Merge branch 'ds/more-index-cleanups'
[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)
41 {
42         char *to_free = NULL;
43         struct strbuf fmt = STRBUF_INIT;
44         struct pretty_print_context ctx = {0};
45         ctx.date_mode.type = DATE_NORMAL;
46         ctx.abbrev = DEFAULT_ABBREV;
47
48         if (src == buf->buf)
49                 to_free = strbuf_detach(buf, NULL);
50         for (;;) {
51                 const char *b, *c;
52
53                 b = memmem(src, len, "$Format:", 8);
54                 if (!b)
55                         break;
56                 c = memchr(b + 8, '$', (src + len) - b - 8);
57                 if (!c)
58                         break;
59
60                 strbuf_reset(&fmt);
61                 strbuf_add(&fmt, b + 8, c - b - 8);
62
63                 strbuf_add(buf, src, b - src);
64                 format_commit_message(commit, fmt.buf, buf, &ctx);
65                 len -= c + 1 - src;
66                 src  = c + 1;
67         }
68         strbuf_add(buf, src, len);
69         strbuf_release(&fmt);
70         free(to_free);
71 }
72
73 static void *object_file_to_archive(const struct archiver_args *args,
74                                     const char *path,
75                                     const struct object_id *oid,
76                                     unsigned int mode,
77                                     enum object_type *type,
78                                     unsigned long *sizep)
79 {
80         void *buffer;
81         const struct commit *commit = args->convert ? args->commit : NULL;
82         struct checkout_metadata meta;
83
84         init_checkout_metadata(&meta, args->refname,
85                                args->commit_oid ? args->commit_oid :
86                                (args->tree ? &args->tree->object.oid : NULL), oid);
87
88         path += args->baselen;
89         buffer = read_object_file(oid, type, sizep);
90         if (buffer && S_ISREG(mode)) {
91                 struct strbuf buf = STRBUF_INIT;
92                 size_t size = 0;
93
94                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
95                 convert_to_working_tree(args->repo->index, path, buf.buf, buf.len, &buf, &meta);
96                 if (commit)
97                         format_subst(commit, buf.buf, buf.len, &buf);
98                 buffer = strbuf_detach(&buf, &size);
99                 *sizep = size;
100         }
101
102         return buffer;
103 }
104
105 struct directory {
106         struct directory *up;
107         struct object_id oid;
108         int baselen, len;
109         unsigned mode;
110         int stage;
111         char path[FLEX_ARRAY];
112 };
113
114 struct archiver_context {
115         struct archiver_args *args;
116         write_archive_entry_fn_t write_entry;
117         struct directory *bottom;
118 };
119
120 static const struct attr_check *get_archive_attrs(struct index_state *istate,
121                                                   const char *path)
122 {
123         static struct attr_check *check;
124         if (!check)
125                 check = attr_check_initl("export-ignore", "export-subst", NULL);
126         git_check_attr(istate, path, check);
127         return check;
128 }
129
130 static int check_attr_export_ignore(const struct attr_check *check)
131 {
132         return check && ATTR_TRUE(check->items[0].value);
133 }
134
135 static int check_attr_export_subst(const struct attr_check *check)
136 {
137         return check && ATTR_TRUE(check->items[1].value);
138 }
139
140 static int write_archive_entry(const struct object_id *oid, const char *base,
141                 int baselen, const char *filename, unsigned mode, int stage,
142                 void *context)
143 {
144         static struct strbuf path = STRBUF_INIT;
145         struct archiver_context *c = context;
146         struct archiver_args *args = c->args;
147         write_archive_entry_fn_t write_entry = c->write_entry;
148         int err;
149         const char *path_without_prefix;
150         unsigned long size;
151         void *buffer;
152         enum object_type type;
153
154         args->convert = 0;
155         strbuf_reset(&path);
156         strbuf_grow(&path, PATH_MAX);
157         strbuf_add(&path, args->base, args->baselen);
158         strbuf_add(&path, base, baselen);
159         strbuf_addstr(&path, filename);
160         if (S_ISDIR(mode) || S_ISGITLINK(mode))
161                 strbuf_addch(&path, '/');
162         path_without_prefix = path.buf + args->baselen;
163
164         if (!S_ISDIR(mode)) {
165                 const struct attr_check *check;
166                 check = get_archive_attrs(args->repo->index, path_without_prefix);
167                 if (check_attr_export_ignore(check))
168                         return 0;
169                 args->convert = check_attr_export_subst(check);
170         }
171
172         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
173                 if (args->verbose)
174                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
175                 err = write_entry(args, oid, path.buf, path.len, mode, NULL, 0);
176                 if (err)
177                         return err;
178                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
179         }
180
181         if (args->verbose)
182                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
183
184         /* Stream it? */
185         if (S_ISREG(mode) && !args->convert &&
186             oid_object_info(args->repo, oid, &size) == OBJ_BLOB &&
187             size > big_file_threshold)
188                 return write_entry(args, oid, path.buf, path.len, mode, NULL, size);
189
190         buffer = object_file_to_archive(args, path.buf, oid, mode, &type, &size);
191         if (!buffer)
192                 return error(_("cannot read %s"), oid_to_hex(oid));
193         err = write_entry(args, oid, path.buf, path.len, mode, buffer, size);
194         free(buffer);
195         return err;
196 }
197
198 static void queue_directory(const unsigned char *sha1,
199                 struct strbuf *base, const char *filename,
200                 unsigned mode, int stage, struct archiver_context *c)
201 {
202         struct directory *d;
203         size_t len = st_add4(base->len, 1, strlen(filename), 1);
204         d = xmalloc(st_add(sizeof(*d), len));
205         d->up      = c->bottom;
206         d->baselen = base->len;
207         d->mode    = mode;
208         d->stage   = stage;
209         c->bottom  = d;
210         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
211         hashcpy(d->oid.hash, sha1);
212 }
213
214 static int write_directory(struct archiver_context *c)
215 {
216         struct directory *d = c->bottom;
217         int ret;
218
219         if (!d)
220                 return 0;
221         c->bottom = d->up;
222         d->path[d->len - 1] = '\0'; /* no trailing slash */
223         ret =
224                 write_directory(c) ||
225                 write_archive_entry(&d->oid, d->path, d->baselen,
226                                     d->path + d->baselen, d->mode,
227                                     d->stage, c) != READ_TREE_RECURSIVE;
228         free(d);
229         return ret ? -1 : 0;
230 }
231
232 static int queue_or_write_archive_entry(const struct object_id *oid,
233                 struct strbuf *base, const char *filename,
234                 unsigned mode, int stage, void *context)
235 {
236         struct archiver_context *c = context;
237
238         while (c->bottom &&
239                !(base->len >= c->bottom->len &&
240                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
241                 struct directory *next = c->bottom->up;
242                 free(c->bottom);
243                 c->bottom = next;
244         }
245
246         if (S_ISDIR(mode)) {
247                 size_t baselen = base->len;
248                 const struct attr_check *check;
249
250                 /* Borrow base, but restore its original value when done. */
251                 strbuf_addstr(base, filename);
252                 strbuf_addch(base, '/');
253                 check = get_archive_attrs(c->args->repo->index, base->buf);
254                 strbuf_setlen(base, baselen);
255
256                 if (check_attr_export_ignore(check))
257                         return 0;
258                 queue_directory(oid->hash, base, filename,
259                                 mode, stage, c);
260                 return READ_TREE_RECURSIVE;
261         }
262
263         if (write_directory(c))
264                 return -1;
265         return write_archive_entry(oid, base->buf, base->len, filename, mode,
266                                    stage, context);
267 }
268
269 struct extra_file_info {
270         char *base;
271         struct stat stat;
272 };
273
274 int write_archive_entries(struct archiver_args *args,
275                 write_archive_entry_fn_t write_entry)
276 {
277         struct archiver_context context;
278         struct unpack_trees_options opts;
279         struct tree_desc t;
280         int err;
281         struct strbuf path_in_archive = STRBUF_INIT;
282         struct strbuf content = STRBUF_INIT;
283         struct object_id fake_oid = null_oid;
284         int i;
285
286         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
287                 size_t len = args->baselen;
288
289                 while (len > 1 && args->base[len - 2] == '/')
290                         len--;
291                 if (args->verbose)
292                         fprintf(stderr, "%.*s\n", (int)len, args->base);
293                 err = write_entry(args, &args->tree->object.oid, args->base,
294                                   len, 040777, NULL, 0);
295                 if (err)
296                         return err;
297         }
298
299         memset(&context, 0, sizeof(context));
300         context.args = args;
301         context.write_entry = write_entry;
302
303         /*
304          * Setup index and instruct attr to read index only
305          */
306         if (!args->worktree_attributes) {
307                 memset(&opts, 0, sizeof(opts));
308                 opts.index_only = 1;
309                 opts.head_idx = -1;
310                 opts.src_index = args->repo->index;
311                 opts.dst_index = args->repo->index;
312                 opts.fn = oneway_merge;
313                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
314                 if (unpack_trees(1, &t, &opts))
315                         return -1;
316                 git_attr_set_direction(GIT_ATTR_INDEX);
317         }
318
319         err = read_tree_recursive(args->repo, args->tree, "",
320                                   0, 0, &args->pathspec,
321                                   queue_or_write_archive_entry,
322                                   &context);
323         if (err == READ_TREE_RECURSIVE)
324                 err = 0;
325         while (context.bottom) {
326                 struct directory *next = context.bottom->up;
327                 free(context.bottom);
328                 context.bottom = next;
329         }
330
331         for (i = 0; i < args->extra_files.nr; i++) {
332                 struct string_list_item *item = args->extra_files.items + i;
333                 char *path = item->string;
334                 struct extra_file_info *info = item->util;
335
336                 put_be64(fake_oid.hash, i + 1);
337
338                 strbuf_reset(&path_in_archive);
339                 if (info->base)
340                         strbuf_addstr(&path_in_archive, info->base);
341                 strbuf_addstr(&path_in_archive, basename(path));
342
343                 strbuf_reset(&content);
344                 if (strbuf_read_file(&content, path, info->stat.st_size) < 0)
345                         err = error_errno(_("could not read '%s'"), path);
346                 else
347                         err = write_entry(args, &fake_oid, path_in_archive.buf,
348                                           path_in_archive.len,
349                                           info->stat.st_mode,
350                                           content.buf, content.len);
351                 if (err)
352                         break;
353         }
354         strbuf_release(&path_in_archive);
355         strbuf_release(&content);
356
357         return err;
358 }
359
360 static const struct archiver *lookup_archiver(const char *name)
361 {
362         int i;
363
364         if (!name)
365                 return NULL;
366
367         for (i = 0; i < nr_archivers; i++) {
368                 if (!strcmp(name, archivers[i]->name))
369                         return archivers[i];
370         }
371         return NULL;
372 }
373
374 struct path_exists_context {
375         struct pathspec pathspec;
376         struct archiver_args *args;
377 };
378
379 static int reject_entry(const struct object_id *oid, struct strbuf *base,
380                         const char *filename, unsigned mode,
381                         int stage, void *context)
382 {
383         int ret = -1;
384         struct path_exists_context *ctx = context;
385
386         if (S_ISDIR(mode)) {
387                 struct strbuf sb = STRBUF_INIT;
388                 strbuf_addbuf(&sb, base);
389                 strbuf_addstr(&sb, filename);
390                 if (!match_pathspec(ctx->args->repo->index,
391                                     &ctx->pathspec,
392                                     sb.buf, sb.len, 0, NULL, 1))
393                         ret = READ_TREE_RECURSIVE;
394                 strbuf_release(&sb);
395         }
396         return ret;
397 }
398
399 static int path_exists(struct archiver_args *args, const char *path)
400 {
401         const char *paths[] = { path, NULL };
402         struct path_exists_context ctx;
403         int ret;
404
405         ctx.args = args;
406         parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
407         ctx.pathspec.recursive = 1;
408         ret = read_tree_recursive(args->repo, args->tree, "",
409                                   0, 0, &ctx.pathspec,
410                                   reject_entry, &ctx);
411         clear_pathspec(&ctx.pathspec);
412         return ret != 0;
413 }
414
415 static void parse_pathspec_arg(const char **pathspec,
416                 struct archiver_args *ar_args)
417 {
418         /*
419          * must be consistent with parse_pathspec in path_exists()
420          * Also if pathspec patterns are dependent, we're in big
421          * trouble as we test each one separately
422          */
423         parse_pathspec(&ar_args->pathspec, 0,
424                        PATHSPEC_PREFER_FULL,
425                        "", pathspec);
426         ar_args->pathspec.recursive = 1;
427         if (pathspec) {
428                 while (*pathspec) {
429                         if (**pathspec && !path_exists(ar_args, *pathspec))
430                                 die(_("pathspec '%s' did not match any files"), *pathspec);
431                         pathspec++;
432                 }
433         }
434 }
435
436 static void parse_treeish_arg(const char **argv,
437                 struct archiver_args *ar_args, const char *prefix,
438                 int remote)
439 {
440         const char *name = argv[0];
441         const struct object_id *commit_oid;
442         time_t archive_time;
443         struct tree *tree;
444         const struct commit *commit;
445         struct object_id oid;
446         char *ref = NULL;
447
448         /* Remotes are only allowed to fetch actual refs */
449         if (remote && !remote_allow_unreachable) {
450                 const char *colon = strchrnul(name, ':');
451                 int refnamelen = colon - name;
452
453                 if (!dwim_ref(name, refnamelen, &oid, &ref, 0))
454                         die(_("no such ref: %.*s"), refnamelen, name);
455         } else {
456                 dwim_ref(name, strlen(name), &oid, &ref, 0);
457         }
458
459         if (get_oid(name, &oid))
460                 die(_("not a valid object name: %s"), name);
461
462         commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
463         if (commit) {
464                 commit_oid = &commit->object.oid;
465                 archive_time = commit->date;
466         } else {
467                 commit_oid = NULL;
468                 archive_time = time(NULL);
469         }
470
471         tree = parse_tree_indirect(&oid);
472         if (tree == NULL)
473                 die(_("not a tree object: %s"), oid_to_hex(&oid));
474
475         if (prefix) {
476                 struct object_id tree_oid;
477                 unsigned short mode;
478                 int err;
479
480                 err = get_tree_entry(ar_args->repo,
481                                      &tree->object.oid,
482                                      prefix, &tree_oid,
483                                      &mode);
484                 if (err || !S_ISDIR(mode))
485                         die(_("current working directory is untracked"));
486
487                 tree = parse_tree_indirect(&tree_oid);
488         }
489         ar_args->refname = ref;
490         ar_args->tree = tree;
491         ar_args->commit_oid = commit_oid;
492         ar_args->commit = commit;
493         ar_args->time = archive_time;
494 }
495
496 static void extra_file_info_clear(void *util, const char *str)
497 {
498         struct extra_file_info *info = util;
499         free(info->base);
500         free(info);
501 }
502
503 static int add_file_cb(const struct option *opt, const char *arg, int unset)
504 {
505         struct archiver_args *args = opt->value;
506         const char **basep = (const char **)opt->defval;
507         const char *base = *basep;
508         char *path;
509         struct string_list_item *item;
510         struct extra_file_info *info;
511
512         if (unset) {
513                 string_list_clear_func(&args->extra_files,
514                                        extra_file_info_clear);
515                 return 0;
516         }
517
518         if (!arg)
519                 return -1;
520
521         path = prefix_filename(args->prefix, arg);
522         item = string_list_append_nodup(&args->extra_files, path);
523         item->util = info = xmalloc(sizeof(*info));
524         info->base = xstrdup_or_null(base);
525         if (stat(path, &info->stat))
526                 die(_("File not found: %s"), path);
527         if (!S_ISREG(info->stat.st_mode))
528                 die(_("Not a regular file: %s"), path);
529         return 0;
530 }
531
532 static int number_callback(const struct option *opt, const char *arg, int unset)
533 {
534         BUG_ON_OPT_NEG(unset);
535         *(int *)opt->value = strtol(arg, NULL, 10);
536         return 0;
537 }
538
539 static int parse_archive_args(int argc, const char **argv,
540                 const struct archiver **ar, struct archiver_args *args,
541                 const char *name_hint, int is_remote)
542 {
543         const char *format = NULL;
544         const char *base = NULL;
545         const char *remote = NULL;
546         const char *exec = NULL;
547         const char *output = NULL;
548         int compression_level = -1;
549         int verbose = 0;
550         int i;
551         int list = 0;
552         int worktree_attributes = 0;
553         struct option opts[] = {
554                 OPT_GROUP(""),
555                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
556                 OPT_STRING(0, "prefix", &base, N_("prefix"),
557                         N_("prepend prefix to each pathname in the archive")),
558                 { OPTION_CALLBACK, 0, "add-file", args, N_("file"),
559                   N_("add untracked file to archive"), 0, add_file_cb,
560                   (intptr_t)&base },
561                 OPT_STRING('o', "output", &output, N_("file"),
562                         N_("write the archive to this file")),
563                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
564                         N_("read .gitattributes in working directory")),
565                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
566                 OPT_NUMBER_CALLBACK(&compression_level,
567                         N_("set compression level"), number_callback),
568                 OPT_GROUP(""),
569                 OPT_BOOL('l', "list", &list,
570                         N_("list supported archive formats")),
571                 OPT_GROUP(""),
572                 OPT_STRING(0, "remote", &remote, N_("repo"),
573                         N_("retrieve the archive from remote repository <repo>")),
574                 OPT_STRING(0, "exec", &exec, N_("command"),
575                         N_("path to the remote git-upload-archive command")),
576                 OPT_END()
577         };
578
579         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
580
581         if (remote)
582                 die(_("Unexpected option --remote"));
583         if (exec)
584                 die(_("Option --exec can only be used together with --remote"));
585         if (output)
586                 die(_("Unexpected option --output"));
587         if (is_remote && args->extra_files.nr)
588                 die(_("Options --add-file and --remote cannot be used together"));
589
590         if (!base)
591                 base = "";
592
593         if (list) {
594                 for (i = 0; i < nr_archivers; i++)
595                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
596                                 printf("%s\n", archivers[i]->name);
597                 exit(0);
598         }
599
600         if (!format && name_hint)
601                 format = archive_format_from_filename(name_hint);
602         if (!format)
603                 format = "tar";
604
605         /* We need at least one parameter -- tree-ish */
606         if (argc < 1)
607                 usage_with_options(archive_usage, opts);
608         *ar = lookup_archiver(format);
609         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
610                 die(_("Unknown archive format '%s'"), format);
611
612         args->compression_level = Z_DEFAULT_COMPRESSION;
613         if (compression_level != -1) {
614                 int levels_ok = (*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS;
615                 int high_ok = (*ar)->flags & ARCHIVER_HIGH_COMPRESSION_LEVELS;
616                 if (levels_ok && (compression_level <= 9 || high_ok))
617                         args->compression_level = compression_level;
618                 else {
619                         die(_("Argument not supported for format '%s': -%d"),
620                                         format, compression_level);
621                 }
622         }
623         args->verbose = verbose;
624         args->base = base;
625         args->baselen = strlen(base);
626         args->worktree_attributes = worktree_attributes;
627
628         return argc;
629 }
630
631 int write_archive(int argc, const char **argv, const char *prefix,
632                   struct repository *repo,
633                   const char *name_hint, int remote)
634 {
635         const struct archiver *ar = NULL;
636         struct archiver_args args;
637         int rc;
638
639         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
640         git_config(git_default_config, NULL);
641
642         args.repo = repo;
643         args.prefix = prefix;
644         string_list_init(&args.extra_files, 1);
645         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
646         if (!startup_info->have_repository) {
647                 /*
648                  * We know this will die() with an error, so we could just
649                  * die ourselves; but its error message will be more specific
650                  * than what we could write here.
651                  */
652                 setup_git_directory();
653         }
654
655         parse_treeish_arg(argv, &args, prefix, remote);
656         parse_pathspec_arg(argv + 1, &args);
657
658         rc = ar->write_archive(ar, &args);
659
660         string_list_clear_func(&args.extra_files, extra_file_info_clear);
661         free(args.refname);
662
663         return rc;
664 }
665
666 static int match_extension(const char *filename, const char *ext)
667 {
668         int prefixlen = strlen(filename) - strlen(ext);
669
670         /*
671          * We need 1 character for the '.', and 1 character to ensure that the
672          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
673          * filename).
674          */
675         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
676                 return 0;
677         return !strcmp(filename + prefixlen, ext);
678 }
679
680 const char *archive_format_from_filename(const char *filename)
681 {
682         int i;
683
684         for (i = 0; i < nr_archivers; i++)
685                 if (match_extension(filename, archivers[i]->name))
686                         return archivers[i]->name;
687         return NULL;
688 }