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 int write_archive_entry(const unsigned char *sha1, const char *base,
 107                 int baselen, const char *filename, unsigned mode, int stage,
 108                 void *context)
 109 {
 110         static struct strbuf path = STRBUF_INIT;
 111         static struct attr_check *check;
 112         struct archiver_context *c = context;
 113         struct archiver_args *args = c->args;
 114         write_archive_entry_fn_t write_entry = c->write_entry;
 115         const char *path_without_prefix;
 116         int err;
 117
 118         args->convert = 0;
 119         strbuf_reset(&path);
 120         strbuf_grow(&path, PATH_MAX);
 121         strbuf_add(&path, args->base, args->baselen);
 122         strbuf_add(&path, base, baselen);
 123         strbuf_addstr(&path, filename);
 124         if (S_ISDIR(mode) || S_ISGITLINK(mode))
 125                 strbuf_addch(&path, '/');
 126         path_without_prefix = path.buf + args->baselen;
 127
 128         if (!check)
 129                 check = attr_check_initl("export-ignore", "export-subst", NULL);
 130         if (!git_check_attr(path_without_prefix, check)) {
 131                 if (ATTR_TRUE(check->items[0].value))
 132                         return 0;
 133                 args->convert = ATTR_TRUE(check->items[1].value);
 134         }
 135
 136         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 137                 if (args->verbose)
 138                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 139                 err = write_entry(args, sha1, path.buf, path.len, mode);
 140                 if (err)
 141                         return err;
 142                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
 143         }
 144
 145         if (args->verbose)
 146                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 147         return write_entry(args, sha1, path.buf, path.len, mode);
 148 }
 149
 150 static int write_archive_entry_buf(const unsigned char *sha1, struct strbuf *base,
 151                 const char *filename, unsigned mode, int stage,
 152                 void *context)
 153 {
 154         return write_archive_entry(sha1, base->buf, base->len,
 155                                      filename, mode, stage, context);
 156 }
 157
 158 static void queue_directory(const unsigned char *sha1,
 159                 struct strbuf *base, const char *filename,
 160                 unsigned mode, int stage, struct archiver_context *c)
 161 {
 162         struct directory *d;
 163         size_t len = st_add4(base->len, 1, strlen(filename), 1);
 164         d = xmalloc(st_add(sizeof(*d), len));
 165         d->up      = c->bottom;
 166         d->baselen = base->len;
 167         d->mode    = mode;
 168         d->stage   = stage;
 169         c->bottom  = d;
 170         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
 171         hashcpy(d->oid.hash, sha1);
 172 }
 173
 174 static int write_directory(struct archiver_context *c)
 175 {
 176         struct directory *d = c->bottom;
 177         int ret;
 178
 179         if (!d)
 180                 return 0;
 181         c->bottom = d->up;
 182         d->path[d->len - 1] = '\0'; /* no trailing slash */
 183         ret =
 184                 write_directory(c) ||
 185                 write_archive_entry(d->oid.hash, d->path, d->baselen,
 186                                     d->path + d->baselen, d->mode,
 187                                     d->stage, c) != READ_TREE_RECURSIVE;
 188         free(d);
 189         return ret ? -1 : 0;
 190 }
 191
 192 static int queue_or_write_archive_entry(const unsigned char *sha1,
 193                 struct strbuf *base, const char *filename,
 194                 unsigned mode, int stage, void *context)
 195 {
 196         struct archiver_context *c = context;
 197
 198         while (c->bottom &&
 199                !(base->len >= c->bottom->len &&
 200                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
 201                 struct directory *next = c->bottom->up;
 202                 free(c->bottom);
 203                 c->bottom = next;
 204         }
 205
 206         if (S_ISDIR(mode)) {
 207                 queue_directory(sha1, base, filename,
 208                                 mode, stage, c);
 209                 return READ_TREE_RECURSIVE;
 210         }
 211
 212         if (write_directory(c))
 213                 return -1;
 214         return write_archive_entry(sha1, base->buf, base->len, filename, mode,
 215                                    stage, context);
 216 }
 217
 218 int write_archive_entries(struct archiver_args *args,
 219                 write_archive_entry_fn_t write_entry)
 220 {
 221         struct archiver_context context;
 222         struct unpack_trees_options opts;
 223         struct tree_desc t;
 224         int err;
 225
 226         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
 227                 size_t len = args->baselen;
 228
 229                 while (len > 1 && args->base[len - 2] == '/')
 230                         len--;
 231                 if (args->verbose)
 232                         fprintf(stderr, "%.*s\n", (int)len, args->base);
 233                 err = write_entry(args, args->tree->object.oid.hash, args->base,
 234                                   len, 040777);
 235                 if (err)
 236                         return err;
 237         }
 238
 239         memset(&context, 0, sizeof(context));
 240         context.args = args;
 241         context.write_entry = write_entry;
 242
 243         /*
 244          * Setup index and instruct attr to read index only
 245          */
 246         if (!args->worktree_attributes) {
 247                 memset(&opts, 0, sizeof(opts));
 248                 opts.index_only = 1;
 249                 opts.head_idx = -1;
 250                 opts.src_index = &the_index;
 251                 opts.dst_index = &the_index;
 252                 opts.fn = oneway_merge;
 253                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
 254                 if (unpack_trees(1, &t, &opts))
 255                         return -1;
 256                 git_attr_set_direction(GIT_ATTR_INDEX, &the_index);
 257         }
 258
 259         err = read_tree_recursive(args->tree, "", 0, 0, &args->pathspec,
 260                                   args->pathspec.has_wildcard ?
 261                                   queue_or_write_archive_entry :
 262                                   write_archive_entry_buf,
 263                                   &context);
 264         if (err == READ_TREE_RECURSIVE)
 265                 err = 0;
 266         while (context.bottom) {
 267                 struct directory *next = context.bottom->up;
 268                 free(context.bottom);
 269                 context.bottom = next;
 270         }
 271         return err;
 272 }
 273
 274 static const struct archiver *lookup_archiver(const char *name)
 275 {
 276         int i;
 277
 278         if (!name)
 279                 return NULL;
 280
 281         for (i = 0; i < nr_archivers; i++) {
 282                 if (!strcmp(name, archivers[i]->name))
 283                         return archivers[i];
 284         }
 285         return NULL;
 286 }
 287
 288 static int reject_entry(const unsigned char *sha1, struct strbuf *base,
 289                         const char *filename, unsigned mode,
 290                         int stage, void *context)
 291 {
 292         int ret = -1;
 293         if (S_ISDIR(mode)) {
 294                 struct strbuf sb = STRBUF_INIT;
 295                 strbuf_addbuf(&sb, base);
 296                 strbuf_addstr(&sb, filename);
 297                 if (!match_pathspec(context, sb.buf, sb.len, 0, NULL, 1))
 298                         ret = READ_TREE_RECURSIVE;
 299                 strbuf_release(&sb);
 300         }
 301         return ret;
 302 }
 303
 304 static int path_exists(struct tree *tree, const char *path)
 305 {
 306         const char *paths[] = { path, NULL };
 307         struct pathspec pathspec;
 308         int ret;
 309
 310         parse_pathspec(&pathspec, 0, 0, "", paths);
 311         pathspec.recursive = 1;
 312         ret = read_tree_recursive(tree, "", 0, 0, &pathspec,
 313                                   reject_entry, &pathspec);
 314         clear_pathspec(&pathspec);
 315         return ret != 0;
 316 }
 317
 318 static void parse_pathspec_arg(const char **pathspec,
 319                 struct archiver_args *ar_args)
 320 {
 321         /*
 322          * must be consistent with parse_pathspec in path_exists()
 323          * Also if pathspec patterns are dependent, we're in big
 324          * trouble as we test each one separately
 325          */
 326         parse_pathspec(&ar_args->pathspec, 0,
 327                        PATHSPEC_PREFER_FULL,
 328                        "", pathspec);
 329         ar_args->pathspec.recursive = 1;
 330         if (pathspec) {
 331                 while (*pathspec) {
 332                         if (**pathspec && !path_exists(ar_args->tree, *pathspec))
 333                                 die(_("pathspec '%s' did not match any files"), *pathspec);
 334                         pathspec++;
 335                 }
 336         }
 337 }
 338
 339 static void parse_treeish_arg(const char **argv,
 340                 struct archiver_args *ar_args, const char *prefix,
 341                 int remote)
 342 {
 343         const char *name = argv[0];
 344         const unsigned char *commit_sha1;
 345         time_t archive_time;
 346         struct tree *tree;
 347         const struct commit *commit;
 348         struct object_id oid;
 349
 350         /* Remotes are only allowed to fetch actual refs */
 351         if (remote && !remote_allow_unreachable) {
 352                 char *ref = NULL;
 353                 const char *colon = strchrnul(name, ':');
 354                 int refnamelen = colon - name;
 355
 356                 if (!dwim_ref(name, refnamelen, oid.hash, &ref))
 357                         die("no such ref: %.*s", refnamelen, name);
 358                 free(ref);
 359         }
 360
 361         if (get_sha1(name, oid.hash))
 362                 die("Not a valid object name");
 363
 364         commit = lookup_commit_reference_gently(&oid, 1);
 365         if (commit) {
 366                 commit_sha1 = commit->object.oid.hash;
 367                 archive_time = commit->date;
 368         } else {
 369                 commit_sha1 = NULL;
 370                 archive_time = time(NULL);
 371         }
 372
 373         tree = parse_tree_indirect(&oid);
 374         if (tree == NULL)
 375                 die("not a tree object");
 376
 377         if (prefix) {
 378                 struct object_id tree_oid;
 379                 unsigned int mode;
 380                 int err;
 381
 382                 err = get_tree_entry(tree->object.oid.hash, prefix,
 383                                      tree_oid.hash, &mode);
 384                 if (err || !S_ISDIR(mode))
 385                         die("current working directory is untracked");
 386
 387                 tree = parse_tree_indirect(&tree_oid);
 388         }
 389         ar_args->tree = tree;
 390         ar_args->commit_sha1 = commit_sha1;
 391         ar_args->commit = commit;
 392         ar_args->time = archive_time;
 393 }
 394
 395 #define OPT__COMPR(s, v, h, p) \
 396         { OPTION_SET_INT, (s), NULL, (v), NULL, (h), \
 397           PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, (p) }
 398 #define OPT__COMPR_HIDDEN(s, v, p) \
 399         { OPTION_SET_INT, (s), NULL, (v), NULL, "", \
 400           PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
 401
 402 static int parse_archive_args(int argc, const char **argv,
 403                 const struct archiver **ar, struct archiver_args *args,
 404                 const char *name_hint, int is_remote)
 405 {
 406         const char *format = NULL;
 407         const char *base = NULL;
 408         const char *remote = NULL;
 409         const char *exec = NULL;
 410         const char *output = NULL;
 411         int compression_level = -1;
 412         int verbose = 0;
 413         int i;
 414         int list = 0;
 415         int worktree_attributes = 0;
 416         struct option opts[] = {
 417                 OPT_GROUP(""),
 418                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
 419                 OPT_STRING(0, "prefix", &base, N_("prefix"),
 420                         N_("prepend prefix to each pathname in the archive")),
 421                 OPT_STRING('o', "output", &output, N_("file"),
 422                         N_("write the archive to this file")),
 423                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
 424                         N_("read .gitattributes in working directory")),
 425                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
 426                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
 427                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
 428                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
 429                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
 430                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
 431                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
 432                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
 433                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
 434                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
 435                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
 436                 OPT_GROUP(""),
 437                 OPT_BOOL('l', "list", &list,
 438                         N_("list supported archive formats")),
 439                 OPT_GROUP(""),
 440                 OPT_STRING(0, "remote", &remote, N_("repo"),
 441                         N_("retrieve the archive from remote repository <repo>")),
 442                 OPT_STRING(0, "exec", &exec, N_("command"),
 443                         N_("path to the remote git-upload-archive command")),
 444                 OPT_END()
 445         };
 446
 447         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
 448
 449         if (remote)
 450                 die(_("Unexpected option --remote"));
 451         if (exec)
 452                 die(_("Option --exec can only be used together with --remote"));
 453         if (output)
 454                 die(_("Unexpected option --output"));
 455
 456         if (!base)
 457                 base = "";
 458
 459         if (list) {
 460                 for (i = 0; i < nr_archivers; i++)
 461                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
 462                                 printf("%s\n", archivers[i]->name);
 463                 exit(0);
 464         }
 465
 466         if (!format && name_hint)
 467                 format = archive_format_from_filename(name_hint);
 468         if (!format)
 469                 format = "tar";
 470
 471         /* We need at least one parameter -- tree-ish */
 472         if (argc < 1)
 473                 usage_with_options(archive_usage, opts);
 474         *ar = lookup_archiver(format);
 475         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
 476                 die(_("Unknown archive format '%s'"), format);
 477
 478         args->compression_level = Z_DEFAULT_COMPRESSION;
 479         if (compression_level != -1) {
 480                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
 481                         args->compression_level = compression_level;
 482                 else {
 483                         die(_("Argument not supported for format '%s': -%d"),
 484                                         format, compression_level);
 485                 }
 486         }
 487         args->verbose = verbose;
 488         args->base = base;
 489         args->baselen = strlen(base);
 490         args->worktree_attributes = worktree_attributes;
 491
 492         return argc;
 493 }
 494
 495 int write_archive(int argc, const char **argv, const char *prefix,
 496                   const char *name_hint, int remote)
 497 {
 498         const struct archiver *ar = NULL;
 499         struct archiver_args args;
 500
 501         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
 502         git_config(git_default_config, NULL);
 503
 504         init_tar_archiver();
 505         init_zip_archiver();
 506
 507         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
 508         if (!startup_info->have_repository) {
 509                 /*
 510                  * We know this will die() with an error, so we could just
 511                  * die ourselves; but its error message will be more specific
 512                  * than what we could write here.
 513                  */
 514                 setup_git_directory();
 515         }
 516
 517         parse_treeish_arg(argv, &args, prefix, remote);
 518         parse_pathspec_arg(argv + 1, &args);
 519
 520         return ar->write_archive(ar, &args);
 521 }
 522
 523 static int match_extension(const char *filename, const char *ext)
 524 {
 525         int prefixlen = strlen(filename) - strlen(ext);
 526
 527         /*
 528          * We need 1 character for the '.', and 1 character to ensure that the
 529          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
 530          * filename).
 531          */
 532         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
 533                 return 0;
 534         return !strcmp(filename + prefixlen, ext);
 535 }
 536
 537 const char *archive_format_from_filename(const char *filename)
 538 {
 539         int i;
 540
 541         for (i = 0; i < nr_archivers; i++)
 542                 if (match_extension(filename, archivers[i]->name))
 543                         return archivers[i]->name;
 544         return NULL;
 545 }