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 static void format_subst(const struct commit *commit,
  33                          const char *src, size_t len,
  34                          struct strbuf *buf)
  35 {
  36         char *to_free = NULL;
  37         struct strbuf fmt = STRBUF_INIT;
  38         struct pretty_print_context ctx = {0};
  39         ctx.date_mode.type = DATE_NORMAL;
  40         ctx.abbrev = DEFAULT_ABBREV;
  41
  42         if (src == buf->buf)
  43                 to_free = strbuf_detach(buf, NULL);
  44         for (;;) {
  45                 const char *b, *c;
  46
  47                 b = memmem(src, len, "$Format:", 8);
  48                 if (!b)
  49                         break;
  50                 c = memchr(b + 8, '$', (src + len) - b - 8);
  51                 if (!c)
  52                         break;
  53
  54                 strbuf_reset(&fmt);
  55                 strbuf_add(&fmt, b + 8, c - b - 8);
  56
  57                 strbuf_add(buf, src, b - src);
  58                 format_commit_message(commit, fmt.buf, buf, &ctx);
  59                 len -= c + 1 - src;
  60                 src  = c + 1;
  61         }
  62         strbuf_add(buf, src, len);
  63         strbuf_release(&fmt);
  64         free(to_free);
  65 }
  66
  67 void *object_file_to_archive(const struct archiver_args *args,
  68                              const char *path, const struct object_id *oid,
  69                              unsigned int mode, enum object_type *type,
  70                              unsigned long *sizep)
  71 {
  72         void *buffer;
  73         const struct commit *commit = args->convert ? args->commit : NULL;
  74
  75         path += args->baselen;
  76         buffer = read_object_file(oid, type, sizep);
  77         if (buffer && S_ISREG(mode)) {
  78                 struct strbuf buf = STRBUF_INIT;
  79                 size_t size = 0;
  80
  81                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
  82                 convert_to_working_tree(args->repo->index, path, buf.buf, buf.len, &buf);
  83                 if (commit)
  84                         format_subst(commit, buf.buf, buf.len, &buf);
  85                 buffer = strbuf_detach(&buf, &size);
  86                 *sizep = size;
  87         }
  88
  89         return buffer;
  90 }
  91
  92 struct directory {
  93         struct directory *up;
  94         struct object_id oid;
  95         int baselen, len;
  96         unsigned mode;
  97         int stage;
  98         char path[FLEX_ARRAY];
  99 };
 100
 101 struct archiver_context {
 102         struct archiver_args *args;
 103         write_archive_entry_fn_t write_entry;
 104         struct directory *bottom;
 105 };
 106
 107 static const struct attr_check *get_archive_attrs(struct index_state *istate,
 108                                                   const char *path)
 109 {
 110         static struct attr_check *check;
 111         if (!check)
 112                 check = attr_check_initl("export-ignore", "export-subst", NULL);
 113         git_check_attr(istate, path, check);
 114         return check;
 115 }
 116
 117 static int check_attr_export_ignore(const struct attr_check *check)
 118 {
 119         return check && ATTR_TRUE(check->items[0].value);
 120 }
 121
 122 static int check_attr_export_subst(const struct attr_check *check)
 123 {
 124         return check && ATTR_TRUE(check->items[1].value);
 125 }
 126
 127 static int write_archive_entry(const struct object_id *oid, const char *base,
 128                 int baselen, const char *filename, unsigned mode, int stage,
 129                 void *context)
 130 {
 131         static struct strbuf path = STRBUF_INIT;
 132         struct archiver_context *c = context;
 133         struct archiver_args *args = c->args;
 134         write_archive_entry_fn_t write_entry = c->write_entry;
 135         int err;
 136         const char *path_without_prefix;
 137
 138         args->convert = 0;
 139         strbuf_reset(&path);
 140         strbuf_grow(&path, PATH_MAX);
 141         strbuf_add(&path, args->base, args->baselen);
 142         strbuf_add(&path, base, baselen);
 143         strbuf_addstr(&path, filename);
 144         if (S_ISDIR(mode) || S_ISGITLINK(mode))
 145                 strbuf_addch(&path, '/');
 146         path_without_prefix = path.buf + args->baselen;
 147
 148         if (!S_ISDIR(mode)) {
 149                 const struct attr_check *check;
 150                 check = get_archive_attrs(args->repo->index, path_without_prefix);
 151                 if (check_attr_export_ignore(check))
 152                         return 0;
 153                 args->convert = check_attr_export_subst(check);
 154         }
 155
 156         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 157                 if (args->verbose)
 158                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 159                 err = write_entry(args, oid, path.buf, path.len, mode);
 160                 if (err)
 161                         return err;
 162                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
 163         }
 164
 165         if (args->verbose)
 166                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 167         return write_entry(args, oid, path.buf, path.len, mode);
 168 }
 169
 170 static void queue_directory(const unsigned char *sha1,
 171                 struct strbuf *base, const char *filename,
 172                 unsigned mode, int stage, struct archiver_context *c)
 173 {
 174         struct directory *d;
 175         size_t len = st_add4(base->len, 1, strlen(filename), 1);
 176         d = xmalloc(st_add(sizeof(*d), len));
 177         d->up      = c->bottom;
 178         d->baselen = base->len;
 179         d->mode    = mode;
 180         d->stage   = stage;
 181         c->bottom  = d;
 182         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
 183         hashcpy(d->oid.hash, sha1);
 184 }
 185
 186 static int write_directory(struct archiver_context *c)
 187 {
 188         struct directory *d = c->bottom;
 189         int ret;
 190
 191         if (!d)
 192                 return 0;
 193         c->bottom = d->up;
 194         d->path[d->len - 1] = '\0'; /* no trailing slash */
 195         ret =
 196                 write_directory(c) ||
 197                 write_archive_entry(&d->oid, d->path, d->baselen,
 198                                     d->path + d->baselen, d->mode,
 199                                     d->stage, c) != READ_TREE_RECURSIVE;
 200         free(d);
 201         return ret ? -1 : 0;
 202 }
 203
 204 static int queue_or_write_archive_entry(const struct object_id *oid,
 205                 struct strbuf *base, const char *filename,
 206                 unsigned mode, int stage, void *context)
 207 {
 208         struct archiver_context *c = context;
 209
 210         while (c->bottom &&
 211                !(base->len >= c->bottom->len &&
 212                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
 213                 struct directory *next = c->bottom->up;
 214                 free(c->bottom);
 215                 c->bottom = next;
 216         }
 217
 218         if (S_ISDIR(mode)) {
 219                 size_t baselen = base->len;
 220                 const struct attr_check *check;
 221
 222                 /* Borrow base, but restore its original value when done. */
 223                 strbuf_addstr(base, filename);
 224                 strbuf_addch(base, '/');
 225                 check = get_archive_attrs(c->args->repo->index, base->buf);
 226                 strbuf_setlen(base, baselen);
 227
 228                 if (check_attr_export_ignore(check))
 229                         return 0;
 230                 queue_directory(oid->hash, base, filename,
 231                                 mode, stage, c);
 232                 return READ_TREE_RECURSIVE;
 233         }
 234
 235         if (write_directory(c))
 236                 return -1;
 237         return write_archive_entry(oid, base->buf, base->len, filename, mode,
 238                                    stage, context);
 239 }
 240
 241 int write_archive_entries(struct archiver_args *args,
 242                 write_archive_entry_fn_t write_entry)
 243 {
 244         struct archiver_context context;
 245         struct unpack_trees_options opts;
 246         struct tree_desc t;
 247         int err;
 248
 249         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
 250                 size_t len = args->baselen;
 251
 252                 while (len > 1 && args->base[len - 2] == '/')
 253                         len--;
 254                 if (args->verbose)
 255                         fprintf(stderr, "%.*s\n", (int)len, args->base);
 256                 err = write_entry(args, &args->tree->object.oid, args->base,
 257                                   len, 040777);
 258                 if (err)
 259                         return err;
 260         }
 261
 262         memset(&context, 0, sizeof(context));
 263         context.args = args;
 264         context.write_entry = write_entry;
 265
 266         /*
 267          * Setup index and instruct attr to read index only
 268          */
 269         if (!args->worktree_attributes) {
 270                 memset(&opts, 0, sizeof(opts));
 271                 opts.index_only = 1;
 272                 opts.head_idx = -1;
 273                 opts.src_index = args->repo->index;
 274                 opts.dst_index = args->repo->index;
 275                 opts.fn = oneway_merge;
 276                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
 277                 if (unpack_trees(1, &t, &opts))
 278                         return -1;
 279                 git_attr_set_direction(GIT_ATTR_INDEX);
 280         }
 281
 282         err = read_tree_recursive(args->tree, "", 0, 0, &args->pathspec,
 283                                   queue_or_write_archive_entry,
 284                                   &context);
 285         if (err == READ_TREE_RECURSIVE)
 286                 err = 0;
 287         while (context.bottom) {
 288                 struct directory *next = context.bottom->up;
 289                 free(context.bottom);
 290                 context.bottom = next;
 291         }
 292         return err;
 293 }
 294
 295 static const struct archiver *lookup_archiver(const char *name)
 296 {
 297         int i;
 298
 299         if (!name)
 300                 return NULL;
 301
 302         for (i = 0; i < nr_archivers; i++) {
 303                 if (!strcmp(name, archivers[i]->name))
 304                         return archivers[i];
 305         }
 306         return NULL;
 307 }
 308
 309 struct path_exists_context {
 310         struct pathspec pathspec;
 311         struct archiver_args *args;
 312 };
 313
 314 static int reject_entry(const struct object_id *oid, struct strbuf *base,
 315                         const char *filename, unsigned mode,
 316                         int stage, void *context)
 317 {
 318         int ret = -1;
 319         struct path_exists_context *ctx = context;
 320
 321         if (S_ISDIR(mode)) {
 322                 struct strbuf sb = STRBUF_INIT;
 323                 strbuf_addbuf(&sb, base);
 324                 strbuf_addstr(&sb, filename);
 325                 if (!match_pathspec(ctx->args->repo->index,
 326                                     &ctx->pathspec,
 327                                     sb.buf, sb.len, 0, NULL, 1))
 328                         ret = READ_TREE_RECURSIVE;
 329                 strbuf_release(&sb);
 330         }
 331         return ret;
 332 }
 333
 334 static int path_exists(struct archiver_args *args, const char *path)
 335 {
 336         const char *paths[] = { path, NULL };
 337         struct path_exists_context ctx;
 338         int ret;
 339
 340         ctx.args = args;
 341         parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
 342         ctx.pathspec.recursive = 1;
 343         ret = read_tree_recursive(args->tree, "", 0, 0, &ctx.pathspec,
 344                                   reject_entry, &ctx);
 345         clear_pathspec(&ctx.pathspec);
 346         return ret != 0;
 347 }
 348
 349 static void parse_pathspec_arg(const char **pathspec,
 350                 struct archiver_args *ar_args)
 351 {
 352         /*
 353          * must be consistent with parse_pathspec in path_exists()
 354          * Also if pathspec patterns are dependent, we're in big
 355          * trouble as we test each one separately
 356          */
 357         parse_pathspec(&ar_args->pathspec, 0,
 358                        PATHSPEC_PREFER_FULL,
 359                        "", pathspec);
 360         ar_args->pathspec.recursive = 1;
 361         if (pathspec) {
 362                 while (*pathspec) {
 363                         if (**pathspec && !path_exists(ar_args, *pathspec))
 364                                 die(_("pathspec '%s' did not match any files"), *pathspec);
 365                         pathspec++;
 366                 }
 367         }
 368 }
 369
 370 static void parse_treeish_arg(const char **argv,
 371                 struct archiver_args *ar_args, const char *prefix,
 372                 int remote)
 373 {
 374         const char *name = argv[0];
 375         const unsigned char *commit_sha1;
 376         time_t archive_time;
 377         struct tree *tree;
 378         const struct commit *commit;
 379         struct object_id oid;
 380
 381         /* Remotes are only allowed to fetch actual refs */
 382         if (remote && !remote_allow_unreachable) {
 383                 char *ref = NULL;
 384                 const char *colon = strchrnul(name, ':');
 385                 int refnamelen = colon - name;
 386
 387                 if (!dwim_ref(name, refnamelen, &oid, &ref))
 388                         die("no such ref: %.*s", refnamelen, name);
 389                 free(ref);
 390         }
 391
 392         if (get_oid(name, &oid))
 393                 die("Not a valid object name");
 394
 395         commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
 396         if (commit) {
 397                 commit_sha1 = commit->object.oid.hash;
 398                 archive_time = commit->date;
 399         } else {
 400                 commit_sha1 = NULL;
 401                 archive_time = time(NULL);
 402         }
 403
 404         tree = parse_tree_indirect(&oid);
 405         if (tree == NULL)
 406                 die("not a tree object");
 407
 408         if (prefix) {
 409                 struct object_id tree_oid;
 410                 unsigned int mode;
 411                 int err;
 412
 413                 err = get_tree_entry(&tree->object.oid, prefix, &tree_oid,
 414                                      &mode);
 415                 if (err || !S_ISDIR(mode))
 416                         die("current working directory is untracked");
 417
 418                 tree = parse_tree_indirect(&tree_oid);
 419         }
 420         ar_args->tree = tree;
 421         ar_args->commit_sha1 = commit_sha1;
 422         ar_args->commit = commit;
 423         ar_args->time = archive_time;
 424 }
 425
 426 #define OPT__COMPR(s, v, h, p) \
 427         OPT_SET_INT_F(s, NULL, v, h, p, PARSE_OPT_NONEG)
 428 #define OPT__COMPR_HIDDEN(s, v, p) \
 429         OPT_SET_INT_F(s, NULL, v, "", p, PARSE_OPT_NONEG | PARSE_OPT_HIDDEN)
 430
 431 static int parse_archive_args(int argc, const char **argv,
 432                 const struct archiver **ar, struct archiver_args *args,
 433                 const char *name_hint, int is_remote)
 434 {
 435         const char *format = NULL;
 436         const char *base = NULL;
 437         const char *remote = NULL;
 438         const char *exec = NULL;
 439         const char *output = NULL;
 440         int compression_level = -1;
 441         int verbose = 0;
 442         int i;
 443         int list = 0;
 444         int worktree_attributes = 0;
 445         struct option opts[] = {
 446                 OPT_GROUP(""),
 447                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
 448                 OPT_STRING(0, "prefix", &base, N_("prefix"),
 449                         N_("prepend prefix to each pathname in the archive")),
 450                 OPT_STRING('o', "output", &output, N_("file"),
 451                         N_("write the archive to this file")),
 452                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
 453                         N_("read .gitattributes in working directory")),
 454                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
 455                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
 456                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
 457                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
 458                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
 459                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
 460                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
 461                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
 462                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
 463                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
 464                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
 465                 OPT_GROUP(""),
 466                 OPT_BOOL('l', "list", &list,
 467                         N_("list supported archive formats")),
 468                 OPT_GROUP(""),
 469                 OPT_STRING(0, "remote", &remote, N_("repo"),
 470                         N_("retrieve the archive from remote repository <repo>")),
 471                 OPT_STRING(0, "exec", &exec, N_("command"),
 472                         N_("path to the remote git-upload-archive command")),
 473                 OPT_END()
 474         };
 475
 476         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
 477
 478         if (remote)
 479                 die(_("Unexpected option --remote"));
 480         if (exec)
 481                 die(_("Option --exec can only be used together with --remote"));
 482         if (output)
 483                 die(_("Unexpected option --output"));
 484
 485         if (!base)
 486                 base = "";
 487
 488         if (list) {
 489                 for (i = 0; i < nr_archivers; i++)
 490                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
 491                                 printf("%s\n", archivers[i]->name);
 492                 exit(0);
 493         }
 494
 495         if (!format && name_hint)
 496                 format = archive_format_from_filename(name_hint);
 497         if (!format)
 498                 format = "tar";
 499
 500         /* We need at least one parameter -- tree-ish */
 501         if (argc < 1)
 502                 usage_with_options(archive_usage, opts);
 503         *ar = lookup_archiver(format);
 504         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
 505                 die(_("Unknown archive format '%s'"), format);
 506
 507         args->compression_level = Z_DEFAULT_COMPRESSION;
 508         if (compression_level != -1) {
 509                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
 510                         args->compression_level = compression_level;
 511                 else {
 512                         die(_("Argument not supported for format '%s': -%d"),
 513                                         format, compression_level);
 514                 }
 515         }
 516         args->verbose = verbose;
 517         args->base = base;
 518         args->baselen = strlen(base);
 519         args->worktree_attributes = worktree_attributes;
 520
 521         return argc;
 522 }
 523
 524 int write_archive(int argc, const char **argv, const char *prefix,
 525                   struct repository *repo,
 526                   const char *name_hint, int remote)
 527 {
 528         const struct archiver *ar = NULL;
 529         struct archiver_args args;
 530
 531         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
 532         git_config(git_default_config, NULL);
 533
 534         init_tar_archiver();
 535         init_zip_archiver();
 536
 537         args.repo = repo;
 538         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
 539         if (!startup_info->have_repository) {
 540                 /*
 541                  * We know this will die() with an error, so we could just
 542                  * die ourselves; but its error message will be more specific
 543                  * than what we could write here.
 544                  */
 545                 setup_git_directory();
 546         }
 547
 548         parse_treeish_arg(argv, &args, prefix, remote);
 549         parse_pathspec_arg(argv + 1, &args);
 550
 551         return ar->write_archive(ar, &args);
 552 }
 553
 554 static int match_extension(const char *filename, const char *ext)
 555 {
 556         int prefixlen = strlen(filename) - strlen(ext);
 557
 558         /*
 559          * We need 1 character for the '.', and 1 character to ensure that the
 560          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
 561          * filename).
 562          */
 563         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
 564                 return 0;
 565         return !strcmp(filename + prefixlen, ext);
 566 }
 567
 568 const char *archive_format_from_filename(const char *filename)
 569 {
 570         int i;
 571
 572         for (i = 0; i < nr_archivers; i++)
 573                 if (match_extension(filename, archivers[i]->name))
 574                         return archivers[i]->name;
 575         return NULL;
 576 }