Merge branch 'ew/git-svn-http-tests'
[git] / archive.c
1 #include "cache.h"
2 #include "refs.h"
3 #include "commit.h"
4 #include "tree-walk.h"
5 #include "attr.h"
6 #include "archive.h"
7 #include "parse-options.h"
8 #include "unpack-trees.h"
9 #include "dir.h"
10
11 static char const * const archive_usage[] = {
12         N_("git archive [<options>] <tree-ish> [<path>...]"),
13         N_("git archive --list"),
14         N_("git archive --remote <repo> [--exec <cmd>] [<options>] <tree-ish> [<path>...]"),
15         N_("git archive --remote <repo> [--exec <cmd>] --list"),
16         NULL
17 };
18
19 static const struct archiver **archivers;
20 static int nr_archivers;
21 static int alloc_archivers;
22 static int remote_allow_unreachable;
23
24 void register_archiver(struct archiver *ar)
25 {
26         ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
27         archivers[nr_archivers++] = ar;
28 }
29
30 static void format_subst(const struct commit *commit,
31                          const char *src, size_t len,
32                          struct strbuf *buf)
33 {
34         char *to_free = NULL;
35         struct strbuf fmt = STRBUF_INIT;
36         struct pretty_print_context ctx = {0};
37         ctx.date_mode.type = DATE_NORMAL;
38         ctx.abbrev = DEFAULT_ABBREV;
39
40         if (src == buf->buf)
41                 to_free = strbuf_detach(buf, NULL);
42         for (;;) {
43                 const char *b, *c;
44
45                 b = memmem(src, len, "$Format:", 8);
46                 if (!b)
47                         break;
48                 c = memchr(b + 8, '$', (src + len) - b - 8);
49                 if (!c)
50                         break;
51
52                 strbuf_reset(&fmt);
53                 strbuf_add(&fmt, b + 8, c - b - 8);
54
55                 strbuf_add(buf, src, b - src);
56                 format_commit_message(commit, fmt.buf, buf, &ctx);
57                 len -= c + 1 - src;
58                 src  = c + 1;
59         }
60         strbuf_add(buf, src, len);
61         strbuf_release(&fmt);
62         free(to_free);
63 }
64
65 void *sha1_file_to_archive(const struct archiver_args *args,
66                            const char *path, const unsigned char *sha1,
67                            unsigned int mode, enum object_type *type,
68                            unsigned long *sizep)
69 {
70         void *buffer;
71         const struct commit *commit = args->convert ? args->commit : NULL;
72
73         path += args->baselen;
74         buffer = read_sha1_file(sha1, type, sizep);
75         if (buffer && S_ISREG(mode)) {
76                 struct strbuf buf = STRBUF_INIT;
77                 size_t size = 0;
78
79                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
80                 convert_to_working_tree(path, buf.buf, buf.len, &buf);
81                 if (commit)
82                         format_subst(commit, buf.buf, buf.len, &buf);
83                 buffer = strbuf_detach(&buf, &size);
84                 *sizep = size;
85         }
86
87         return buffer;
88 }
89
90 static void setup_archive_check(struct git_attr_check *check)
91 {
92         static struct git_attr *attr_export_ignore;
93         static struct git_attr *attr_export_subst;
94
95         if (!attr_export_ignore) {
96                 attr_export_ignore = git_attr("export-ignore");
97                 attr_export_subst = git_attr("export-subst");
98         }
99         check[0].attr = attr_export_ignore;
100         check[1].attr = attr_export_subst;
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 int write_archive_entry(const unsigned char *sha1, const char *base,
119                 int baselen, const char *filename, unsigned mode, int stage,
120                 void *context)
121 {
122         static struct strbuf path = STRBUF_INIT;
123         struct archiver_context *c = context;
124         struct archiver_args *args = c->args;
125         write_archive_entry_fn_t write_entry = c->write_entry;
126         struct git_attr_check check[2];
127         const char *path_without_prefix;
128         int err;
129
130         args->convert = 0;
131         strbuf_reset(&path);
132         strbuf_grow(&path, PATH_MAX);
133         strbuf_add(&path, args->base, args->baselen);
134         strbuf_add(&path, base, baselen);
135         strbuf_addstr(&path, filename);
136         if (S_ISDIR(mode) || S_ISGITLINK(mode))
137                 strbuf_addch(&path, '/');
138         path_without_prefix = path.buf + args->baselen;
139
140         setup_archive_check(check);
141         if (!git_check_attr(path_without_prefix, ARRAY_SIZE(check), check)) {
142                 if (ATTR_TRUE(check[0].value))
143                         return 0;
144                 args->convert = ATTR_TRUE(check[1].value);
145         }
146
147         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
148                 if (args->verbose)
149                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
150                 err = write_entry(args, sha1, path.buf, path.len, mode);
151                 if (err)
152                         return err;
153                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
154         }
155
156         if (args->verbose)
157                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
158         return write_entry(args, sha1, path.buf, path.len, mode);
159 }
160
161 static int write_archive_entry_buf(const unsigned char *sha1, struct strbuf *base,
162                 const char *filename, unsigned mode, int stage,
163                 void *context)
164 {
165         return write_archive_entry(sha1, base->buf, base->len,
166                                      filename, mode, stage, context);
167 }
168
169 static void queue_directory(const unsigned char *sha1,
170                 struct strbuf *base, const char *filename,
171                 unsigned mode, int stage, struct archiver_context *c)
172 {
173         struct directory *d;
174         size_t len = st_add4(base->len, 1, strlen(filename), 1);
175         d = xmalloc(st_add(sizeof(*d), len));
176         d->up      = c->bottom;
177         d->baselen = base->len;
178         d->mode    = mode;
179         d->stage   = stage;
180         c->bottom  = d;
181         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
182         hashcpy(d->oid.hash, sha1);
183 }
184
185 static int write_directory(struct archiver_context *c)
186 {
187         struct directory *d = c->bottom;
188         int ret;
189
190         if (!d)
191                 return 0;
192         c->bottom = d->up;
193         d->path[d->len - 1] = '\0'; /* no trailing slash */
194         ret =
195                 write_directory(c) ||
196                 write_archive_entry(d->oid.hash, d->path, d->baselen,
197                                     d->path + d->baselen, d->mode,
198                                     d->stage, c) != READ_TREE_RECURSIVE;
199         free(d);
200         return ret ? -1 : 0;
201 }
202
203 static int queue_or_write_archive_entry(const unsigned char *sha1,
204                 struct strbuf *base, const char *filename,
205                 unsigned mode, int stage, void *context)
206 {
207         struct archiver_context *c = context;
208
209         while (c->bottom &&
210                !(base->len >= c->bottom->len &&
211                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
212                 struct directory *next = c->bottom->up;
213                 free(c->bottom);
214                 c->bottom = next;
215         }
216
217         if (S_ISDIR(mode)) {
218                 queue_directory(sha1, base, filename,
219                                 mode, stage, c);
220                 return READ_TREE_RECURSIVE;
221         }
222
223         if (write_directory(c))
224                 return -1;
225         return write_archive_entry(sha1, base->buf, base->len, filename, mode,
226                                    stage, context);
227 }
228
229 int write_archive_entries(struct archiver_args *args,
230                 write_archive_entry_fn_t write_entry)
231 {
232         struct archiver_context context;
233         struct unpack_trees_options opts;
234         struct tree_desc t;
235         int err;
236
237         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
238                 size_t len = args->baselen;
239
240                 while (len > 1 && args->base[len - 2] == '/')
241                         len--;
242                 if (args->verbose)
243                         fprintf(stderr, "%.*s\n", (int)len, args->base);
244                 err = write_entry(args, args->tree->object.oid.hash, args->base,
245                                   len, 040777);
246                 if (err)
247                         return err;
248         }
249
250         memset(&context, 0, sizeof(context));
251         context.args = args;
252         context.write_entry = write_entry;
253
254         /*
255          * Setup index and instruct attr to read index only
256          */
257         if (!args->worktree_attributes) {
258                 memset(&opts, 0, sizeof(opts));
259                 opts.index_only = 1;
260                 opts.head_idx = -1;
261                 opts.src_index = &the_index;
262                 opts.dst_index = &the_index;
263                 opts.fn = oneway_merge;
264                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
265                 if (unpack_trees(1, &t, &opts))
266                         return -1;
267                 git_attr_set_direction(GIT_ATTR_INDEX, &the_index);
268         }
269
270         err = read_tree_recursive(args->tree, "", 0, 0, &args->pathspec,
271                                   args->pathspec.has_wildcard ?
272                                   queue_or_write_archive_entry :
273                                   write_archive_entry_buf,
274                                   &context);
275         if (err == READ_TREE_RECURSIVE)
276                 err = 0;
277         while (context.bottom) {
278                 struct directory *next = context.bottom->up;
279                 free(context.bottom);
280                 context.bottom = next;
281         }
282         return err;
283 }
284
285 static const struct archiver *lookup_archiver(const char *name)
286 {
287         int i;
288
289         if (!name)
290                 return NULL;
291
292         for (i = 0; i < nr_archivers; i++) {
293                 if (!strcmp(name, archivers[i]->name))
294                         return archivers[i];
295         }
296         return NULL;
297 }
298
299 static int reject_entry(const unsigned char *sha1, struct strbuf *base,
300                         const char *filename, unsigned mode,
301                         int stage, void *context)
302 {
303         int ret = -1;
304         if (S_ISDIR(mode)) {
305                 struct strbuf sb = STRBUF_INIT;
306                 strbuf_addbuf(&sb, base);
307                 strbuf_addstr(&sb, filename);
308                 if (!match_pathspec(context, sb.buf, sb.len, 0, NULL, 1))
309                         ret = READ_TREE_RECURSIVE;
310                 strbuf_release(&sb);
311         }
312         return ret;
313 }
314
315 static int path_exists(struct tree *tree, const char *path)
316 {
317         const char *paths[] = { path, NULL };
318         struct pathspec pathspec;
319         int ret;
320
321         parse_pathspec(&pathspec, 0, 0, "", paths);
322         pathspec.recursive = 1;
323         ret = read_tree_recursive(tree, "", 0, 0, &pathspec,
324                                   reject_entry, &pathspec);
325         clear_pathspec(&pathspec);
326         return ret != 0;
327 }
328
329 static void parse_pathspec_arg(const char **pathspec,
330                 struct archiver_args *ar_args)
331 {
332         /*
333          * must be consistent with parse_pathspec in path_exists()
334          * Also if pathspec patterns are dependent, we're in big
335          * trouble as we test each one separately
336          */
337         parse_pathspec(&ar_args->pathspec, 0,
338                        PATHSPEC_PREFER_FULL,
339                        "", pathspec);
340         ar_args->pathspec.recursive = 1;
341         if (pathspec) {
342                 while (*pathspec) {
343                         if (**pathspec && !path_exists(ar_args->tree, *pathspec))
344                                 die(_("pathspec '%s' did not match any files"), *pathspec);
345                         pathspec++;
346                 }
347         }
348 }
349
350 static void parse_treeish_arg(const char **argv,
351                 struct archiver_args *ar_args, const char *prefix,
352                 int remote)
353 {
354         const char *name = argv[0];
355         const unsigned char *commit_sha1;
356         time_t archive_time;
357         struct tree *tree;
358         const struct commit *commit;
359         struct object_id oid;
360
361         /* Remotes are only allowed to fetch actual refs */
362         if (remote && !remote_allow_unreachable) {
363                 char *ref = NULL;
364                 const char *colon = strchrnul(name, ':');
365                 int refnamelen = colon - name;
366
367                 if (!dwim_ref(name, refnamelen, oid.hash, &ref))
368                         die("no such ref: %.*s", refnamelen, name);
369                 free(ref);
370         }
371
372         if (get_sha1(name, oid.hash))
373                 die("Not a valid object name");
374
375         commit = lookup_commit_reference_gently(oid.hash, 1);
376         if (commit) {
377                 commit_sha1 = commit->object.oid.hash;
378                 archive_time = commit->date;
379         } else {
380                 commit_sha1 = NULL;
381                 archive_time = time(NULL);
382         }
383
384         tree = parse_tree_indirect(oid.hash);
385         if (tree == NULL)
386                 die("not a tree object");
387
388         if (prefix) {
389                 struct object_id tree_oid;
390                 unsigned int mode;
391                 int err;
392
393                 err = get_tree_entry(tree->object.oid.hash, prefix,
394                                      tree_oid.hash, &mode);
395                 if (err || !S_ISDIR(mode))
396                         die("current working directory is untracked");
397
398                 tree = parse_tree_indirect(tree_oid.hash);
399         }
400         ar_args->tree = tree;
401         ar_args->commit_sha1 = commit_sha1;
402         ar_args->commit = commit;
403         ar_args->time = archive_time;
404 }
405
406 #define OPT__COMPR(s, v, h, p) \
407         { OPTION_SET_INT, (s), NULL, (v), NULL, (h), \
408           PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, (p) }
409 #define OPT__COMPR_HIDDEN(s, v, p) \
410         { OPTION_SET_INT, (s), NULL, (v), NULL, "", \
411           PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
412
413 static int parse_archive_args(int argc, const char **argv,
414                 const struct archiver **ar, struct archiver_args *args,
415                 const char *name_hint, int is_remote)
416 {
417         const char *format = NULL;
418         const char *base = NULL;
419         const char *remote = NULL;
420         const char *exec = NULL;
421         const char *output = NULL;
422         int compression_level = -1;
423         int verbose = 0;
424         int i;
425         int list = 0;
426         int worktree_attributes = 0;
427         struct option opts[] = {
428                 OPT_GROUP(""),
429                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
430                 OPT_STRING(0, "prefix", &base, N_("prefix"),
431                         N_("prepend prefix to each pathname in the archive")),
432                 OPT_STRING('o', "output", &output, N_("file"),
433                         N_("write the archive to this file")),
434                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
435                         N_("read .gitattributes in working directory")),
436                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
437                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
438                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
439                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
440                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
441                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
442                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
443                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
444                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
445                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
446                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
447                 OPT_GROUP(""),
448                 OPT_BOOL('l', "list", &list,
449                         N_("list supported archive formats")),
450                 OPT_GROUP(""),
451                 OPT_STRING(0, "remote", &remote, N_("repo"),
452                         N_("retrieve the archive from remote repository <repo>")),
453                 OPT_STRING(0, "exec", &exec, N_("command"),
454                         N_("path to the remote git-upload-archive command")),
455                 OPT_END()
456         };
457
458         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
459
460         if (remote)
461                 die("Unexpected option --remote");
462         if (exec)
463                 die("Option --exec can only be used together with --remote");
464         if (output)
465                 die("Unexpected option --output");
466
467         if (!base)
468                 base = "";
469
470         if (list) {
471                 for (i = 0; i < nr_archivers; i++)
472                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
473                                 printf("%s\n", archivers[i]->name);
474                 exit(0);
475         }
476
477         if (!format && name_hint)
478                 format = archive_format_from_filename(name_hint);
479         if (!format)
480                 format = "tar";
481
482         /* We need at least one parameter -- tree-ish */
483         if (argc < 1)
484                 usage_with_options(archive_usage, opts);
485         *ar = lookup_archiver(format);
486         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
487                 die("Unknown archive format '%s'", format);
488
489         args->compression_level = Z_DEFAULT_COMPRESSION;
490         if (compression_level != -1) {
491                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
492                         args->compression_level = compression_level;
493                 else {
494                         die("Argument not supported for format '%s': -%d",
495                                         format, compression_level);
496                 }
497         }
498         args->verbose = verbose;
499         args->base = base;
500         args->baselen = strlen(base);
501         args->worktree_attributes = worktree_attributes;
502
503         return argc;
504 }
505
506 int write_archive(int argc, const char **argv, const char *prefix,
507                   int setup_prefix, const char *name_hint, int remote)
508 {
509         int nongit = 0;
510         const struct archiver *ar = NULL;
511         struct archiver_args args;
512
513         if (setup_prefix && prefix == NULL)
514                 prefix = setup_git_directory_gently(&nongit);
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 (nongit) {
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 }