Merge branch 'jt/pretend-object-never-come-from-elsewhere'
[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 void *object_file_to_archive(const struct archiver_args *args,
74                              const char *path, const struct object_id *oid,
75                              unsigned int mode, enum object_type *type,
76                              unsigned long *sizep)
77 {
78         void *buffer;
79         const struct commit *commit = args->convert ? args->commit : NULL;
80         struct checkout_metadata meta;
81
82         init_checkout_metadata(&meta, args->refname,
83                                args->commit_oid ? args->commit_oid :
84                                (args->tree ? &args->tree->object.oid : NULL), oid);
85
86         path += args->baselen;
87         buffer = read_object_file(oid, type, sizep);
88         if (buffer && S_ISREG(mode)) {
89                 struct strbuf buf = STRBUF_INIT;
90                 size_t size = 0;
91
92                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
93                 convert_to_working_tree(args->repo->index, path, buf.buf, buf.len, &buf, &meta);
94                 if (commit)
95                         format_subst(commit, buf.buf, buf.len, &buf);
96                 buffer = strbuf_detach(&buf, &size);
97                 *sizep = size;
98         }
99
100         return buffer;
101 }
102
103 struct directory {
104         struct directory *up;
105         struct object_id oid;
106         int baselen, len;
107         unsigned mode;
108         int stage;
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, 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, int stage,
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
149         args->convert = 0;
150         strbuf_reset(&path);
151         strbuf_grow(&path, PATH_MAX);
152         strbuf_add(&path, args->base, args->baselen);
153         strbuf_add(&path, base, baselen);
154         strbuf_addstr(&path, filename);
155         if (S_ISDIR(mode) || S_ISGITLINK(mode))
156                 strbuf_addch(&path, '/');
157         path_without_prefix = path.buf + args->baselen;
158
159         if (!S_ISDIR(mode)) {
160                 const struct attr_check *check;
161                 check = get_archive_attrs(args->repo->index, path_without_prefix);
162                 if (check_attr_export_ignore(check))
163                         return 0;
164                 args->convert = check_attr_export_subst(check);
165         }
166
167         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
168                 if (args->verbose)
169                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
170                 err = write_entry(args, oid, path.buf, path.len, mode);
171                 if (err)
172                         return err;
173                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
174         }
175
176         if (args->verbose)
177                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
178         return write_entry(args, oid, path.buf, path.len, mode);
179 }
180
181 static void queue_directory(const unsigned char *sha1,
182                 struct strbuf *base, const char *filename,
183                 unsigned mode, int stage, struct archiver_context *c)
184 {
185         struct directory *d;
186         size_t len = st_add4(base->len, 1, strlen(filename), 1);
187         d = xmalloc(st_add(sizeof(*d), len));
188         d->up      = c->bottom;
189         d->baselen = base->len;
190         d->mode    = mode;
191         d->stage   = stage;
192         c->bottom  = d;
193         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
194         hashcpy(d->oid.hash, sha1);
195 }
196
197 static int write_directory(struct archiver_context *c)
198 {
199         struct directory *d = c->bottom;
200         int ret;
201
202         if (!d)
203                 return 0;
204         c->bottom = d->up;
205         d->path[d->len - 1] = '\0'; /* no trailing slash */
206         ret =
207                 write_directory(c) ||
208                 write_archive_entry(&d->oid, d->path, d->baselen,
209                                     d->path + d->baselen, d->mode,
210                                     d->stage, c) != READ_TREE_RECURSIVE;
211         free(d);
212         return ret ? -1 : 0;
213 }
214
215 static int queue_or_write_archive_entry(const struct object_id *oid,
216                 struct strbuf *base, const char *filename,
217                 unsigned mode, int stage, void *context)
218 {
219         struct archiver_context *c = context;
220
221         while (c->bottom &&
222                !(base->len >= c->bottom->len &&
223                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
224                 struct directory *next = c->bottom->up;
225                 free(c->bottom);
226                 c->bottom = next;
227         }
228
229         if (S_ISDIR(mode)) {
230                 size_t baselen = base->len;
231                 const struct attr_check *check;
232
233                 /* Borrow base, but restore its original value when done. */
234                 strbuf_addstr(base, filename);
235                 strbuf_addch(base, '/');
236                 check = get_archive_attrs(c->args->repo->index, base->buf);
237                 strbuf_setlen(base, baselen);
238
239                 if (check_attr_export_ignore(check))
240                         return 0;
241                 queue_directory(oid->hash, base, filename,
242                                 mode, stage, c);
243                 return READ_TREE_RECURSIVE;
244         }
245
246         if (write_directory(c))
247                 return -1;
248         return write_archive_entry(oid, base->buf, base->len, filename, mode,
249                                    stage, context);
250 }
251
252 int write_archive_entries(struct archiver_args *args,
253                 write_archive_entry_fn_t write_entry)
254 {
255         struct archiver_context context;
256         struct unpack_trees_options opts;
257         struct tree_desc t;
258         int err;
259
260         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
261                 size_t len = args->baselen;
262
263                 while (len > 1 && args->base[len - 2] == '/')
264                         len--;
265                 if (args->verbose)
266                         fprintf(stderr, "%.*s\n", (int)len, args->base);
267                 err = write_entry(args, &args->tree->object.oid, args->base,
268                                   len, 040777);
269                 if (err)
270                         return err;
271         }
272
273         memset(&context, 0, sizeof(context));
274         context.args = args;
275         context.write_entry = write_entry;
276
277         /*
278          * Setup index and instruct attr to read index only
279          */
280         if (!args->worktree_attributes) {
281                 memset(&opts, 0, sizeof(opts));
282                 opts.index_only = 1;
283                 opts.head_idx = -1;
284                 opts.src_index = args->repo->index;
285                 opts.dst_index = args->repo->index;
286                 opts.fn = oneway_merge;
287                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
288                 if (unpack_trees(1, &t, &opts))
289                         return -1;
290                 git_attr_set_direction(GIT_ATTR_INDEX);
291         }
292
293         err = read_tree_recursive(args->repo, args->tree, "",
294                                   0, 0, &args->pathspec,
295                                   queue_or_write_archive_entry,
296                                   &context);
297         if (err == READ_TREE_RECURSIVE)
298                 err = 0;
299         while (context.bottom) {
300                 struct directory *next = context.bottom->up;
301                 free(context.bottom);
302                 context.bottom = next;
303         }
304         return err;
305 }
306
307 static const struct archiver *lookup_archiver(const char *name)
308 {
309         int i;
310
311         if (!name)
312                 return NULL;
313
314         for (i = 0; i < nr_archivers; i++) {
315                 if (!strcmp(name, archivers[i]->name))
316                         return archivers[i];
317         }
318         return NULL;
319 }
320
321 struct path_exists_context {
322         struct pathspec pathspec;
323         struct archiver_args *args;
324 };
325
326 static int reject_entry(const struct object_id *oid, struct strbuf *base,
327                         const char *filename, unsigned mode,
328                         int stage, void *context)
329 {
330         int ret = -1;
331         struct path_exists_context *ctx = context;
332
333         if (S_ISDIR(mode)) {
334                 struct strbuf sb = STRBUF_INIT;
335                 strbuf_addbuf(&sb, base);
336                 strbuf_addstr(&sb, filename);
337                 if (!match_pathspec(ctx->args->repo->index,
338                                     &ctx->pathspec,
339                                     sb.buf, sb.len, 0, NULL, 1))
340                         ret = READ_TREE_RECURSIVE;
341                 strbuf_release(&sb);
342         }
343         return ret;
344 }
345
346 static int path_exists(struct archiver_args *args, const char *path)
347 {
348         const char *paths[] = { path, NULL };
349         struct path_exists_context ctx;
350         int ret;
351
352         ctx.args = args;
353         parse_pathspec(&ctx.pathspec, 0, 0, "", paths);
354         ctx.pathspec.recursive = 1;
355         ret = read_tree_recursive(args->repo, args->tree, "",
356                                   0, 0, &ctx.pathspec,
357                                   reject_entry, &ctx);
358         clear_pathspec(&ctx.pathspec);
359         return ret != 0;
360 }
361
362 static void parse_pathspec_arg(const char **pathspec,
363                 struct archiver_args *ar_args)
364 {
365         /*
366          * must be consistent with parse_pathspec in path_exists()
367          * Also if pathspec patterns are dependent, we're in big
368          * trouble as we test each one separately
369          */
370         parse_pathspec(&ar_args->pathspec, 0,
371                        PATHSPEC_PREFER_FULL,
372                        "", pathspec);
373         ar_args->pathspec.recursive = 1;
374         if (pathspec) {
375                 while (*pathspec) {
376                         if (**pathspec && !path_exists(ar_args, *pathspec))
377                                 die(_("pathspec '%s' did not match any files"), *pathspec);
378                         pathspec++;
379                 }
380         }
381 }
382
383 static void parse_treeish_arg(const char **argv,
384                 struct archiver_args *ar_args, const char *prefix,
385                 int remote)
386 {
387         const char *name = argv[0];
388         const struct object_id *commit_oid;
389         time_t archive_time;
390         struct tree *tree;
391         const struct commit *commit;
392         struct object_id oid;
393         char *ref = NULL;
394
395         /* Remotes are only allowed to fetch actual refs */
396         if (remote && !remote_allow_unreachable) {
397                 const char *colon = strchrnul(name, ':');
398                 int refnamelen = colon - name;
399
400                 if (!dwim_ref(name, refnamelen, &oid, &ref))
401                         die(_("no such ref: %.*s"), refnamelen, name);
402         } else {
403                 dwim_ref(name, strlen(name), &oid, &ref);
404         }
405
406         if (get_oid(name, &oid))
407                 die(_("not a valid object name: %s"), name);
408
409         commit = lookup_commit_reference_gently(ar_args->repo, &oid, 1);
410         if (commit) {
411                 commit_oid = &commit->object.oid;
412                 archive_time = commit->date;
413         } else {
414                 commit_oid = NULL;
415                 archive_time = time(NULL);
416         }
417
418         tree = parse_tree_indirect(&oid);
419         if (tree == NULL)
420                 die(_("not a tree object: %s"), oid_to_hex(&oid));
421
422         if (prefix) {
423                 struct object_id tree_oid;
424                 unsigned short mode;
425                 int err;
426
427                 err = get_tree_entry(ar_args->repo,
428                                      &tree->object.oid,
429                                      prefix, &tree_oid,
430                                      &mode);
431                 if (err || !S_ISDIR(mode))
432                         die(_("current working directory is untracked"));
433
434                 tree = parse_tree_indirect(&tree_oid);
435         }
436         ar_args->refname = ref;
437         ar_args->tree = tree;
438         ar_args->commit_oid = commit_oid;
439         ar_args->commit = commit;
440         ar_args->time = archive_time;
441 }
442
443 #define OPT__COMPR(s, v, h, p) \
444         OPT_SET_INT_F(s, NULL, v, h, p, PARSE_OPT_NONEG)
445 #define OPT__COMPR_HIDDEN(s, v, p) \
446         OPT_SET_INT_F(s, NULL, v, "", p, PARSE_OPT_NONEG | PARSE_OPT_HIDDEN)
447
448 static int parse_archive_args(int argc, const char **argv,
449                 const struct archiver **ar, struct archiver_args *args,
450                 const char *name_hint, int is_remote)
451 {
452         const char *format = NULL;
453         const char *base = NULL;
454         const char *remote = NULL;
455         const char *exec = NULL;
456         const char *output = NULL;
457         int compression_level = -1;
458         int verbose = 0;
459         int i;
460         int list = 0;
461         int worktree_attributes = 0;
462         struct option opts[] = {
463                 OPT_GROUP(""),
464                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
465                 OPT_STRING(0, "prefix", &base, N_("prefix"),
466                         N_("prepend prefix to each pathname in the archive")),
467                 OPT_STRING('o', "output", &output, N_("file"),
468                         N_("write the archive to this file")),
469                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
470                         N_("read .gitattributes in working directory")),
471                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
472                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
473                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
474                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
475                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
476                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
477                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
478                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
479                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
480                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
481                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
482                 OPT_GROUP(""),
483                 OPT_BOOL('l', "list", &list,
484                         N_("list supported archive formats")),
485                 OPT_GROUP(""),
486                 OPT_STRING(0, "remote", &remote, N_("repo"),
487                         N_("retrieve the archive from remote repository <repo>")),
488                 OPT_STRING(0, "exec", &exec, N_("command"),
489                         N_("path to the remote git-upload-archive command")),
490                 OPT_END()
491         };
492
493         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
494
495         if (remote)
496                 die(_("Unexpected option --remote"));
497         if (exec)
498                 die(_("Option --exec can only be used together with --remote"));
499         if (output)
500                 die(_("Unexpected option --output"));
501
502         if (!base)
503                 base = "";
504
505         if (list) {
506                 for (i = 0; i < nr_archivers; i++)
507                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
508                                 printf("%s\n", archivers[i]->name);
509                 exit(0);
510         }
511
512         if (!format && name_hint)
513                 format = archive_format_from_filename(name_hint);
514         if (!format)
515                 format = "tar";
516
517         /* We need at least one parameter -- tree-ish */
518         if (argc < 1)
519                 usage_with_options(archive_usage, opts);
520         *ar = lookup_archiver(format);
521         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
522                 die(_("Unknown archive format '%s'"), format);
523
524         args->compression_level = Z_DEFAULT_COMPRESSION;
525         if (compression_level != -1) {
526                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
527                         args->compression_level = compression_level;
528                 else {
529                         die(_("Argument not supported for format '%s': -%d"),
530                                         format, compression_level);
531                 }
532         }
533         args->verbose = verbose;
534         args->base = base;
535         args->baselen = strlen(base);
536         args->worktree_attributes = worktree_attributes;
537
538         return argc;
539 }
540
541 int write_archive(int argc, const char **argv, const char *prefix,
542                   struct repository *repo,
543                   const char *name_hint, int remote)
544 {
545         const struct archiver *ar = NULL;
546         struct archiver_args args;
547
548         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
549         git_config(git_default_config, NULL);
550
551         args.repo = repo;
552         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
553         if (!startup_info->have_repository) {
554                 /*
555                  * We know this will die() with an error, so we could just
556                  * die ourselves; but its error message will be more specific
557                  * than what we could write here.
558                  */
559                 setup_git_directory();
560         }
561
562         parse_treeish_arg(argv, &args, prefix, remote);
563         parse_pathspec_arg(argv + 1, &args);
564
565         return ar->write_archive(ar, &args);
566 }
567
568 static int match_extension(const char *filename, const char *ext)
569 {
570         int prefixlen = strlen(filename) - strlen(ext);
571
572         /*
573          * We need 1 character for the '.', and 1 character to ensure that the
574          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
575          * filename).
576          */
577         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
578                 return 0;
579         return !strcmp(filename + prefixlen, ext);
580 }
581
582 const char *archive_format_from_filename(const char *filename)
583 {
584         int i;
585
586         for (i = 0; i < nr_archivers; i++)
587                 if (match_extension(filename, archivers[i]->name))
588                         return archivers[i]->name;
589         return NULL;
590 }