treewide: remove cache.h inclusion due to git-zlib changes
[alt-git.git] / archive.c
blobab8966d73a517ece5e5cb0eab55d23dbe7b8ef30
1 #include "git-compat-util.h"
2 #include "abspath.h"
3 #include "alloc.h"
4 #include "config.h"
5 #include "convert.h"
6 #include "environment.h"
7 #include "gettext.h"
8 #include "hex.h"
9 #include "setup.h"
10 #include "refs.h"
11 #include "object-store.h"
12 #include "commit.h"
13 #include "tree-walk.h"
14 #include "attr.h"
15 #include "archive.h"
16 #include "parse-options.h"
17 #include "unpack-trees.h"
18 #include "dir.h"
19 #include "quote.h"
21 static char const * const archive_usage[] = {
22 N_("git archive [<options>] <tree-ish> [<path>...]"),
23 "git archive --list",
24 N_("git archive --remote <repo> [--exec <cmd>] [<options>] <tree-ish> [<path>...]"),
25 N_("git archive --remote <repo> [--exec <cmd>] --list"),
26 NULL
29 static const struct archiver **archivers;
30 static int nr_archivers;
31 static int alloc_archivers;
32 static int remote_allow_unreachable;
34 void register_archiver(struct archiver *ar)
36 ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
37 archivers[nr_archivers++] = ar;
40 void init_archivers(void)
42 init_tar_archiver();
43 init_zip_archiver();
46 static void format_subst(const struct commit *commit,
47 const char *src, size_t len,
48 struct strbuf *buf, struct pretty_print_context *ctx)
50 char *to_free = NULL;
51 struct strbuf fmt = STRBUF_INIT;
53 if (src == buf->buf)
54 to_free = strbuf_detach(buf, NULL);
55 for (;;) {
56 const char *b, *c;
58 b = memmem(src, len, "$Format:", 8);
59 if (!b)
60 break;
61 c = memchr(b + 8, '$', (src + len) - b - 8);
62 if (!c)
63 break;
65 strbuf_reset(&fmt);
66 strbuf_add(&fmt, b + 8, c - b - 8);
68 strbuf_add(buf, src, b - src);
69 repo_format_commit_message(the_repository, commit, fmt.buf,
70 buf, ctx);
71 len -= c + 1 - src;
72 src = c + 1;
74 strbuf_add(buf, src, len);
75 strbuf_release(&fmt);
76 free(to_free);
79 static void *object_file_to_archive(const struct archiver_args *args,
80 const char *path,
81 const struct object_id *oid,
82 unsigned int mode,
83 enum object_type *type,
84 unsigned long *sizep)
86 void *buffer;
87 const struct commit *commit = args->convert ? args->commit : NULL;
88 struct checkout_metadata meta;
90 init_checkout_metadata(&meta, args->refname,
91 args->commit_oid ? args->commit_oid :
92 (args->tree ? &args->tree->object.oid : NULL), oid);
94 path += args->baselen;
95 buffer = repo_read_object_file(the_repository, oid, type, sizep);
96 if (buffer && S_ISREG(mode)) {
97 struct strbuf buf = STRBUF_INIT;
98 size_t size = 0;
100 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
101 convert_to_working_tree(args->repo->index, path, buf.buf, buf.len, &buf, &meta);
102 if (commit)
103 format_subst(commit, buf.buf, buf.len, &buf, args->pretty_ctx);
104 buffer = strbuf_detach(&buf, &size);
105 *sizep = size;
108 return buffer;
111 struct directory {
112 struct directory *up;
113 struct object_id oid;
114 int baselen, len;
115 unsigned mode;
116 char path[FLEX_ARRAY];
119 struct archiver_context {
120 struct archiver_args *args;
121 write_archive_entry_fn_t write_entry;
122 struct directory *bottom;
125 static const struct attr_check *get_archive_attrs(struct index_state *istate,
126 const char *path)
128 static struct attr_check *check;
129 if (!check)
130 check = attr_check_initl("export-ignore", "export-subst", NULL);
131 git_check_attr(istate, NULL, path, check);
132 return check;
135 static int check_attr_export_ignore(const struct attr_check *check)
137 return check && ATTR_TRUE(check->items[0].value);
140 static int check_attr_export_subst(const struct attr_check *check)
142 return check && ATTR_TRUE(check->items[1].value);
145 static int write_archive_entry(const struct object_id *oid, const char *base,
146 int baselen, const char *filename, unsigned mode,
147 void *context)
149 static struct strbuf path = STRBUF_INIT;
150 struct archiver_context *c = context;
151 struct archiver_args *args = c->args;
152 write_archive_entry_fn_t write_entry = c->write_entry;
153 int err;
154 const char *path_without_prefix;
155 unsigned long size;
156 void *buffer;
157 enum object_type type;
159 args->convert = 0;
160 strbuf_reset(&path);
161 strbuf_grow(&path, PATH_MAX);
162 strbuf_add(&path, args->base, args->baselen);
163 strbuf_add(&path, base, baselen);
164 strbuf_addstr(&path, filename);
165 if (S_ISDIR(mode) || S_ISGITLINK(mode))
166 strbuf_addch(&path, '/');
167 path_without_prefix = path.buf + args->baselen;
169 if (!S_ISDIR(mode)) {
170 const struct attr_check *check;
171 check = get_archive_attrs(args->repo->index, path_without_prefix);
172 if (check_attr_export_ignore(check))
173 return 0;
174 args->convert = check_attr_export_subst(check);
177 if (args->verbose)
178 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
180 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
181 err = write_entry(args, oid, path.buf, path.len, mode, NULL, 0);
182 if (err)
183 return err;
184 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
187 /* Stream it? */
188 if (S_ISREG(mode) && !args->convert &&
189 oid_object_info(args->repo, oid, &size) == OBJ_BLOB &&
190 size > big_file_threshold)
191 return write_entry(args, oid, path.buf, path.len, mode, NULL, size);
193 buffer = object_file_to_archive(args, path.buf, oid, mode, &type, &size);
194 if (!buffer)
195 return error(_("cannot read '%s'"), oid_to_hex(oid));
196 err = write_entry(args, oid, path.buf, path.len, mode, buffer, size);
197 free(buffer);
198 return err;
201 static void queue_directory(const struct object_id *oid,
202 struct strbuf *base, const char *filename,
203 unsigned mode, struct archiver_context *c)
205 struct directory *d;
206 size_t len = st_add4(base->len, 1, strlen(filename), 1);
207 d = xmalloc(st_add(sizeof(*d), len));
208 d->up = c->bottom;
209 d->baselen = base->len;
210 d->mode = mode;
211 c->bottom = d;
212 d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
213 oidcpy(&d->oid, oid);
216 static int write_directory(struct archiver_context *c)
218 struct directory *d = c->bottom;
219 int ret;
221 if (!d)
222 return 0;
223 c->bottom = d->up;
224 d->path[d->len - 1] = '\0'; /* no trailing slash */
225 ret =
226 write_directory(c) ||
227 write_archive_entry(&d->oid, d->path, d->baselen,
228 d->path + d->baselen, d->mode,
229 c) != READ_TREE_RECURSIVE;
230 free(d);
231 return ret ? -1 : 0;
234 static int queue_or_write_archive_entry(const struct object_id *oid,
235 struct strbuf *base, const char *filename,
236 unsigned mode, void *context)
238 struct archiver_context *c = context;
240 while (c->bottom &&
241 !(base->len >= c->bottom->len &&
242 !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
243 struct directory *next = c->bottom->up;
244 free(c->bottom);
245 c->bottom = next;
248 if (S_ISDIR(mode)) {
249 size_t baselen = base->len;
250 const struct attr_check *check;
252 /* Borrow base, but restore its original value when done. */
253 strbuf_addstr(base, filename);
254 strbuf_addch(base, '/');
255 check = get_archive_attrs(c->args->repo->index, base->buf);
256 strbuf_setlen(base, baselen);
258 if (check_attr_export_ignore(check))
259 return 0;
260 queue_directory(oid, base, filename, mode, c);
261 return READ_TREE_RECURSIVE;
264 if (write_directory(c))
265 return -1;
266 return write_archive_entry(oid, base->buf, base->len, filename, mode,
267 context);
270 struct extra_file_info {
271 char *base;
272 struct stat stat;
273 void *content;
276 int write_archive_entries(struct archiver_args *args,
277 write_archive_entry_fn_t write_entry)
279 struct archiver_context context;
280 struct unpack_trees_options opts;
281 struct tree_desc t;
282 int err;
283 struct strbuf path_in_archive = STRBUF_INIT;
284 struct strbuf content = STRBUF_INIT;
285 struct object_id fake_oid;
286 int i;
288 oidcpy(&fake_oid, null_oid());
290 if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
291 size_t len = args->baselen;
293 while (len > 1 && args->base[len - 2] == '/')
294 len--;
295 if (args->verbose)
296 fprintf(stderr, "%.*s\n", (int)len, args->base);
297 err = write_entry(args, &args->tree->object.oid, args->base,
298 len, 040777, NULL, 0);
299 if (err)
300 return err;
303 memset(&context, 0, sizeof(context));
304 context.args = args;
305 context.write_entry = write_entry;
308 * Setup index and instruct attr to read index only
310 if (!args->worktree_attributes) {
311 memset(&opts, 0, sizeof(opts));
312 opts.index_only = 1;
313 opts.head_idx = -1;
314 opts.src_index = args->repo->index;
315 opts.dst_index = args->repo->index;
316 opts.fn = oneway_merge;
317 init_tree_desc(&t, args->tree->buffer, args->tree->size);
318 if (unpack_trees(1, &t, &opts))
319 return -1;
320 git_attr_set_direction(GIT_ATTR_INDEX);
323 err = read_tree(args->repo, args->tree,
324 &args->pathspec,
325 queue_or_write_archive_entry,
326 &context);
327 if (err == READ_TREE_RECURSIVE)
328 err = 0;
329 while (context.bottom) {
330 struct directory *next = context.bottom->up;
331 free(context.bottom);
332 context.bottom = next;
335 for (i = 0; i < args->extra_files.nr; i++) {
336 struct string_list_item *item = args->extra_files.items + i;
337 char *path = item->string;
338 struct extra_file_info *info = item->util;
340 put_be64(fake_oid.hash, i + 1);
342 if (!info->content) {
343 strbuf_reset(&path_in_archive);
344 if (info->base)
345 strbuf_addstr(&path_in_archive, info->base);
346 strbuf_addstr(&path_in_archive, basename(path));
348 strbuf_reset(&content);
349 if (strbuf_read_file(&content, path, info->stat.st_size) < 0)
350 err = error_errno(_("cannot read '%s'"), path);
351 else
352 err = write_entry(args, &fake_oid, path_in_archive.buf,
353 path_in_archive.len,
354 canon_mode(info->stat.st_mode),
355 content.buf, content.len);
356 } else {
357 err = write_entry(args, &fake_oid,
358 path, strlen(path),
359 canon_mode(info->stat.st_mode),
360 info->content, info->stat.st_size);
363 if (err)
364 break;
366 strbuf_release(&path_in_archive);
367 strbuf_release(&content);
369 return err;
372 static const struct archiver *lookup_archiver(const char *name)
374 int i;
376 if (!name)
377 return NULL;
379 for (i = 0; i < nr_archivers; i++) {
380 if (!strcmp(name, archivers[i]->name))
381 return archivers[i];
383 return NULL;
386 struct path_exists_context {
387 struct pathspec pathspec;
388 struct archiver_args *args;
391 static int reject_entry(const struct object_id *oid UNUSED,
392 struct strbuf *base,
393 const char *filename, unsigned mode,
394 void *context)
396 int ret = -1;
397 struct path_exists_context *ctx = context;
399 if (S_ISDIR(mode)) {
400 struct strbuf sb = STRBUF_INIT;
401 strbuf_addbuf(&sb, base);
402 strbuf_addstr(&sb, filename);
403 if (!match_pathspec(ctx->args->repo->index,
404 &ctx->pathspec,
405 sb.buf, sb.len, 0, NULL, 1))
406 ret = READ_TREE_RECURSIVE;
407 strbuf_release(&sb);
409 return ret;
412 static int path_exists(struct archiver_args *args, const char *path)
414 const char *paths[] = { path, NULL };
415 struct path_exists_context ctx;
416 int ret;
418 ctx.args = args;
419 parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
420 ctx.pathspec.recursive = 1;
421 ret = read_tree(args->repo, args->tree,
422 &ctx.pathspec,
423 reject_entry, &ctx);
424 clear_pathspec(&ctx.pathspec);
425 return ret != 0;
428 static void parse_pathspec_arg(const char **pathspec,
429 struct archiver_args *ar_args)
432 * must be consistent with parse_pathspec in path_exists()
433 * Also if pathspec patterns are dependent, we're in big
434 * trouble as we test each one separately
436 parse_pathspec(&ar_args->pathspec, 0,
437 PATHSPEC_PREFER_FULL,
438 "", pathspec);
439 ar_args->pathspec.recursive = 1;
440 if (pathspec) {
441 while (*pathspec) {
442 if (**pathspec && !path_exists(ar_args, *pathspec))
443 die(_("pathspec '%s' did not match any files"), *pathspec);
444 pathspec++;
449 static void parse_treeish_arg(const char **argv,
450 struct archiver_args *ar_args, const char *prefix,
451 int remote)
453 const char *name = argv[0];
454 const struct object_id *commit_oid;
455 time_t archive_time;
456 struct tree *tree;
457 const struct commit *commit;
458 struct object_id oid;
459 char *ref = NULL;
461 /* Remotes are only allowed to fetch actual refs */
462 if (remote && !remote_allow_unreachable) {
463 const char *colon = strchrnul(name, ':');
464 int refnamelen = colon - name;
466 if (!repo_dwim_ref(the_repository, name, refnamelen, &oid, &ref, 0))
467 die(_("no such ref: %.*s"), refnamelen, name);
468 } else {
469 repo_dwim_ref(the_repository, name, strlen(name), &oid, &ref,
473 if (repo_get_oid(the_repository, name, &oid))
474 die(_("not a valid object name: %s"), name);
476 commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
477 if (commit) {
478 commit_oid = &commit->object.oid;
479 archive_time = commit->date;
480 } else {
481 commit_oid = NULL;
482 archive_time = time(NULL);
484 if (ar_args->mtime_option)
485 archive_time = approxidate(ar_args->mtime_option);
487 tree = parse_tree_indirect(&oid);
488 if (!tree)
489 die(_("not a tree object: %s"), oid_to_hex(&oid));
491 if (prefix) {
492 struct object_id tree_oid;
493 unsigned short mode;
494 int err;
496 err = get_tree_entry(ar_args->repo,
497 &tree->object.oid,
498 prefix, &tree_oid,
499 &mode);
500 if (err || !S_ISDIR(mode))
501 die(_("current working directory is untracked"));
503 tree = parse_tree_indirect(&tree_oid);
505 ar_args->refname = ref;
506 ar_args->tree = tree;
507 ar_args->commit_oid = commit_oid;
508 ar_args->commit = commit;
509 ar_args->time = archive_time;
512 static void extra_file_info_clear(void *util, const char *str UNUSED)
514 struct extra_file_info *info = util;
515 free(info->base);
516 free(info->content);
517 free(info);
520 static int add_file_cb(const struct option *opt, const char *arg, int unset)
522 struct archiver_args *args = opt->value;
523 const char **basep = (const char **)opt->defval;
524 const char *base = *basep;
525 char *path;
526 struct string_list_item *item;
527 struct extra_file_info *info;
529 if (unset) {
530 string_list_clear_func(&args->extra_files,
531 extra_file_info_clear);
532 return 0;
535 if (!arg)
536 return -1;
538 info = xmalloc(sizeof(*info));
539 info->base = xstrdup_or_null(base);
541 if (!strcmp(opt->long_name, "add-file")) {
542 path = prefix_filename(args->prefix, arg);
543 if (stat(path, &info->stat))
544 die(_("File not found: %s"), path);
545 if (!S_ISREG(info->stat.st_mode))
546 die(_("Not a regular file: %s"), path);
547 info->content = NULL; /* read the file later */
548 } else if (!strcmp(opt->long_name, "add-virtual-file")) {
549 struct strbuf buf = STRBUF_INIT;
550 const char *p = arg;
552 if (*p != '"')
553 p = strchr(p, ':');
554 else if (unquote_c_style(&buf, p, &p) < 0)
555 die(_("unclosed quote: '%s'"), arg);
557 if (!p || *p != ':')
558 die(_("missing colon: '%s'"), arg);
560 if (p == arg)
561 die(_("empty file name: '%s'"), arg);
563 path = buf.len ?
564 strbuf_detach(&buf, NULL) : xstrndup(arg, p - arg);
566 if (args->prefix) {
567 char *save = path;
568 path = prefix_filename(args->prefix, path);
569 free(save);
571 memset(&info->stat, 0, sizeof(info->stat));
572 info->stat.st_mode = S_IFREG | 0644;
573 info->content = xstrdup(p + 1);
574 info->stat.st_size = strlen(info->content);
575 } else {
576 BUG("add_file_cb() called for %s", opt->long_name);
578 item = string_list_append_nodup(&args->extra_files, path);
579 item->util = info;
581 return 0;
584 static int number_callback(const struct option *opt, const char *arg, int unset)
586 BUG_ON_OPT_NEG(unset);
587 *(int *)opt->value = strtol(arg, NULL, 10);
588 return 0;
591 static int parse_archive_args(int argc, const char **argv,
592 const struct archiver **ar, struct archiver_args *args,
593 const char *name_hint, int is_remote)
595 const char *format = NULL;
596 const char *base = NULL;
597 const char *remote = NULL;
598 const char *exec = NULL;
599 const char *output = NULL;
600 const char *mtime_option = NULL;
601 int compression_level = -1;
602 int verbose = 0;
603 int i;
604 int list = 0;
605 int worktree_attributes = 0;
606 struct option opts[] = {
607 OPT_GROUP(""),
608 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
609 OPT_STRING(0, "prefix", &base, N_("prefix"),
610 N_("prepend prefix to each pathname in the archive")),
611 { OPTION_CALLBACK, 0, "add-file", args, N_("file"),
612 N_("add untracked file to archive"), 0, add_file_cb,
613 (intptr_t)&base },
614 { OPTION_CALLBACK, 0, "add-virtual-file", args,
615 N_("path:content"), N_("add untracked file to archive"), 0,
616 add_file_cb, (intptr_t)&base },
617 OPT_STRING('o', "output", &output, N_("file"),
618 N_("write the archive to this file")),
619 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
620 N_("read .gitattributes in working directory")),
621 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
622 { OPTION_STRING, 0, "mtime", &mtime_option, N_("time"),
623 N_("set modification time of archive entries"),
624 PARSE_OPT_NONEG },
625 OPT_NUMBER_CALLBACK(&compression_level,
626 N_("set compression level"), number_callback),
627 OPT_GROUP(""),
628 OPT_BOOL('l', "list", &list,
629 N_("list supported archive formats")),
630 OPT_GROUP(""),
631 OPT_STRING(0, "remote", &remote, N_("repo"),
632 N_("retrieve the archive from remote repository <repo>")),
633 OPT_STRING(0, "exec", &exec, N_("command"),
634 N_("path to the remote git-upload-archive command")),
635 OPT_END()
638 argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
640 if (remote)
641 die(_("Unexpected option --remote"));
642 if (exec)
643 die(_("the option '%s' requires '%s'"), "--exec", "--remote");
644 if (output)
645 die(_("Unexpected option --output"));
646 if (is_remote && args->extra_files.nr)
647 die(_("options '%s' and '%s' cannot be used together"), "--add-file", "--remote");
649 if (!base)
650 base = "";
652 if (list) {
653 for (i = 0; i < nr_archivers; i++)
654 if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
655 printf("%s\n", archivers[i]->name);
656 exit(0);
659 if (!format && name_hint)
660 format = archive_format_from_filename(name_hint);
661 if (!format)
662 format = "tar";
664 /* We need at least one parameter -- tree-ish */
665 if (argc < 1)
666 usage_with_options(archive_usage, opts);
667 *ar = lookup_archiver(format);
668 if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
669 die(_("Unknown archive format '%s'"), format);
671 args->compression_level = Z_DEFAULT_COMPRESSION;
672 if (compression_level != -1) {
673 int levels_ok = (*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS;
674 int high_ok = (*ar)->flags & ARCHIVER_HIGH_COMPRESSION_LEVELS;
675 if (levels_ok && (compression_level <= 9 || high_ok))
676 args->compression_level = compression_level;
677 else {
678 die(_("Argument not supported for format '%s': -%d"),
679 format, compression_level);
682 args->verbose = verbose;
683 args->base = base;
684 args->baselen = strlen(base);
685 args->worktree_attributes = worktree_attributes;
686 args->mtime_option = mtime_option;
688 return argc;
691 int write_archive(int argc, const char **argv, const char *prefix,
692 struct repository *repo,
693 const char *name_hint, int remote)
695 const struct archiver *ar = NULL;
696 struct pretty_print_describe_status describe_status = {0};
697 struct pretty_print_context ctx = {0};
698 struct archiver_args args;
699 int rc;
701 git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
702 git_config(git_default_config, NULL);
704 describe_status.max_invocations = 1;
705 ctx.date_mode.type = DATE_NORMAL;
706 ctx.abbrev = DEFAULT_ABBREV;
707 ctx.describe_status = &describe_status;
708 args.pretty_ctx = &ctx;
709 args.repo = repo;
710 args.prefix = prefix;
711 string_list_init_dup(&args.extra_files);
712 argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
713 if (!startup_info->have_repository) {
715 * We know this will die() with an error, so we could just
716 * die ourselves; but its error message will be more specific
717 * than what we could write here.
719 setup_git_directory();
722 parse_treeish_arg(argv, &args, prefix, remote);
723 parse_pathspec_arg(argv + 1, &args);
725 rc = ar->write_archive(ar, &args);
727 string_list_clear_func(&args.extra_files, extra_file_info_clear);
728 free(args.refname);
729 clear_pathspec(&args.pathspec);
731 return rc;
734 static int match_extension(const char *filename, const char *ext)
736 int prefixlen = strlen(filename) - strlen(ext);
739 * We need 1 character for the '.', and 1 character to ensure that the
740 * prefix is non-empty (k.e., we don't match .tar.gz with no actual
741 * filename).
743 if (prefixlen < 2 || filename[prefixlen - 1] != '.')
744 return 0;
745 return !strcmp(filename + prefixlen, ext);
748 const char *archive_format_from_filename(const char *filename)
750 int i;
752 for (i = 0; i < nr_archivers; i++)
753 if (match_extension(filename, archivers[i]->name))
754 return archivers[i]->name;
755 return NULL;