Merge branch 'jk/am-leakfix' into maint
[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 struct directory {
91         struct directory *up;
92         struct object_id oid;
93         int baselen, len;
94         unsigned mode;
95         int stage;
96         char path[FLEX_ARRAY];
97 };
98
99 struct archiver_context {
100         struct archiver_args *args;
101         write_archive_entry_fn_t write_entry;
102         struct directory *bottom;
103 };
104
105 static int write_archive_entry(const unsigned char *sha1, const char *base,
106                 int baselen, const char *filename, unsigned mode, int stage,
107                 void *context)
108 {
109         static struct strbuf path = STRBUF_INIT;
110         static struct attr_check *check;
111         struct archiver_context *c = context;
112         struct archiver_args *args = c->args;
113         write_archive_entry_fn_t write_entry = c->write_entry;
114         const char *path_without_prefix;
115         int err;
116
117         args->convert = 0;
118         strbuf_reset(&path);
119         strbuf_grow(&path, PATH_MAX);
120         strbuf_add(&path, args->base, args->baselen);
121         strbuf_add(&path, base, baselen);
122         strbuf_addstr(&path, filename);
123         if (S_ISDIR(mode) || S_ISGITLINK(mode))
124                 strbuf_addch(&path, '/');
125         path_without_prefix = path.buf + args->baselen;
126
127         if (!check)
128                 check = attr_check_initl("export-ignore", "export-subst", NULL);
129         if (!git_check_attr(path_without_prefix, check)) {
130                 if (ATTR_TRUE(check->items[0].value))
131                         return 0;
132                 args->convert = ATTR_TRUE(check->items[1].value);
133         }
134
135         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
136                 if (args->verbose)
137                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
138                 err = write_entry(args, sha1, path.buf, path.len, mode);
139                 if (err)
140                         return err;
141                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
142         }
143
144         if (args->verbose)
145                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
146         return write_entry(args, sha1, path.buf, path.len, mode);
147 }
148
149 static int write_archive_entry_buf(const unsigned char *sha1, struct strbuf *base,
150                 const char *filename, unsigned mode, int stage,
151                 void *context)
152 {
153         return write_archive_entry(sha1, base->buf, base->len,
154                                      filename, mode, stage, context);
155 }
156
157 static void queue_directory(const unsigned char *sha1,
158                 struct strbuf *base, const char *filename,
159                 unsigned mode, int stage, struct archiver_context *c)
160 {
161         struct directory *d;
162         size_t len = st_add4(base->len, 1, strlen(filename), 1);
163         d = xmalloc(st_add(sizeof(*d), len));
164         d->up      = c->bottom;
165         d->baselen = base->len;
166         d->mode    = mode;
167         d->stage   = stage;
168         c->bottom  = d;
169         d->len = xsnprintf(d->path, len, "%.*s%s/", (int)base->len, base->buf, filename);
170         hashcpy(d->oid.hash, sha1);
171 }
172
173 static int write_directory(struct archiver_context *c)
174 {
175         struct directory *d = c->bottom;
176         int ret;
177
178         if (!d)
179                 return 0;
180         c->bottom = d->up;
181         d->path[d->len - 1] = '\0'; /* no trailing slash */
182         ret =
183                 write_directory(c) ||
184                 write_archive_entry(d->oid.hash, d->path, d->baselen,
185                                     d->path + d->baselen, d->mode,
186                                     d->stage, c) != READ_TREE_RECURSIVE;
187         free(d);
188         return ret ? -1 : 0;
189 }
190
191 static int queue_or_write_archive_entry(const unsigned char *sha1,
192                 struct strbuf *base, const char *filename,
193                 unsigned mode, int stage, void *context)
194 {
195         struct archiver_context *c = context;
196
197         while (c->bottom &&
198                !(base->len >= c->bottom->len &&
199                  !strncmp(base->buf, c->bottom->path, c->bottom->len))) {
200                 struct directory *next = c->bottom->up;
201                 free(c->bottom);
202                 c->bottom = next;
203         }
204
205         if (S_ISDIR(mode)) {
206                 queue_directory(sha1, base, filename,
207                                 mode, stage, c);
208                 return READ_TREE_RECURSIVE;
209         }
210
211         if (write_directory(c))
212                 return -1;
213         return write_archive_entry(sha1, base->buf, base->len, filename, mode,
214                                    stage, context);
215 }
216
217 int write_archive_entries(struct archiver_args *args,
218                 write_archive_entry_fn_t write_entry)
219 {
220         struct archiver_context context;
221         struct unpack_trees_options opts;
222         struct tree_desc t;
223         int err;
224
225         if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
226                 size_t len = args->baselen;
227
228                 while (len > 1 && args->base[len - 2] == '/')
229                         len--;
230                 if (args->verbose)
231                         fprintf(stderr, "%.*s\n", (int)len, args->base);
232                 err = write_entry(args, args->tree->object.oid.hash, args->base,
233                                   len, 040777);
234                 if (err)
235                         return err;
236         }
237
238         memset(&context, 0, sizeof(context));
239         context.args = args;
240         context.write_entry = write_entry;
241
242         /*
243          * Setup index and instruct attr to read index only
244          */
245         if (!args->worktree_attributes) {
246                 memset(&opts, 0, sizeof(opts));
247                 opts.index_only = 1;
248                 opts.head_idx = -1;
249                 opts.src_index = &the_index;
250                 opts.dst_index = &the_index;
251                 opts.fn = oneway_merge;
252                 init_tree_desc(&t, args->tree->buffer, args->tree->size);
253                 if (unpack_trees(1, &t, &opts))
254                         return -1;
255                 git_attr_set_direction(GIT_ATTR_INDEX, &the_index);
256         }
257
258         err = read_tree_recursive(args->tree, "", 0, 0, &args->pathspec,
259                                   args->pathspec.has_wildcard ?
260                                   queue_or_write_archive_entry :
261                                   write_archive_entry_buf,
262                                   &context);
263         if (err == READ_TREE_RECURSIVE)
264                 err = 0;
265         while (context.bottom) {
266                 struct directory *next = context.bottom->up;
267                 free(context.bottom);
268                 context.bottom = next;
269         }
270         return err;
271 }
272
273 static const struct archiver *lookup_archiver(const char *name)
274 {
275         int i;
276
277         if (!name)
278                 return NULL;
279
280         for (i = 0; i < nr_archivers; i++) {
281                 if (!strcmp(name, archivers[i]->name))
282                         return archivers[i];
283         }
284         return NULL;
285 }
286
287 static int reject_entry(const unsigned char *sha1, struct strbuf *base,
288                         const char *filename, unsigned mode,
289                         int stage, void *context)
290 {
291         int ret = -1;
292         if (S_ISDIR(mode)) {
293                 struct strbuf sb = STRBUF_INIT;
294                 strbuf_addbuf(&sb, base);
295                 strbuf_addstr(&sb, filename);
296                 if (!match_pathspec(context, sb.buf, sb.len, 0, NULL, 1))
297                         ret = READ_TREE_RECURSIVE;
298                 strbuf_release(&sb);
299         }
300         return ret;
301 }
302
303 static int path_exists(struct tree *tree, const char *path)
304 {
305         const char *paths[] = { path, NULL };
306         struct pathspec pathspec;
307         int ret;
308
309         parse_pathspec(&pathspec, 0, 0, "", paths);
310         pathspec.recursive = 1;
311         ret = read_tree_recursive(tree, "", 0, 0, &pathspec,
312                                   reject_entry, &pathspec);
313         clear_pathspec(&pathspec);
314         return ret != 0;
315 }
316
317 static void parse_pathspec_arg(const char **pathspec,
318                 struct archiver_args *ar_args)
319 {
320         /*
321          * must be consistent with parse_pathspec in path_exists()
322          * Also if pathspec patterns are dependent, we're in big
323          * trouble as we test each one separately
324          */
325         parse_pathspec(&ar_args->pathspec, 0,
326                        PATHSPEC_PREFER_FULL,
327                        "", pathspec);
328         ar_args->pathspec.recursive = 1;
329         if (pathspec) {
330                 while (*pathspec) {
331                         if (**pathspec && !path_exists(ar_args->tree, *pathspec))
332                                 die(_("pathspec '%s' did not match any files"), *pathspec);
333                         pathspec++;
334                 }
335         }
336 }
337
338 static void parse_treeish_arg(const char **argv,
339                 struct archiver_args *ar_args, const char *prefix,
340                 int remote)
341 {
342         const char *name = argv[0];
343         const unsigned char *commit_sha1;
344         time_t archive_time;
345         struct tree *tree;
346         const struct commit *commit;
347         struct object_id oid;
348
349         /* Remotes are only allowed to fetch actual refs */
350         if (remote && !remote_allow_unreachable) {
351                 char *ref = NULL;
352                 const char *colon = strchrnul(name, ':');
353                 int refnamelen = colon - name;
354
355                 if (!dwim_ref(name, refnamelen, oid.hash, &ref))
356                         die("no such ref: %.*s", refnamelen, name);
357                 free(ref);
358         }
359
360         if (get_sha1(name, oid.hash))
361                 die("Not a valid object name");
362
363         commit = lookup_commit_reference_gently(oid.hash, 1);
364         if (commit) {
365                 commit_sha1 = commit->object.oid.hash;
366                 archive_time = commit->date;
367         } else {
368                 commit_sha1 = NULL;
369                 archive_time = time(NULL);
370         }
371
372         tree = parse_tree_indirect(oid.hash);
373         if (tree == NULL)
374                 die("not a tree object");
375
376         if (prefix) {
377                 struct object_id tree_oid;
378                 unsigned int mode;
379                 int err;
380
381                 err = get_tree_entry(tree->object.oid.hash, prefix,
382                                      tree_oid.hash, &mode);
383                 if (err || !S_ISDIR(mode))
384                         die("current working directory is untracked");
385
386                 tree = parse_tree_indirect(tree_oid.hash);
387         }
388         ar_args->tree = tree;
389         ar_args->commit_sha1 = commit_sha1;
390         ar_args->commit = commit;
391         ar_args->time = archive_time;
392 }
393
394 #define OPT__COMPR(s, v, h, p) \
395         { OPTION_SET_INT, (s), NULL, (v), NULL, (h), \
396           PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, (p) }
397 #define OPT__COMPR_HIDDEN(s, v, p) \
398         { OPTION_SET_INT, (s), NULL, (v), NULL, "", \
399           PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
400
401 static int parse_archive_args(int argc, const char **argv,
402                 const struct archiver **ar, struct archiver_args *args,
403                 const char *name_hint, int is_remote)
404 {
405         const char *format = NULL;
406         const char *base = NULL;
407         const char *remote = NULL;
408         const char *exec = NULL;
409         const char *output = NULL;
410         int compression_level = -1;
411         int verbose = 0;
412         int i;
413         int list = 0;
414         int worktree_attributes = 0;
415         struct option opts[] = {
416                 OPT_GROUP(""),
417                 OPT_STRING(0, "format", &format, N_("fmt"), N_("archive format")),
418                 OPT_STRING(0, "prefix", &base, N_("prefix"),
419                         N_("prepend prefix to each pathname in the archive")),
420                 OPT_STRING('o', "output", &output, N_("file"),
421                         N_("write the archive to this file")),
422                 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
423                         N_("read .gitattributes in working directory")),
424                 OPT__VERBOSE(&verbose, N_("report archived files on stderr")),
425                 OPT__COMPR('0', &compression_level, N_("store only"), 0),
426                 OPT__COMPR('1', &compression_level, N_("compress faster"), 1),
427                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
428                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
429                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
430                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
431                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
432                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
433                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
434                 OPT__COMPR('9', &compression_level, N_("compress better"), 9),
435                 OPT_GROUP(""),
436                 OPT_BOOL('l', "list", &list,
437                         N_("list supported archive formats")),
438                 OPT_GROUP(""),
439                 OPT_STRING(0, "remote", &remote, N_("repo"),
440                         N_("retrieve the archive from remote repository <repo>")),
441                 OPT_STRING(0, "exec", &exec, N_("command"),
442                         N_("path to the remote git-upload-archive command")),
443                 OPT_END()
444         };
445
446         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
447
448         if (remote)
449                 die(_("Unexpected option --remote"));
450         if (exec)
451                 die(_("Option --exec can only be used together with --remote"));
452         if (output)
453                 die(_("Unexpected option --output"));
454
455         if (!base)
456                 base = "";
457
458         if (list) {
459                 for (i = 0; i < nr_archivers; i++)
460                         if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
461                                 printf("%s\n", archivers[i]->name);
462                 exit(0);
463         }
464
465         if (!format && name_hint)
466                 format = archive_format_from_filename(name_hint);
467         if (!format)
468                 format = "tar";
469
470         /* We need at least one parameter -- tree-ish */
471         if (argc < 1)
472                 usage_with_options(archive_usage, opts);
473         *ar = lookup_archiver(format);
474         if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
475                 die(_("Unknown archive format '%s'"), format);
476
477         args->compression_level = Z_DEFAULT_COMPRESSION;
478         if (compression_level != -1) {
479                 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
480                         args->compression_level = compression_level;
481                 else {
482                         die(_("Argument not supported for format '%s': -%d"),
483                                         format, compression_level);
484                 }
485         }
486         args->verbose = verbose;
487         args->base = base;
488         args->baselen = strlen(base);
489         args->worktree_attributes = worktree_attributes;
490
491         return argc;
492 }
493
494 int write_archive(int argc, const char **argv, const char *prefix,
495                   const char *name_hint, int remote)
496 {
497         const struct archiver *ar = NULL;
498         struct archiver_args args;
499
500         git_config_get_bool("uploadarchive.allowunreachable", &remote_allow_unreachable);
501         git_config(git_default_config, NULL);
502
503         init_tar_archiver();
504         init_zip_archiver();
505
506         argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
507         if (!startup_info->have_repository) {
508                 /*
509                  * We know this will die() with an error, so we could just
510                  * die ourselves; but its error message will be more specific
511                  * than what we could write here.
512                  */
513                 setup_git_directory();
514         }
515
516         parse_treeish_arg(argv, &args, prefix, remote);
517         parse_pathspec_arg(argv + 1, &args);
518
519         return ar->write_archive(ar, &args);
520 }
521
522 static int match_extension(const char *filename, const char *ext)
523 {
524         int prefixlen = strlen(filename) - strlen(ext);
525
526         /*
527          * We need 1 character for the '.', and 1 character to ensure that the
528          * prefix is non-empty (k.e., we don't match .tar.gz with no actual
529          * filename).
530          */
531         if (prefixlen < 2 || filename[prefixlen - 1] != '.')
532                 return 0;
533         return !strcmp(filename + prefixlen, ext);
534 }
535
536 const char *archive_format_from_filename(const char *filename)
537 {
538         int i;
539
540         for (i = 0; i < nr_archivers; i++)
541                 if (match_extension(filename, archivers[i]->name))
542                         return archivers[i]->name;
543         return NULL;
544 }