archive.c

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "refs.h"
   4 #include "commit.h"
   5 #include "tree-walk.h"
   6 #include "attr.h"
   7 #include "archive.h"
   8 #include "parse-options.h"
   9 #include "unpack-trees.h"
  10 #include "dir.h"
  11
  12 static char const * const archive_usage[] = {
  13         N_("git archive [<options>] <tree-ish> [<path>...]"),
  14         N_("git archive --list"),
  15         N_("git archive --remote <repo> [--exec <cmd>] [<options>] <tree-ish> [<path>...]"),
  16         N_("git archive --remote <repo> [--exec <cmd>] --list"),
  17         NULL
  18 };
  19
  20 static const struct archiver **archivers;
  21 static int nr_archivers;
  22 static int alloc_archivers;
  23 static int remote_allow_unreachable;
  24
  25 void register_archiver(struct archiver *ar)
  26 {
  27         ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
  28         archivers[nr_archivers++] = ar;
  29 }
  30
  31 static void format_subst(const struct commit *commit,
  32                          const char *src, size_t len,
  33                          struct strbuf *buf)
  34 {
  35         char *to_free = NULL;
  36         struct strbuf fmt = STRBUF_INIT;
  37         struct pretty_print_context ctx = {0};
  38         ctx.date_mode.type = DATE_NORMAL;
  39         ctx.abbrev = DEFAULT_ABBREV;
  40
  41         if (src == buf->buf)
  42                 to_free = strbuf_detach(buf, NULL);
  43         for (;;) {
  44                 const char *b, *c;
  45
  46                 b = memmem(src, len, "$Format:", 8);
  47                 if (!b)
  48                         break;
  49                 c = memchr(b + 8, '$', (src + len) - b - 8);
  50                 if (!c)
  51                         break;
  52
  53                 strbuf_reset(&fmt);
  54                 strbuf_add(&fmt, b + 8, c - b - 8);
  55
  56                 strbuf_add(buf, src, b - src);
  57                 format_commit_message(commit, fmt.buf, buf, &ctx);
  58                 len -= c + 1 - src;
  59                 src  = c + 1;
  60         }
  61         strbuf_add(buf, src, len);
  62         strbuf_release(&fmt);
  63         free(to_free);
  64 }
  65
  66 void *sha1_file_to_archive(const struct archiver_args *args,
  67                            const char *path, const unsigned char *sha1,
  68                            unsigned int mode, enum object_type *type,
  69                            unsigned long *sizep)
  70 {
  71         void *buffer;
  72         const struct commit *commit = args->convert ? args->commit : NULL;
  73
  74         path += args->baselen;
  75         buffer = read_sha1_file(sha1, type, sizep);
  76         if (buffer && S_ISREG(mode)) {
  77                 struct strbuf buf = STRBUF_INIT;
  78                 size_t size = 0;
  79
  80                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
  81                 convert_to_working_tree(path, buf.buf, buf.len, &buf);
  82                 if (commit)
  83                         format_subst(commit, buf.buf, buf.len, &buf);
  84                 buffer = strbuf_detach(&buf, &size);
  85                 *sizep = size;
  86         }
  87
  88         return buffer;
  89 }
  90
  91 struct directory {
  92         struct directory *up;
  93         struct object_id oid;
  94         int baselen, len;
  95         unsigned mode;
  96         int stage;
  97         char path[FLEX_ARRAY];
  98 };
  99
 100 struct archiver_context {
 101         struct archiver_args *args;
 102         write_archive_entry_fn_t write_entry;
 103         struct directory *bottom;
 104 };
 105
 106 static const struct attr_check *get_archive_attrs(const char *path)
 107 {
 108         static struct attr_check *check;
 109         if (!check)
 110                 check = attr_check_initl("export-ignore", "export-subst", NULL);
 111         return git_check_attr(path, check) ? NULL : check;
 112 }
 113
 114 static int check_attr_export_ignore(const struct attr_check *check)
 115 {
 116         return check && ATTR_TRUE(check->items[0].value);
 117 }
 118
 119 static int check_attr_export_subst(const struct attr_check *check)
 120 {
 121         return check && ATTR_TRUE(check->items[1].value);
 122 }
 123
 124 static int should_queue_directories(const struct archiver_args *args)
 125 {
 126         return args->pathspec.has_wildcard;
 127 }
 128
 129 static int write_archive_entry(const unsigned char *sha1, const char *base,
 130                 int baselen, const char *filename, unsigned mode, int stage,
 131                 void *context)
 132 {
 133         static struct strbuf path = STRBUF_INIT;
 134         struct archiver_context *c = context;
 135         struct archiver_args *args = c->args;
 136         write_archive_entry_fn_t write_entry = c->write_entry;
 137         int err;
 138         const char *path_without_prefix;
 139
 140         args->convert = 0;
 141         strbuf_reset(&path);
 142         strbuf_grow(&path, PATH_MAX);
 143         strbuf_add(&path, args->base, args->baselen);
 144         strbuf_add(&path, base, baselen);
 145         strbuf_addstr(&path, filename);
 146         if (S_ISDIR(mode) || S_ISGITLINK(mode))
 147                 strbuf_addch(&path, '/');
 148         path_without_prefix = path.buf + args->baselen;
 149
 150         if (!S_ISDIR(mode) || !should_queue_directories(args)) {
 151                 const struct attr_check *check;
 152                 check = get_archive_attrs(path_without_prefix);
 153                 if (check_attr_export_ignore(check))
 154                         return 0;
 155                 args->convert = check_attr_export_subst(check);
 156         }
 157
 158         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 159                 if (args->verbose)
 160                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 161                 err = write_entry(args, sha1, path.buf, path.len, mode);
 162                 if (err)
 163                         return err;
 164                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
 165         }
 166
 167         if (args->verbose)
 168                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 169         return write_entry(args, sha1, path.buf, path.len, mode);
 170 }
 171
 172 static int write_archive_entry_buf(const unsigned char *sha1, struct strbuf *base,
 173                 const char *filename, unsigned mode, int stage,
 174                 void *context)
 175 {
 176         return write_archive_entry(sha1, base->buf, base->len,
 177                                      filename, mode, stage, context);
 178 }
 179
 180 static void queue_directory(const unsigned char *sha1,
 181                 struct strbuf *base, const char *filename,
 182                 unsigned mode, int stage, struct archiver_context *c)
 183 {
 184         struct directory *d;
 185         size_t len = st_add4(base->len, 1, strlen(filename), 1);
 186         d = xmalloc(st_add(sizeof(*d), len));
 187         d->up      = c->bottom;
 188         d->baselen = base->len;
 189         d->mode    = mode;
 190         d->stage   = stage;
 191         c->bottom  = d;
 192         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
 193         hashcpy(d->oid.hash, sha1);
 194 }
 195
 196 static int write_directory(struct archiver_context *c)
 197 {
 198         struct directory *d = c->bottom;
 199         int ret;
 200
 201         if (!d)
 202                 return 0;
 203         c->bottom = d->up;
 204         d->path[d->len - 1] = '\0'; /* no trailing slash */
 205         ret =
 206                 write_directory(c) ||
 207                 write_archive_entry(d->oid.hash, d->path, d->baselen,
 208                                     d->path + d->baselen, d->mode,
 209                                     d->stage, c) != READ_TREE_RECURSIVE;
 210         free(d);
 211         return ret ? -1 : 0;
 212 }
 213
 214 static int queue_or_write_archive_entry(const unsigned char *sha1,
 215                 struct strbuf *base, const char *filename,
 216                 unsigned mode, int stage, void *context)
 217 {
 218         struct archiver_context *c = context;
 219
 220         while (c->bottom &&
 221                !(base->len >= c->bottom->len &&
 222                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
 223                 struct directory *next = c->bottom->up;
 224                 free(c->bottom);
 225                 c->bottom = next;
 226         }
 227
 228         if (S_ISDIR(mode)) {
 229                 size_t baselen = base->len;
 230                 const struct attr_check *check;
 231
 232                 /* Borrow base, but restore its original value when done. */
 233                 strbuf_addstr(base, filename);
 234                 strbuf_addch(base, '/');
 235                 check = get_archive_attrs(base->buf);
 236                 strbuf_setlen(base, baselen);
 237
 238                 if (check_attr_export_ignore(check))
 239                         return 0;
 240                 queue_directory(sha1, base, filename,
 241                                 mode, stage, c);
 242                 return READ_TREE_RECURSIVE;
 243         }
 244
 245         if (write_directory(c))
 246                 return -1;
 247         return write_archive_entry(sha1, base->buf, base->len, filename, mode,
 248                                    stage, context);
 249 }
 250
 251 int write_archive_entries(struct archiver_args *args,
 252                 write_archive_entry_fn_t write_entry)
 253 {
 254         struct archiver_context context;
 255         struct unpack_trees_options opts;
 256         struct tree_desc t;
 257         int err;
 258
 259         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
 260                 size_t len = args->baselen;
 261
 262                 while (len > 1 && args->base[len - 2] == '/')
 263                         len--;
 264                 if (args->verbose)
 265                         fprintf(stderr, "%.*s\n", (int)len, args->base);
 266                 err = write_entry(args, args->tree->object.oid.hash, args->base,
 267                                   len, 040777);
 268                 if (err)
 269                         return err;
 270         }
 271
 272         memset(&context, 0, sizeof(context));
 273         context.args = args;
 274         context.write_entry = write_entry;
 275
 276         /*
 277          * Setup index and instruct attr to read index only
 278          */
 279         if (!args->worktree_attributes) {
 280                 memset(&opts, 0, sizeof(opts));
 281                 opts.index_only = 1;
 282                 opts.head_idx = -1;
 283                 opts.src_index = &the_index;
 284                 opts.dst_index = &the_index;
 285                 opts.fn = oneway_merge;
 286                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
 287                 if (unpack_trees(1, &t, &opts))
 288                         return -1;
 289                 git_attr_set_direction(GIT_ATTR_INDEX, &the_index);
 290         }
 291
 292         err = read_tree_recursive(args->tree, "", 0, 0, &args->pathspec,
 293                                   should_queue_directories(args) ?
 294                                   queue_or_write_archive_entry :
 295                                   write_archive_entry_buf,
 296                                   &context);
 297         if (err == READ_TREE_RECURSIVE)
 298                 err = 0;
 299         while (context.bottom) {
 300                 struct directory *next = context.bottom->up;
 301                 free(context.bottom);
 302                 context.bottom = next;
 303         }
 304         return err;
 305 }
 306
 307 static const struct archiver *lookup_archiver(const char *name)
 308 {
 309         int i;
 310
 311         if (!name)
 312                 return NULL;
 313
 314         for (i = 0; i < nr_archivers; i++) {
 315                 if (!strcmp(name, archivers[i]->name))
 316                         return archivers[i];
 317         }
 318         return NULL;
 319 }
 320
 321 static int reject_entry(const unsigned char *sha1, struct strbuf *base,
 322                         const char *filename, unsigned mode,
 323                         int stage, void *context)
 324 {
 325         int ret = -1;
 326         if (S_ISDIR(mode)) {
 327                 struct strbuf sb = STRBUF_INIT;
 328                 strbuf_addbuf(&sb, base);
 329                 strbuf_addstr(&sb, filename);
 330                 if (!match_pathspec(context, sb.buf, sb.len, 0, NULL, 1))
 331                         ret = READ_TREE_RECURSIVE;
 332                 strbuf_release(&sb);
 333         }
 334         return ret;
 335 }
 336
 337 static int path_exists(struct tree *tree, const char *path)
 338 {
 339         const char *paths[] = { path, NULL };
 340         struct pathspec pathspec;
 341         int ret;
 342
 343         parse_pathspec(&pathspec, 0, 0, "", paths);
 344         pathspec.recursive = 1;
 345         ret = read_tree_recursive(tree, "", 0, 0, &pathspec,
 346                                   reject_entry, &pathspec);
 347         clear_pathspec(&pathspec);
 348         return ret != 0;
 349 }
 350
 351 static void parse_pathspec_arg(const char **pathspec,
 352                 struct archiver_args *ar_args)
 353 {
 354         /*
 355          * must be consistent with parse_pathspec in path_exists()
 356          * Also if pathspec patterns are dependent, we're in big
 357          * trouble as we test each one separately
 358          */
 359         parse_pathspec(&ar_args->pathspec, 0,
 360                        PATHSPEC_PREFER_FULL,
 361                        "", pathspec);
 362         ar_args->pathspec.recursive = 1;
 363         if (pathspec) {
 364                 while (*pathspec) {
 365                         if (**pathspec && !path_exists(ar_args->tree, *pathspec))
 366                                 die(_("pathspec '%s' did not match any files"), *pathspec);
 367                         pathspec++;
 368                 }
 369         }
 370 }
 371
 372 static void parse_treeish_arg(const char **argv,
 373                 struct archiver_args *ar_args, const char *prefix,
 374                 int remote)
 375 {
 376         const char *name = argv[0];
 377         const unsigned char *commit_sha1;
 378         time_t archive_time;
 379         struct tree *tree;
 380         const struct commit *commit;
 381         struct object_id oid;
 382
 383         /* Remotes are only allowed to fetch actual refs */
 384         if (remote && !remote_allow_unreachable) {
 385                 char *ref = NULL;
 386                 const char *colon = strchrnul(name, ':');
 387                 int refnamelen = colon - name;
 388
 389                 if (!dwim_ref(name, refnamelen, oid.hash, &ref))
 390                         die("no such ref: %.*s", refnamelen, name);
 391                 free(ref);
 392         }
 393
 394         if (get_oid(name, &oid))
 395                 die("Not a valid object name");
 396
 397         commit = lookup_commit_reference_gently(&oid, 1);
 398         if (commit) {
 399                 commit_sha1 = commit->object.oid.hash;
 400                 archive_time = commit->date;
 401         } else {
 402                 commit_sha1 = NULL;
 403                 archive_time = time(NULL);
 404         }
 405
 406         tree = parse_tree_indirect(&oid);
 407         if (tree == NULL)
 408                 die("not a tree object");
 409
 410         if (prefix) {
 411                 struct object_id tree_oid;
 412                 unsigned int mode;
 413                 int err;
 414
 415                 err = get_tree_entry(tree->object.oid.hash, prefix,
 416                                      tree_oid.hash, &mode);
 417                 if (err || !S_ISDIR(mode))
 418                         die("current working directory is untracked");
 419
 420                 tree = parse_tree_indirect(&tree_oid);
 421         }
 422         ar_args->tree = tree;
 423         ar_args->commit_sha1 = commit_sha1;
 424         ar_args->commit = commit;
 425         ar_args->time = archive_time;
 426 }
 427
 428 #define OPT__COMPR(s, v, h, p) \
 429         { OPTION_SET_INT, (s), NULL, (v), NULL, (h), \
 430           PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, (p) }
 431 #define OPT__COMPR_HIDDEN(s, v, p) \
 432         { OPTION_SET_INT, (s), NULL, (v), NULL, "", \
 433           PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
 434
 435 static int parse_archive_args(int argc, const char **argv,
 436                 const struct archiver **ar, struct archiver_args *args,
 437                 const char *name_hint, int is_remote)
 438 {
 439         const char *format = NULL;
 440         const char *base = NULL;
 441         const char *remote = NULL;
 442         const char *exec = NULL;
 443         const char *output = NULL;
 444         int compression_level = -1;
 445         int verbose = 0;
 446         int i;
 447         int list = 0;
 448         int worktree_attributes = 0;
 449         struct option opts[] = {
 450                 OPT_GROUP(""),
 451                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
 452                 OPT_STRING(0, "prefix", &base, N_("prefix"),
 453                         N_("prepend prefix to each pathname in the archive")),
 454                 OPT_STRING('o', "output", &output, N_("file"),
 455                         N_("write the archive to this file")),
 456                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
 457                         N_("read .gitattributes in working directory")),
 458                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
 459                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
 460                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
 461                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
 462                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
 463                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
 464                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
 465                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
 466                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
 467                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
 468                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
 469                 OPT_GROUP(""),
 470                 OPT_BOOL('l', "list", &list,
 471                         N_("list supported archive formats")),
 472                 OPT_GROUP(""),
 473                 OPT_STRING(0, "remote", &remote, N_("repo"),
 474                         N_("retrieve the archive from remote repository <repo>")),
 475                 OPT_STRING(0, "exec", &exec, N_("command"),
 476                         N_("path to the remote git-upload-archive command")),
 477                 OPT_END()
 478         };
 479
 480         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
 481
 482         if (remote)
 483                 die(_("Unexpected option --remote"));
 484         if (exec)
 485                 die(_("Option --exec can only be used together with --remote"));
 486         if (output)
 487                 die(_("Unexpected option --output"));
 488
 489         if (!base)
 490                 base = "";
 491
 492         if (list) {
 493                 for (i = 0; i < nr_archivers; i++)
 494                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
 495                                 printf("%s\n", archivers[i]->name);
 496                 exit(0);
 497         }
 498
 499         if (!format && name_hint)
 500                 format = archive_format_from_filename(name_hint);
 501         if (!format)
 502                 format = "tar";
 503
 504         /* We need at least one parameter -- tree-ish */
 505         if (argc < 1)
 506                 usage_with_options(archive_usage, opts);
 507         *ar = lookup_archiver(format);
 508         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
 509                 die(_("Unknown archive format '%s'"), format);
 510
 511         args->compression_level = Z_DEFAULT_COMPRESSION;
 512         if (compression_level != -1) {
 513                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
 514                         args->compression_level = compression_level;
 515                 else {
 516                         die(_("Argument not supported for format '%s': -%d"),
 517                                         format, compression_level);
 518                 }
 519         }
 520         args->verbose = verbose;
 521         args->base = base;
 522         args->baselen = strlen(base);
 523         args->worktree_attributes = worktree_attributes;
 524
 525         return argc;
 526 }
 527
 528 int write_archive(int argc, const char **argv, const char *prefix,
 529                   const char *name_hint, int remote)
 530 {
 531         const struct archiver *ar = NULL;
 532         struct archiver_args args;
 533
 534         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
 535         git_config(git_default_config, NULL);
 536
 537         init_tar_archiver();
 538         init_zip_archiver();
 539
 540         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
 541         if (!startup_info->have_repository) {
 542                 /*
 543                  * We know this will die() with an error, so we could just
 544                  * die ourselves; but its error message will be more specific
 545                  * than what we could write here.
 546                  */
 547                 setup_git_directory();
 548         }
 549
 550         parse_treeish_arg(argv, &args, prefix, remote);
 551         parse_pathspec_arg(argv + 1, &args);
 552
 553         return ar->write_archive(ar, &args);
 554 }
 555
 556 static int match_extension(const char *filename, const char *ext)
 557 {
 558         int prefixlen = strlen(filename) - strlen(ext);
 559
 560         /*
 561          * We need 1 character for the '.', and 1 character to ensure that the
 562          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
 563          * filename).
 564          */
 565         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
 566                 return 0;
 567         return !strcmp(filename + prefixlen, ext);
 568 }
 569
 570 const char *archive_format_from_filename(const char *filename)
 571 {
 572         int i;
 573
 574         for (i = 0; i < nr_archivers; i++)
 575                 if (match_extension(filename, archivers[i]->name))
 576                         return archivers[i]->name;
 577         return NULL;
 578 }