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