archive.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "tree-walk.h"
   4 #include "attr.h"
   5 #include "archive.h"
   6 #include "parse-options.h"
   7 #include "unpack-trees.h"
   8 #include "dir.h"
   9
  10 static char const * const archive_usage[] = {
  11         N_("git archive [options] <tree-ish> [<path>...]"),
  12         N_("git archive --list"),
  13         N_("git archive --remote <repo> [--exec <cmd>] [options] <tree-ish> [<path>...]"),
  14         N_("git archive --remote <repo> [--exec <cmd>] --list"),
  15         NULL
  16 };
  17
  18 static const struct archiver **archivers;
  19 static int nr_archivers;
  20 static int alloc_archivers;
  21 static int remote_allow_unreachable;
  22
  23 void register_archiver(struct archiver *ar)
  24 {
  25         ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
  26         archivers[nr_archivers++] = ar;
  27 }
  28
  29 static void format_subst(const struct commit *commit,
  30                          const char *src, size_t len,
  31                          struct strbuf *buf)
  32 {
  33         char *to_free = NULL;
  34         struct strbuf fmt = STRBUF_INIT;
  35         struct pretty_print_context ctx = {0};
  36         ctx.date_mode = DATE_NORMAL;
  37         ctx.abbrev = DEFAULT_ABBREV;
  38
  39         if (src == buf->buf)
  40                 to_free = strbuf_detach(buf, NULL);
  41         for (;;) {
  42                 const char *b, *c;
  43
  44                 b = memmem(src, len, "$Format:", 8);
  45                 if (!b)
  46                         break;
  47                 c = memchr(b + 8, '$', (src + len) - b - 8);
  48                 if (!c)
  49                         break;
  50
  51                 strbuf_reset(&fmt);
  52                 strbuf_add(&fmt, b + 8, c - b - 8);
  53
  54                 strbuf_add(buf, src, b - src);
  55                 format_commit_message(commit, fmt.buf, buf, &ctx);
  56                 len -= c + 1 - src;
  57                 src  = c + 1;
  58         }
  59         strbuf_add(buf, src, len);
  60         strbuf_release(&fmt);
  61         free(to_free);
  62 }
  63
  64 void *sha1_file_to_archive(const struct archiver_args *args,
  65                            const char *path, const unsigned char *sha1,
  66                            unsigned int mode, enum object_type *type,
  67                            unsigned long *sizep)
  68 {
  69         void *buffer;
  70         const struct commit *commit = args->convert ? args->commit : NULL;
  71
  72         path += args->baselen;
  73         buffer = read_sha1_file(sha1, type, sizep);
  74         if (buffer && S_ISREG(mode)) {
  75                 struct strbuf buf = STRBUF_INIT;
  76                 size_t size = 0;
  77
  78                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
  79                 convert_to_working_tree(path, buf.buf, buf.len, &buf);
  80                 if (commit)
  81                         format_subst(commit, buf.buf, buf.len, &buf);
  82                 buffer = strbuf_detach(&buf, &size);
  83                 *sizep = size;
  84         }
  85
  86         return buffer;
  87 }
  88
  89 static void setup_archive_check(struct git_attr_check *check)
  90 {
  91         static struct git_attr *attr_export_ignore;
  92         static struct git_attr *attr_export_subst;
  93
  94         if (!attr_export_ignore) {
  95                 attr_export_ignore = git_attr("export-ignore");
  96                 attr_export_subst = git_attr("export-subst");
  97         }
  98         check[0].attr = attr_export_ignore;
  99         check[1].attr = attr_export_subst;
 100 }
 101
 102 struct directory {
 103         struct directory *up;
 104         unsigned char sha1[20];
 105         int baselen, len;
 106         unsigned mode;
 107         int stage;
 108         char path[FLEX_ARRAY];
 109 };
 110
 111 struct archiver_context {
 112         struct archiver_args *args;
 113         write_archive_entry_fn_t write_entry;
 114         struct directory *bottom;
 115 };
 116
 117 static int write_archive_entry(const unsigned char *sha1, const char *base,
 118                 int baselen, const char *filename, unsigned mode, int stage,
 119                 void *context)
 120 {
 121         static struct strbuf path = STRBUF_INIT;
 122         struct archiver_context *c = context;
 123         struct archiver_args *args = c->args;
 124         write_archive_entry_fn_t write_entry = c->write_entry;
 125         struct git_attr_check check[2];
 126         const char *path_without_prefix;
 127         int err;
 128
 129         args->convert = 0;
 130         strbuf_reset(&path);
 131         strbuf_grow(&path, PATH_MAX);
 132         strbuf_add(&path, args->base, args->baselen);
 133         strbuf_add(&path, base, baselen);
 134         strbuf_addstr(&path, filename);
 135         if (S_ISDIR(mode) || S_ISGITLINK(mode))
 136                 strbuf_addch(&path, '/');
 137         path_without_prefix = path.buf + args->baselen;
 138
 139         setup_archive_check(check);
 140         if (!git_check_attr(path_without_prefix, ARRAY_SIZE(check), check)) {
 141                 if (ATTR_TRUE(check[0].value))
 142                         return 0;
 143                 args->convert = ATTR_TRUE(check[1].value);
 144         }
 145
 146         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 147                 if (args->verbose)
 148                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 149                 err = write_entry(args, sha1, path.buf, path.len, mode);
 150                 if (err)
 151                         return err;
 152                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
 153         }
 154
 155         if (args->verbose)
 156                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 157         return write_entry(args, sha1, path.buf, path.len, mode);
 158 }
 159
 160 static void queue_directory(const unsigned char *sha1,
 161                 const char *base, int baselen, const char *filename,
 162                 unsigned mode, int stage, struct archiver_context *c)
 163 {
 164         struct directory *d;
 165         d = xmallocz(sizeof(*d) + baselen + 1 + strlen(filename));
 166         d->up      = c->bottom;
 167         d->baselen = baselen;
 168         d->mode    = mode;
 169         d->stage   = stage;
 170         c->bottom  = d;
 171         d->len = sprintf(d->path, "%.*s%s/", baselen, base, filename);
 172         hashcpy(d->sha1, sha1);
 173 }
 174
 175 static int write_directory(struct archiver_context *c)
 176 {
 177         struct directory *d = c->bottom;
 178         int ret;
 179
 180         if (!d)
 181                 return 0;
 182         c->bottom = d->up;
 183         d->path[d->len - 1] = '\0'; /* no trailing slash */
 184         ret =
 185                 write_directory(c) ||
 186                 write_archive_entry(d->sha1, d->path, d->baselen,
 187                                     d->path + d->baselen, d->mode,
 188                                     d->stage, c) != READ_TREE_RECURSIVE;
 189         free(d);
 190         return ret ? -1 : 0;
 191 }
 192
 193 static int queue_or_write_archive_entry(const unsigned char *sha1,
 194                 const char *base, int baselen, const char *filename,
 195                 unsigned mode, int stage, void *context)
 196 {
 197         struct archiver_context *c = context;
 198
 199         while (c->bottom &&
 200                !(baselen >= c->bottom->len &&
 201                  !strncmp(base, c->bottom->path, c->bottom->len))) {
 202                 struct directory *next = c->bottom->up;
 203                 free(c->bottom);
 204                 c->bottom = next;
 205         }
 206
 207         if (S_ISDIR(mode)) {
 208                 queue_directory(sha1, base, baselen, filename,
 209                                 mode, stage, c);
 210                 return READ_TREE_RECURSIVE;
 211         }
 212
 213         if (write_directory(c))
 214                 return -1;
 215         return write_archive_entry(sha1, base, baselen, filename, mode,
 216                                    stage, context);
 217 }
 218
 219 int write_archive_entries(struct archiver_args *args,
 220                 write_archive_entry_fn_t write_entry)
 221 {
 222         struct archiver_context context;
 223         struct unpack_trees_options opts;
 224         struct tree_desc t;
 225         int err;
 226
 227         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
 228                 size_t len = args->baselen;
 229
 230                 while (len > 1 && args->base[len - 2] == '/')
 231                         len--;
 232                 if (args->verbose)
 233                         fprintf(stderr, "%.*s\n", (int)len, args->base);
 234                 err = write_entry(args, args->tree->object.sha1, args->base,
 235                                   len, 040777);
 236                 if (err)
 237                         return err;
 238         }
 239
 240         memset(&context, 0, sizeof(context));
 241         context.args = args;
 242         context.write_entry = write_entry;
 243
 244         /*
 245          * Setup index and instruct attr to read index only
 246          */
 247         if (!args->worktree_attributes) {
 248                 memset(&opts, 0, sizeof(opts));
 249                 opts.index_only = 1;
 250                 opts.head_idx = -1;
 251                 opts.src_index = &the_index;
 252                 opts.dst_index = &the_index;
 253                 opts.fn = oneway_merge;
 254                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
 255                 if (unpack_trees(1, &t, &opts))
 256                         return -1;
 257                 git_attr_set_direction(GIT_ATTR_INDEX, &the_index);
 258         }
 259
 260         err = read_tree_recursive(args->tree, "", 0, 0, &args->pathspec,
 261                                   args->pathspec.has_wildcard ?
 262                                   queue_or_write_archive_entry :
 263                                   write_archive_entry,
 264                                   &context);
 265         if (err == READ_TREE_RECURSIVE)
 266                 err = 0;
 267         while (context.bottom) {
 268                 struct directory *next = context.bottom->up;
 269                 free(context.bottom);
 270                 context.bottom = next;
 271         }
 272         return err;
 273 }
 274
 275 static const struct archiver *lookup_archiver(const char *name)
 276 {
 277         int i;
 278
 279         if (!name)
 280                 return NULL;
 281
 282         for (i = 0; i < nr_archivers; i++) {
 283                 if (!strcmp(name, archivers[i]->name))
 284                         return archivers[i];
 285         }
 286         return NULL;
 287 }
 288
 289 static int reject_entry(const unsigned char *sha1, const char *base,
 290                         int baselen, const char *filename, unsigned mode,
 291                         int stage, void *context)
 292 {
 293         int ret = -1;
 294         if (S_ISDIR(mode)) {
 295                 struct strbuf sb = STRBUF_INIT;
 296                 strbuf_addstr(&sb, base);
 297                 strbuf_addstr(&sb, filename);
 298                 if (!match_pathspec(context, sb.buf, sb.len, 0, NULL, 1))
 299                         ret = READ_TREE_RECURSIVE;
 300                 strbuf_release(&sb);
 301         }
 302         return ret;
 303 }
 304
 305 static int path_exists(struct tree *tree, const char *path)
 306 {
 307         const char *paths[] = { path, NULL };
 308         struct pathspec pathspec;
 309         int ret;
 310
 311         parse_pathspec(&pathspec, 0, 0, "", paths);
 312         pathspec.recursive = 1;
 313         ret = read_tree_recursive(tree, "", 0, 0, &pathspec,
 314                                   reject_entry, &pathspec);
 315         free_pathspec(&pathspec);
 316         return ret != 0;
 317 }
 318
 319 static void parse_pathspec_arg(const char **pathspec,
 320                 struct archiver_args *ar_args)
 321 {
 322         /*
 323          * must be consistent with parse_pathspec in path_exists()
 324          * Also if pathspec patterns are dependent, we're in big
 325          * trouble as we test each one separately
 326          */
 327         parse_pathspec(&ar_args->pathspec, 0,
 328                        PATHSPEC_PREFER_FULL,
 329                        "", pathspec);
 330         ar_args->pathspec.recursive = 1;
 331         if (pathspec) {
 332                 while (*pathspec) {
 333                         if (**pathspec && !path_exists(ar_args->tree, *pathspec))
 334                                 die(_("pathspec '%s' did not match any files"), *pathspec);
 335                         pathspec++;
 336                 }
 337         }
 338 }
 339
 340 static void parse_treeish_arg(const char **argv,
 341                 struct archiver_args *ar_args, const char *prefix,
 342                 int remote)
 343 {
 344         const char *name = argv[0];
 345         const unsigned char *commit_sha1;
 346         time_t archive_time;
 347         struct tree *tree;
 348         const struct commit *commit;
 349         unsigned char sha1[20];
 350
 351         /* Remotes are only allowed to fetch actual refs */
 352         if (remote && !remote_allow_unreachable) {
 353                 char *ref = NULL;
 354                 const char *colon = strchrnul(name, ':');
 355                 int refnamelen = colon - name;
 356
 357                 if (!dwim_ref(name, refnamelen, sha1, &ref))
 358                         die("no such ref: %.*s", refnamelen, name);
 359                 free(ref);
 360         }
 361
 362         if (get_sha1(name, sha1))
 363                 die("Not a valid object name");
 364
 365         commit = lookup_commit_reference_gently(sha1, 1);
 366         if (commit) {
 367                 commit_sha1 = commit->object.sha1;
 368                 archive_time = commit->date;
 369         } else {
 370                 commit_sha1 = NULL;
 371                 archive_time = time(NULL);
 372         }
 373
 374         tree = parse_tree_indirect(sha1);
 375         if (tree == NULL)
 376                 die("not a tree object");
 377
 378         if (prefix) {
 379                 unsigned char tree_sha1[20];
 380                 unsigned int mode;
 381                 int err;
 382
 383                 err = get_tree_entry(tree->object.sha1, prefix,
 384                                      tree_sha1, &mode);
 385                 if (err || !S_ISDIR(mode))
 386                         die("current working directory is untracked");
 387
 388                 tree = parse_tree_indirect(tree_sha1);
 389         }
 390         ar_args->tree = tree;
 391         ar_args->commit_sha1 = commit_sha1;
 392         ar_args->commit = commit;
 393         ar_args->time = archive_time;
 394 }
 395
 396 #define OPT__COMPR(s, v, h, p) \
 397         { OPTION_SET_INT, (s), NULL, (v), NULL, (h), \
 398           PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, (p) }
 399 #define OPT__COMPR_HIDDEN(s, v, p) \
 400         { OPTION_SET_INT, (s), NULL, (v), NULL, "", \
 401           PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
 402
 403 static int parse_archive_args(int argc, const char **argv,
 404                 const struct archiver **ar, struct archiver_args *args,
 405                 const char *name_hint, int is_remote)
 406 {
 407         const char *format = NULL;
 408         const char *base = NULL;
 409         const char *remote = NULL;
 410         const char *exec = NULL;
 411         const char *output = NULL;
 412         int compression_level = -1;
 413         int verbose = 0;
 414         int i;
 415         int list = 0;
 416         int worktree_attributes = 0;
 417         struct option opts[] = {
 418                 OPT_GROUP(""),
 419                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
 420                 OPT_STRING(0, "prefix", &base, N_("prefix"),
 421                         N_("prepend prefix to each pathname in the archive")),
 422                 OPT_STRING('o', "output", &output, N_("file"),
 423                         N_("write the archive to this file")),
 424                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
 425                         N_("read .gitattributes in working directory")),
 426                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
 427                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
 428                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
 429                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
 430                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
 431                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
 432                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
 433                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
 434                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
 435                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
 436                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
 437                 OPT_GROUP(""),
 438                 OPT_BOOL('l', "list", &list,
 439                         N_("list supported archive formats")),
 440                 OPT_GROUP(""),
 441                 OPT_STRING(0, "remote", &remote, N_("repo"),
 442                         N_("retrieve the archive from remote repository <repo>")),
 443                 OPT_STRING(0, "exec", &exec, N_("command"),
 444                         N_("path to the remote git-upload-archive command")),
 445                 OPT_END()
 446         };
 447
 448         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
 449
 450         if (remote)
 451                 die("Unexpected option --remote");
 452         if (exec)
 453                 die("Option --exec can only be used together with --remote");
 454         if (output)
 455                 die("Unexpected option --output");
 456
 457         if (!base)
 458                 base = "";
 459
 460         if (list) {
 461                 for (i = 0; i < nr_archivers; i++)
 462                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
 463                                 printf("%s\n", archivers[i]->name);
 464                 exit(0);
 465         }
 466
 467         if (!format && name_hint)
 468                 format = archive_format_from_filename(name_hint);
 469         if (!format)
 470                 format = "tar";
 471
 472         /* We need at least one parameter -- tree-ish */
 473         if (argc < 1)
 474                 usage_with_options(archive_usage, opts);
 475         *ar = lookup_archiver(format);
 476         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
 477                 die("Unknown archive format '%s'", format);
 478
 479         args->compression_level = Z_DEFAULT_COMPRESSION;
 480         if (compression_level != -1) {
 481                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
 482                         args->compression_level = compression_level;
 483                 else {
 484                         die("Argument not supported for format '%s': -%d",
 485                                         format, compression_level);
 486                 }
 487         }
 488         args->verbose = verbose;
 489         args->base = base;
 490         args->baselen = strlen(base);
 491         args->worktree_attributes = worktree_attributes;
 492
 493         return argc;
 494 }
 495
 496 int write_archive(int argc, const char **argv, const char *prefix,
 497                   int setup_prefix, const char *name_hint, int remote)
 498 {
 499         int nongit = 0;
 500         const struct archiver *ar = NULL;
 501         struct archiver_args args;
 502
 503         if (setup_prefix && prefix == NULL)
 504                 prefix = setup_git_directory_gently(&nongit);
 505
 506         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
 507         git_config(git_default_config, NULL);
 508
 509         init_tar_archiver();
 510         init_zip_archiver();
 511
 512         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
 513         if (nongit) {
 514                 /*
 515                  * We know this will die() with an error, so we could just
 516                  * die ourselves; but its error message will be more specific
 517                  * than what we could write here.
 518                  */
 519                 setup_git_directory();
 520         }
 521
 522         parse_treeish_arg(argv, &args, prefix, remote);
 523         parse_pathspec_arg(argv + 1, &args);
 524
 525         return ar->write_archive(ar, &args);
 526 }
 527
 528 static int match_extension(const char *filename, const char *ext)
 529 {
 530         int prefixlen = strlen(filename) - strlen(ext);
 531
 532         /*
 533          * We need 1 character for the '.', and 1 character to ensure that the
 534          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
 535          * filename).
 536          */
 537         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
 538                 return 0;
 539         return !strcmp(filename + prefixlen, ext);
 540 }
 541
 542 const char *archive_format_from_filename(const char *filename)
 543 {
 544         int i;
 545
 546         for (i = 0; i < nr_archivers; i++)
 547                 if (match_extension(filename, archivers[i]->name))
 548                         return archivers[i]->name;
 549         return NULL;
 550 }