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