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 static int parse_bundle_header(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 parse_bundle_header(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 = parse_bundle_header(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;
 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                 return error(_("need a repository to verify a bundle"));
 207
 208         repo_init_revisions(r, &revs, NULL);
 209         for (i = 0; i < p->nr; i++) {
 210                 struct string_list_item *e = p->items + i;
 211                 const char *name = e->string;
 212                 struct object_id *oid = e->util;
 213                 struct object *o = parse_object(r, oid);
 214                 if (o) {
 215                         o->flags |= PREREQ_MARK;
 216                         add_pending_object(&revs, o, name);
 217                         continue;
 218                 }
 219                 if (++ret == 1)
 220                         error("%s", message);
 221                 error("%s %s", oid_to_hex(oid), name);
 222         }
 223         if (revs.pending.nr != p->nr)
 224                 return ret;
 225         req_nr = revs.pending.nr;
 226         setup_revisions(2, argv, &revs, NULL);
 227
 228         list_objects_filter_copy(&revs.filter, &header->filter);
 229
 230         if (prepare_revision_walk(&revs))
 231                 die(_("revision walk setup failed"));
 232
 233         i = req_nr;
 234         while (i && (commit = get_revision(&revs)))
 235                 if (commit->object.flags & PREREQ_MARK)
 236                         i--;
 237
 238         for (i = 0; i < p->nr; i++) {
 239                 struct string_list_item *e = p->items + i;
 240                 const char *name = e->string;
 241                 const struct object_id *oid = e->util;
 242                 struct object *o = parse_object(r, oid);
 243                 assert(o); /* otherwise we'd have returned early */
 244                 if (o->flags & SHOWN)
 245                         continue;
 246                 if (++ret == 1)
 247                         error("%s", message);
 248                 error("%s %s", oid_to_hex(oid), name);
 249         }
 250
 251         /* Clean up objects used, as they will be reused. */
 252         for (i = 0; i < p->nr; i++) {
 253                 struct string_list_item *e = p->items + i;
 254                 struct object_id *oid = e->util;
 255                 commit = lookup_commit_reference_gently(r, oid, 1);
 256                 if (commit)
 257                         clear_commit_marks(commit, ALL_REV_FLAGS);
 258         }
 259
 260         if (verbose) {
 261                 struct string_list *r;
 262
 263                 r = &header->references;
 264                 printf_ln(Q_("The bundle contains this ref:",
 265                              "The bundle contains these %"PRIuMAX" refs:",
 266                              r->nr),
 267                           (uintmax_t)r->nr);
 268                 list_refs(r, 0, NULL);
 269
 270                 r = &header->prerequisites;
 271                 if (!r->nr) {
 272                         printf_ln(_("The bundle records a complete history."));
 273                 } else {
 274                         printf_ln(Q_("The bundle requires this ref:",
 275                                      "The bundle requires these %"PRIuMAX" refs:",
 276                                      r->nr),
 277                                   (uintmax_t)r->nr);
 278                         list_refs(r, 0, NULL);
 279                 }
 280
 281                 printf_ln("The bundle uses this hash algorithm: %s",
 282                           header->hash_algo->name);
 283                 if (header->filter.choice)
 284                         printf_ln("The bundle uses this filter: %s",
 285                                   list_objects_filter_spec(&header->filter));
 286         }
 287         return ret;
 288 }
 289
 290 int list_bundle_refs(struct bundle_header *header, int argc, const char **argv)
 291 {
 292         return list_refs(&header->references, argc, argv);
 293 }
 294
 295 static int is_tag_in_date_range(struct object *tag, struct rev_info *revs)
 296 {
 297         unsigned long size;
 298         enum object_type type;
 299         char *buf = NULL, *line, *lineend;
 300         timestamp_t date;
 301         int result = 1;
 302
 303         if (revs->max_age == -1 && revs->min_age == -1)
 304                 goto out;
 305
 306         buf = read_object_file(&tag->oid, &type, &size);
 307         if (!buf)
 308                 goto out;
 309         line = memmem(buf, size, "\ntagger ", 8);
 310         if (!line++)
 311                 goto out;
 312         lineend = memchr(line, '\n', buf + size - line);
 313         line = memchr(line, '>', lineend ? lineend - line : buf + size - line);
 314         if (!line++)
 315                 goto out;
 316         date = parse_timestamp(line, NULL, 10);
 317         result = (revs->max_age == -1 || revs->max_age < date) &&
 318                 (revs->min_age == -1 || revs->min_age > date);
 319 out:
 320         free(buf);
 321         return result;
 322 }
 323
 324
 325 /* Write the pack data to bundle_fd */
 326 static int write_pack_data(int bundle_fd, struct rev_info *revs, struct strvec *pack_options)
 327 {
 328         struct child_process pack_objects = CHILD_PROCESS_INIT;
 329         int i;
 330
 331         strvec_pushl(&pack_objects.args,
 332                      "pack-objects",
 333                      "--stdout", "--thin", "--delta-base-offset",
 334                      NULL);
 335         strvec_pushv(&pack_objects.args, pack_options->v);
 336         if (revs->filter.choice)
 337                 strvec_pushf(&pack_objects.args, "--filter=%s",
 338                              list_objects_filter_spec(&revs->filter));
 339         pack_objects.in = -1;
 340         pack_objects.out = bundle_fd;
 341         pack_objects.git_cmd = 1;
 342
 343         /*
 344          * start_command() will close our descriptor if it's >1. Duplicate it
 345          * to avoid surprising the caller.
 346          */
 347         if (pack_objects.out > 1) {
 348                 pack_objects.out = dup(pack_objects.out);
 349                 if (pack_objects.out < 0) {
 350                         error_errno(_("unable to dup bundle descriptor"));
 351                         child_process_clear(&pack_objects);
 352                         return -1;
 353                 }
 354         }
 355
 356         if (start_command(&pack_objects))
 357                 return error(_("Could not spawn pack-objects"));
 358
 359         for (i = 0; i < revs->pending.nr; i++) {
 360                 struct object *object = revs->pending.objects[i].item;
 361                 if (object->flags & UNINTERESTING)
 362                         write_or_die(pack_objects.in, "^", 1);
 363                 write_or_die(pack_objects.in, oid_to_hex(&object->oid), the_hash_algo->hexsz);
 364                 write_or_die(pack_objects.in, "\n", 1);
 365         }
 366         close(pack_objects.in);
 367         if (finish_command(&pack_objects))
 368                 return error(_("pack-objects died"));
 369         return 0;
 370 }
 371
 372 /*
 373  * Write out bundle refs based on the tips already
 374  * parsed into revs.pending. As a side effect, may
 375  * manipulate revs.pending to include additional
 376  * necessary objects (like tags).
 377  *
 378  * Returns the number of refs written, or negative
 379  * on error.
 380  */
 381 static int write_bundle_refs(int bundle_fd, struct rev_info *revs)
 382 {
 383         int i;
 384         int ref_count = 0;
 385
 386         for (i = 0; i < revs->pending.nr; i++) {
 387                 struct object_array_entry *e = revs->pending.objects + i;
 388                 struct object_id oid;
 389                 char *ref;
 390                 const char *display_ref;
 391                 int flag;
 392
 393                 if (e->item->flags & UNINTERESTING)
 394                         continue;
 395                 if (dwim_ref(e->name, strlen(e->name), &oid, &ref, 0) != 1)
 396                         goto skip_write_ref;
 397                 if (read_ref_full(e->name, RESOLVE_REF_READING, &oid, &flag))
 398                         flag = 0;
 399                 display_ref = (flag & REF_ISSYMREF) ? e->name : ref;
 400
 401                 if (e->item->type == OBJ_TAG &&
 402                                 !is_tag_in_date_range(e->item, revs)) {
 403                         e->item->flags |= UNINTERESTING;
 404                         goto skip_write_ref;
 405                 }
 406
 407                 /*
 408                  * Make sure the refs we wrote out is correct; --max-count and
 409                  * other limiting options could have prevented all the tips
 410                  * from getting output.
 411                  *
 412                  * Non commit objects such as tags and blobs do not have
 413                  * this issue as they are not affected by those extra
 414                  * constraints.
 415                  */
 416                 if (!(e->item->flags & SHOWN) && e->item->type == OBJ_COMMIT) {
 417                         warning(_("ref '%s' is excluded by the rev-list options"),
 418                                 e->name);
 419                         goto skip_write_ref;
 420                 }
 421                 /*
 422                  * If you run "git bundle create bndl v1.0..v2.0", the
 423                  * name of the positive ref is "v2.0" but that is the
 424                  * commit that is referenced by the tag, and not the tag
 425                  * itself.
 426                  */
 427                 if (!oideq(&oid, &e->item->oid)) {
 428                         /*
 429                          * Is this the positive end of a range expressed
 430                          * in terms of a tag (e.g. v2.0 from the range
 431                          * "v1.0..v2.0")?
 432                          */
 433                         struct commit *one = lookup_commit_reference(revs->repo, &oid);
 434                         struct object *obj;
 435
 436                         if (e->item == &(one->object)) {
 437                                 /*
 438                                  * Need to include e->name as an
 439                                  * independent ref to the pack-objects
 440                                  * input, so that the tag is included
 441                                  * in the output; otherwise we would
 442                                  * end up triggering "empty bundle"
 443                                  * error.
 444                                  */
 445                                 obj = parse_object_or_die(&oid, e->name);
 446                                 obj->flags |= SHOWN;
 447                                 add_pending_object(revs, obj, e->name);
 448                         }
 449                         goto skip_write_ref;
 450                 }
 451
 452                 ref_count++;
 453                 write_or_die(bundle_fd, oid_to_hex(&e->item->oid), the_hash_algo->hexsz);
 454                 write_or_die(bundle_fd, " ", 1);
 455                 write_or_die(bundle_fd, display_ref, strlen(display_ref));
 456                 write_or_die(bundle_fd, "\n", 1);
 457  skip_write_ref:
 458                 free(ref);
 459         }
 460
 461         /* end header */
 462         write_or_die(bundle_fd, "\n", 1);
 463         return ref_count;
 464 }
 465
 466 struct bundle_prerequisites_info {
 467         struct object_array *pending;
 468         int fd;
 469 };
 470
 471 static void write_bundle_prerequisites(struct commit *commit, void *data)
 472 {
 473         struct bundle_prerequisites_info *bpi = data;
 474         struct object *object;
 475         struct pretty_print_context ctx = { 0 };
 476         struct strbuf buf = STRBUF_INIT;
 477
 478         if (!(commit->object.flags & BOUNDARY))
 479                 return;
 480         strbuf_addf(&buf, "-%s ", oid_to_hex(&commit->object.oid));
 481         write_or_die(bpi->fd, buf.buf, buf.len);
 482
 483         ctx.fmt = CMIT_FMT_ONELINE;
 484         ctx.output_encoding = get_log_output_encoding();
 485         strbuf_reset(&buf);
 486         pretty_print_commit(&ctx, commit, &buf);
 487         strbuf_trim(&buf);
 488
 489         object = (struct object *)commit;
 490         object->flags |= UNINTERESTING;
 491         add_object_array_with_path(object, buf.buf, bpi->pending, S_IFINVALID,
 492                                    NULL);
 493         strbuf_addch(&buf, '\n');
 494         write_or_die(bpi->fd, buf.buf, buf.len);
 495         strbuf_release(&buf);
 496 }
 497
 498 int create_bundle(struct repository *r, const char *path,
 499                   int argc, const char **argv, struct strvec *pack_options, int version)
 500 {
 501         struct lock_file lock = LOCK_INIT;
 502         int bundle_fd = -1;
 503         int bundle_to_stdout;
 504         int ref_count = 0;
 505         struct rev_info revs, revs_copy;
 506         int min_version = 2;
 507         struct bundle_prerequisites_info bpi;
 508         int i;
 509
 510         /* init revs to list objects for pack-objects later */
 511         save_commit_buffer = 0;
 512         repo_init_revisions(r, &revs, NULL);
 513
 514         /*
 515          * Pre-initialize the '--objects' flag so we can parse a
 516          * --filter option successfully.
 517          */
 518         revs.tree_objects = revs.blob_objects = 1;
 519
 520         argc = setup_revisions(argc, argv, &revs, NULL);
 521
 522         /*
 523          * Reasons to require version 3:
 524          *
 525          * 1. @object-format is required because our hash algorithm is not
 526          *    SHA1.
 527          * 2. @filter is required because we parsed an object filter.
 528          */
 529         if (the_hash_algo != &hash_algos[GIT_HASH_SHA1] || revs.filter.choice)
 530                 min_version = 3;
 531
 532         if (argc > 1) {
 533                 error(_("unrecognized argument: %s"), argv[1]);
 534                 goto err;
 535         }
 536
 537         bundle_to_stdout = !strcmp(path, "-");
 538         if (bundle_to_stdout)
 539                 bundle_fd = 1;
 540         else
 541                 bundle_fd = hold_lock_file_for_update(&lock, path,
 542                                                       LOCK_DIE_ON_ERROR);
 543
 544         if (version == -1)
 545                 version = min_version;
 546
 547         if (version < 2 || version > 3) {
 548                 die(_("unsupported bundle version %d"), version);
 549         } else if (version < min_version) {
 550                 die(_("cannot write bundle version %d with algorithm %s"), version, the_hash_algo->name);
 551         } else if (version == 2) {
 552                 write_or_die(bundle_fd, v2_bundle_signature, strlen(v2_bundle_signature));
 553         } else {
 554                 const char *capability = "@object-format=";
 555                 write_or_die(bundle_fd, v3_bundle_signature, strlen(v3_bundle_signature));
 556                 write_or_die(bundle_fd, capability, strlen(capability));
 557                 write_or_die(bundle_fd, the_hash_algo->name, strlen(the_hash_algo->name));
 558                 write_or_die(bundle_fd, "\n", 1);
 559
 560                 if (revs.filter.choice) {
 561                         const char *value = expand_list_objects_filter_spec(&revs.filter);
 562                         capability = "@filter=";
 563                         write_or_die(bundle_fd, capability, strlen(capability));
 564                         write_or_die(bundle_fd, value, strlen(value));
 565                         write_or_die(bundle_fd, "\n", 1);
 566                 }
 567         }
 568
 569         /* save revs.pending in revs_copy for later use */
 570         memcpy(&revs_copy, &revs, sizeof(revs));
 571         revs_copy.pending.nr = 0;
 572         revs_copy.pending.alloc = 0;
 573         revs_copy.pending.objects = NULL;
 574         for (i = 0; i < revs.pending.nr; i++) {
 575                 struct object_array_entry *e = revs.pending.objects + i;
 576                 if (e)
 577                         add_object_array_with_path(e->item, e->name,
 578                                                    &revs_copy.pending,
 579                                                    e->mode, e->path);
 580         }
 581
 582         /* write prerequisites */
 583         revs.boundary = 1;
 584         if (prepare_revision_walk(&revs))
 585                 die("revision walk setup failed");
 586         bpi.fd = bundle_fd;
 587         bpi.pending = &revs_copy.pending;
 588
 589         /*
 590          * Remove any object walking here. We only care about commits and
 591          * tags here. The revs_copy has the right instances of these values.
 592          */
 593         revs.blob_objects = revs.tree_objects = 0;
 594         traverse_commit_list(&revs, write_bundle_prerequisites, NULL, &bpi);
 595         object_array_remove_duplicates(&revs_copy.pending);
 596
 597         /* write bundle refs */
 598         ref_count = write_bundle_refs(bundle_fd, &revs_copy);
 599         if (!ref_count)
 600                 die(_("Refusing to create empty bundle."));
 601         else if (ref_count < 0)
 602                 goto err;
 603
 604         /* write pack */
 605         if (write_pack_data(bundle_fd, &revs_copy, pack_options))
 606                 goto err;
 607
 608         if (!bundle_to_stdout) {
 609                 if (commit_lock_file(&lock))
 610                         die_errno(_("cannot create '%s'"), path);
 611         }
 612         return 0;
 613 err:
 614         rollback_lock_file(&lock);
 615         return -1;
 616 }
 617
 618 int unbundle(struct repository *r, struct bundle_header *header,
 619              int bundle_fd, struct strvec *extra_index_pack_args)
 620 {
 621         struct child_process ip = CHILD_PROCESS_INIT;
 622         strvec_pushl(&ip.args, "index-pack", "--fix-thin", "--stdin", NULL);
 623
 624         /* If there is a filter, then we need to create the promisor pack. */
 625         if (header->filter.choice)
 626                 strvec_push(&ip.args, "--promisor=from-bundle");
 627
 628         if (extra_index_pack_args) {
 629                 strvec_pushv(&ip.args, extra_index_pack_args->v);
 630                 strvec_clear(extra_index_pack_args);
 631         }
 632
 633         if (verify_bundle(r, header, 0))
 634                 return -1;
 635         ip.in = bundle_fd;
 636         ip.no_stdout = 1;
 637         ip.git_cmd = 1;
 638         if (run_command(&ip))
 639                 return error(_("index-pack died"));
 640         return 0;
 641 }