archive: factor out helper functions for handling attributes
[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 *sha1_file_to_archive(const struct archiver_args *args,
67                            const char *path, const unsigned char *sha1,
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_sha1_file(sha1, 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 unsigned char *sha1, 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         const struct attr_check *check;
130         struct archiver_context *c = context;
131         struct archiver_args *args = c->args;
132         write_archive_entry_fn_t write_entry = c->write_entry;
133         const char *path_without_prefix;
134         int err;
135
136         args->convert = 0;
137         strbuf_reset(&path);
138         strbuf_grow(&path, PATH_MAX);
139         strbuf_add(&path, args->base, args->baselen);
140         strbuf_add(&path, base, baselen);
141         strbuf_addstr(&path, filename);
142         if (S_ISDIR(mode) || S_ISGITLINK(mode))
143                 strbuf_addch(&path, '/');
144         path_without_prefix = path.buf + args->baselen;
145
146         check = get_archive_attrs(path_without_prefix);
147         if (check_attr_export_ignore(check))
148                 return 0;
149         args->convert = check_attr_export_subst(check);
150
151         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
152                 if (args->verbose)
153                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
154                 err = write_entry(args, sha1, path.buf, path.len, mode);
155                 if (err)
156                         return err;
157                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
158         }
159
160         if (args->verbose)
161                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
162         return write_entry(args, sha1, path.buf, path.len, mode);
163 }
164
165 static int write_archive_entry_buf(const unsigned char *sha1, struct strbuf *base,
166                 const char *filename, unsigned mode, int stage,
167                 void *context)
168 {
169         return write_archive_entry(sha1, base->buf, base->len,
170                                      filename, mode, stage, context);
171 }
172
173 static void queue_directory(const unsigned char *sha1,
174                 struct strbuf *base, const char *filename,
175                 unsigned mode, int stage, struct archiver_context *c)
176 {
177         struct directory *d;
178         size_t len = st_add4(base->len, 1, strlen(filename), 1);
179         d = xmalloc(st_add(sizeof(*d), len));
180         d->up      = c->bottom;
181         d->baselen = base->len;
182         d->mode    = mode;
183         d->stage   = stage;
184         c->bottom  = d;
185         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
186         hashcpy(d->oid.hash, sha1);
187 }
188
189 static int write_directory(struct archiver_context *c)
190 {
191         struct directory *d = c->bottom;
192         int ret;
193
194         if (!d)
195                 return 0;
196         c->bottom = d->up;
197         d->path[d->len - 1] = '\0'; /* no trailing slash */
198         ret =
199                 write_directory(c) ||
200                 write_archive_entry(d->oid.hash, d->path, d->baselen,
201                                     d->path + d->baselen, d->mode,
202                                     d->stage, c) != READ_TREE_RECURSIVE;
203         free(d);
204         return ret ? -1 : 0;
205 }
206
207 static int queue_or_write_archive_entry(const unsigned char *sha1,
208                 struct strbuf *base, const char *filename,
209                 unsigned mode, int stage, void *context)
210 {
211         struct archiver_context *c = context;
212
213         while (c->bottom &&
214                !(base->len >= c->bottom->len &&
215                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
216                 struct directory *next = c->bottom->up;
217                 free(c->bottom);
218                 c->bottom = next;
219         }
220
221         if (S_ISDIR(mode)) {
222                 queue_directory(sha1, base, filename,
223                                 mode, stage, c);
224                 return READ_TREE_RECURSIVE;
225         }
226
227         if (write_directory(c))
228                 return -1;
229         return write_archive_entry(sha1, base->buf, base->len, filename, mode,
230                                    stage, context);
231 }
232
233 int write_archive_entries(struct archiver_args *args,
234                 write_archive_entry_fn_t write_entry)
235 {
236         struct archiver_context context;
237         struct unpack_trees_options opts;
238         struct tree_desc t;
239         int err;
240
241         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
242                 size_t len = args->baselen;
243
244                 while (len > 1 && args->base[len - 2] == '/')
245                         len--;
246                 if (args->verbose)
247                         fprintf(stderr, "%.*s\n", (int)len, args->base);
248                 err = write_entry(args, args->tree->object.oid.hash, args->base,
249                                   len, 040777);
250                 if (err)
251                         return err;
252         }
253
254         memset(&context, 0, sizeof(context));
255         context.args = args;
256         context.write_entry = write_entry;
257
258         /*
259          * Setup index and instruct attr to read index only
260          */
261         if (!args->worktree_attributes) {
262                 memset(&opts, 0, sizeof(opts));
263                 opts.index_only = 1;
264                 opts.head_idx = -1;
265                 opts.src_index = &the_index;
266                 opts.dst_index = &the_index;
267                 opts.fn = oneway_merge;
268                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
269                 if (unpack_trees(1, &t, &opts))
270                         return -1;
271                 git_attr_set_direction(GIT_ATTR_INDEX, &the_index);
272         }
273
274         err = read_tree_recursive(args->tree, "", 0, 0, &args->pathspec,
275                                   args->pathspec.has_wildcard ?
276                                   queue_or_write_archive_entry :
277                                   write_archive_entry_buf,
278                                   &context);
279         if (err == READ_TREE_RECURSIVE)
280                 err = 0;
281         while (context.bottom) {
282                 struct directory *next = context.bottom->up;
283                 free(context.bottom);
284                 context.bottom = next;
285         }
286         return err;
287 }
288
289 static const struct archiver *lookup_archiver(const char *name)
290 {
291         int i;
292
293         if (!name)
294                 return NULL;
295
296         for (i = 0; i < nr_archivers; i++) {
297                 if (!strcmp(name, archivers[i]->name))
298                         return archivers[i];
299         }
300         return NULL;
301 }
302
303 static int reject_entry(const unsigned char *sha1, struct strbuf *base,
304                         const char *filename, unsigned mode,
305                         int stage, void *context)
306 {
307         int ret = -1;
308         if (S_ISDIR(mode)) {
309                 struct strbuf sb = STRBUF_INIT;
310                 strbuf_addbuf(&sb, base);
311                 strbuf_addstr(&sb, filename);
312                 if (!match_pathspec(context, sb.buf, sb.len, 0, NULL, 1))
313                         ret = READ_TREE_RECURSIVE;
314                 strbuf_release(&sb);
315         }
316         return ret;
317 }
318
319 static int path_exists(struct tree *tree, const char *path)
320 {
321         const char *paths[] = { path, NULL };
322         struct pathspec pathspec;
323         int ret;
324
325         parse_pathspec(&pathspec, 0, 0, "", paths);
326         pathspec.recursive = 1;
327         ret = read_tree_recursive(tree, "", 0, 0, &pathspec,
328                                   reject_entry, &pathspec);
329         clear_pathspec(&pathspec);
330         return ret != 0;
331 }
332
333 static void parse_pathspec_arg(const char **pathspec,
334                 struct archiver_args *ar_args)
335 {
336         /*
337          * must be consistent with parse_pathspec in path_exists()
338          * Also if pathspec patterns are dependent, we're in big
339          * trouble as we test each one separately
340          */
341         parse_pathspec(&ar_args->pathspec, 0,
342                        PATHSPEC_PREFER_FULL,
343                        "", pathspec);
344         ar_args->pathspec.recursive = 1;
345         if (pathspec) {
346                 while (*pathspec) {
347                         if (**pathspec && !path_exists(ar_args->tree, *pathspec))
348                                 die(_("pathspec '%s' did not match any files"), *pathspec);
349                         pathspec++;
350                 }
351         }
352 }
353
354 static void parse_treeish_arg(const char **argv,
355                 struct archiver_args *ar_args, const char *prefix,
356                 int remote)
357 {
358         const char *name = argv[0];
359         const unsigned char *commit_sha1;
360         time_t archive_time;
361         struct tree *tree;
362         const struct commit *commit;
363         struct object_id oid;
364
365         /* Remotes are only allowed to fetch actual refs */
366         if (remote && !remote_allow_unreachable) {
367                 char *ref = NULL;
368                 const char *colon = strchrnul(name, ':');
369                 int refnamelen = colon - name;
370
371                 if (!dwim_ref(name, refnamelen, oid.hash, &ref))
372                         die("no such ref: %.*s", refnamelen, name);
373                 free(ref);
374         }
375
376         if (get_sha1(name, oid.hash))
377                 die("Not a valid object name");
378
379         commit = lookup_commit_reference_gently(&oid, 1);
380         if (commit) {
381                 commit_sha1 = commit->object.oid.hash;
382                 archive_time = commit->date;
383         } else {
384                 commit_sha1 = NULL;
385                 archive_time = time(NULL);
386         }
387
388         tree = parse_tree_indirect(&oid);
389         if (tree == NULL)
390                 die("not a tree object");
391
392         if (prefix) {
393                 struct object_id tree_oid;
394                 unsigned int mode;
395                 int err;
396
397                 err = get_tree_entry(tree->object.oid.hash, prefix,
398                                      tree_oid.hash, &mode);
399                 if (err || !S_ISDIR(mode))
400                         die("current working directory is untracked");
401
402                 tree = parse_tree_indirect(&tree_oid);
403         }
404         ar_args->tree = tree;
405         ar_args->commit_sha1 = commit_sha1;
406         ar_args->commit = commit;
407         ar_args->time = archive_time;
408 }
409
410 #define OPT__COMPR(s, v, h, p) \
411         { OPTION_SET_INT, (s), NULL, (v), NULL, (h), \
412           PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, (p) }
413 #define OPT__COMPR_HIDDEN(s, v, p) \
414         { OPTION_SET_INT, (s), NULL, (v), NULL, "", \
415           PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
416
417 static int parse_archive_args(int argc, const char **argv,
418                 const struct archiver **ar, struct archiver_args *args,
419                 const char *name_hint, int is_remote)
420 {
421         const char *format = NULL;
422         const char *base = NULL;
423         const char *remote = NULL;
424         const char *exec = NULL;
425         const char *output = NULL;
426         int compression_level = -1;
427         int verbose = 0;
428         int i;
429         int list = 0;
430         int worktree_attributes = 0;
431         struct option opts[] = {
432                 OPT_GROUP(""),
433                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
434                 OPT_STRING(0, "prefix", &base, N_("prefix"),
435                         N_("prepend prefix to each pathname in the archive")),
436                 OPT_STRING('o', "output", &output, N_("file"),
437                         N_("write the archive to this file")),
438                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
439                         N_("read .gitattributes in working directory")),
440                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
441                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
442                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
443                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
444                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
445                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
446                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
447                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
448                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
449                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
450                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
451                 OPT_GROUP(""),
452                 OPT_BOOL('l', "list", &list,
453                         N_("list supported archive formats")),
454                 OPT_GROUP(""),
455                 OPT_STRING(0, "remote", &remote, N_("repo"),
456                         N_("retrieve the archive from remote repository <repo>")),
457                 OPT_STRING(0, "exec", &exec, N_("command"),
458                         N_("path to the remote git-upload-archive command")),
459                 OPT_END()
460         };
461
462         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
463
464         if (remote)
465                 die(_("Unexpected option --remote"));
466         if (exec)
467                 die(_("Option --exec can only be used together with --remote"));
468         if (output)
469                 die(_("Unexpected option --output"));
470
471         if (!base)
472                 base = "";
473
474         if (list) {
475                 for (i = 0; i < nr_archivers; i++)
476                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
477                                 printf("%s\n", archivers[i]->name);
478                 exit(0);
479         }
480
481         if (!format && name_hint)
482                 format = archive_format_from_filename(name_hint);
483         if (!format)
484                 format = "tar";
485
486         /* We need at least one parameter -- tree-ish */
487         if (argc < 1)
488                 usage_with_options(archive_usage, opts);
489         *ar = lookup_archiver(format);
490         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
491                 die(_("Unknown archive format '%s'"), format);
492
493         args->compression_level = Z_DEFAULT_COMPRESSION;
494         if (compression_level != -1) {
495                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
496                         args->compression_level = compression_level;
497                 else {
498                         die(_("Argument not supported for format '%s': -%d"),
499                                         format, compression_level);
500                 }
501         }
502         args->verbose = verbose;
503         args->base = base;
504         args->baselen = strlen(base);
505         args->worktree_attributes = worktree_attributes;
506
507         return argc;
508 }
509
510 int write_archive(int argc, const char **argv, const char *prefix,
511                   const char *name_hint, int remote)
512 {
513         const struct archiver *ar = NULL;
514         struct archiver_args args;
515
516         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
517         git_config(git_default_config, NULL);
518
519         init_tar_archiver();
520         init_zip_archiver();
521
522         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
523         if (!startup_info->have_repository) {
524                 /*
525                  * We know this will die() with an error, so we could just
526                  * die ourselves; but its error message will be more specific
527                  * than what we could write here.
528                  */
529                 setup_git_directory();
530         }
531
532         parse_treeish_arg(argv, &args, prefix, remote);
533         parse_pathspec_arg(argv + 1, &args);
534
535         return ar->write_archive(ar, &args);
536 }
537
538 static int match_extension(const char *filename, const char *ext)
539 {
540         int prefixlen = strlen(filename) - strlen(ext);
541
542         /*
543          * We need 1 character for the '.', and 1 character to ensure that the
544          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
545          * filename).
546          */
547         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
548                 return 0;
549         return !strcmp(filename + prefixlen, ext);
550 }
551
552 const char *archive_format_from_filename(const char *filename)
553 {
554         int i;
555
556         for (i = 0; i < nr_archivers; i++)
557                 if (match_extension(filename, archivers[i]->name))
558                         return archivers[i]->name;
559         return NULL;
560 }