;
 639         ip.in = bundle_fd;
 640         ip.no_stdout = 1;
 641         ip.git_cmd = 1;
 642         if (run_command(&ip))
 643                 return error(_("index-pack died"));
 644         return 0;
 645 }