Merge branch 'mt/add-rm-in-sparse-checkout'
[git] / archive.c
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
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
19 };
20
21 static const struct archiver **archivers;
22 static int nr_archivers;
23 static int alloc_archivers;
24 static int remote_allow_unreachable;
25
26 void register_archiver(struct archiver *ar)
27 {
28         ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
29         archivers[nr_archivers++] = ar;
30 }
31
32 void init_archivers(void)
33 {
34         init_tar_archiver();
35         init_zip_archiver();
36 }
37
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)
41 {
42         char *to_free = NULL;
43         struct strbuf fmt = STRBUF_INIT;
44
45         if (src == buf->buf)
46                 to_free = strbuf_detach(buf, NULL);
47         for (;;) {
48                 const char *b, *c;
49
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;
56
57                 strbuf_reset(&fmt);
58                 strbuf_add(&fmt, b + 8, c - b - 8);
59
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;
64         }
65         strbuf_add(buf, src, len);
66         strbuf_release(&fmt);
67         free(to_free);
68 }
69
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)
76 {
77         void *buffer;
78         const struct commit *commit = args->convert ? args->commit : NULL;
79         struct checkout_metadata meta;
80
81         init_checkout_metadata(&meta, args->refname,
82                                args->commit_oid ? args->commit_oid :
83                                (args->tree ? &args->tree->object.oid : NULL), oid);
84
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;
90
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;
97         }
98
99         return buffer;
100 }
101
102 struct directory {
103         struct directory *up;
104         struct object_id oid;
105         int baselen, len;
106         unsigned mode;
107         char path[FLEX_ARRAY];
108 };
109
110 struct archiver_context {
111         struct archiver_args *args;
112         write_archive_entry_fn_t write_entry;
113         struct directory *bottom;
114 };
115
116 static const struct attr_check *get_archive_attrs(struct index_state *istate,
117                                                   const char *path)
118 {
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;
124 }
125
126 static int check_attr_export_ignore(const struct attr_check *check)
127 {
128         return check && ATTR_TRUE(check->items[0].value);
129 }
130
131 static int check_attr_export_subst(const struct attr_check *check)
132 {
133         return check && ATTR_TRUE(check->items[1].value);
134 }
135
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)
139 {
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;
149
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;
159
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);
166         }
167
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);
175         }
176
177         if (args->verbose)
178                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
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 unsigned char *sha1,
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         hashcpy(d->oid.hash, sha1);
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->hash, base, filename,
254                                 mode, c);
255                 return READ_TREE_RECURSIVE;
256         }
257
258         if (write_directory(c))
259                 return -1;
260         return write_archive_entry(oid, base->buf, base->len, filename, mode,
261                                    context);
262 }
263
264 struct extra_file_info {
265         char *base;
266         struct stat stat;
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 = null_oid;
279         int i;
280
281         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
282                 size_t len = args->baselen;
283
284                 while (len > 1 && args->base[len - 2] == '/')
285                         len--;
286                 if (args->verbose)
287                         fprintf(stderr, "%.*s\n", (int)len, args->base);
288                 err = write_entry(args, &args->tree->object.oid, args->base,
289                                   len, 040777, NULL, 0);
290                 if (err)
291                         return err;
292         }
293
294         memset(&context, 0, sizeof(context));
295         context.args = args;
296         context.write_entry = write_entry;
297
298         /*
299          * Setup index and instruct attr to read index only
300          */
301         if (!args->worktree_attributes) {
302                 memset(&opts, 0, sizeof(opts));
303                 opts.index_only = 1;
304                 opts.head_idx = -1;
305                 opts.src_index = args->repo->index;
306                 opts.dst_index = args->repo->index;
307                 opts.fn = oneway_merge;
308                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
309                 if (unpack_trees(1, &t, &opts))
310                         return -1;
311                 git_attr_set_direction(GIT_ATTR_INDEX);
312         }
313
314         err = read_tree(args->repo, args->tree,
315                         &args->pathspec,
316                         queue_or_write_archive_entry,
317                         &context);
318         if (err == READ_TREE_RECURSIVE)
319                 err = 0;
320         while (context.bottom) {
321                 struct directory *next = context.bottom->up;
322                 free(context.bottom);
323                 context.bottom = next;
324         }
325
326         for (i = 0; i < args->extra_files.nr; i++) {
327                 struct string_list_item *item = args->extra_files.items + i;
328                 char *path = item->string;
329                 struct extra_file_info *info = item->util;
330
331                 put_be64(fake_oid.hash, i + 1);
332
333                 strbuf_reset(&path_in_archive);
334                 if (info->base)
335                         strbuf_addstr(&path_in_archive, info->base);
336                 strbuf_addstr(&path_in_archive, basename(path));
337
338                 strbuf_reset(&content);
339                 if (strbuf_read_file(&content, path, info->stat.st_size) < 0)
340                         err = error_errno(_("could not read '%s'"), path);
341                 else
342                         err = write_entry(args, &fake_oid, path_in_archive.buf,
343                                           path_in_archive.len,
344                                           info->stat.st_mode,
345                                           content.buf, content.len);
346                 if (err)
347                         break;
348         }
349         strbuf_release(&path_in_archive);
350         strbuf_release(&content);
351
352         return err;
353 }
354
355 static const struct archiver *lookup_archiver(const char *name)
356 {
357         int i;
358
359         if (!name)
360                 return NULL;
361
362         for (i = 0; i < nr_archivers; i++) {
363                 if (!strcmp(name, archivers[i]->name))
364                         return archivers[i];
365         }
366         return NULL;
367 }
368
369 struct path_exists_context {
370         struct pathspec pathspec;
371         struct archiver_args *args;
372 };
373
374 static int reject_entry(const struct object_id *oid, struct strbuf *base,
375                         const char *filename, unsigned mode,
376                         void *context)
377 {
378         int ret = -1;
379         struct path_exists_context *ctx = context;
380
381         if (S_ISDIR(mode)) {
382                 struct strbuf sb = STRBUF_INIT;
383                 strbuf_addbuf(&sb, base);
384                 strbuf_addstr(&sb, filename);
385                 if (!match_pathspec(ctx->args->repo->index,
386                                     &ctx->pathspec,
387                                     sb.buf, sb.len, 0, NULL, 1))
388                         ret = READ_TREE_RECURSIVE;
389                 strbuf_release(&sb);
390         }
391         return ret;
392 }
393
394 static int path_exists(struct archiver_args *args, const char *path)
395 {
396         const char *paths[] = { path, NULL };
397         struct path_exists_context ctx;
398         int ret;
399
400         ctx.args = args;
401         parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
402         ctx.pathspec.recursive = 1;
403         ret = read_tree(args->repo, args->tree,
404                         &ctx.pathspec,
405                         reject_entry, &ctx);
406         clear_pathspec(&ctx.pathspec);
407         return ret != 0;
408 }
409
410 static void parse_pathspec_arg(const char **pathspec,
411                 struct archiver_args *ar_args)
412 {
413         /*
414          * must be consistent with parse_pathspec in path_exists()
415          * Also if pathspec patterns are dependent, we're in big
416          * trouble as we test each one separately
417          */
418         parse_pathspec(&ar_args->pathspec, 0,
419                        PATHSPEC_PREFER_FULL,
420                        "", pathspec);
421         ar_args->pathspec.recursive = 1;
422         if (pathspec) {
423                 while (*pathspec) {
424                         if (**pathspec && !path_exists(ar_args, *pathspec))
425                                 die(_("pathspec '%s' did not match any files"), *pathspec);
426                         pathspec++;
427                 }
428         }
429 }
430
431 static void parse_treeish_arg(const char **argv,
432                 struct archiver_args *ar_args, const char *prefix,
433                 int remote)
434 {
435         const char *name = argv[0];
436         const struct object_id *commit_oid;
437         time_t archive_time;
438         struct tree *tree;
439         const struct commit *commit;
440         struct object_id oid;
441         char *ref = NULL;
442
443         /* Remotes are only allowed to fetch actual refs */
444         if (remote && !remote_allow_unreachable) {
445                 const char *colon = strchrnul(name, ':');
446                 int refnamelen = colon - name;
447
448                 if (!dwim_ref(name, refnamelen, &oid, &ref, 0))
449                         die(_("no such ref: %.*s"), refnamelen, name);
450         } else {
451                 dwim_ref(name, strlen(name), &oid, &ref, 0);
452         }
453
454         if (get_oid(name, &oid))
455                 die(_("not a valid object name: %s"), name);
456
457         commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
458         if (commit) {
459                 commit_oid = &commit->object.oid;
460                 archive_time = commit->date;
461         } else {
462                 commit_oid = NULL;
463                 archive_time = time(NULL);
464         }
465
466         tree = parse_tree_indirect(&oid);
467         if (tree == NULL)
468                 die(_("not a tree object: %s"), oid_to_hex(&oid));
469
470         if (prefix) {
471                 struct object_id tree_oid;
472                 unsigned short mode;
473                 int err;
474
475                 err = get_tree_entry(ar_args->repo,
476                                      &tree->object.oid,
477                                      prefix, &tree_oid,
478                                      &mode);
479                 if (err || !S_ISDIR(mode))
480                         die(_("current working directory is untracked"));
481
482                 tree = parse_tree_indirect(&tree_oid);
483         }
484         ar_args->refname = ref;
485         ar_args->tree = tree;
486         ar_args->commit_oid = commit_oid;
487         ar_args->commit = commit;
488         ar_args->time = archive_time;
489 }
490
491 static void extra_file_info_clear(void *util, const char *str)
492 {
493         struct extra_file_info *info = util;
494         free(info->base);
495         free(info);
496 }
497
498 static int add_file_cb(const struct option *opt, const char *arg, int unset)
499 {
500         struct archiver_args *args = opt->value;
501         const char **basep = (const char **)opt->defval;
502         const char *base = *basep;
503         char *path;
504         struct string_list_item *item;
505         struct extra_file_info *info;
506
507         if (unset) {
508                 string_list_clear_func(&args->extra_files,
509                                        extra_file_info_clear);
510                 return 0;
511         }
512
513         if (!arg)
514                 return -1;
515
516         path = prefix_filename(args->prefix, arg);
517         item = string_list_append_nodup(&args->extra_files, path);
518         item->util = info = xmalloc(sizeof(*info));
519         info->base = xstrdup_or_null(base);
520         if (stat(path, &info->stat))
521                 die(_("File not found: %s"), path);
522         if (!S_ISREG(info->stat.st_mode))
523                 die(_("Not a regular file: %s"), path);
524         return 0;
525 }
526
527 static int number_callback(const struct option *opt, const char *arg, int unset)
528 {
529         BUG_ON_OPT_NEG(unset);
530         *(int *)opt->value = strtol(arg, NULL, 10);
531         return 0;
532 }
533
534 static int parse_archive_args(int argc, const char **argv,
535                 const struct archiver **ar, struct archiver_args *args,
536                 const char *name_hint, int is_remote)
537 {
538         const char *format = NULL;
539         const char *base = NULL;
540         const char *remote = NULL;
541         const char *exec = NULL;
542         const char *output = NULL;
543         int compression_level = -1;
544         int verbose = 0;
545         int i;
546         int list = 0;
547         int worktree_attributes = 0;
548         struct option opts[] = {
549                 OPT_GROUP(""),
550                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
551                 OPT_STRING(0, "prefix", &base, N_("prefix"),
552                         N_("prepend prefix to each pathname in the archive")),
553                 { OPTION_CALLBACK, 0, "add-file", args, N_("file"),
554                   N_("add untracked file to archive"), 0, add_file_cb,
555                   (intptr_t)&base },
556                 OPT_STRING('o', "output", &output, N_("file"),
557                         N_("write the archive to this file")),
558                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
559                         N_("read .gitattributes in working directory")),
560                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
561                 OPT_NUMBER_CALLBACK(&compression_level,
562                         N_("set compression level"), number_callback),
563                 OPT_GROUP(""),
564                 OPT_BOOL('l', "list", &list,
565                         N_("list supported archive formats")),
566                 OPT_GROUP(""),
567                 OPT_STRING(0, "remote", &remote, N_("repo"),
568                         N_("retrieve the archive from remote repository <repo>")),
569                 OPT_STRING(0, "exec", &exec, N_("command"),
570                         N_("path to the remote git-upload-archive command")),
571                 OPT_END()
572         };
573
574         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
575
576         if (remote)
577                 die(_("Unexpected option --remote"));
578         if (exec)
579                 die(_("Option --exec can only be used together with --remote"));
580         if (output)
581                 die(_("Unexpected option --output"));
582         if (is_remote && args->extra_files.nr)
583                 die(_("Options --add-file and --remote cannot be used together"));
584
585         if (!base)
586                 base = "";
587
588         if (list) {
589                 for (i = 0; i < nr_archivers; i++)
590                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
591                                 printf("%s\n", archivers[i]->name);
592                 exit(0);
593         }
594
595         if (!format && name_hint)
596                 format = archive_format_from_filename(name_hint);
597         if (!format)
598                 format = "tar";
599
600         /* We need at least one parameter -- tree-ish */
601         if (argc < 1)
602                 usage_with_options(archive_usage, opts);
603         *ar = lookup_archiver(format);
604         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
605                 die(_("Unknown archive format '%s'"), format);
606
607         args->compression_level = Z_DEFAULT_COMPRESSION;
608         if (compression_level != -1) {
609                 int levels_ok = (*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS;
610                 int high_ok = (*ar)->flags & ARCHIVER_HIGH_COMPRESSION_LEVELS;
611                 if (levels_ok && (compression_level <= 9 || high_ok))
612                         args->compression_level = compression_level;
613                 else {
614                         die(_("Argument not supported for format '%s': -%d"),
615                                         format, compression_level);
616                 }
617         }
618         args->verbose = verbose;
619         args->base = base;
620         args->baselen = strlen(base);
621         args->worktree_attributes = worktree_attributes;
622
623         return argc;
624 }
625
626 int write_archive(int argc, const char **argv, const char *prefix,
627                   struct repository *repo,
628                   const char *name_hint, int remote)
629 {
630         const struct archiver *ar = NULL;
631         struct pretty_print_describe_status describe_status = {0};
632         struct pretty_print_context ctx = {0};
633         struct archiver_args args;
634         int rc;
635
636         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
637         git_config(git_default_config, NULL);
638
639         describe_status.max_invocations = 1;
640         ctx.date_mode.type = DATE_NORMAL;
641         ctx.abbrev = DEFAULT_ABBREV;
642         ctx.describe_status = &describe_status;
643         args.pretty_ctx = &ctx;
644         args.repo = repo;
645         args.prefix = prefix;
646         string_list_init(&args.extra_files, 1);
647         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
648         if (!startup_info->have_repository) {
649                 /*
650                  * We know this will die() with an error, so we could just
651                  * die ourselves; but its error message will be more specific
652                  * than what we could write here.
653                  */
654                 setup_git_directory();
655         }
656
657         parse_treeish_arg(argv, &args, prefix, remote);
658         parse_pathspec_arg(argv + 1, &args);
659
660         rc = ar->write_archive(ar, &args);
661
662         string_list_clear_func(&args.extra_files, extra_file_info_clear);
663         free(args.refname);
664
665         return rc;
666 }
667
668 static int match_extension(const char *filename, const char *ext)
669 {
670         int prefixlen = strlen(filename) - strlen(ext);
671
672         /*
673          * We need 1 character for the '.', and 1 character to ensure that the
674          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
675          * filename).
676          */
677         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
678                 return 0;
679         return !strcmp(filename + prefixlen, ext);
680 }
681
682 const char *archive_format_from_filename(const char *filename)
683 {
684         int i;
685
686         for (i = 0; i < nr_archivers; i++)
687                 if (match_extension(filename, archivers[i]->name))
688                         return archivers[i]->name;
689         return NULL;
690 }