bundle.c

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