bundle.c

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