bundle.c

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