apply: reject patches larger than ~1 GiB
[alt-git.git] / archive.c
blob61a79e4a2270dfc0e738ee97dd5cb5a9054ba79f
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"
12 #include "quote.h"
14 static char const * const archive_usage[] = {
15 N_("git archive [<options>] <tree-ish> [<path>...]"),
16 "git archive --list",
17 N_("git archive --remote <repo> [--exec <cmd>] [<options>] <tree-ish> [<path>...]"),
18 N_("git archive --remote <repo> [--exec <cmd>] --list"),
19 NULL
22 static const struct archiver **archivers;
23 static int nr_archivers;
24 static int alloc_archivers;
25 static int remote_allow_unreachable;
27 void register_archiver(struct archiver *ar)
29 ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
30 archivers[nr_archivers++] = ar;
33 void init_archivers(void)
35 init_tar_archiver();
36 init_zip_archiver();
39 static void format_subst(const struct commit *commit,
40 const char *src, size_t len,
41 struct strbuf *buf, struct pretty_print_context *ctx)
43 char *to_free = NULL;
44 struct strbuf fmt = STRBUF_INIT;
46 if (src == buf->buf)
47 to_free = strbuf_detach(buf, NULL);
48 for (;;) {
49 const char *b, *c;
51 b = memmem(src, len, "$Format:", 8);
52 if (!b)
53 break;
54 c = memchr(b + 8, '$', (src + len) - b - 8);
55 if (!c)
56 break;
58 strbuf_reset(&fmt);
59 strbuf_add(&fmt, b + 8, c - b - 8);
61 strbuf_add(buf, src, b - src);
62 format_commit_message(commit, fmt.buf, buf, ctx);
63 len -= c + 1 - src;
64 src = c + 1;
66 strbuf_add(buf, src, len);
67 strbuf_release(&fmt);
68 free(to_free);
71 static void *object_file_to_archive(const struct archiver_args *args,
72 const char *path,
73 const struct object_id *oid,
74 unsigned int mode,
75 enum object_type *type,
76 unsigned long *sizep)
78 void *buffer;
79 const struct commit *commit = args->convert ? args->commit : NULL;
80 struct checkout_metadata meta;
82 init_checkout_metadata(&meta, args->refname,
83 args->commit_oid ? args->commit_oid :
84 (args->tree ? &args->tree->object.oid : NULL), oid);
86 path += args->baselen;
87 buffer = read_object_file(oid, type, sizep);
88 if (buffer && S_ISREG(mode)) {
89 struct strbuf buf = STRBUF_INIT;
90 size_t size = 0;
92 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
93 convert_to_working_tree(args->repo->index, path, buf.buf, buf.len, &buf, &meta);
94 if (commit)
95 format_subst(commit, buf.buf, buf.len, &buf, args->pretty_ctx);
96 buffer = strbuf_detach(&buf, &size);
97 *sizep = size;
100 return buffer;
103 struct directory {
104 struct directory *up;
105 struct object_id oid;
106 int baselen, len;
107 unsigned mode;
108 char path[FLEX_ARRAY];
111 struct archiver_context {
112 struct archiver_args *args;
113 write_archive_entry_fn_t write_entry;
114 struct directory *bottom;
117 static const struct attr_check *get_archive_attrs(struct index_state *istate,
118 const char *path)
120 static struct attr_check *check;
121 if (!check)
122 check = attr_check_initl("export-ignore", "export-subst", NULL);
123 git_check_attr(istate, path, check);
124 return check;
127 static int check_attr_export_ignore(const struct attr_check *check)
129 return check && ATTR_TRUE(check->items[0].value);
132 static int check_attr_export_subst(const struct attr_check *check)
134 return check && ATTR_TRUE(check->items[1].value);
137 static int write_archive_entry(const struct object_id *oid, const char *base,
138 int baselen, const char *filename, unsigned mode,
139 void *context)
141 static struct strbuf path = STRBUF_INIT;
142 struct archiver_context *c = context;
143 struct archiver_args *args = c->args;
144 write_archive_entry_fn_t write_entry = c->write_entry;
145 int err;
146 const char *path_without_prefix;
147 unsigned long size;
148 void *buffer;
149 enum object_type type;
151 args->convert = 0;
152 strbuf_reset(&path);
153 strbuf_grow(&path, PATH_MAX);
154 strbuf_add(&path, args->base, args->baselen);
155 strbuf_add(&path, base, baselen);
156 strbuf_addstr(&path, filename);
157 if (S_ISDIR(mode) || S_ISGITLINK(mode))
158 strbuf_addch(&path, '/');
159 path_without_prefix = path.buf + args->baselen;
161 if (!S_ISDIR(mode)) {
162 const struct attr_check *check;
163 check = get_archive_attrs(args->repo->index, path_without_prefix);
164 if (check_attr_export_ignore(check))
165 return 0;
166 args->convert = check_attr_export_subst(check);
169 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
170 if (args->verbose)
171 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
172 err = write_entry(args, oid, path.buf, path.len, mode, NULL, 0);
173 if (err)
174 return err;
175 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
178 if (args->verbose)
179 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
181 /* Stream it? */
182 if (S_ISREG(mode) && !args->convert &&
183 oid_object_info(args->repo, oid, &size) == OBJ_BLOB &&
184 size > big_file_threshold)
185 return write_entry(args, oid, path.buf, path.len, mode, NULL, size);
187 buffer = object_file_to_archive(args, path.buf, oid, mode, &type, &size);
188 if (!buffer)
189 return error(_("cannot read '%s'"), oid_to_hex(oid));
190 err = write_entry(args, oid, path.buf, path.len, mode, buffer, size);
191 free(buffer);
192 return err;
195 static void queue_directory(const struct object_id *oid,
196 struct strbuf *base, const char *filename,
197 unsigned mode, struct archiver_context *c)
199 struct directory *d;
200 size_t len = st_add4(base->len, 1, strlen(filename), 1);
201 d = xmalloc(st_add(sizeof(*d), len));
202 d->up = c->bottom;
203 d->baselen = base->len;
204 d->mode = mode;
205 c->bottom = d;
206 d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
207 oidcpy(&d->oid, oid);
210 static int write_directory(struct archiver_context *c)
212 struct directory *d = c->bottom;
213 int ret;
215 if (!d)
216 return 0;
217 c->bottom = d->up;
218 d->path[d->len - 1] = '\0'; /* no trailing slash */
219 ret =
220 write_directory(c) ||
221 write_archive_entry(&d->oid, d->path, d->baselen,
222 d->path + d->baselen, d->mode,
223 c) != READ_TREE_RECURSIVE;
224 free(d);
225 return ret ? -1 : 0;
228 static int queue_or_write_archive_entry(const struct object_id *oid,
229 struct strbuf *base, const char *filename,
230 unsigned mode, void *context)
232 struct archiver_context *c = context;
234 while (c->bottom &&
235 !(base->len >= c->bottom->len &&
236 !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
237 struct directory *next = c->bottom->up;
238 free(c->bottom);
239 c->bottom = next;
242 if (S_ISDIR(mode)) {
243 size_t baselen = base->len;
244 const struct attr_check *check;
246 /* Borrow base, but restore its original value when done. */
247 strbuf_addstr(base, filename);
248 strbuf_addch(base, '/');
249 check = get_archive_attrs(c->args->repo->index, base->buf);
250 strbuf_setlen(base, baselen);
252 if (check_attr_export_ignore(check))
253 return 0;
254 queue_directory(oid, base, filename, mode, c);
255 return READ_TREE_RECURSIVE;
258 if (write_directory(c))
259 return -1;
260 return write_archive_entry(oid, base->buf, base->len, filename, mode,
261 context);
264 struct extra_file_info {
265 char *base;
266 struct stat stat;
267 void *content;
270 int write_archive_entries(struct archiver_args *args,
271 write_archive_entry_fn_t write_entry)
273 struct archiver_context context;
274 struct unpack_trees_options opts;
275 struct tree_desc t;
276 int err;
277 struct strbuf path_in_archive = STRBUF_INIT;
278 struct strbuf content = STRBUF_INIT;
279 struct object_id fake_oid;
280 int i;
282 oidcpy(&fake_oid, null_oid());
284 if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
285 size_t len = args->baselen;
287 while (len > 1 && args->base[len - 2] == '/')
288 len--;
289 if (args->verbose)
290 fprintf(stderr, "%.*s\n", (int)len, args->base);
291 err = write_entry(args, &args->tree->object.oid, args->base,
292 len, 040777, NULL, 0);
293 if (err)
294 return err;
297 memset(&context, 0, sizeof(context));
298 context.args = args;
299 context.write_entry = write_entry;
302 * Setup index and instruct attr to read index only
304 if (!args->worktree_attributes) {
305 memset(&opts, 0, sizeof(opts));
306 opts.index_only = 1;
307 opts.head_idx = -1;
308 opts.src_index = args->repo->index;
309 opts.dst_index = args->repo->index;
310 opts.fn = oneway_merge;
311 init_tree_desc(&t, args->tree->buffer, args->tree->size);
312 if (unpack_trees(1, &t, &opts))
313 return -1;
314 git_attr_set_direction(GIT_ATTR_INDEX);
317 err = read_tree(args->repo, args->tree,
318 &args->pathspec,
319 queue_or_write_archive_entry,
320 &context);
321 if (err == READ_TREE_RECURSIVE)
322 err = 0;
323 while (context.bottom) {
324 struct directory *next = context.bottom->up;
325 free(context.bottom);
326 context.bottom = next;
329 for (i = 0; i < args->extra_files.nr; i++) {
330 struct string_list_item *item = args->extra_files.items + i;
331 char *path = item->string;
332 struct extra_file_info *info = item->util;
334 put_be64(fake_oid.hash, i + 1);
336 if (!info->content) {
337 strbuf_reset(&path_in_archive);
338 if (info->base)
339 strbuf_addstr(&path_in_archive, info->base);
340 strbuf_addstr(&path_in_archive, basename(path));
342 strbuf_reset(&content);
343 if (strbuf_read_file(&content, path, info->stat.st_size) < 0)
344 err = error_errno(_("cannot read '%s'"), path);
345 else
346 err = write_entry(args, &fake_oid, path_in_archive.buf,
347 path_in_archive.len,
348 canon_mode(info->stat.st_mode),
349 content.buf, content.len);
350 } else {
351 err = write_entry(args, &fake_oid,
352 path, strlen(path),
353 canon_mode(info->stat.st_mode),
354 info->content, info->stat.st_size);
357 if (err)
358 break;
360 strbuf_release(&path_in_archive);
361 strbuf_release(&content);
363 return err;
366 static const struct archiver *lookup_archiver(const char *name)
368 int i;
370 if (!name)
371 return NULL;
373 for (i = 0; i < nr_archivers; i++) {
374 if (!strcmp(name, archivers[i]->name))
375 return archivers[i];
377 return NULL;
380 struct path_exists_context {
381 struct pathspec pathspec;
382 struct archiver_args *args;
385 static int reject_entry(const struct object_id *oid UNUSED,
386 struct strbuf *base,
387 const char *filename, unsigned mode,
388 void *context)
390 int ret = -1;
391 struct path_exists_context *ctx = context;
393 if (S_ISDIR(mode)) {
394 struct strbuf sb = STRBUF_INIT;
395 strbuf_addbuf(&sb, base);
396 strbuf_addstr(&sb, filename);
397 if (!match_pathspec(ctx->args->repo->index,
398 &ctx->pathspec,
399 sb.buf, sb.len, 0, NULL, 1))
400 ret = READ_TREE_RECURSIVE;
401 strbuf_release(&sb);
403 return ret;
406 static int path_exists(struct archiver_args *args, const char *path)
408 const char *paths[] = { path, NULL };
409 struct path_exists_context ctx;
410 int ret;
412 ctx.args = args;
413 parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
414 ctx.pathspec.recursive = 1;
415 ret = read_tree(args->repo, args->tree,
416 &ctx.pathspec,
417 reject_entry, &ctx);
418 clear_pathspec(&ctx.pathspec);
419 return ret != 0;
422 static void parse_pathspec_arg(const char **pathspec,
423 struct archiver_args *ar_args)
426 * must be consistent with parse_pathspec in path_exists()
427 * Also if pathspec patterns are dependent, we're in big
428 * trouble as we test each one separately
430 parse_pathspec(&ar_args->pathspec, 0,
431 PATHSPEC_PREFER_FULL,
432 "", pathspec);
433 ar_args->pathspec.recursive = 1;
434 if (pathspec) {
435 while (*pathspec) {
436 if (**pathspec && !path_exists(ar_args, *pathspec))
437 die(_("pathspec '%s' did not match any files"), *pathspec);
438 pathspec++;
443 static void parse_treeish_arg(const char **argv,
444 struct archiver_args *ar_args, const char *prefix,
445 int remote)
447 const char *name = argv[0];
448 const struct object_id *commit_oid;
449 time_t archive_time;
450 struct tree *tree;
451 const struct commit *commit;
452 struct object_id oid;
453 char *ref = NULL;
455 /* Remotes are only allowed to fetch actual refs */
456 if (remote && !remote_allow_unreachable) {
457 const char *colon = strchrnul(name, ':');
458 int refnamelen = colon - name;
460 if (!dwim_ref(name, refnamelen, &oid, &ref, 0))
461 die(_("no such ref: %.*s"), refnamelen, name);
462 } else {
463 dwim_ref(name, strlen(name), &oid, &ref, 0);
466 if (get_oid(name, &oid))
467 die(_("not a valid object name: %s"), name);
469 commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
470 if (commit) {
471 commit_oid = &commit->object.oid;
472 archive_time = commit->date;
473 } else {
474 commit_oid = NULL;
475 archive_time = time(NULL);
478 tree = parse_tree_indirect(&oid);
479 if (!tree)
480 die(_("not a tree object: %s"), oid_to_hex(&oid));
482 if (prefix) {
483 struct object_id tree_oid;
484 unsigned short mode;
485 int err;
487 err = get_tree_entry(ar_args->repo,
488 &tree->object.oid,
489 prefix, &tree_oid,
490 &mode);
491 if (err || !S_ISDIR(mode))
492 die(_("current working directory is untracked"));
494 tree = parse_tree_indirect(&tree_oid);
496 ar_args->refname = ref;
497 ar_args->tree = tree;
498 ar_args->commit_oid = commit_oid;
499 ar_args->commit = commit;
500 ar_args->time = archive_time;
503 static void extra_file_info_clear(void *util, const char *str)
505 struct extra_file_info *info = util;
506 free(info->base);
507 free(info->content);
508 free(info);
511 static int add_file_cb(const struct option *opt, const char *arg, int unset)
513 struct archiver_args *args = opt->value;
514 const char **basep = (const char **)opt->defval;
515 const char *base = *basep;
516 char *path;
517 struct string_list_item *item;
518 struct extra_file_info *info;
520 if (unset) {
521 string_list_clear_func(&args->extra_files,
522 extra_file_info_clear);
523 return 0;
526 if (!arg)
527 return -1;
529 info = xmalloc(sizeof(*info));
530 info->base = xstrdup_or_null(base);
532 if (!strcmp(opt->long_name, "add-file")) {
533 path = prefix_filename(args->prefix, arg);
534 if (stat(path, &info->stat))
535 die(_("File not found: %s"), path);
536 if (!S_ISREG(info->stat.st_mode))
537 die(_("Not a regular file: %s"), path);
538 info->content = NULL; /* read the file later */
539 } else if (!strcmp(opt->long_name, "add-virtual-file")) {
540 struct strbuf buf = STRBUF_INIT;
541 const char *p = arg;
543 if (*p != '"')
544 p = strchr(p, ':');
545 else if (unquote_c_style(&buf, p, &p) < 0)
546 die(_("unclosed quote: '%s'"), arg);
548 if (!p || *p != ':')
549 die(_("missing colon: '%s'"), arg);
551 if (p == arg)
552 die(_("empty file name: '%s'"), arg);
554 path = buf.len ?
555 strbuf_detach(&buf, NULL) : xstrndup(arg, p - arg);
557 if (args->prefix) {
558 char *save = path;
559 path = prefix_filename(args->prefix, path);
560 free(save);
562 memset(&info->stat, 0, sizeof(info->stat));
563 info->stat.st_mode = S_IFREG | 0644;
564 info->content = xstrdup(p + 1);
565 info->stat.st_size = strlen(info->content);
566 } else {
567 BUG("add_file_cb() called for %s", opt->long_name);
569 item = string_list_append_nodup(&args->extra_files, path);
570 item->util = info;
572 return 0;
575 static int number_callback(const struct option *opt, const char *arg, int unset)
577 BUG_ON_OPT_NEG(unset);
578 *(int *)opt->value = strtol(arg, NULL, 10);
579 return 0;
582 static int parse_archive_args(int argc, const char **argv,
583 const struct archiver **ar, struct archiver_args *args,
584 const char *name_hint, int is_remote)
586 const char *format = NULL;
587 const char *base = NULL;
588 const char *remote = NULL;
589 const char *exec = NULL;
590 const char *output = NULL;
591 int compression_level = -1;
592 int verbose = 0;
593 int i;
594 int list = 0;
595 int worktree_attributes = 0;
596 struct option opts[] = {
597 OPT_GROUP(""),
598 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
599 OPT_STRING(0, "prefix", &base, N_("prefix"),
600 N_("prepend prefix to each pathname in the archive")),
601 { OPTION_CALLBACK, 0, "add-file", args, N_("file"),
602 N_("add untracked file to archive"), 0, add_file_cb,
603 (intptr_t)&base },
604 { OPTION_CALLBACK, 0, "add-virtual-file", args,
605 N_("path:content"), N_("add untracked file to archive"), 0,
606 add_file_cb, (intptr_t)&base },
607 OPT_STRING('o', "output", &output, N_("file"),
608 N_("write the archive to this file")),
609 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
610 N_("read .gitattributes in working directory")),
611 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
612 OPT_NUMBER_CALLBACK(&compression_level,
613 N_("set compression level"), number_callback),
614 OPT_GROUP(""),
615 OPT_BOOL('l', "list", &list,
616 N_("list supported archive formats")),
617 OPT_GROUP(""),
618 OPT_STRING(0, "remote", &remote, N_("repo"),
619 N_("retrieve the archive from remote repository <repo>")),
620 OPT_STRING(0, "exec", &exec, N_("command"),
621 N_("path to the remote git-upload-archive command")),
622 OPT_END()
625 argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
627 if (remote)
628 die(_("Unexpected option --remote"));
629 if (exec)
630 die(_("the option '%s' requires '%s'"), "--exec", "--remote");
631 if (output)
632 die(_("Unexpected option --output"));
633 if (is_remote && args->extra_files.nr)
634 die(_("options '%s' and '%s' cannot be used together"), "--add-file", "--remote");
636 if (!base)
637 base = "";
639 if (list) {
640 for (i = 0; i < nr_archivers; i++)
641 if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
642 printf("%s\n", archivers[i]->name);
643 exit(0);
646 if (!format && name_hint)
647 format = archive_format_from_filename(name_hint);
648 if (!format)
649 format = "tar";
651 /* We need at least one parameter -- tree-ish */
652 if (argc < 1)
653 usage_with_options(archive_usage, opts);
654 *ar = lookup_archiver(format);
655 if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
656 die(_("Unknown archive format '%s'"), format);
658 args->compression_level = Z_DEFAULT_COMPRESSION;
659 if (compression_level != -1) {
660 int levels_ok = (*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS;
661 int high_ok = (*ar)->flags & ARCHIVER_HIGH_COMPRESSION_LEVELS;
662 if (levels_ok && (compression_level <= 9 || high_ok))
663 args->compression_level = compression_level;
664 else {
665 die(_("Argument not supported for format '%s': -%d"),
666 format, compression_level);
669 args->verbose = verbose;
670 args->base = base;
671 args->baselen = strlen(base);
672 args->worktree_attributes = worktree_attributes;
674 return argc;
677 int write_archive(int argc, const char **argv, const char *prefix,
678 struct repository *repo,
679 const char *name_hint, int remote)
681 const struct archiver *ar = NULL;
682 struct pretty_print_describe_status describe_status = {0};
683 struct pretty_print_context ctx = {0};
684 struct archiver_args args;
685 int rc;
687 git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
688 git_config(git_default_config, NULL);
690 describe_status.max_invocations = 1;
691 ctx.date_mode.type = DATE_NORMAL;
692 ctx.abbrev = DEFAULT_ABBREV;
693 ctx.describe_status = &describe_status;
694 args.pretty_ctx = &ctx;
695 args.repo = repo;
696 args.prefix = prefix;
697 string_list_init_dup(&args.extra_files);
698 argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
699 if (!startup_info->have_repository) {
701 * We know this will die() with an error, so we could just
702 * die ourselves; but its error message will be more specific
703 * than what we could write here.
705 setup_git_directory();
708 parse_treeish_arg(argv, &args, prefix, remote);
709 parse_pathspec_arg(argv + 1, &args);
711 rc = ar->write_archive(ar, &args);
713 string_list_clear_func(&args.extra_files, extra_file_info_clear);
714 free(args.refname);
716 return rc;
719 static int match_extension(const char *filename, const char *ext)
721 int prefixlen = strlen(filename) - strlen(ext);
724 * We need 1 character for the '.', and 1 character to ensure that the
725 * prefix is non-empty (k.e., we don't match .tar.gz with no actual
726 * filename).
728 if (prefixlen < 2 || filename[prefixlen - 1] != '.')
729 return 0;
730 return !strcmp(filename + prefixlen, ext);
733 const char *archive_format_from_filename(const char *filename)
735 int i;
737 for (i = 0; i < nr_archivers; i++)
738 if (match_extension(filename, archivers[i]->name))
739 return archivers[i]->name;
740 return NULL;