Merge branch 'mt/fix-add-rm-with-sparse-index'
[git.git] / archive.c
bloba3bbb091256dd7077e257f20f532d47516e4a960
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"
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
21 static const struct archiver **archivers;
22 static int nr_archivers;
23 static int alloc_archivers;
24 static int remote_allow_unreachable;
26 void register_archiver(struct archiver *ar)
28 ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
29 archivers[nr_archivers++] = ar;
32 void init_archivers(void)
34 init_tar_archiver();
35 init_zip_archiver();
38 static void format_subst(const struct commit *commit,
39 const char *src, size_t len,
40 struct strbuf *buf, struct pretty_print_context *ctx)
42 char *to_free = NULL;
43 struct strbuf fmt = STRBUF_INIT;
45 if (src == buf->buf)
46 to_free = strbuf_detach(buf, NULL);
47 for (;;) {
48 const char *b, *c;
50 b = memmem(src, len, "$Format:", 8);
51 if (!b)
52 break;
53 c = memchr(b + 8, '$', (src + len) - b - 8);
54 if (!c)
55 break;
57 strbuf_reset(&fmt);
58 strbuf_add(&fmt, b + 8, c - b - 8);
60 strbuf_add(buf, src, b - src);
61 format_commit_message(commit, fmt.buf, buf, ctx);
62 len -= c + 1 - src;
63 src = c + 1;
65 strbuf_add(buf, src, len);
66 strbuf_release(&fmt);
67 free(to_free);
70 static void *object_file_to_archive(const struct archiver_args *args,
71 const char *path,
72 const struct object_id *oid,
73 unsigned int mode,
74 enum object_type *type,
75 unsigned long *sizep)
77 void *buffer;
78 const struct commit *commit = args->convert ? args->commit : NULL;
79 struct checkout_metadata meta;
81 init_checkout_metadata(&meta, args->refname,
82 args->commit_oid ? args->commit_oid :
83 (args->tree ? &args->tree->object.oid : NULL), oid);
85 path += args->baselen;
86 buffer = read_object_file(oid, type, sizep);
87 if (buffer && S_ISREG(mode)) {
88 struct strbuf buf = STRBUF_INIT;
89 size_t size = 0;
91 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
92 convert_to_working_tree(args->repo->index, path, buf.buf, buf.len, &buf, &meta);
93 if (commit)
94 format_subst(commit, buf.buf, buf.len, &buf, args->pretty_ctx);
95 buffer = strbuf_detach(&buf, &size);
96 *sizep = size;
99 return buffer;
102 struct directory {
103 struct directory *up;
104 struct object_id oid;
105 int baselen, len;
106 unsigned mode;
107 char path[FLEX_ARRAY];
110 struct archiver_context {
111 struct archiver_args *args;
112 write_archive_entry_fn_t write_entry;
113 struct directory *bottom;
116 static const struct attr_check *get_archive_attrs(struct index_state *istate,
117 const char *path)
119 static struct attr_check *check;
120 if (!check)
121 check = attr_check_initl("export-ignore", "export-subst", NULL);
122 git_check_attr(istate, path, check);
123 return check;
126 static int check_attr_export_ignore(const struct attr_check *check)
128 return check && ATTR_TRUE(check->items[0].value);
131 static int check_attr_export_subst(const struct attr_check *check)
133 return check && ATTR_TRUE(check->items[1].value);
136 static int write_archive_entry(const struct object_id *oid, const char *base,
137 int baselen, const char *filename, unsigned mode,
138 void *context)
140 static struct strbuf path = STRBUF_INIT;
141 struct archiver_context *c = context;
142 struct archiver_args *args = c->args;
143 write_archive_entry_fn_t write_entry = c->write_entry;
144 int err;
145 const char *path_without_prefix;
146 unsigned long size;
147 void *buffer;
148 enum object_type type;
150 args->convert = 0;
151 strbuf_reset(&path);
152 strbuf_grow(&path, PATH_MAX);
153 strbuf_add(&path, args->base, args->baselen);
154 strbuf_add(&path, base, baselen);
155 strbuf_addstr(&path, filename);
156 if (S_ISDIR(mode) || S_ISGITLINK(mode))
157 strbuf_addch(&path, '/');
158 path_without_prefix = path.buf + args->baselen;
160 if (!S_ISDIR(mode)) {
161 const struct attr_check *check;
162 check = get_archive_attrs(args->repo->index, path_without_prefix);
163 if (check_attr_export_ignore(check))
164 return 0;
165 args->convert = check_attr_export_subst(check);
168 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
169 if (args->verbose)
170 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
171 err = write_entry(args, oid, path.buf, path.len, mode, NULL, 0);
172 if (err)
173 return err;
174 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
177 if (args->verbose)
178 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
180 /* Stream it? */
181 if (S_ISREG(mode) && !args->convert &&
182 oid_object_info(args->repo, oid, &size) == OBJ_BLOB &&
183 size > big_file_threshold)
184 return write_entry(args, oid, path.buf, path.len, mode, NULL, size);
186 buffer = object_file_to_archive(args, path.buf, oid, mode, &type, &size);
187 if (!buffer)
188 return error(_("cannot read %s"), oid_to_hex(oid));
189 err = write_entry(args, oid, path.buf, path.len, mode, buffer, size);
190 free(buffer);
191 return err;
194 static void queue_directory(const struct object_id *oid,
195 struct strbuf *base, const char *filename,
196 unsigned mode, struct archiver_context *c)
198 struct directory *d;
199 size_t len = st_add4(base->len, 1, strlen(filename), 1);
200 d = xmalloc(st_add(sizeof(*d), len));
201 d->up = c->bottom;
202 d->baselen = base->len;
203 d->mode = mode;
204 c->bottom = d;
205 d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
206 oidcpy(&d->oid, oid);
209 static int write_directory(struct archiver_context *c)
211 struct directory *d = c->bottom;
212 int ret;
214 if (!d)
215 return 0;
216 c->bottom = d->up;
217 d->path[d->len - 1] = '\0'; /* no trailing slash */
218 ret =
219 write_directory(c) ||
220 write_archive_entry(&d->oid, d->path, d->baselen,
221 d->path + d->baselen, d->mode,
222 c) != READ_TREE_RECURSIVE;
223 free(d);
224 return ret ? -1 : 0;
227 static int queue_or_write_archive_entry(const struct object_id *oid,
228 struct strbuf *base, const char *filename,
229 unsigned mode, void *context)
231 struct archiver_context *c = context;
233 while (c->bottom &&
234 !(base->len >= c->bottom->len &&
235 !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
236 struct directory *next = c->bottom->up;
237 free(c->bottom);
238 c->bottom = next;
241 if (S_ISDIR(mode)) {
242 size_t baselen = base->len;
243 const struct attr_check *check;
245 /* Borrow base, but restore its original value when done. */
246 strbuf_addstr(base, filename);
247 strbuf_addch(base, '/');
248 check = get_archive_attrs(c->args->repo->index, base->buf);
249 strbuf_setlen(base, baselen);
251 if (check_attr_export_ignore(check))
252 return 0;
253 queue_directory(oid, base, filename, mode, c);
254 return READ_TREE_RECURSIVE;
257 if (write_directory(c))
258 return -1;
259 return write_archive_entry(oid, base->buf, base->len, filename, mode,
260 context);
263 struct extra_file_info {
264 char *base;
265 struct stat stat;
268 int write_archive_entries(struct archiver_args *args,
269 write_archive_entry_fn_t write_entry)
271 struct archiver_context context;
272 struct unpack_trees_options opts;
273 struct tree_desc t;
274 int err;
275 struct strbuf path_in_archive = STRBUF_INIT;
276 struct strbuf content = STRBUF_INIT;
277 struct object_id fake_oid;
278 int i;
280 oidcpy(&fake_oid, null_oid());
282 if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
283 size_t len = args->baselen;
285 while (len > 1 && args->base[len - 2] == '/')
286 len--;
287 if (args->verbose)
288 fprintf(stderr, "%.*s\n", (int)len, args->base);
289 err = write_entry(args, &args->tree->object.oid, args->base,
290 len, 040777, NULL, 0);
291 if (err)
292 return err;
295 memset(&context, 0, sizeof(context));
296 context.args = args;
297 context.write_entry = write_entry;
300 * Setup index and instruct attr to read index only
302 if (!args->worktree_attributes) {
303 memset(&opts, 0, sizeof(opts));
304 opts.index_only = 1;
305 opts.head_idx = -1;
306 opts.src_index = args->repo->index;
307 opts.dst_index = args->repo->index;
308 opts.fn = oneway_merge;
309 init_tree_desc(&t, args->tree->buffer, args->tree->size);
310 if (unpack_trees(1, &t, &opts))
311 return -1;
312 git_attr_set_direction(GIT_ATTR_INDEX);
315 err = read_tree(args->repo, args->tree,
316 &args->pathspec,
317 queue_or_write_archive_entry,
318 &context);
319 if (err == READ_TREE_RECURSIVE)
320 err = 0;
321 while (context.bottom) {
322 struct directory *next = context.bottom->up;
323 free(context.bottom);
324 context.bottom = next;
327 for (i = 0; i < args->extra_files.nr; i++) {
328 struct string_list_item *item = args->extra_files.items + i;
329 char *path = item->string;
330 struct extra_file_info *info = item->util;
332 put_be64(fake_oid.hash, i + 1);
334 strbuf_reset(&path_in_archive);
335 if (info->base)
336 strbuf_addstr(&path_in_archive, info->base);
337 strbuf_addstr(&path_in_archive, basename(path));
339 strbuf_reset(&content);
340 if (strbuf_read_file(&content, path, info->stat.st_size) < 0)
341 err = error_errno(_("could not read '%s'"), path);
342 else
343 err = write_entry(args, &fake_oid, path_in_archive.buf,
344 path_in_archive.len,
345 info->stat.st_mode,
346 content.buf, content.len);
347 if (err)
348 break;
350 strbuf_release(&path_in_archive);
351 strbuf_release(&content);
353 return err;
356 static const struct archiver *lookup_archiver(const char *name)
358 int i;
360 if (!name)
361 return NULL;
363 for (i = 0; i < nr_archivers; i++) {
364 if (!strcmp(name, archivers[i]->name))
365 return archivers[i];
367 return NULL;
370 struct path_exists_context {
371 struct pathspec pathspec;
372 struct archiver_args *args;
375 static int reject_entry(const struct object_id *oid, struct strbuf *base,
376 const char *filename, unsigned mode,
377 void *context)
379 int ret = -1;
380 struct path_exists_context *ctx = context;
382 if (S_ISDIR(mode)) {
383 struct strbuf sb = STRBUF_INIT;
384 strbuf_addbuf(&sb, base);
385 strbuf_addstr(&sb, filename);
386 if (!match_pathspec(ctx->args->repo->index,
387 &ctx->pathspec,
388 sb.buf, sb.len, 0, NULL, 1))
389 ret = READ_TREE_RECURSIVE;
390 strbuf_release(&sb);
392 return ret;
395 static int path_exists(struct archiver_args *args, const char *path)
397 const char *paths[] = { path, NULL };
398 struct path_exists_context ctx;
399 int ret;
401 ctx.args = args;
402 parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
403 ctx.pathspec.recursive = 1;
404 ret = read_tree(args->repo, args->tree,
405 &ctx.pathspec,
406 reject_entry, &ctx);
407 clear_pathspec(&ctx.pathspec);
408 return ret != 0;
411 static void parse_pathspec_arg(const char **pathspec,
412 struct archiver_args *ar_args)
415 * must be consistent with parse_pathspec in path_exists()
416 * Also if pathspec patterns are dependent, we're in big
417 * trouble as we test each one separately
419 parse_pathspec(&ar_args->pathspec, 0,
420 PATHSPEC_PREFER_FULL,
421 "", pathspec);
422 ar_args->pathspec.recursive = 1;
423 if (pathspec) {
424 while (*pathspec) {
425 if (**pathspec && !path_exists(ar_args, *pathspec))
426 die(_("pathspec '%s' did not match any files"), *pathspec);
427 pathspec++;
432 static void parse_treeish_arg(const char **argv,
433 struct archiver_args *ar_args, const char *prefix,
434 int remote)
436 const char *name = argv[0];
437 const struct object_id *commit_oid;
438 time_t archive_time;
439 struct tree *tree;
440 const struct commit *commit;
441 struct object_id oid;
442 char *ref = NULL;
444 /* Remotes are only allowed to fetch actual refs */
445 if (remote && !remote_allow_unreachable) {
446 const char *colon = strchrnul(name, ':');
447 int refnamelen = colon - name;
449 if (!dwim_ref(name, refnamelen, &oid, &ref, 0))
450 die(_("no such ref: %.*s"), refnamelen, name);
451 } else {
452 dwim_ref(name, strlen(name), &oid, &ref, 0);
455 if (get_oid(name, &oid))
456 die(_("not a valid object name: %s"), name);
458 commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
459 if (commit) {
460 commit_oid = &commit->object.oid;
461 archive_time = commit->date;
462 } else {
463 commit_oid = NULL;
464 archive_time = time(NULL);
467 tree = parse_tree_indirect(&oid);
468 if (tree == NULL)
469 die(_("not a tree object: %s"), oid_to_hex(&oid));
471 if (prefix) {
472 struct object_id tree_oid;
473 unsigned short mode;
474 int err;
476 err = get_tree_entry(ar_args->repo,
477 &tree->object.oid,
478 prefix, &tree_oid,
479 &mode);
480 if (err || !S_ISDIR(mode))
481 die(_("current working directory is untracked"));
483 tree = parse_tree_indirect(&tree_oid);
485 ar_args->refname = ref;
486 ar_args->tree = tree;
487 ar_args->commit_oid = commit_oid;
488 ar_args->commit = commit;
489 ar_args->time = archive_time;
492 static void extra_file_info_clear(void *util, const char *str)
494 struct extra_file_info *info = util;
495 free(info->base);
496 free(info);
499 static int add_file_cb(const struct option *opt, const char *arg, int unset)
501 struct archiver_args *args = opt->value;
502 const char **basep = (const char **)opt->defval;
503 const char *base = *basep;
504 char *path;
505 struct string_list_item *item;
506 struct extra_file_info *info;
508 if (unset) {
509 string_list_clear_func(&args->extra_files,
510 extra_file_info_clear);
511 return 0;
514 if (!arg)
515 return -1;
517 path = prefix_filename(args->prefix, arg);
518 item = string_list_append_nodup(&args->extra_files, path);
519 item->util = info = xmalloc(sizeof(*info));
520 info->base = xstrdup_or_null(base);
521 if (stat(path, &info->stat))
522 die(_("File not found: %s"), path);
523 if (!S_ISREG(info->stat.st_mode))
524 die(_("Not a regular file: %s"), path);
525 return 0;
528 static int number_callback(const struct option *opt, const char *arg, int unset)
530 BUG_ON_OPT_NEG(unset);
531 *(int *)opt->value = strtol(arg, NULL, 10);
532 return 0;
535 static int parse_archive_args(int argc, const char **argv,
536 const struct archiver **ar, struct archiver_args *args,
537 const char *name_hint, int is_remote)
539 const char *format = NULL;
540 const char *base = NULL;
541 const char *remote = NULL;
542 const char *exec = NULL;
543 const char *output = NULL;
544 int compression_level = -1;
545 int verbose = 0;
546 int i;
547 int list = 0;
548 int worktree_attributes = 0;
549 struct option opts[] = {
550 OPT_GROUP(""),
551 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
552 OPT_STRING(0, "prefix", &base, N_("prefix"),
553 N_("prepend prefix to each pathname in the archive")),
554 { OPTION_CALLBACK, 0, "add-file", args, N_("file"),
555 N_("add untracked file to archive"), 0, add_file_cb,
556 (intptr_t)&base },
557 OPT_STRING('o', "output", &output, N_("file"),
558 N_("write the archive to this file")),
559 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
560 N_("read .gitattributes in working directory")),
561 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
562 OPT_NUMBER_CALLBACK(&compression_level,
563 N_("set compression level"), number_callback),
564 OPT_GROUP(""),
565 OPT_BOOL('l', "list", &list,
566 N_("list supported archive formats")),
567 OPT_GROUP(""),
568 OPT_STRING(0, "remote", &remote, N_("repo"),
569 N_("retrieve the archive from remote repository <repo>")),
570 OPT_STRING(0, "exec", &exec, N_("command"),
571 N_("path to the remote git-upload-archive command")),
572 OPT_END()
575 argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
577 if (remote)
578 die(_("Unexpected option --remote"));
579 if (exec)
580 die(_("Option --exec can only be used together with --remote"));
581 if (output)
582 die(_("Unexpected option --output"));
583 if (is_remote && args->extra_files.nr)
584 die(_("Options --add-file and --remote cannot be used together"));
586 if (!base)
587 base = "";
589 if (list) {
590 for (i = 0; i < nr_archivers; i++)
591 if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
592 printf("%s\n", archivers[i]->name);
593 exit(0);
596 if (!format && name_hint)
597 format = archive_format_from_filename(name_hint);
598 if (!format)
599 format = "tar";
601 /* We need at least one parameter -- tree-ish */
602 if (argc < 1)
603 usage_with_options(archive_usage, opts);
604 *ar = lookup_archiver(format);
605 if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
606 die(_("Unknown archive format '%s'"), format);
608 args->compression_level = Z_DEFAULT_COMPRESSION;
609 if (compression_level != -1) {
610 int levels_ok = (*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS;
611 int high_ok = (*ar)->flags & ARCHIVER_HIGH_COMPRESSION_LEVELS;
612 if (levels_ok && (compression_level <= 9 || high_ok))
613 args->compression_level = compression_level;
614 else {
615 die(_("Argument not supported for format '%s': -%d"),
616 format, compression_level);
619 args->verbose = verbose;
620 args->base = base;
621 args->baselen = strlen(base);
622 args->worktree_attributes = worktree_attributes;
624 return argc;
627 int write_archive(int argc, const char **argv, const char *prefix,
628 struct repository *repo,
629 const char *name_hint, int remote)
631 const struct archiver *ar = NULL;
632 struct pretty_print_describe_status describe_status = {0};
633 struct pretty_print_context ctx = {0};
634 struct archiver_args args;
635 int rc;
637 git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
638 git_config(git_default_config, NULL);
640 describe_status.max_invocations = 1;
641 ctx.date_mode.type = DATE_NORMAL;
642 ctx.abbrev = DEFAULT_ABBREV;
643 ctx.describe_status = &describe_status;
644 args.pretty_ctx = &ctx;
645 args.repo = repo;
646 args.prefix = prefix;
647 string_list_init_dup(&args.extra_files);
648 argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
649 if (!startup_info->have_repository) {
651 * We know this will die() with an error, so we could just
652 * die ourselves; but its error message will be more specific
653 * than what we could write here.
655 setup_git_directory();
658 parse_treeish_arg(argv, &args, prefix, remote);
659 parse_pathspec_arg(argv + 1, &args);
661 rc = ar->write_archive(ar, &args);
663 string_list_clear_func(&args.extra_files, extra_file_info_clear);
664 free(args.refname);
666 return rc;
669 static int match_extension(const char *filename, const char *ext)
671 int prefixlen = strlen(filename) - strlen(ext);
674 * We need 1 character for the '.', and 1 character to ensure that the
675 * prefix is non-empty (k.e., we don't match .tar.gz with no actual
676 * filename).
678 if (prefixlen < 2 || filename[prefixlen - 1] != '.')
679 return 0;
680 return !strcmp(filename + prefixlen, ext);
683 const char *archive_format_from_filename(const char *filename)
685 int i;
687 for (i = 0; i < nr_archivers; i++)
688 if (match_extension(filename, archivers[i]->name))
689 return archivers[i]->name;
690 return NULL;