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
   9 static char const * const archive_usage[] = {
  10         "git archive [options] <tree-ish> [<path>...]",
  11         "git archive --list",
  12         "git archive --remote <repo> [--exec <cmd>] [options] <tree-ish> [<path>...]",
  13         "git archive --remote <repo> [--exec <cmd>] --list",
  14         NULL
  15 };
  16
  17 static const struct archiver **archivers;
  18 static int nr_archivers;
  19 static int alloc_archivers;
  20
  21 void register_archiver(struct archiver *ar)
  22 {
  23         ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
  24         archivers[nr_archivers++] = ar;
  25 }
  26
  27 static void format_subst(const struct commit *commit,
  28                          const char *src, size_t len,
  29                          struct strbuf *buf)
  30 {
  31         char *to_free = NULL;
  32         struct strbuf fmt = STRBUF_INIT;
  33         struct pretty_print_context ctx = {0};
  34         ctx.date_mode = DATE_NORMAL;
  35         ctx.abbrev = DEFAULT_ABBREV;
  36
  37         if (src == buf->buf)
  38                 to_free = strbuf_detach(buf, NULL);
  39         for (;;) {
  40                 const char *b, *c;
  41
  42                 b = memmem(src, len, "$Format:", 8);
  43                 if (!b)
  44                         break;
  45                 c = memchr(b + 8, '$', (src + len) - b - 8);
  46                 if (!c)
  47                         break;
  48
  49                 strbuf_reset(&fmt);
  50                 strbuf_add(&fmt, b + 8, c - b - 8);
  51
  52                 strbuf_add(buf, src, b - src);
  53                 format_commit_message(commit, fmt.buf, buf, &ctx);
  54                 len -= c + 1 - src;
  55                 src  = c + 1;
  56         }
  57         strbuf_add(buf, src, len);
  58         strbuf_release(&fmt);
  59         free(to_free);
  60 }
  61
  62 static void *sha1_file_to_archive(const char *path, const unsigned char *sha1,
  63                 unsigned int mode, enum object_type *type,
  64                 unsigned long *sizep, const struct commit *commit)
  65 {
  66         void *buffer;
  67
  68         buffer = read_sha1_file(sha1, type, sizep);
  69         if (buffer && S_ISREG(mode)) {
  70                 struct strbuf buf = STRBUF_INIT;
  71                 size_t size = 0;
  72
  73                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
  74                 convert_to_working_tree(path, buf.buf, buf.len, &buf);
  75                 if (commit)
  76                         format_subst(commit, buf.buf, buf.len, &buf);
  77                 buffer = strbuf_detach(&buf, &size);
  78                 *sizep = size;
  79         }
  80
  81         return buffer;
  82 }
  83
  84 static void setup_archive_check(struct git_attr_check *check)
  85 {
  86         static struct git_attr *attr_export_ignore;
  87         static struct git_attr *attr_export_subst;
  88
  89         if (!attr_export_ignore) {
  90                 attr_export_ignore = git_attr("export-ignore");
  91                 attr_export_subst = git_attr("export-subst");
  92         }
  93         check[0].attr = attr_export_ignore;
  94         check[1].attr = attr_export_subst;
  95 }
  96
  97 struct archiver_context {
  98         struct archiver_args *args;
  99         write_archive_entry_fn_t write_entry;
 100 };
 101
 102 static int write_archive_entry(const unsigned char *sha1, const char *base,
 103                 int baselen, const char *filename, unsigned mode, int stage,
 104                 void *context)
 105 {
 106         static struct strbuf path = STRBUF_INIT;
 107         struct archiver_context *c = context;
 108         struct archiver_args *args = c->args;
 109         write_archive_entry_fn_t write_entry = c->write_entry;
 110         struct git_attr_check check[2];
 111         const char *path_without_prefix;
 112         int convert = 0;
 113         int err;
 114         enum object_type type;
 115         unsigned long size;
 116         void *buffer;
 117
 118         strbuf_reset(&path);
 119         strbuf_grow(&path, PATH_MAX);
 120         strbuf_add(&path, args->base, args->baselen);
 121         strbuf_add(&path, base, baselen);
 122         strbuf_addstr(&path, filename);
 123         path_without_prefix = path.buf + args->baselen;
 124
 125         setup_archive_check(check);
 126         if (!git_check_attr(path_without_prefix, ARRAY_SIZE(check), check)) {
 127                 if (ATTR_TRUE(check[0].value))
 128                         return 0;
 129                 convert = ATTR_TRUE(check[1].value);
 130         }
 131
 132         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
 133                 strbuf_addch(&path, '/');
 134                 if (args->verbose)
 135                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 136                 err = write_entry(args, sha1, path.buf, path.len, mode, NULL, 0);
 137                 if (err)
 138                         return err;
 139                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
 140         }
 141
 142         buffer = sha1_file_to_archive(path_without_prefix, sha1, mode,
 143                         &type, &size, convert ? args->commit : NULL);
 144         if (!buffer)
 145                 return error("cannot read %s", sha1_to_hex(sha1));
 146         if (args->verbose)
 147                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
 148         err = write_entry(args, sha1, path.buf, path.len, mode, buffer, size);
 149         free(buffer);
 150         return err;
 151 }
 152
 153 int write_archive_entries(struct archiver_args *args,
 154                 write_archive_entry_fn_t write_entry)
 155 {
 156         struct archiver_context context;
 157         struct unpack_trees_options opts;
 158         struct tree_desc t;
 159         struct pathspec pathspec;
 160         int err;
 161
 162         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
 163                 size_t len = args->baselen;
 164
 165                 while (len > 1 && args->base[len - 2] == '/')
 166                         len--;
 167                 if (args->verbose)
 168                         fprintf(stderr, "%.*s\n", (int)len, args->base);
 169                 err = write_entry(args, args->tree->object.sha1, args->base,
 170                                 len, 040777, NULL, 0);
 171                 if (err)
 172                         return err;
 173         }
 174
 175         context.args = args;
 176         context.write_entry = write_entry;
 177
 178         /*
 179          * Setup index and instruct attr to read index only
 180          */
 181         if (!args->worktree_attributes) {
 182                 memset(&opts, 0, sizeof(opts));
 183                 opts.index_only = 1;
 184                 opts.head_idx = -1;
 185                 opts.src_index = &the_index;
 186                 opts.dst_index = &the_index;
 187                 opts.fn = oneway_merge;
 188                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
 189                 if (unpack_trees(1, &t, &opts))
 190                         return -1;
 191                 git_attr_set_direction(GIT_ATTR_INDEX, &the_index);
 192         }
 193
 194         init_pathspec(&pathspec, args->pathspec);
 195         err = read_tree_recursive(args->tree, "", 0, 0, &pathspec,
 196                                   write_archive_entry, &context);
 197         free_pathspec(&pathspec);
 198         if (err == READ_TREE_RECURSIVE)
 199                 err = 0;
 200         return err;
 201 }
 202
 203 static const struct archiver *lookup_archiver(const char *name)
 204 {
 205         int i;
 206
 207         if (!name)
 208                 return NULL;
 209
 210         for (i = 0; i < nr_archivers; i++) {
 211                 if (!strcmp(name, archivers[i]->name))
 212                         return archivers[i];
 213         }
 214         return NULL;
 215 }
 216
 217 static int reject_entry(const unsigned char *sha1, const char *base,
 218                         int baselen, const char *filename, unsigned mode,
 219                         int stage, void *context)
 220 {
 221         return -1;
 222 }
 223
 224 static int path_exists(struct tree *tree, const char *path)
 225 {
 226         const char *paths[] = { path, NULL };
 227         struct pathspec pathspec;
 228         int ret;
 229
 230         init_pathspec(&pathspec, paths);
 231         ret = read_tree_recursive(tree, "", 0, 0, &pathspec, reject_entry, NULL);
 232         free_pathspec(&pathspec);
 233         return ret != 0;
 234 }
 235
 236 static void parse_pathspec_arg(const char **pathspec,
 237                 struct archiver_args *ar_args)
 238 {
 239         ar_args->pathspec = pathspec = get_pathspec("", pathspec);
 240         if (pathspec) {
 241                 while (*pathspec) {
 242                         if (!path_exists(ar_args->tree, *pathspec))
 243                                 die("path not found: %s", *pathspec);
 244                         pathspec++;
 245                 }
 246         }
 247 }
 248
 249 static void parse_treeish_arg(const char **argv,
 250                 struct archiver_args *ar_args, const char *prefix)
 251 {
 252         const char *name = argv[0];
 253         const unsigned char *commit_sha1;
 254         time_t archive_time;
 255         struct tree *tree;
 256         const struct commit *commit;
 257         unsigned char sha1[20];
 258
 259         if (get_sha1(name, sha1))
 260                 die("Not a valid object name");
 261
 262         commit = lookup_commit_reference_gently(sha1, 1);
 263         if (commit) {
 264                 commit_sha1 = commit->object.sha1;
 265                 archive_time = commit->date;
 266         } else {
 267                 commit_sha1 = NULL;
 268                 archive_time = time(NULL);
 269         }
 270
 271         tree = parse_tree_indirect(sha1);
 272         if (tree == NULL)
 273                 die("not a tree object");
 274
 275         if (prefix) {
 276                 unsigned char tree_sha1[20];
 277                 unsigned int mode;
 278                 int err;
 279
 280                 err = get_tree_entry(tree->object.sha1, prefix,
 281                                      tree_sha1, &mode);
 282                 if (err || !S_ISDIR(mode))
 283                         die("current working directory is untracked");
 284
 285                 tree = parse_tree_indirect(tree_sha1);
 286         }
 287         ar_args->tree = tree;
 288         ar_args->commit_sha1 = commit_sha1;
 289         ar_args->commit = commit;
 290         ar_args->time = archive_time;
 291 }
 292
 293 #define OPT__COMPR(s, v, h, p) \
 294         { OPTION_SET_INT, (s), NULL, (v), NULL, (h), \
 295           PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, (p) }
 296 #define OPT__COMPR_HIDDEN(s, v, p) \
 297         { OPTION_SET_INT, (s), NULL, (v), NULL, "", \
 298           PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
 299
 300 static int parse_archive_args(int argc, const char **argv,
 301                 const struct archiver **ar, struct archiver_args *args,
 302                 const char *name_hint, int is_remote)
 303 {
 304         const char *format = NULL;
 305         const char *base = NULL;
 306         const char *remote = NULL;
 307         const char *exec = NULL;
 308         const char *output = NULL;
 309         int compression_level = -1;
 310         int verbose = 0;
 311         int i;
 312         int list = 0;
 313         int worktree_attributes = 0;
 314         struct option opts[] = {
 315                 OPT_GROUP(""),
 316                 OPT_STRING(0, "format", &format, "fmt", "archive format"),
 317                 OPT_STRING(0, "prefix", &base, "prefix",
 318                         "prepend prefix to each pathname in the archive"),
 319                 OPT_STRING('o', "output", &output, "file",
 320                         "write the archive to this file"),
 321                 OPT_BOOLEAN(0, "worktree-attributes", &worktree_attributes,
 322                         "read .gitattributes in working directory"),
 323                 OPT__VERBOSE(&verbose, "report archived files on stderr"),
 324                 OPT__COMPR('0', &compression_level, "store only", 0),
 325                 OPT__COMPR('1', &compression_level, "compress faster", 1),
 326                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
 327                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
 328                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
 329                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
 330                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
 331                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
 332                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
 333                 OPT__COMPR('9', &compression_level, "compress better", 9),
 334                 OPT_GROUP(""),
 335                 OPT_BOOLEAN('l', "list", &list,
 336                         "list supported archive formats"),
 337                 OPT_GROUP(""),
 338                 OPT_STRING(0, "remote", &remote, "repo",
 339                         "retrieve the archive from remote repository <repo>"),
 340                 OPT_STRING(0, "exec", &exec, "cmd",
 341                         "path to the remote git-upload-archive command"),
 342                 OPT_END()
 343         };
 344
 345         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
 346
 347         if (remote)
 348                 die("Unexpected option --remote");
 349         if (exec)
 350                 die("Option --exec can only be used together with --remote");
 351         if (output)
 352                 die("Unexpected option --output");
 353
 354         if (!base)
 355                 base = "";
 356
 357         if (list) {
 358                 for (i = 0; i < nr_archivers; i++)
 359                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
 360                                 printf("%s\n", archivers[i]->name);
 361                 exit(0);
 362         }
 363
 364         if (!format && name_hint)
 365                 format = archive_format_from_filename(name_hint);
 366         if (!format)
 367                 format = "tar";
 368
 369         /* We need at least one parameter -- tree-ish */
 370         if (argc < 1)
 371                 usage_with_options(archive_usage, opts);
 372         *ar = lookup_archiver(format);
 373         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
 374                 die("Unknown archive format '%s'", format);
 375
 376         args->compression_level = Z_DEFAULT_COMPRESSION;
 377         if (compression_level != -1) {
 378                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
 379                         args->compression_level = compression_level;
 380                 else {
 381                         die("Argument not supported for format '%s': -%d",
 382                                         format, compression_level);
 383                 }
 384         }
 385         args->verbose = verbose;
 386         args->base = base;
 387         args->baselen = strlen(base);
 388         args->worktree_attributes = worktree_attributes;
 389
 390         return argc;
 391 }
 392
 393 int write_archive(int argc, const char **argv, const char *prefix,
 394                   int setup_prefix, const char *name_hint, int remote)
 395 {
 396         int nongit = 0;
 397         const struct archiver *ar = NULL;
 398         struct archiver_args args;
 399
 400         if (setup_prefix && prefix == NULL)
 401                 prefix = setup_git_directory_gently(&nongit);
 402
 403         git_config(git_default_config, NULL);
 404         init_tar_archiver();
 405         init_zip_archiver();
 406
 407         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
 408         if (nongit) {
 409                 /*
 410                  * We know this will die() with an error, so we could just
 411                  * die ourselves; but its error message will be more specific
 412                  * than what we could write here.
 413                  */
 414                 setup_git_directory();
 415         }
 416
 417         parse_treeish_arg(argv, &args, prefix);
 418         parse_pathspec_arg(argv + 1, &args);
 419
 420         return ar->write_archive(ar, &args);
 421 }
 422
 423 static int match_extension(const char *filename, const char *ext)
 424 {
 425         int prefixlen = strlen(filename) - strlen(ext);
 426
 427         /*
 428          * We need 1 character for the '.', and 1 character to ensure that the
 429          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
 430          * filename).
 431          */
 432         if (prefixlen < 2 || filename[prefixlen-1] != '.')
 433                 return 0;
 434         return !strcmp(filename + prefixlen, ext);
 435 }
 436
 437 const char *archive_format_from_filename(const char *filename)
 438 {
 439         int i;
 440
 441         for (i = 0; i < nr_archivers; i++)
 442                 if (match_extension(filename, archivers[i]->name))
 443                         return archivers[i]->name;
 444         return NULL;
 445 }