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