bundle.c

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