]> git.ipfire.org Git - thirdparty/git.git/blob - archive.c
alloc.h: move ALLOC_GROW() functions from cache.h
[thirdparty/git.git] / archive.c
1 #include "git-compat-util.h"
2 #include "alloc.h"
3 #include "config.h"
4 #include "refs.h"
5 #include "object-store.h"
6 #include "commit.h"
7 #include "tree-walk.h"
8 #include "attr.h"
9 #include "archive.h"
10 #include "parse-options.h"
11 #include "unpack-trees.h"
12 #include "dir.h"
13 #include "quote.h"
14
15 static char const * const archive_usage[] = {
16 N_("git archive [<options>] <tree-ish> [<path>...]"),
17 "git archive --list",
18 N_("git archive --remote <repo> [--exec <cmd>] [<options>] <tree-ish> [<path>...]"),
19 N_("git archive --remote <repo> [--exec <cmd>] --list"),
20 NULL
21 };
22
23 static const struct archiver **archivers;
24 static int nr_archivers;
25 static int alloc_archivers;
26 static int remote_allow_unreachable;
27
28 void register_archiver(struct archiver *ar)
29 {
30 ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
31 archivers[nr_archivers++] = ar;
32 }
33
34 void init_archivers(void)
35 {
36 init_tar_archiver();
37 init_zip_archiver();
38 }
39
40 static void format_subst(const struct commit *commit,
41 const char *src, size_t len,
42 struct strbuf *buf, struct pretty_print_context *ctx)
43 {
44 char *to_free = NULL;
45 struct strbuf fmt = STRBUF_INIT;
46
47 if (src == buf->buf)
48 to_free = strbuf_detach(buf, NULL);
49 for (;;) {
50 const char *b, *c;
51
52 b = memmem(src, len, "$Format:", 8);
53 if (!b)
54 break;
55 c = memchr(b + 8, '$', (src + len) - b - 8);
56 if (!c)
57 break;
58
59 strbuf_reset(&fmt);
60 strbuf_add(&fmt, b + 8, c - b - 8);
61
62 strbuf_add(buf, src, b - src);
63 format_commit_message(commit, fmt.buf, buf, ctx);
64 len -= c + 1 - src;
65 src = c + 1;
66 }
67 strbuf_add(buf, src, len);
68 strbuf_release(&fmt);
69 free(to_free);
70 }
71
72 static void *object_file_to_archive(const struct archiver_args *args,
73 const char *path,
74 const struct object_id *oid,
75 unsigned int mode,
76 enum object_type *type,
77 unsigned long *sizep)
78 {
79 void *buffer;
80 const struct commit *commit = args->convert ? args->commit : NULL;
81 struct checkout_metadata meta;
82
83 init_checkout_metadata(&meta, args->refname,
84 args->commit_oid ? args->commit_oid :
85 (args->tree ? &args->tree->object.oid : NULL), oid);
86
87 path += args->baselen;
88 buffer = read_object_file(oid, type, sizep);
89 if (buffer && S_ISREG(mode)) {
90 struct strbuf buf = STRBUF_INIT;
91 size_t size = 0;
92
93 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
94 convert_to_working_tree(args->repo->index, path, buf.buf, buf.len, &buf, &meta);
95 if (commit)
96 format_subst(commit, buf.buf, buf.len, &buf, args->pretty_ctx);
97 buffer = strbuf_detach(&buf, &size);
98 *sizep = size;
99 }
100
101 return buffer;
102 }
103
104 struct directory {
105 struct directory *up;
106 struct object_id oid;
107 int baselen, len;
108 unsigned mode;
109 char path[FLEX_ARRAY];
110 };
111
112 struct archiver_context {
113 struct archiver_args *args;
114 write_archive_entry_fn_t write_entry;
115 struct directory *bottom;
116 };
117
118 static const struct attr_check *get_archive_attrs(struct index_state *istate,
119 const char *path)
120 {
121 static struct attr_check *check;
122 if (!check)
123 check = attr_check_initl("export-ignore", "export-subst", NULL);
124 git_check_attr(istate, NULL, path, check);
125 return check;
126 }
127
128 static int check_attr_export_ignore(const struct attr_check *check)
129 {
130 return check && ATTR_TRUE(check->items[0].value);
131 }
132
133 static int check_attr_export_subst(const struct attr_check *check)
134 {
135 return check && ATTR_TRUE(check->items[1].value);
136 }
137
138 static int write_archive_entry(const struct object_id *oid, const char *base,
139 int baselen, const char *filename, unsigned mode,
140 void *context)
141 {
142 static struct strbuf path = STRBUF_INIT;
143 struct archiver_context *c = context;
144 struct archiver_args *args = c->args;
145 write_archive_entry_fn_t write_entry = c->write_entry;
146 int err;
147 const char *path_without_prefix;
148 unsigned long size;
149 void *buffer;
150 enum object_type type;
151
152 args->convert = 0;
153 strbuf_reset(&path);
154 strbuf_grow(&path, PATH_MAX);
155 strbuf_add(&path, args->base, args->baselen);
156 strbuf_add(&path, base, baselen);
157 strbuf_addstr(&path, filename);
158 if (S_ISDIR(mode) || S_ISGITLINK(mode))
159 strbuf_addch(&path, '/');
160 path_without_prefix = path.buf + args->baselen;
161
162 if (!S_ISDIR(mode)) {
163 const struct attr_check *check;
164 check = get_archive_attrs(args->repo->index, path_without_prefix);
165 if (check_attr_export_ignore(check))
166 return 0;
167 args->convert = check_attr_export_subst(check);
168 }
169
170 if (args->verbose)
171 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
172
173 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
174 err = write_entry(args, oid, path.buf, path.len, mode, NULL, 0);
175 if (err)
176 return err;
177 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
178 }
179
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);
185
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;
192 }
193
194 static void queue_directory(const struct object_id *oid,
195 struct strbuf *base, const char *filename,
196 unsigned mode, struct archiver_context *c)
197 {
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);
207 }
208
209 static int write_directory(struct archiver_context *c)
210 {
211 struct directory *d = c->bottom;
212 int ret;
213
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;
225 }
226
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)
230 {
231 struct archiver_context *c = context;
232
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;
239 }
240
241 if (S_ISDIR(mode)) {
242 size_t baselen = base->len;
243 const struct attr_check *check;
244
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);
250
251 if (check_attr_export_ignore(check))
252 return 0;
253 queue_directory(oid, base, filename, mode, c);
254 return READ_TREE_RECURSIVE;
255 }
256
257 if (write_directory(c))
258 return -1;
259 return write_archive_entry(oid, base->buf, base->len, filename, mode,
260 context);
261 }
262
263 struct extra_file_info {
264 char *base;
265 struct stat stat;
266 void *content;
267 };
268
269 int write_archive_entries(struct archiver_args *args,
270 write_archive_entry_fn_t write_entry)
271 {
272 struct archiver_context context;
273 struct unpack_trees_options opts;
274 struct tree_desc t;
275 int err;
276 struct strbuf path_in_archive = STRBUF_INIT;
277 struct strbuf content = STRBUF_INIT;
278 struct object_id fake_oid;
279 int i;
280
281 oidcpy(&fake_oid, null_oid());
282
283 if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
284 size_t len = args->baselen;
285
286 while (len > 1 && args->base[len - 2] == '/')
287 len--;
288 if (args->verbose)
289 fprintf(stderr, "%.*s\n", (int)len, args->base);
290 err = write_entry(args, &args->tree->object.oid, args->base,
291 len, 040777, NULL, 0);
292 if (err)
293 return err;
294 }
295
296 memset(&context, 0, sizeof(context));
297 context.args = args;
298 context.write_entry = write_entry;
299
300 /*
301 * Setup index and instruct attr to read index only
302 */
303 if (!args->worktree_attributes) {
304 memset(&opts, 0, sizeof(opts));
305 opts.index_only = 1;
306 opts.head_idx = -1;
307 opts.src_index = args->repo->index;
308 opts.dst_index = args->repo->index;
309 opts.fn = oneway_merge;
310 init_tree_desc(&t, args->tree->buffer, args->tree->size);
311 if (unpack_trees(1, &t, &opts))
312 return -1;
313 git_attr_set_direction(GIT_ATTR_INDEX);
314 }
315
316 err = read_tree(args->repo, args->tree,
317 &args->pathspec,
318 queue_or_write_archive_entry,
319 &context);
320 if (err == READ_TREE_RECURSIVE)
321 err = 0;
322 while (context.bottom) {
323 struct directory *next = context.bottom->up;
324 free(context.bottom);
325 context.bottom = next;
326 }
327
328 for (i = 0; i < args->extra_files.nr; i++) {
329 struct string_list_item *item = args->extra_files.items + i;
330 char *path = item->string;
331 struct extra_file_info *info = item->util;
332
333 put_be64(fake_oid.hash, i + 1);
334
335 if (!info->content) {
336 strbuf_reset(&path_in_archive);
337 if (info->base)
338 strbuf_addstr(&path_in_archive, info->base);
339 strbuf_addstr(&path_in_archive, basename(path));
340
341 strbuf_reset(&content);
342 if (strbuf_read_file(&content, path, info->stat.st_size) < 0)
343 err = error_errno(_("cannot read '%s'"), path);
344 else
345 err = write_entry(args, &fake_oid, path_in_archive.buf,
346 path_in_archive.len,
347 canon_mode(info->stat.st_mode),
348 content.buf, content.len);
349 } else {
350 err = write_entry(args, &fake_oid,
351 path, strlen(path),
352 canon_mode(info->stat.st_mode),
353 info->content, info->stat.st_size);
354 }
355
356 if (err)
357 break;
358 }
359 strbuf_release(&path_in_archive);
360 strbuf_release(&content);
361
362 return err;
363 }
364
365 static const struct archiver *lookup_archiver(const char *name)
366 {
367 int i;
368
369 if (!name)
370 return NULL;
371
372 for (i = 0; i < nr_archivers; i++) {
373 if (!strcmp(name, archivers[i]->name))
374 return archivers[i];
375 }
376 return NULL;
377 }
378
379 struct path_exists_context {
380 struct pathspec pathspec;
381 struct archiver_args *args;
382 };
383
384 static int reject_entry(const struct object_id *oid UNUSED,
385 struct strbuf *base,
386 const char *filename, unsigned mode,
387 void *context)
388 {
389 int ret = -1;
390 struct path_exists_context *ctx = context;
391
392 if (S_ISDIR(mode)) {
393 struct strbuf sb = STRBUF_INIT;
394 strbuf_addbuf(&sb, base);
395 strbuf_addstr(&sb, filename);
396 if (!match_pathspec(ctx->args->repo->index,
397 &ctx->pathspec,
398 sb.buf, sb.len, 0, NULL, 1))
399 ret = READ_TREE_RECURSIVE;
400 strbuf_release(&sb);
401 }
402 return ret;
403 }
404
405 static int path_exists(struct archiver_args *args, const char *path)
406 {
407 const char *paths[] = { path, NULL };
408 struct path_exists_context ctx;
409 int ret;
410
411 ctx.args = args;
412 parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
413 ctx.pathspec.recursive = 1;
414 ret = read_tree(args->repo, args->tree,
415 &ctx.pathspec,
416 reject_entry, &ctx);
417 clear_pathspec(&ctx.pathspec);
418 return ret != 0;
419 }
420
421 static void parse_pathspec_arg(const char **pathspec,
422 struct archiver_args *ar_args)
423 {
424 /*
425 * must be consistent with parse_pathspec in path_exists()
426 * Also if pathspec patterns are dependent, we're in big
427 * trouble as we test each one separately
428 */
429 parse_pathspec(&ar_args->pathspec, 0,
430 PATHSPEC_PREFER_FULL,
431 "", pathspec);
432 ar_args->pathspec.recursive = 1;
433 if (pathspec) {
434 while (*pathspec) {
435 if (**pathspec && !path_exists(ar_args, *pathspec))
436 die(_("pathspec '%s' did not match any files"), *pathspec);
437 pathspec++;
438 }
439 }
440 }
441
442 static void parse_treeish_arg(const char **argv,
443 struct archiver_args *ar_args, const char *prefix,
444 int remote)
445 {
446 const char *name = argv[0];
447 const struct object_id *commit_oid;
448 time_t archive_time;
449 struct tree *tree;
450 const struct commit *commit;
451 struct object_id oid;
452 char *ref = NULL;
453
454 /* Remotes are only allowed to fetch actual refs */
455 if (remote && !remote_allow_unreachable) {
456 const char *colon = strchrnul(name, ':');
457 int refnamelen = colon - name;
458
459 if (!dwim_ref(name, refnamelen, &oid, &ref, 0))
460 die(_("no such ref: %.*s"), refnamelen, name);
461 } else {
462 dwim_ref(name, strlen(name), &oid, &ref, 0);
463 }
464
465 if (get_oid(name, &oid))
466 die(_("not a valid object name: %s"), name);
467
468 commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
469 if (commit) {
470 commit_oid = &commit->object.oid;
471 archive_time = commit->date;
472 } else {
473 commit_oid = NULL;
474 archive_time = time(NULL);
475 }
476
477 tree = parse_tree_indirect(&oid);
478 if (!tree)
479 die(_("not a tree object: %s"), oid_to_hex(&oid));
480
481 if (prefix) {
482 struct object_id tree_oid;
483 unsigned short mode;
484 int err;
485
486 err = get_tree_entry(ar_args->repo,
487 &tree->object.oid,
488 prefix, &tree_oid,
489 &mode);
490 if (err || !S_ISDIR(mode))
491 die(_("current working directory is untracked"));
492
493 tree = parse_tree_indirect(&tree_oid);
494 }
495 ar_args->refname = ref;
496 ar_args->tree = tree;
497 ar_args->commit_oid = commit_oid;
498 ar_args->commit = commit;
499 ar_args->time = archive_time;
500 }
501
502 static void extra_file_info_clear(void *util, const char *str UNUSED)
503 {
504 struct extra_file_info *info = util;
505 free(info->base);
506 free(info->content);
507 free(info);
508 }
509
510 static int add_file_cb(const struct option *opt, const char *arg, int unset)
511 {
512 struct archiver_args *args = opt->value;
513 const char **basep = (const char **)opt->defval;
514 const char *base = *basep;
515 char *path;
516 struct string_list_item *item;
517 struct extra_file_info *info;
518
519 if (unset) {
520 string_list_clear_func(&args->extra_files,
521 extra_file_info_clear);
522 return 0;
523 }
524
525 if (!arg)
526 return -1;
527
528 info = xmalloc(sizeof(*info));
529 info->base = xstrdup_or_null(base);
530
531 if (!strcmp(opt->long_name, "add-file")) {
532 path = prefix_filename(args->prefix, arg);
533 if (stat(path, &info->stat))
534 die(_("File not found: %s"), path);
535 if (!S_ISREG(info->stat.st_mode))
536 die(_("Not a regular file: %s"), path);
537 info->content = NULL; /* read the file later */
538 } else if (!strcmp(opt->long_name, "add-virtual-file")) {
539 struct strbuf buf = STRBUF_INIT;
540 const char *p = arg;
541
542 if (*p != '"')
543 p = strchr(p, ':');
544 else if (unquote_c_style(&buf, p, &p) < 0)
545 die(_("unclosed quote: '%s'"), arg);
546
547 if (!p || *p != ':')
548 die(_("missing colon: '%s'"), arg);
549
550 if (p == arg)
551 die(_("empty file name: '%s'"), arg);
552
553 path = buf.len ?
554 strbuf_detach(&buf, NULL) : xstrndup(arg, p - arg);
555
556 if (args->prefix) {
557 char *save = path;
558 path = prefix_filename(args->prefix, path);
559 free(save);
560 }
561 memset(&info->stat, 0, sizeof(info->stat));
562 info->stat.st_mode = S_IFREG | 0644;
563 info->content = xstrdup(p + 1);
564 info->stat.st_size = strlen(info->content);
565 } else {
566 BUG("add_file_cb() called for %s", opt->long_name);
567 }
568 item = string_list_append_nodup(&args->extra_files, path);
569 item->util = info;
570
571 return 0;
572 }
573
574 static int number_callback(const struct option *opt, const char *arg, int unset)
575 {
576 BUG_ON_OPT_NEG(unset);
577 *(int *)opt->value = strtol(arg, NULL, 10);
578 return 0;
579 }
580
581 static int parse_archive_args(int argc, const char **argv,
582 const struct archiver **ar, struct archiver_args *args,
583 const char *name_hint, int is_remote)
584 {
585 const char *format = NULL;
586 const char *base = NULL;
587 const char *remote = NULL;
588 const char *exec = NULL;
589 const char *output = NULL;
590 int compression_level = -1;
591 int verbose = 0;
592 int i;
593 int list = 0;
594 int worktree_attributes = 0;
595 struct option opts[] = {
596 OPT_GROUP(""),
597 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
598 OPT_STRING(0, "prefix", &base, N_("prefix"),
599 N_("prepend prefix to each pathname in the archive")),
600 { OPTION_CALLBACK, 0, "add-file", args, N_("file"),
601 N_("add untracked file to archive"), 0, add_file_cb,
602 (intptr_t)&base },
603 { OPTION_CALLBACK, 0, "add-virtual-file", args,
604 N_("path:content"), N_("add untracked file to archive"), 0,
605 add_file_cb, (intptr_t)&base },
606 OPT_STRING('o', "output", &output, N_("file"),
607 N_("write the archive to this file")),
608 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
609 N_("read .gitattributes in working directory")),
610 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
611 OPT_NUMBER_CALLBACK(&compression_level,
612 N_("set compression level"), number_callback),
613 OPT_GROUP(""),
614 OPT_BOOL('l', "list", &list,
615 N_("list supported archive formats")),
616 OPT_GROUP(""),
617 OPT_STRING(0, "remote", &remote, N_("repo"),
618 N_("retrieve the archive from remote repository <repo>")),
619 OPT_STRING(0, "exec", &exec, N_("command"),
620 N_("path to the remote git-upload-archive command")),
621 OPT_END()
622 };
623
624 argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
625
626 if (remote)
627 die(_("Unexpected option --remote"));
628 if (exec)
629 die(_("the option '%s' requires '%s'"), "--exec", "--remote");
630 if (output)
631 die(_("Unexpected option --output"));
632 if (is_remote && args->extra_files.nr)
633 die(_("options '%s' and '%s' cannot be used together"), "--add-file", "--remote");
634
635 if (!base)
636 base = "";
637
638 if (list) {
639 for (i = 0; i < nr_archivers; i++)
640 if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
641 printf("%s\n", archivers[i]->name);
642 exit(0);
643 }
644
645 if (!format && name_hint)
646 format = archive_format_from_filename(name_hint);
647 if (!format)
648 format = "tar";
649
650 /* We need at least one parameter -- tree-ish */
651 if (argc < 1)
652 usage_with_options(archive_usage, opts);
653 *ar = lookup_archiver(format);
654 if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
655 die(_("Unknown archive format '%s'"), format);
656
657 args->compression_level = Z_DEFAULT_COMPRESSION;
658 if (compression_level != -1) {
659 int levels_ok = (*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS;
660 int high_ok = (*ar)->flags & ARCHIVER_HIGH_COMPRESSION_LEVELS;
661 if (levels_ok && (compression_level <= 9 || high_ok))
662 args->compression_level = compression_level;
663 else {
664 die(_("Argument not supported for format '%s': -%d"),
665 format, compression_level);
666 }
667 }
668 args->verbose = verbose;
669 args->base = base;
670 args->baselen = strlen(base);
671 args->worktree_attributes = worktree_attributes;
672
673 return argc;
674 }
675
676 int write_archive(int argc, const char **argv, const char *prefix,
677 struct repository *repo,
678 const char *name_hint, int remote)
679 {
680 const struct archiver *ar = NULL;
681 struct pretty_print_describe_status describe_status = {0};
682 struct pretty_print_context ctx = {0};
683 struct archiver_args args;
684 int rc;
685
686 git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
687 git_config(git_default_config, NULL);
688
689 describe_status.max_invocations = 1;
690 ctx.date_mode.type = DATE_NORMAL;
691 ctx.abbrev = DEFAULT_ABBREV;
692 ctx.describe_status = &describe_status;
693 args.pretty_ctx = &ctx;
694 args.repo = repo;
695 args.prefix = prefix;
696 string_list_init_dup(&args.extra_files);
697 argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
698 if (!startup_info->have_repository) {
699 /*
700 * We know this will die() with an error, so we could just
701 * die ourselves; but its error message will be more specific
702 * than what we could write here.
703 */
704 setup_git_directory();
705 }
706
707 parse_treeish_arg(argv, &args, prefix, remote);
708 parse_pathspec_arg(argv + 1, &args);
709
710 rc = ar->write_archive(ar, &args);
711
712 string_list_clear_func(&args.extra_files, extra_file_info_clear);
713 free(args.refname);
714 clear_pathspec(&args.pathspec);
715
716 return rc;
717 }
718
719 static int match_extension(const char *filename, const char *ext)
720 {
721 int prefixlen = strlen(filename) - strlen(ext);
722
723 /*
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).
727 */
728 if (prefixlen < 2 || filename[prefixlen - 1] != '.')
729 return 0;
730 return !strcmp(filename + prefixlen, ext);
731 }
732
733 const char *archive_format_from_filename(const char *filename)
734 {
735 int i;
736
737 for (i = 0; i < nr_archivers; i++)
738 if (match_extension(filename, archivers[i]->name))
739 return archivers[i]->name;
740 return NULL;
741 }