bundle.c

   1 #define USE_THE_REPOSITORY_VARIABLE
   2
   3 #include "git-compat-util.h"
   4 #include "lockfile.h"
   5 #include "bundle.h"
   6 #include "environment.h"
   7 #include "gettext.h"
   8 #include "hex.h"
   9 #include "object-store-ll.h"
  10 #include "repository.h"
  11 #include "object.h"
  12 #include "commit.h"
  13 #include "diff.h"
  14 #include "revision.h"
  15 #include "list-objects.h"
  16 #include "run-command.h"
  17 #include "refs.h"
  18 #include "strvec.h"
  19 #include "list-objects-filter-options.h"
  20 #include "connected.h"
  21 #include "write-or-die.h"
  22
  23 static const char v2_bundle_signature[] = "# v2 git bundle\n";
  24 static const char v3_bundle_signature[] = "# v3 git bundle\n";
  25 static struct {
  26         int version;
  27         const char *signature;
  28 } bundle_sigs[] = {
  29         { 2, v2_bundle_signature },
  30         { 3, v3_bundle_signature },
  31 };
  32
  33 void bundle_header_init(struct bundle_header *header)
  34 {
  35         struct bundle_header blank = BUNDLE_HEADER_INIT;
  36         memcpy(header, &blank, sizeof(*header));
  37 }
  38
  39 void bundle_header_release(struct bundle_header *header)
  40 {
  41         string_list_clear(&header->prerequisites, 1);
  42         string_list_clear(&header->references, 1);
  43         list_objects_filter_release(&header->filter);
  44 }
  45
  46 static int parse_capability(struct bundle_header *header, const char *capability)
  47 {
  48         const char *arg;
  49         if (skip_prefix(capability, "object-format=", &arg)) {
  50                 int algo = hash_algo_by_name(arg);
  51                 if (algo == GIT_HASH_UNKNOWN)
  52                         return error(_("unrecognized bundle hash algorithm: %s"), arg);
  53                 header->hash_algo = &hash_algos[algo];
  54                 return 0;
  55         }
  56         if (skip_prefix(capability, "filter=", &arg)) {
  57                 parse_list_objects_filter(&header->filter, arg);
  58                 return 0;
  59         }
  60         return error(_("unknown capability '%s'"), capability);
  61 }
  62
  63 static int parse_bundle_signature(struct bundle_header *header, const char *line)
  64 {
  65         int i;
  66
  67         for (i = 0; i < ARRAY_SIZE(bundle_sigs); i++) {
  68                 if (!strcmp(line, bundle_sigs[i].signature)) {
  69                         header->version = bundle_sigs[i].version;
  70                         return 0;
  71                 }
  72         }
  73         return -1;
  74 }
  75
  76 int read_bundle_header_fd(int fd, struct bundle_header *header,
  77                           const char *report_path)
  78 {
  79         struct strbuf buf = STRBUF_INIT;
  80         int status = 0;
  81
  82         /* The bundle header begins with the signature */
  83         if (strbuf_getwholeline_fd(&buf, fd, '\n') ||
  84             parse_bundle_signature(header, buf.buf)) {
  85                 if (report_path)
  86                         error(_("'%s' does not look like a v2 or v3 bundle file"),
  87                               report_path);
  88                 status = -1;
  89                 goto abort;
  90         }
  91
  92         header->hash_algo = the_hash_algo;
  93
  94         /* The bundle header ends with an empty line */
  95         while (!strbuf_getwholeline_fd(&buf, fd, '\n') &&
  96                buf.len && buf.buf[0] != '\n') {
  97                 struct object_id oid;
  98                 int is_prereq = 0;
  99                 const char *p;
 100
 101                 strbuf_rtrim(&buf);
 102
 103                 if (header->version == 3 && *buf.buf == '@') {
 104                         if (parse_capability(header, buf.buf + 1)) {
 105                                 status = -1;
 106                                 break;
 107                         }
 108                         continue;
 109                 }
 110
 111                 if (*buf.buf == '-') {
 112                         is_prereq = 1;
 113                         strbuf_remove(&buf, 0, 1);
 114                 }
 115
 116                 /*
 117                  * Tip lines have object name, SP, and refname.
 118                  * Prerequisites have object name that is optionally
 119                  * followed by SP and subject line.
 120                  */
 121                 if (parse_oid_hex_algop(buf.buf, &oid, &p, header->hash_algo) ||
 122                     (*p && !isspace(*p)) ||
 123                     (!is_prereq && !*p)) {
 124                         if (report_path)
 125                                 error(_("unrecognized header: %s%s (%d)"),
 126                                       (is_prereq ? "-" : ""), buf.buf, (int)buf.len);
 127                         status = -1;
 128                         break;
 129                 } else {
 130                         struct object_id *dup = oiddup(&oid);
 131                         if (is_prereq)
 132                                 string_list_append(&header->prerequisites, "")->util = dup;
 133                         else
 134                                 string_list_append(&header->references, p + 1)->util = dup;
 135                 }
 136         }
 137
 138  abort:
 139         if (status) {
 140                 close(fd);
 141                 fd = -1;
 142         }
 143         strbuf_release(&buf);
 144         return fd;
 145 }
 146
 147 int read_bundle_header(const char *path, struct bundle_header *header)
 148 {
 149         int fd = open(path, O_RDONLY);
 150
 151         if (fd < 0)
 152                 return error(_("could not open '%s'"), path);
 153         return read_bundle_header_fd(fd, header, path);
 154 }
 155
 156 int is_bundle(const char *path, int quiet)
 157 {
 158         struct bundle_header header = BUNDLE_HEADER_INIT;
 159         int fd = open(path, O_RDONLY);
 160
 161         if (fd < 0)
 162                 return 0;
 163         fd = read_bundle_header_fd(fd, &header, quiet ? NULL : path);
 164         if (fd >= 0)
 165                 close(fd);
 166         bundle_header_release(&header);
 167         return (fd >= 0);
 168 }
 169
 170 static int list_refs(struct string_list *r, int argc, const char **argv)
 171 {
 172         int i;
 173
 174         for (i = 0; i < r->nr; i++) {
 175                 struct object_id *oid;
 176                 const char *name;
 177
 178                 if (argc > 1) {
 179                         int j;
 180                         for (j = 1; j < argc; j++)
 181                                 if (!strcmp(r->items[i].string, argv[j]))
 182                                         break;
 183                         if (j == argc)
 184                                 continue;
 185                 }
 186
 187                 oid = r->items[i].util;
 188                 name = r->items[i].string;
 189                 printf("%s %s\n", oid_to_hex(oid), name);
 190         }
 191         return 0;
 192 }
 193
 194 /* Remember to update object flag allocation in object.h */
 195 #define PREREQ_MARK (1u<<16)
 196
 197 struct string_list_iterator {
 198         struct string_list *list;
 199         size_t cur;
 200 };
 201
 202 static const struct object_id *iterate_ref_map(void *cb_data)
 203 {
 204         struct string_list_iterator *iter = cb_data;
 205
 206         if (iter->cur >= iter->list->nr)
 207                 return NULL;
 208
 209         return iter->list->items[iter->cur++].util;
 210 }
 211
 212 int verify_bundle(struct repository *r,
 213                   struct bundle_header *header,
 214                   enum verify_bundle_flags flags)
 215 {
 216         /*
 217          * Do fast check, then if any prereqs are missing then go line by line
 218          * to be verbose about the errors
 219          */
 220         struct string_list *p = &header->prerequisites;
 221         int i, ret = 0;
 222         const char *message = _("Repository lacks these prerequisite commits:");
 223         struct string_list_iterator iter = {
 224                 .list = p,
 225         };
 226         struct check_connected_options opts = {
 227                 .quiet = 1,
 228         };
 229
 230         if (!r || !r->objects || !r->objects->odb)
 231                 return error(_("need a repository to verify a bundle"));
 232
 233         for (i = 0; i < p->nr; i++) {
 234                 struct string_list_item *e = p->items + i;
 235                 const char *name = e->string;
 236                 struct object_id *oid = e->util;
 237                 struct object *o = parse_object(r, oid);
 238                 if (o)
 239                         continue;
 240                 ret++;
 241                 if (flags & VERIFY_BUNDLE_QUIET)
 242                         continue;
 243                 if (ret == 1)
 244                         error("%s", message);
 245                 error("%s %s", oid_to_hex(oid), name);
 246         }
 247         if (ret)
 248                 goto cleanup;
 249
 250         if ((ret = check_connected(iterate_ref_map, &iter, &opts)))
 251                 error(_("some prerequisite commits exist in the object store, "
 252                         "but are not connected to the repository's history"));
 253
 254         /* TODO: preserve this verbose language. */
 255         if (flags & VERIFY_BUNDLE_VERBOSE) {
 256                 struct string_list *r;
 257
 258                 r = &header->references;
 259                 printf_ln(Q_("The bundle contains this ref:",
 260                              "The bundle contains these %"PRIuMAX" refs:",
 261                              r->nr),
 262                           (uintmax_t)r->nr);
 263                 list_refs(r, 0, NULL);
 264
 265                 r = &header->prerequisites;
 266                 if (!r->nr) {
 267                         printf_ln(_("The bundle records a complete history."));
 268                 } else {
 269                         printf_ln(Q_("The bundle requires this ref:",
 270                                      "The bundle requires these %"PRIuMAX" refs:",
 271                                      r->nr),
 272                                   (uintmax_t)r->nr);
 273                         list_refs(r, 0, NULL);
 274                 }
 275
 276                 printf_ln(_("The bundle uses this hash algorithm: %s"),
 277                           header->hash_algo->name);
 278                 if (header->filter.choice)
 279                         printf_ln(_("The bundle uses this filter: %s"),
 280                                   list_objects_filter_spec(&header->filter));
 281         }
 282 cleanup:
 283         return ret;
 284 }
 285
 286 int list_bundle_refs(struct bundle_header *header, int argc, const char **argv)
 287 {
 288         return list_refs(&header->references, argc, argv);
 289 }
 290
 291 static int is_tag_in_date_range(struct object *tag, struct rev_info *revs)
 292 {
 293         unsigned long size;
 294         enum object_type type;
 295         char *buf = NULL, *line, *lineend;
 296         timestamp_t date;
 297         int result = 1;
 298
 299         if (revs->max_age == -1 && revs->min_age == -1)
 300                 goto out;
 301
 302         buf = repo_read_object_file(the_repository, &tag->oid, &type, &size);
 303         if (!buf)
 304                 goto out;
 305         line = memmem(buf, size, "\ntagger ", 8);
 306         if (!line++)
 307                 goto out;
 308         lineend = memchr(line, '\n', buf + size - line);
 309         line = memchr(line, '>', lineend ? lineend - line : buf + size - line);
 310         if (!line++)
 311                 goto out;
 312         date = parse_timestamp(line, NULL, 10);
 313         result = (revs->max_age == -1 || revs->max_age < date) &&
 314                 (revs->min_age == -1 || revs->min_age > date);
 315 out:
 316         free(buf);
 317         return result;
 318 }
 319
 320
 321 /* Write the pack data to bundle_fd */
 322 static int write_pack_data(int bundle_fd, struct rev_info *revs, struct strvec *pack_options)
 323 {
 324         struct child_process pack_objects = CHILD_PROCESS_INIT;
 325         int i;
 326
 327         strvec_pushl(&pack_objects.args,
 328                      "pack-objects",
 329                      "--stdout", "--thin", "--delta-base-offset",
 330                      NULL);
 331         strvec_pushv(&pack_objects.args, pack_options->v);
 332         if (revs->filter.choice)
 333                 strvec_pushf(&pack_objects.args, "--filter=%s",
 334                              list_objects_filter_spec(&revs->filter));
 335         pack_objects.in = -1;
 336         pack_objects.out = bundle_fd;
 337         pack_objects.git_cmd = 1;
 338
 339         /*
 340          * start_command() will close our descriptor if it's >1. Duplicate it
 341          * to avoid surprising the caller.
 342          */
 343         if (pack_objects.out > 1) {
 344                 pack_objects.out = dup(pack_objects.out);
 345                 if (pack_objects.out < 0) {
 346                         error_errno(_("unable to dup bundle descriptor"));
 347                         child_process_clear(&pack_objects);
 348                         return -1;
 349                 }
 350         }
 351
 352         if (start_command(&pack_objects))
 353                 return error(_("Could not spawn pack-objects"));
 354
 355         for (i = 0; i < revs->pending.nr; i++) {
 356                 struct object *object = revs->pending.objects[i].item;
 357                 if (object->flags & UNINTERESTING)
 358                         write_or_die(pack_objects.in, "^", 1);
 359                 write_or_die(pack_objects.in, oid_to_hex(&object->oid), the_hash_algo->hexsz);
 360                 write_or_die(pack_objects.in, "\n", 1);
 361         }
 362         close(pack_objects.in);
 363         if (finish_command(&pack_objects))
 364                 return error(_("pack-objects died"));
 365         return 0;
 366 }
 367
 368 /*
 369  * Write out bundle refs based on the tips already
 370  * parsed into revs.pending. As a side effect, may
 371  * manipulate revs.pending to include additional
 372  * necessary objects (like tags).
 373  *
 374  * Returns the number of refs written, or negative
 375  * on error.
 376  */
 377 static int write_bundle_refs(int bundle_fd, struct rev_info *revs)
 378 {
 379         int i;
 380         int ref_count = 0;
 381
 382         for (i = 0; i < revs->pending.nr; i++) {
 383                 struct object_array_entry *e = revs->pending.objects + i;
 384                 struct object_id oid;
 385                 char *ref;
 386                 const char *display_ref;
 387                 int flag;
 388
 389                 if (e->item->flags & UNINTERESTING)
 390                         continue;
 391                 if (repo_dwim_ref(the_repository, e->name, strlen(e->name),
 392                                   &oid, &ref, 0) != 1)
 393                         goto skip_write_ref;
 394                 if (refs_read_ref_full(get_main_ref_store(the_repository), e->name, RESOLVE_REF_READING, &oid, &flag))
 395                         flag = 0;
 396                 display_ref = (flag & REF_ISSYMREF) ? e->name : ref;
 397
 398                 if (e->item->type == OBJ_TAG &&
 399                                 !is_tag_in_date_range(e->item, revs)) {
 400                         e->item->flags |= UNINTERESTING;
 401                         goto skip_write_ref;
 402                 }
 403
 404                 /*
 405                  * Make sure the refs we wrote out is correct; --max-count and
 406                  * other limiting options could have prevented all the tips
 407                  * from getting output.
 408                  *
 409                  * Non commit objects such as tags and blobs do not have
 410                  * this issue as they are not affected by those extra
 411                  * constraints.
 412                  */
 413                 if (!(e->item->flags & SHOWN) && e->item->type == OBJ_COMMIT) {
 414                         warning(_("ref '%s' is excluded by the rev-list options"),
 415                                 e->name);
 416                         goto skip_write_ref;
 417                 }
 418                 /*
 419                  * If you run "git bundle create bndl v1.0..v2.0", the
 420                  * name of the positive ref is "v2.0" but that is the
 421                  * commit that is referenced by the tag, and not the tag
 422                  * itself.
 423                  */
 424                 if (!oideq(&oid, &e->item->oid)) {
 425                         /*
 426                          * Is this the positive end of a range expressed
 427                          * in terms of a tag (e.g. v2.0 from the range
 428                          * "v1.0..v2.0")?
 429                          */
 430                         struct commit *one = lookup_commit_reference(revs->repo, &oid);
 431                         struct object *obj;
 432
 433                         if (e->item == &(one->object)) {
 434                                 /*
 435                                  * Need to include e->name as an
 436                                  * independent ref to the pack-objects
 437                                  * input, so that the tag is included
 438                                  * in the output; otherwise we would
 439                                  * end up triggering "empty bundle"
 440                                  * error.
 441                                  */
 442                                 obj = parse_object_or_die(&oid, e->name);
 443                                 obj->flags |= SHOWN;
 444                                 add_pending_object(revs, obj, e->name);
 445                         }
 446                         goto skip_write_ref;
 447                 }
 448
 449                 ref_count++;
 450                 write_or_die(bundle_fd, oid_to_hex(&e->item->oid), the_hash_algo->hexsz);
 451                 write_or_die(bundle_fd, " ", 1);
 452                 write_or_die(bundle_fd, display_ref, strlen(display_ref));
 453                 write_or_die(bundle_fd, "\n", 1);
 454  skip_write_ref:
 455                 free(ref);
 456         }
 457
 458         /* end header */
 459         write_or_die(bundle_fd, "\n", 1);
 460         return ref_count;
 461 }
 462
 463 struct bundle_prerequisites_info {
 464         struct object_array *pending;
 465         int fd;
 466 };
 467
 468 static void write_bundle_prerequisites(struct commit *commit, void *data)
 469 {
 470         struct bundle_prerequisites_info *bpi = data;
 471         struct object *object;
 472         struct pretty_print_context ctx = { 0 };
 473         struct strbuf buf = STRBUF_INIT;
 474
 475         if (!(commit->object.flags & BOUNDARY))
 476                 return;
 477         strbuf_addf(&buf, "-%s ", oid_to_hex(&commit->object.oid));
 478         write_or_die(bpi->fd, buf.buf, buf.len);
 479
 480         ctx.fmt = CMIT_FMT_ONELINE;
 481         ctx.output_encoding = get_log_output_encoding();
 482         strbuf_reset(&buf);
 483         pretty_print_commit(&ctx, commit, &buf);
 484         strbuf_trim(&buf);
 485
 486         object = (struct object *)commit;
 487         object->flags |= UNINTERESTING;
 488         add_object_array_with_path(object, buf.buf, bpi->pending, S_IFINVALID,
 489                                    NULL);
 490         strbuf_addch(&buf, '\n');
 491         write_or_die(bpi->fd, buf.buf, buf.len);
 492         strbuf_release(&buf);
 493 }
 494
 495 int create_bundle(struct repository *r, const char *path,
 496                   int argc, const char **argv, struct strvec *pack_options, int version)
 497 {
 498         struct lock_file lock = LOCK_INIT;
 499         int bundle_fd = -1;
 500         int bundle_to_stdout;
 501         int ref_count = 0;
 502         struct rev_info revs, revs_copy;
 503         int min_version = 2;
 504         struct bundle_prerequisites_info bpi;
 505         int ret;
 506         int i;
 507
 508         /* init revs to list objects for pack-objects later */
 509         save_commit_buffer = 0;
 510         repo_init_revisions(r, &revs, NULL);
 511
 512         /*
 513          * Pre-initialize the '--objects' flag so we can parse a
 514          * --filter option successfully.
 515          */
 516         revs.tree_objects = revs.blob_objects = 1;
 517
 518         argc = setup_revisions(argc, argv, &revs, NULL);
 519
 520         /*
 521          * Reasons to require version 3:
 522          *
 523          * 1. @object-format is required because our hash algorithm is not
 524          *    SHA1.
 525          * 2. @filter is required because we parsed an object filter.
 526          */
 527         if (the_hash_algo != &hash_algos[GIT_HASH_SHA1] || revs.filter.choice)
 528                 min_version = 3;
 529
 530         if (argc > 1) {
 531                 ret = error(_("unrecognized argument: %s"), argv[1]);
 532                 goto out;
 533         }
 534
 535         bundle_to_stdout = !strcmp(path, "-");
 536         if (bundle_to_stdout)
 537                 bundle_fd = 1;
 538         else
 539                 bundle_fd = hold_lock_file_for_update(&lock, path,
 540                                                       LOCK_DIE_ON_ERROR);
 541
 542         if (version == -1)
 543                 version = min_version;
 544
 545         if (version < 2 || version > 3) {
 546                 die(_("unsupported bundle version %d"), version);
 547         } else if (version < min_version) {
 548                 die(_("cannot write bundle version %d with algorithm %s"), version, the_hash_algo->name);
 549         } else if (version == 2) {
 550                 write_or_die(bundle_fd, v2_bundle_signature, strlen(v2_bundle_signature));
 551         } else {
 552                 const char *capability = "@object-format=";
 553                 write_or_die(bundle_fd, v3_bundle_signature, strlen(v3_bundle_signature));
 554                 write_or_die(bundle_fd, capability, strlen(capability));
 555                 write_or_die(bundle_fd, the_hash_algo->name, strlen(the_hash_algo->name));
 556                 write_or_die(bundle_fd, "\n", 1);
 557
 558                 if (revs.filter.choice) {
 559                         const char *value = expand_list_objects_filter_spec(&revs.filter);
 560                         capability = "@filter=";
 561                         write_or_die(bundle_fd, capability, strlen(capability));
 562                         write_or_die(bundle_fd, value, strlen(value));
 563                         write_or_die(bundle_fd, "\n", 1);
 564                 }
 565         }
 566
 567         /* save revs.pending in revs_copy for later use */
 568         memcpy(&revs_copy, &revs, sizeof(revs));
 569         revs_copy.pending.nr = 0;
 570         revs_copy.pending.alloc = 0;
 571         revs_copy.pending.objects = NULL;
 572         for (i = 0; i < revs.pending.nr; i++) {
 573                 struct object_array_entry *e = revs.pending.objects + i;
 574                 if (e)
 575                         add_object_array_with_path(e->item, e->name,
 576                                                    &revs_copy.pending,
 577                                                    e->mode, e->path);
 578         }
 579
 580         /* write prerequisites */
 581         revs.boundary = 1;
 582         if (prepare_revision_walk(&revs))
 583                 die("revision walk setup failed");
 584         bpi.fd = bundle_fd;
 585         bpi.pending = &revs_copy.pending;
 586
 587         /*
 588          * Remove any object walking here. We only care about commits and
 589          * tags here. The revs_copy has the right instances of these values.
 590          */
 591         revs.blob_objects = revs.tree_objects = 0;
 592         traverse_commit_list(&revs, write_bundle_prerequisites, NULL, &bpi);
 593         object_array_remove_duplicates(&revs_copy.pending);
 594
 595         /* write bundle refs */
 596         ref_count = write_bundle_refs(bundle_fd, &revs_copy);
 597         if (!ref_count) {
 598                 die(_("Refusing to create empty bundle."));
 599         } else if (ref_count < 0) {
 600                 ret = -1;
 601                 goto out;
 602         }
 603
 604         /* write pack */
 605         if (write_pack_data(bundle_fd, &revs_copy, pack_options)) {
 606                 ret = -1;
 607                 goto out;
 608         }
 609
 610         if (!bundle_to_stdout) {
 611                 if (commit_lock_file(&lock))
 612                         die_errno(_("cannot create '%s'"), path);
 613         }
 614
 615         ret = 0;
 616
 617 out:
 618         object_array_clear(&revs_copy.pending);
 619         release_revisions(&revs);
 620         rollback_lock_file(&lock);
 621         return ret;
 622 }
 623
 624 int unbundle(struct repository *r, struct bundle_header *header,
 625              int bundle_fd, struct strvec *extra_index_pack_args,
 626              enum verify_bundle_flags flags)
 627 {
 628         struct child_process ip = CHILD_PROCESS_INIT;
 629
 630         if (verify_bundle(r, header, flags))
 631                 return -1;
 632
 633         strvec_pushl(&ip.args, "index-pack", "--fix-thin", "--stdin", NULL);
 634
 635         /* If there is a filter, then we need to create the promisor pack. */
 636         if (header->filter.choice)
 637                 strvec_push(&ip.args, "--promisor=from-bundle");
 638
 639         if (flags & VERIFY_BUNDLE_FSCK)
 640                 strvec_push(&ip.args, "--fsck-objects");
 641
 642         if (extra_index_pack_args) {
 643                 strvec_pushv(&ip.args, extra_index_pack_args->v);
 644                 strvec_clear(extra_index_pack_args);
 645         }
 646
 647         ip.in = bundle_fd;
 648         ip.no_stdout = 1;
 649         ip.git_cmd = 1;
 650         if (run_command(&ip))
 651                 return error(_("index-pack died"));
 652         return 0;
 653 }