Merge branch 'jk/submodule-fsck-loose-fixup'
[git] / archive.c
1 #include "cache.h"
2 #include "config.h"
3 #include "refs.h"
4 #include "commit.h"
5 #include "tree-walk.h"
6 #include "attr.h"
7 #include "archive.h"
8 #include "parse-options.h"
9 #include "unpack-trees.h"
10 #include "dir.h"
11
12 static char const * const archive_usage[] = {
13         N_("git archive [<options>] <tree-ish> [<path>...]"),
14         N_("git archive --list"),
15         N_("git archive --remote <repo> [--exec <cmd>] [<options>] <tree-ish> [<path>...]"),
16         N_("git archive --remote <repo> [--exec <cmd>] --list"),
17         NULL
18 };
19
20 static const struct archiver **archivers;
21 static int nr_archivers;
22 static int alloc_archivers;
23 static int remote_allow_unreachable;
24
25 void register_archiver(struct archiver *ar)
26 {
27         ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
28         archivers[nr_archivers++] = ar;
29 }
30
31 static void format_subst(const struct commit *commit,
32                          const char *src, size_t len,
33                          struct strbuf *buf)
34 {
35         char *to_free = NULL;
36         struct strbuf fmt = STRBUF_INIT;
37         struct pretty_print_context ctx = {0};
38         ctx.date_mode.type = DATE_NORMAL;
39         ctx.abbrev = DEFAULT_ABBREV;
40
41         if (src == buf->buf)
42                 to_free = strbuf_detach(buf, NULL);
43         for (;;) {
44                 const char *b, *c;
45
46                 b = memmem(src, len, "$Format:", 8);
47                 if (!b)
48                         break;
49                 c = memchr(b + 8, '$', (src + len) - b - 8);
50                 if (!c)
51                         break;
52
53                 strbuf_reset(&fmt);
54                 strbuf_add(&fmt, b + 8, c - b - 8);
55
56                 strbuf_add(buf, src, b - src);
57                 format_commit_message(commit, fmt.buf, buf, &ctx);
58                 len -= c + 1 - src;
59                 src  = c + 1;
60         }
61         strbuf_add(buf, src, len);
62         strbuf_release(&fmt);
63         free(to_free);
64 }
65
66 void *object_file_to_archive(const struct archiver_args *args,
67                              const char *path, const struct object_id *oid,
68                              unsigned int mode, enum object_type *type,
69                              unsigned long *sizep)
70 {
71         void *buffer;
72         const struct commit *commit = args->convert ? args->commit : NULL;
73
74         path += args->baselen;
75         buffer = read_object_file(oid, type, sizep);
76         if (buffer && S_ISREG(mode)) {
77                 struct strbuf buf = STRBUF_INIT;
78                 size_t size = 0;
79
80                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
81                 convert_to_working_tree(path, buf.buf, buf.len, &buf);
82                 if (commit)
83                         format_subst(commit, buf.buf, buf.len, &buf);
84                 buffer = strbuf_detach(&buf, &size);
85                 *sizep = size;
86         }
87
88         return buffer;
89 }
90
91 struct directory {
92         struct directory *up;
93         struct object_id oid;
94         int baselen, len;
95         unsigned mode;
96         int stage;
97         char path[FLEX_ARRAY];
98 };
99
100 struct archiver_context {
101         struct archiver_args *args;
102         write_archive_entry_fn_t write_entry;
103         struct directory *bottom;
104 };
105
106 static const struct attr_check *get_archive_attrs(const char *path)
107 {
108         static struct attr_check *check;
109         if (!check)
110                 check = attr_check_initl("export-ignore", "export-subst", NULL);
111         return git_check_attr(path, check) ? NULL : check;
112 }
113
114 static int check_attr_export_ignore(const struct attr_check *check)
115 {
116         return check && ATTR_TRUE(check->items[0].value);
117 }
118
119 static int check_attr_export_subst(const struct attr_check *check)
120 {
121         return check && ATTR_TRUE(check->items[1].value);
122 }
123
124 static int write_archive_entry(const struct object_id *oid, const char *base,
125                 int baselen, const char *filename, unsigned mode, int stage,
126                 void *context)
127 {
128         static struct strbuf path = STRBUF_INIT;
129         struct archiver_context *c = context;
130         struct archiver_args *args = c->args;
131         write_archive_entry_fn_t write_entry = c->write_entry;
132         int err;
133         const char *path_without_prefix;
134
135         args->convert = 0;
136         strbuf_reset(&path);
137         strbuf_grow(&path, PATH_MAX);
138         strbuf_add(&path, args->base, args->baselen);
139         strbuf_add(&path, base, baselen);
140         strbuf_addstr(&path, filename);
141         if (S_ISDIR(mode) || S_ISGITLINK(mode))
142                 strbuf_addch(&path, '/');
143         path_without_prefix = path.buf + args->baselen;
144
145         if (!S_ISDIR(mode)) {
146                 const struct attr_check *check;
147                 check = get_archive_attrs(path_without_prefix);
148                 if (check_attr_export_ignore(check))
149                         return 0;
150                 args->convert = check_attr_export_subst(check);
151         }
152
153         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
154                 if (args->verbose)
155                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
156                 err = write_entry(args, oid, path.buf, path.len, mode);
157                 if (err)
158                         return err;
159                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
160         }
161
162         if (args->verbose)
163                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
164         return write_entry(args, oid, path.buf, path.len, mode);
165 }
166
167 static void queue_directory(const unsigned char *sha1,
168                 struct strbuf *base, const char *filename,
169                 unsigned mode, int stage, struct archiver_context *c)
170 {
171         struct directory *d;
172         size_t len = st_add4(base->len, 1, strlen(filename), 1);
173         d = xmalloc(st_add(sizeof(*d), len));
174         d->up      = c->bottom;
175         d->baselen = base->len;
176         d->mode    = mode;
177         d->stage   = stage;
178         c->bottom  = d;
179         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
180         hashcpy(d->oid.hash, sha1);
181 }
182
183 static int write_directory(struct archiver_context *c)
184 {
185         struct directory *d = c->bottom;
186         int ret;
187
188         if (!d)
189                 return 0;
190         c->bottom = d->up;
191         d->path[d->len - 1] = '\0'; /* no trailing slash */
192         ret =
193                 write_directory(c) ||
194                 write_archive_entry(&d->oid, d->path, d->baselen,
195                                     d->path + d->baselen, d->mode,
196                                     d->stage, c) != READ_TREE_RECURSIVE;
197         free(d);
198         return ret ? -1 : 0;
199 }
200
201 static int queue_or_write_archive_entry(const struct object_id *oid,
202                 struct strbuf *base, const char *filename,
203                 unsigned mode, int stage, void *context)
204 {
205         struct archiver_context *c = context;
206
207         while (c->bottom &&
208                !(base->len >= c->bottom->len &&
209                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
210                 struct directory *next = c->bottom->up;
211                 free(c->bottom);
212                 c->bottom = next;
213         }
214
215         if (S_ISDIR(mode)) {
216                 size_t baselen = base->len;
217                 const struct attr_check *check;
218
219                 /* Borrow base, but restore its original value when done. */
220                 strbuf_addstr(base, filename);
221                 strbuf_addch(base, '/');
222                 check = get_archive_attrs(base->buf);
223                 strbuf_setlen(base, baselen);
224
225                 if (check_attr_export_ignore(check))
226                         return 0;
227                 queue_directory(oid->hash, base, filename,
228                                 mode, stage, c);
229                 return READ_TREE_RECURSIVE;
230         }
231
232         if (write_directory(c))
233                 return -1;
234         return write_archive_entry(oid, base->buf, base->len, filename, mode,
235                                    stage, context);
236 }
237
238 int write_archive_entries(struct archiver_args *args,
239                 write_archive_entry_fn_t write_entry)
240 {
241         struct archiver_context context;
242         struct unpack_trees_options opts;
243         struct tree_desc t;
244         int err;
245
246         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
247                 size_t len = args->baselen;
248
249                 while (len > 1 && args->base[len - 2] == '/')
250                         len--;
251                 if (args->verbose)
252                         fprintf(stderr, "%.*s\n", (int)len, args->base);
253                 err = write_entry(args, &args->tree->object.oid, args->base,
254                                   len, 040777);
255                 if (err)
256                         return err;
257         }
258
259         memset(&context, 0, sizeof(context));
260         context.args = args;
261         context.write_entry = write_entry;
262
263         /*
264          * Setup index and instruct attr to read index only
265          */
266         if (!args->worktree_attributes) {
267                 memset(&opts, 0, sizeof(opts));
268                 opts.index_only = 1;
269                 opts.head_idx = -1;
270                 opts.src_index = &the_index;
271                 opts.dst_index = &the_index;
272                 opts.fn = oneway_merge;
273                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
274                 if (unpack_trees(1, &t, &opts))
275                         return -1;
276                 git_attr_set_direction(GIT_ATTR_INDEX, &the_index);
277         }
278
279         err = read_tree_recursive(args->tree, "", 0, 0, &args->pathspec,
280                                   queue_or_write_archive_entry,
281                                   &context);
282         if (err == READ_TREE_RECURSIVE)
283                 err = 0;
284         while (context.bottom) {
285                 struct directory *next = context.bottom->up;
286                 free(context.bottom);
287                 context.bottom = next;
288         }
289         return err;
290 }
291
292 static const struct archiver *lookup_archiver(const char *name)
293 {
294         int i;
295
296         if (!name)
297                 return NULL;
298
299         for (i = 0; i < nr_archivers; i++) {
300                 if (!strcmp(name, archivers[i]->name))
301                         return archivers[i];
302         }
303         return NULL;
304 }
305
306 static int reject_entry(const struct object_id *oid, struct strbuf *base,
307                         const char *filename, unsigned mode,
308                         int stage, void *context)
309 {
310         int ret = -1;
311         if (S_ISDIR(mode)) {
312                 struct strbuf sb = STRBUF_INIT;
313                 strbuf_addbuf(&sb, base);
314                 strbuf_addstr(&sb, filename);
315                 if (!match_pathspec(context, sb.buf, sb.len, 0, NULL, 1))
316                         ret = READ_TREE_RECURSIVE;
317                 strbuf_release(&sb);
318         }
319         return ret;
320 }
321
322 static int path_exists(struct tree *tree, const char *path)
323 {
324         const char *paths[] = { path, NULL };
325         struct pathspec pathspec;
326         int ret;
327
328         parse_pathspec(&pathspec, 0, 0, "", paths);
329         pathspec.recursive = 1;
330         ret = read_tree_recursive(tree, "", 0, 0, &pathspec,
331                                   reject_entry, &pathspec);
332         clear_pathspec(&pathspec);
333         return ret != 0;
334 }
335
336 static void parse_pathspec_arg(const char **pathspec,
337                 struct archiver_args *ar_args)
338 {
339         /*
340          * must be consistent with parse_pathspec in path_exists()
341          * Also if pathspec patterns are dependent, we're in big
342          * trouble as we test each one separately
343          */
344         parse_pathspec(&ar_args->pathspec, 0,
345                        PATHSPEC_PREFER_FULL,
346                        "", pathspec);
347         ar_args->pathspec.recursive = 1;
348         if (pathspec) {
349                 while (*pathspec) {
350                         if (**pathspec && !path_exists(ar_args->tree, *pathspec))
351                                 die(_("pathspec '%s' did not match any files"), *pathspec);
352                         pathspec++;
353                 }
354         }
355 }
356
357 static void parse_treeish_arg(const char **argv,
358                 struct archiver_args *ar_args, const char *prefix,
359                 int remote)
360 {
361         const char *name = argv[0];
362         const unsigned char *commit_sha1;
363         time_t archive_time;
364         struct tree *tree;
365         const struct commit *commit;
366         struct object_id oid;
367
368         /* Remotes are only allowed to fetch actual refs */
369         if (remote && !remote_allow_unreachable) {
370                 char *ref = NULL;
371                 const char *colon = strchrnul(name, ':');
372                 int refnamelen = colon - name;
373
374                 if (!dwim_ref(name, refnamelen, &oid, &ref))
375                         die("no such ref: %.*s", refnamelen, name);
376                 free(ref);
377         }
378
379         if (get_oid(name, &oid))
380                 die("Not a valid object name");
381
382         commit = lookup_commit_reference_gently(&oid, 1);
383         if (commit) {
384                 commit_sha1 = commit->object.oid.hash;
385                 archive_time = commit->date;
386         } else {
387                 commit_sha1 = NULL;
388                 archive_time = time(NULL);
389         }
390
391         tree = parse_tree_indirect(&oid);
392         if (tree == NULL)
393                 die("not a tree object");
394
395         if (prefix) {
396                 struct object_id tree_oid;
397                 unsigned int mode;
398                 int err;
399
400                 err = get_tree_entry(&tree->object.oid, prefix, &tree_oid,
401                                      &mode);
402                 if (err || !S_ISDIR(mode))
403                         die("current working directory is untracked");
404
405                 tree = parse_tree_indirect(&tree_oid);
406         }
407         ar_args->tree = tree;
408         ar_args->commit_sha1 = commit_sha1;
409         ar_args->commit = commit;
410         ar_args->time = archive_time;
411 }
412
413 #define OPT__COMPR(s, v, h, p) \
414         OPT_SET_INT_F(s, NULL, v, h, p, PARSE_OPT_NONEG)
415 #define OPT__COMPR_HIDDEN(s, v, p) \
416         OPT_SET_INT_F(s, NULL, v, "", p, PARSE_OPT_NONEG | PARSE_OPT_HIDDEN)
417
418 static int parse_archive_args(int argc, const char **argv,
419                 const struct archiver **ar, struct archiver_args *args,
420                 const char *name_hint, int is_remote)
421 {
422         const char *format = NULL;
423         const char *base = NULL;
424         const char *remote = NULL;
425         const char *exec = NULL;
426         const char *output = NULL;
427         int compression_level = -1;
428         int verbose = 0;
429         int i;
430         int list = 0;
431         int worktree_attributes = 0;
432         struct option opts[] = {
433                 OPT_GROUP(""),
434                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
435                 OPT_STRING(0, "prefix", &base, N_("prefix"),
436                         N_("prepend prefix to each pathname in the archive")),
437                 OPT_STRING('o', "output", &output, N_("file"),
438                         N_("write the archive to this file")),
439                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
440                         N_("read .gitattributes in working directory")),
441                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
442                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
443                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
444                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
445                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
446                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
447                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
448                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
449                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
450                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
451                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
452                 OPT_GROUP(""),
453                 OPT_BOOL('l', "list", &list,
454                         N_("list supported archive formats")),
455                 OPT_GROUP(""),
456                 OPT_STRING(0, "remote", &remote, N_("repo"),
457                         N_("retrieve the archive from remote repository <repo>")),
458                 OPT_STRING(0, "exec", &exec, N_("command"),
459                         N_("path to the remote git-upload-archive command")),
460                 OPT_END()
461         };
462
463         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
464
465         if (remote)
466                 die(_("Unexpected option --remote"));
467         if (exec)
468                 die(_("Option --exec can only be used together with --remote"));
469         if (output)
470                 die(_("Unexpected option --output"));
471
472         if (!base)
473                 base = "";
474
475         if (list) {
476                 for (i = 0; i < nr_archivers; i++)
477                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
478                                 printf("%s\n", archivers[i]->name);
479                 exit(0);
480         }
481
482         if (!format && name_hint)
483                 format = archive_format_from_filename(name_hint);
484         if (!format)
485                 format = "tar";
486
487         /* We need at least one parameter -- tree-ish */
488         if (argc < 1)
489                 usage_with_options(archive_usage, opts);
490         *ar = lookup_archiver(format);
491         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
492                 die(_("Unknown archive format '%s'"), format);
493
494         args->compression_level = Z_DEFAULT_COMPRESSION;
495         if (compression_level != -1) {
496                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
497                         args->compression_level = compression_level;
498                 else {
499                         die(_("Argument not supported for format '%s': -%d"),
500                                         format, compression_level);
501                 }
502         }
503         args->verbose = verbose;
504         args->base = base;
505         args->baselen = strlen(base);
506         args->worktree_attributes = worktree_attributes;
507
508         return argc;
509 }
510
511 int write_archive(int argc, const char **argv, const char *prefix,
512                   const char *name_hint, int remote)
513 {
514         const struct archiver *ar = NULL;
515         struct archiver_args args;
516
517         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
518         git_config(git_default_config, NULL);
519
520         init_tar_archiver();
521         init_zip_archiver();
522
523         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
524         if (!startup_info->have_repository) {
525                 /*
526                  * We know this will die() with an error, so we could just
527                  * die ourselves; but its error message will be more specific
528                  * than what we could write here.
529                  */
530                 setup_git_directory();
531         }
532
533         parse_treeish_arg(argv, &args, prefix, remote);
534         parse_pathspec_arg(argv + 1, &args);
535
536         return ar->write_archive(ar, &args);
537 }
538
539 static int match_extension(const char *filename, const char *ext)
540 {
541         int prefixlen = strlen(filename) - strlen(ext);
542
543         /*
544          * We need 1 character for the '.', and 1 character to ensure that the
545          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
546          * filename).
547          */
548         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
549                 return 0;
550         return !strcmp(filename + prefixlen, ext);
551 }
552
553 const char *archive_format_from_filename(const char *filename)
554 {
555         int i;
556
557         for (i = 0; i < nr_archivers; i++)
558                 if (match_extension(filename, archivers[i]->name))
559                         return archivers[i]->name;
560         return NULL;
561 }