The second batch
[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         oidread(&d->oid, 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;
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                 strbuf_reset(&path_in_archive);
336                 if (info->base)
337                         strbuf_addstr(&path_in_archive, info->base);
338                 strbuf_addstr(&path_in_archive, basename(path));
339
340                 strbuf_reset(&content);
341                 if (strbuf_read_file(&content, path, info->stat.st_size) < 0)
342                         err = error_errno(_("could not read '%s'"), path);
343                 else
344                         err = write_entry(args, &fake_oid, path_in_archive.buf,
345                                           path_in_archive.len,
346                                           info->stat.st_mode,
347                                           content.buf, content.len);
348                 if (err)
349                         break;
350         }
351         strbuf_release(&path_in_archive);
352         strbuf_release(&content);
353
354         return err;
355 }
356
357 static const struct archiver *lookup_archiver(const char *name)
358 {
359         int i;
360
361         if (!name)
362                 return NULL;
363
364         for (i = 0; i < nr_archivers; i++) {
365                 if (!strcmp(name, archivers[i]->name))
366                         return archivers[i];
367         }
368         return NULL;
369 }
370
371 struct path_exists_context {
372         struct pathspec pathspec;
373         struct archiver_args *args;
374 };
375
376 static int reject_entry(const struct object_id *oid, struct strbuf *base,
377                         const char *filename, unsigned mode,
378                         void *context)
379 {
380         int ret = -1;
381         struct path_exists_context *ctx = context;
382
383         if (S_ISDIR(mode)) {
384                 struct strbuf sb = STRBUF_INIT;
385                 strbuf_addbuf(&sb, base);
386                 strbuf_addstr(&sb, filename);
387                 if (!match_pathspec(ctx->args->repo->index,
388                                     &ctx->pathspec,
389                                     sb.buf, sb.len, 0, NULL, 1))
390                         ret = READ_TREE_RECURSIVE;
391                 strbuf_release(&sb);
392         }
393         return ret;
394 }
395
396 static int path_exists(struct archiver_args *args, const char *path)
397 {
398         const char *paths[] = { path, NULL };
399         struct path_exists_context ctx;
400         int ret;
401
402         ctx.args = args;
403         parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
404         ctx.pathspec.recursive = 1;
405         ret = read_tree(args->repo, args->tree,
406                         &ctx.pathspec,
407                         reject_entry, &ctx);
408         clear_pathspec(&ctx.pathspec);
409         return ret != 0;
410 }
411
412 static void parse_pathspec_arg(const char **pathspec,
413                 struct archiver_args *ar_args)
414 {
415         /*
416          * must be consistent with parse_pathspec in path_exists()
417          * Also if pathspec patterns are dependent, we're in big
418          * trouble as we test each one separately
419          */
420         parse_pathspec(&ar_args->pathspec, 0,
421                        PATHSPEC_PREFER_FULL,
422                        "", pathspec);
423         ar_args->pathspec.recursive = 1;
424         if (pathspec) {
425                 while (*pathspec) {
426                         if (**pathspec && !path_exists(ar_args, *pathspec))
427                                 die(_("pathspec '%s' did not match any files"), *pathspec);
428                         pathspec++;
429                 }
430         }
431 }
432
433 static void parse_treeish_arg(const char **argv,
434                 struct archiver_args *ar_args, const char *prefix,
435                 int remote)
436 {
437         const char *name = argv[0];
438         const struct object_id *commit_oid;
439         time_t archive_time;
440         struct tree *tree;
441         const struct commit *commit;
442         struct object_id oid;
443         char *ref = NULL;
444
445         /* Remotes are only allowed to fetch actual refs */
446         if (remote && !remote_allow_unreachable) {
447                 const char *colon = strchrnul(name, ':');
448                 int refnamelen = colon - name;
449
450                 if (!dwim_ref(name, refnamelen, &oid, &ref, 0))
451                         die(_("no such ref: %.*s"), refnamelen, name);
452         } else {
453                 dwim_ref(name, strlen(name), &oid, &ref, 0);
454         }
455
456         if (get_oid(name, &oid))
457                 die(_("not a valid object name: %s"), name);
458
459         commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
460         if (commit) {
461                 commit_oid = &commit->object.oid;
462                 archive_time = commit->date;
463         } else {
464                 commit_oid = NULL;
465                 archive_time = time(NULL);
466         }
467
468         tree = parse_tree_indirect(&oid);
469         if (tree == NULL)
470                 die(_("not a tree object: %s"), oid_to_hex(&oid));
471
472         if (prefix) {
473                 struct object_id tree_oid;
474                 unsigned short mode;
475                 int err;
476
477                 err = get_tree_entry(ar_args->repo,
478                                      &tree->object.oid,
479                                      prefix, &tree_oid,
480                                      &mode);
481                 if (err || !S_ISDIR(mode))
482                         die(_("current working directory is untracked"));
483
484                 tree = parse_tree_indirect(&tree_oid);
485         }
486         ar_args->refname = ref;
487         ar_args->tree = tree;
488         ar_args->commit_oid = commit_oid;
489         ar_args->commit = commit;
490         ar_args->time = archive_time;
491 }
492
493 static void extra_file_info_clear(void *util, const char *str)
494 {
495         struct extra_file_info *info = util;
496         free(info->base);
497         free(info);
498 }
499
500 static int add_file_cb(const struct option *opt, const char *arg, int unset)
501 {
502         struct archiver_args *args = opt->value;
503         const char **basep = (const char **)opt->defval;
504         const char *base = *basep;
505         char *path;
506         struct string_list_item *item;
507         struct extra_file_info *info;
508
509         if (unset) {
510                 string_list_clear_func(&args->extra_files,
511                                        extra_file_info_clear);
512                 return 0;
513         }
514
515         if (!arg)
516                 return -1;
517
518         path = prefix_filename(args->prefix, arg);
519         item = string_list_append_nodup(&args->extra_files, path);
520         item->util = info = xmalloc(sizeof(*info));
521         info->base = xstrdup_or_null(base);
522         if (stat(path, &info->stat))
523                 die(_("File not found: %s"), path);
524         if (!S_ISREG(info->stat.st_mode))
525                 die(_("Not a regular file: %s"), path);
526         return 0;
527 }
528
529 static int number_callback(const struct option *opt, const char *arg, int unset)
530 {
531         BUG_ON_OPT_NEG(unset);
532         *(int *)opt->value = strtol(arg, NULL, 10);
533         return 0;
534 }
535
536 static int parse_archive_args(int argc, const char **argv,
537                 const struct archiver **ar, struct archiver_args *args,
538                 const char *name_hint, int is_remote)
539 {
540         const char *format = NULL;
541         const char *base = NULL;
542         const char *remote = NULL;
543         const char *exec = NULL;
544         const char *output = NULL;
545         int compression_level = -1;
546         int verbose = 0;
547         int i;
548         int list = 0;
549         int worktree_attributes = 0;
550         struct option opts[] = {
551                 OPT_GROUP(""),
552                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
553                 OPT_STRING(0, "prefix", &base, N_("prefix"),
554                         N_("prepend prefix to each pathname in the archive")),
555                 { OPTION_CALLBACK, 0, "add-file", args, N_("file"),
556                   N_("add untracked file to archive"), 0, add_file_cb,
557                   (intptr_t)&base },
558                 OPT_STRING('o', "output", &output, N_("file"),
559                         N_("write the archive to this file")),
560                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
561                         N_("read .gitattributes in working directory")),
562                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
563                 OPT_NUMBER_CALLBACK(&compression_level,
564                         N_("set compression level"), number_callback),
565                 OPT_GROUP(""),
566                 OPT_BOOL('l', "list", &list,
567                         N_("list supported archive formats")),
568                 OPT_GROUP(""),
569                 OPT_STRING(0, "remote", &remote, N_("repo"),
570                         N_("retrieve the archive from remote repository <repo>")),
571                 OPT_STRING(0, "exec", &exec, N_("command"),
572                         N_("path to the remote git-upload-archive command")),
573                 OPT_END()
574         };
575
576         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
577
578         if (remote)
579                 die(_("Unexpected option --remote"));
580         if (exec)
581                 die(_("Option --exec can only be used together with --remote"));
582         if (output)
583                 die(_("Unexpected option --output"));
584         if (is_remote && args->extra_files.nr)
585                 die(_("Options --add-file and --remote cannot be used together"));
586
587         if (!base)
588                 base = "";
589
590         if (list) {
591                 for (i = 0; i < nr_archivers; i++)
592                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
593                                 printf("%s\n", archivers[i]->name);
594                 exit(0);
595         }
596
597         if (!format && name_hint)
598                 format = archive_format_from_filename(name_hint);
599         if (!format)
600                 format = "tar";
601
602         /* We need at least one parameter -- tree-ish */
603         if (argc < 1)
604                 usage_with_options(archive_usage, opts);
605         *ar = lookup_archiver(format);
606         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
607                 die(_("Unknown archive format '%s'"), format);
608
609         args->compression_level = Z_DEFAULT_COMPRESSION;
610         if (compression_level != -1) {
611                 int levels_ok = (*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS;
612                 int high_ok = (*ar)->flags & ARCHIVER_HIGH_COMPRESSION_LEVELS;
613                 if (levels_ok && (compression_level <= 9 || high_ok))
614                         args->compression_level = compression_level;
615                 else {
616                         die(_("Argument not supported for format '%s': -%d"),
617                                         format, compression_level);
618                 }
619         }
620         args->verbose = verbose;
621         args->base = base;
622         args->baselen = strlen(base);
623         args->worktree_attributes = worktree_attributes;
624
625         return argc;
626 }
627
628 int write_archive(int argc, const char **argv, const char *prefix,
629                   struct repository *repo,
630                   const char *name_hint, int remote)
631 {
632         const struct archiver *ar = NULL;
633         struct pretty_print_describe_status describe_status = {0};
634         struct pretty_print_context ctx = {0};
635         struct archiver_args args;
636         int rc;
637
638         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
639         git_config(git_default_config, NULL);
640
641         describe_status.max_invocations = 1;
642         ctx.date_mode.type = DATE_NORMAL;
643         ctx.abbrev = DEFAULT_ABBREV;
644         ctx.describe_status = &describe_status;
645         args.pretty_ctx = &ctx;
646         args.repo = repo;
647         args.prefix = prefix;
648         string_list_init(&args.extra_files, 1);
649         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
650         if (!startup_info->have_repository) {
651                 /*
652                  * We know this will die() with an error, so we could just
653                  * die ourselves; but its error message will be more specific
654                  * than what we could write here.
655                  */
656                 setup_git_directory();
657         }
658
659         parse_treeish_arg(argv, &args, prefix, remote);
660         parse_pathspec_arg(argv + 1, &args);
661
662         rc = ar->write_archive(ar, &args);
663
664         string_list_clear_func(&args.extra_files, extra_file_info_clear);
665         free(args.refname);
666
667         return rc;
668 }
669
670 static int match_extension(const char *filename, const char *ext)
671 {
672         int prefixlen = strlen(filename) - strlen(ext);
673
674         /*
675          * We need 1 character for the '.', and 1 character to ensure that the
676          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
677          * filename).
678          */
679         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
680                 return 0;
681         return !strcmp(filename + prefixlen, ext);
682 }
683
684 const char *archive_format_from_filename(const char *filename)
685 {
686         int i;
687
688         for (i = 0; i < nr_archivers; i++)
689                 if (match_extension(filename, archivers[i]->name))
690                         return archivers[i]->name;
691         return NULL;
692 }