git.oblomov.eu Git - git/blob - archive.c

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "refs.h"
   4 #include "object-store.h"
   5 #include "commit.h"
   6 #include "tree-walk.h"
   7 #include "attr.h"
   8 #include "archive.h"
   9 #include "parse-options.h"
  10 #include "unpack-trees.h"
  11 #include "dir.h"
  12
  13 static char const * const archive_usage[] = {
  14         N_("git archive [<options>] <tree-ish> [<path>...]"),
  15         N_("git archive --list"),
  16         N_("git archive --remote <repo> [--exec <cmd>] [<options>] <tree-ish> [<path>...]"),
  17         N_("git archive --remote <repo> [--exec <cmd>] --list"),
  18         NULL
  19 };
  20
  21 static const struct archiver **archivers;
  22 static int nr_archivers;
  23 static int alloc_archivers;
  24 static int remote_allow_unreachable;
  25
  26 void register_archiver(struct archiver *ar)
  27 {
  28         ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
  29         archivers[nr_archivers++] = ar;
  30 }
  31
  32 void init_archivers(void)
  33 {
  34         init_tar_archiver();
  35         init_zip_archiver();
  36 }
  37
  38 static void format_subst(const struct commit *commit,
  39                          const char *src, size_t len,
  40                          struct strbuf *buf)
  41 {
  42         char *to_free = NULL;
  43         struct strbuf fmt = STRBUF_INIT;
  44         struct pretty_print_context ctx = {0};
  45         ctx.date_mode.type = DATE_NORMAL;
  46         ctx.abbrev = DEFAULT_ABBREV;
  47
  48         if (src == buf->buf)
  49                 to_free = strbuf_detach(buf, NULL);
  50         for (;;) {
  51                 const char *b, *c;
  52
  53                 b = memmem(src, len, "$Format:", 8);
  54                 if (!b)
  55                         break;
  56                 c = memchr(b + 8, '$', (src + len) - b - 8);
  57                 if (!c)
  58                         break;
  59
  60                 strbuf_reset(&fmt);
  61                 strbuf_add(&fmt, b + 8, c - b - 8);
  62
  63                 strbuf_add(buf, src, b - src);
  64                 format_commit_message(commit, fmt.buf, buf, &ctx);
  65                 len -= c + 1 - src;
  66                 src  = c + 1;
  67         }
  68         strbuf_add(buf, src, len);
  69         strbuf_release(&fmt);
  70         free(to_free);
  71 }
  72
  73 static void *object_file_to_archive(const struct archiver_args *args,
  74                                     const char *path,
  75                                     const struct object_id *oid,
  76                                     unsigned int mode,
  77                                     enum object_type *type,
  78                                     unsigned long *sizep)
  79 {
  80         void *buffer;
  81         const struct commit *commit = args->convert ? args->commit : NULL;
  82         struct checkout_metadata meta;
  83
  84         init_checkout_metadata(&meta, args->refname,
  85                                args->commit_oid ? args->commit_oid :
  86                                (args->tree ? &args->tree->object.oid : NULL), oid);
  87
  88         path += args->baselen;
  89         buffer = read_object_file(oid, type, sizep);
  90         if (buffer && S_ISREG(mode)) {
  91                 struct strbuf buf = STRBUF_INIT;
  92                 size_t size = 0;
  93
  94                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
  95                 convert_to_working_tree(args->repo->index, path, buf.buf, buf.len, &buf, &meta);
  96                 if (commit)
  97                         format_subst(commit, buf.buf, buf.len, &buf);
  98                 buffer = strbuf_detach(&buf, &size);
  99                 *sizep = size;
 100         }
 101
 102         return buffer;
 103 }
 104
 105 struct directory {
 106         struct directory *up;
 107         struct object_id oid;
 108         int baselen, len;
 109         unsigned mode;
 110         int stage;
 111         char path[FLEX_ARRAY];
 112 };
 113
 114 struct archiver_context {
 115         struct archiver_args *args;
 116         write_archive_entry_fn_t write_entry;
 117         struct directory *bottom;
 118 };
 119
 120 static const struct attr_check *get_archive_attrs(struct index_state *istate,
 121                                                   const char *path)
 122 {
 123         static struct attr_check *check;
 124         if (!check)
 125                 check = attr_check_initl("export-ignore", "export-subst", NULL);
 126         git_check_attr(istate, path, check);
 127         return check;
 128 }
 129
 130 static int check_attr_export_ignore(const struct attr_check *check)
 131 {
 132         return check && ATTR_TRUE(check->items[0].value);
 133 }
 134
 135 static int check_attr_export_subst(const struct attr_check *check)
 136 {
 137         return check && ATTR_TRUE(check->items[1].value);
 138 }
 139
 140 static int write_archive_entry(const struct object_id *oid, const char *base,
 141                 int baselen, const char *filename, unsigned mode, int stage,
 142                 void *context)
 143 {
 144         static struct strbuf path = STRBUF_INIT;
 145         struct archiver_context *c = context;
 146         struct archiver_args *args = c->args;
 147         write_archive_entry_fn_t write_entry = c->write_entry;
 148         int err;
 149         const char *path_without_prefix;
 150         unsigned long size;
 151         void *buffer;
 152         enum object_type type;
 153
 154         args->convert = 0;
 155         strbuf_reset(&path);
 156         strbuf_grow(&path, PATH_MAX);
 157         strbuf_add(&path, args->base, args->baselen);
 158         strbuf_add(&path, base, baselen);
 159         strbuf_addstr(&path, filename);
 160         if (S_ISDIR(mode) || S_ISGITLINK(mode))
 161                 strbuf_addch(&path, '/');
 162         path_without_prefix = path.buf + args->baselen;
 163
 164         if (!S_ISDIR(mode)) {
 165                 const struct attr_check *check;
 166                 check = get_archive_attrs(args->repo->index, path_without_prefix);
 167                 if (check_attr_export_ignore(check))
 168                         return 0;
 169                 args->convert = check_attr_export_subst(check);
 170         }
 171
 172         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 173                 if (args->verbose)
 174                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 175                 err = write_entry(args, oid, path.buf, path.len, mode, NULL, 0);
 176                 if (err)
 177                         return err;
 178                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
 179         }
 180
 181         if (args->verbose)
 182                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 183
 184         /* Stream it? */
 185         if (S_ISREG(mode) && !args->convert &&
 186             oid_object_info(args->repo, oid, &size) == OBJ_BLOB &&
 187             size > big_file_threshold)
 188                 return write_entry(args, oid, path.buf, path.len, mode, NULL, size);
 189
 190         buffer = object_file_to_archive(args, path.buf, oid, mode, &type, &size);
 191         if (!buffer)
 192                 return error(_("cannot read %s"), oid_to_hex(oid));
 193         err = write_entry(args, oid, path.buf, path.len, mode, buffer, size);
 194         free(buffer);
 195         return err;
 196 }
 197
 198 static void queue_directory(const unsigned char *sha1,
 199                 struct strbuf *base, const char *filename,
 200                 unsigned mode, int stage, struct archiver_context *c)
 201 {
 202         struct directory *d;
 203         size_t len = st_add4(base->len, 1, strlen(filename), 1);
 204         d = xmalloc(st_add(sizeof(*d), len));
 205         d->up      = c->bottom;
 206         d->baselen = base->len;
 207         d->mode    = mode;
 208         d->stage   = stage;
 209         c->bottom  = d;
 210         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
 211         hashcpy(d->oid.hash, sha1);
 212 }
 213
 214 static int write_directory(struct archiver_context *c)
 215 {
 216         struct directory *d = c->bottom;
 217         int ret;
 218
 219         if (!d)
 220                 return 0;
 221         c->bottom = d->up;
 222         d->path[d->len - 1] = '\0'; /* no trailing slash */
 223         ret =
 224                 write_directory(c) ||
 225                 write_archive_entry(&d->oid, d->path, d->baselen,
 226                                     d->path + d->baselen, d->mode,
 227                                     d->stage, c) != READ_TREE_RECURSIVE;
 228         free(d);
 229         return ret ? -1 : 0;
 230 }
 231
 232 static int queue_or_write_archive_entry(const struct object_id *oid,
 233                 struct strbuf *base, const char *filename,
 234                 unsigned mode, int stage, void *context)
 235 {
 236         struct archiver_context *c = context;
 237
 238         while (c->bottom &&
 239                !(base->len >= c->bottom->len &&
 240                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
 241                 struct directory *next = c->bottom->up;
 242                 free(c->bottom);
 243                 c->bottom = next;
 244         }
 245
 246         if (S_ISDIR(mode)) {
 247                 size_t baselen = base->len;
 248                 const struct attr_check *check;
 249
 250                 /* Borrow base, but restore its original value when done. */
 251                 strbuf_addstr(base, filename);
 252                 strbuf_addch(base, '/');
 253                 check = get_archive_attrs(c->args->repo->index, base->buf);
 254                 strbuf_setlen(base, baselen);
 255
 256                 if (check_attr_export_ignore(check))
 257                         return 0;
 258                 queue_directory(oid->hash, base, filename,
 259                                 mode, stage, c);
 260                 return READ_TREE_RECURSIVE;
 261         }
 262
 263         if (write_directory(c))
 264                 return -1;
 265         return write_archive_entry(oid, base->buf, base->len, filename, mode,
 266                                    stage, context);
 267 }
 268
 269 int write_archive_entries(struct archiver_args *args,
 270                 write_archive_entry_fn_t write_entry)
 271 {
 272         struct archiver_context context;
 273         struct unpack_trees_options opts;
 274         struct tree_desc t;
 275         int err;
 276
 277         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
 278                 size_t len = args->baselen;
 279
 280                 while (len > 1 && args->base[len - 2] == '/')
 281                         len--;
 282                 if (args->verbose)
 283                         fprintf(stderr, "%.*s\n", (int)len, args->base);
 284                 err = write_entry(args, &args->tree->object.oid, args->base,
 285                                   len, 040777, NULL, 0);
 286                 if (err)
 287                         return err;
 288         }
 289
 290         memset(&context, 0, sizeof(context));
 291         context.args = args;
 292         context.write_entry = write_entry;
 293
 294         /*
 295          * Setup index and instruct attr to read index only
 296          */
 297         if (!args->worktree_attributes) {
 298                 memset(&opts, 0, sizeof(opts));
 299                 opts.index_only = 1;
 300                 opts.head_idx = -1;
 301                 opts.src_index = args->repo->index;
 302                 opts.dst_index = args->repo->index;
 303                 opts.fn = oneway_merge;
 304                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
 305                 if (unpack_trees(1, &t, &opts))
 306                         return -1;
 307                 git_attr_set_direction(GIT_ATTR_INDEX);
 308         }
 309
 310         err = read_tree_recursive(args->repo, args->tree, "",
 311                                   0, 0, &args->pathspec,
 312                                   queue_or_write_archive_entry,
 313                                   &context);
 314         if (err == READ_TREE_RECURSIVE)
 315                 err = 0;
 316         while (context.bottom) {
 317                 struct directory *next = context.bottom->up;
 318                 free(context.bottom);
 319                 context.bottom = next;
 320         }
 321         return err;
 322 }
 323
 324 static const struct archiver *lookup_archiver(const char *name)
 325 {
 326         int i;
 327
 328         if (!name)
 329                 return NULL;
 330
 331         for (i = 0; i < nr_archivers; i++) {
 332                 if (!strcmp(name, archivers[i]->name))
 333                         return archivers[i];
 334         }
 335         return NULL;
 336 }
 337
 338 struct path_exists_context {
 339         struct pathspec pathspec;
 340         struct archiver_args *args;
 341 };
 342
 343 static int reject_entry(const struct object_id *oid, struct strbuf *base,
 344                         const char *filename, unsigned mode,
 345                         int stage, void *context)
 346 {
 347         int ret = -1;
 348         struct path_exists_context *ctx = context;
 349
 350         if (S_ISDIR(mode)) {
 351                 struct strbuf sb = STRBUF_INIT;
 352                 strbuf_addbuf(&sb, base);
 353                 strbuf_addstr(&sb, filename);
 354                 if (!match_pathspec(ctx->args->repo->index,
 355                                     &ctx->pathspec,
 356                                     sb.buf, sb.len, 0, NULL, 1))
 357                         ret = READ_TREE_RECURSIVE;
 358                 strbuf_release(&sb);
 359         }
 360         return ret;
 361 }
 362
 363 static int path_exists(struct archiver_args *args, const char *path)
 364 {
 365         const char *paths[] = { path, NULL };
 366         struct path_exists_context ctx;
 367         int ret;
 368
 369         ctx.args = args;
 370         parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
 371         ctx.pathspec.recursive = 1;
 372         ret = read_tree_recursive(args->repo, args->tree, "",
 373                                   0, 0, &ctx.pathspec,
 374                                   reject_entry, &ctx);
 375         clear_pathspec(&ctx.pathspec);
 376         return ret != 0;
 377 }
 378
 379 static void parse_pathspec_arg(const char **pathspec,
 380                 struct archiver_args *ar_args)
 381 {
 382         /*
 383          * must be consistent with parse_pathspec in path_exists()
 384          * Also if pathspec patterns are dependent, we're in big
 385          * trouble as we test each one separately
 386          */
 387         parse_pathspec(&ar_args->pathspec, 0,
 388                        PATHSPEC_PREFER_FULL,
 389                        "", pathspec);
 390         ar_args->pathspec.recursive = 1;
 391         if (pathspec) {
 392                 while (*pathspec) {
 393                         if (**pathspec && !path_exists(ar_args, *pathspec))
 394                                 die(_("pathspec '%s' did not match any files"), *pathspec);
 395                         pathspec++;
 396                 }
 397         }
 398 }
 399
 400 static void parse_treeish_arg(const char **argv,
 401                 struct archiver_args *ar_args, const char *prefix,
 402                 int remote)
 403 {
 404         const char *name = argv[0];
 405         const struct object_id *commit_oid;
 406         time_t archive_time;
 407         struct tree *tree;
 408         const struct commit *commit;
 409         struct object_id oid;
 410         char *ref = NULL;
 411
 412         /* Remotes are only allowed to fetch actual refs */
 413         if (remote && !remote_allow_unreachable) {
 414                 const char *colon = strchrnul(name, ':');
 415                 int refnamelen = colon - name;
 416
 417                 if (!dwim_ref(name, refnamelen, &oid, &ref, 0))
 418                         die(_("no such ref: %.*s"), refnamelen, name);
 419         } else {
 420                 dwim_ref(name, strlen(name), &oid, &ref, 0);
 421         }
 422
 423         if (get_oid(name, &oid))
 424                 die(_("not a valid object name: %s"), name);
 425
 426         commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
 427         if (commit) {
 428                 commit_oid = &commit->object.oid;
 429                 archive_time = commit->date;
 430         } else {
 431                 commit_oid = NULL;
 432                 archive_time = time(NULL);
 433         }
 434
 435         tree = parse_tree_indirect(&oid);
 436         if (tree == NULL)
 437                 die(_("not a tree object: %s"), oid_to_hex(&oid));
 438
 439         if (prefix) {
 440                 struct object_id tree_oid;
 441                 unsigned short mode;
 442                 int err;
 443
 444                 err = get_tree_entry(ar_args->repo,
 445                                      &tree->object.oid,
 446                                      prefix, &tree_oid,
 447                                      &mode);
 448                 if (err || !S_ISDIR(mode))
 449                         die(_("current working directory is untracked"));
 450
 451                 tree = parse_tree_indirect(&tree_oid);
 452         }
 453         ar_args->refname = ref;
 454         ar_args->tree = tree;
 455         ar_args->commit_oid = commit_oid;
 456         ar_args->commit = commit;
 457         ar_args->time = archive_time;
 458 }
 459
 460 #define OPT__COMPR(s, v, h, p) \
 461         OPT_SET_INT_F(s, NULL, v, h, p, PARSE_OPT_NONEG)
 462 #define OPT__COMPR_HIDDEN(s, v, p) \
 463         OPT_SET_INT_F(s, NULL, v, "", p, PARSE_OPT_NONEG | PARSE_OPT_HIDDEN)
 464
 465 static int parse_archive_args(int argc, const char **argv,
 466                 const struct archiver **ar, struct archiver_args *args,
 467                 const char *name_hint, int is_remote)
 468 {
 469         const char *format = NULL;
 470         const char *base = NULL;
 471         const char *remote = NULL;
 472         const char *exec = NULL;
 473         const char *output = NULL;
 474         int compression_level = -1;
 475         int verbose = 0;
 476         int i;
 477         int list = 0;
 478         int worktree_attributes = 0;
 479         struct option opts[] = {
 480                 OPT_GROUP(""),
 481                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
 482                 OPT_STRING(0, "prefix", &base, N_("prefix"),
 483                         N_("prepend prefix to each pathname in the archive")),
 484                 OPT_STRING('o', "output", &output, N_("file"),
 485                         N_("write the archive to this file")),
 486                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
 487                         N_("read .gitattributes in working directory")),
 488                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
 489                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
 490                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
 491                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
 492                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
 493                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
 494                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
 495                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
 496                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
 497                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
 498                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
 499                 OPT_GROUP(""),
 500                 OPT_BOOL('l', "list", &list,
 501                         N_("list supported archive formats")),
 502                 OPT_GROUP(""),
 503                 OPT_STRING(0, "remote", &remote, N_("repo"),
 504                         N_("retrieve the archive from remote repository <repo>")),
 505                 OPT_STRING(0, "exec", &exec, N_("command"),
 506                         N_("path to the remote git-upload-archive command")),
 507                 OPT_END()
 508         };
 509
 510         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
 511
 512         if (remote)
 513                 die(_("Unexpected option --remote"));
 514         if (exec)
 515                 die(_("Option --exec can only be used together with --remote"));
 516         if (output)
 517                 die(_("Unexpected option --output"));
 518
 519         if (!base)
 520                 base = "";
 521
 522         if (list) {
 523                 for (i = 0; i < nr_archivers; i++)
 524                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
 525                                 printf("%s\n", archivers[i]->name);
 526                 exit(0);
 527         }
 528
 529         if (!format && name_hint)
 530                 format = archive_format_from_filename(name_hint);
 531         if (!format)
 532                 format = "tar";
 533
 534         /* We need at least one parameter -- tree-ish */
 535         if (argc < 1)
 536                 usage_with_options(archive_usage, opts);
 537         *ar = lookup_archiver(format);
 538         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
 539                 die(_("Unknown archive format '%s'"), format);
 540
 541         args->compression_level = Z_DEFAULT_COMPRESSION;
 542         if (compression_level != -1) {
 543                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
 544                         args->compression_level = compression_level;
 545                 else {
 546                         die(_("Argument not supported for format '%s': -%d"),
 547                                         format, compression_level);
 548                 }
 549         }
 550         args->verbose = verbose;
 551         args->base = base;
 552         args->baselen = strlen(base);
 553         args->worktree_attributes = worktree_attributes;
 554
 555         return argc;
 556 }
 557
 558 int write_archive(int argc, const char **argv, const char *prefix,
 559                   struct repository *repo,
 560                   const char *name_hint, int remote)
 561 {
 562         const struct archiver *ar = NULL;
 563         struct archiver_args args;
 564
 565         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
 566         git_config(git_default_config, NULL);
 567
 568         args.repo = repo;
 569         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
 570         if (!startup_info->have_repository) {
 571                 /*
 572                  * We know this will die() with an error, so we could just
 573                  * die ourselves; but its error message will be more specific
 574                  * than what we could write here.
 575                  */
 576                 setup_git_directory();
 577         }
 578
 579         parse_treeish_arg(argv, &args, prefix, remote);
 580         parse_pathspec_arg(argv + 1, &args);
 581
 582         return ar->write_archive(ar, &args);
 583 }
 584
 585 static int match_extension(const char *filename, const char *ext)
 586 {
 587         int prefixlen = strlen(filename) - strlen(ext);
 588
 589         /*
 590          * We need 1 character for the '.', and 1 character to ensure that the
 591          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
 592          * filename).
 593          */
 594         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
 595                 return 0;
 596         return !strcmp(filename + prefixlen, ext);
 597 }
 598
 599 const char *archive_format_from_filename(const char *filename)
 600 {
 601         int i;
 602
 603         for (i = 0; i < nr_archivers; i++)
 604                 if (match_extension(filename, archivers[i]->name))
 605                         return archivers[i]->name;
 606         return NULL;
 607 }