1#include "cache.h"
2#include "commit.h"
3#include "tree-walk.h"
4#include "attr.h"
5#include "archive.h"
6#include "parse-options.h"
7#include "unpack-trees.h"
8
9static char const * const archive_usage[] = {
10 "git archive [options] <tree-ish> [<path>...]",
11 "git archive --list",
12 "git archive --remote <repo> [--exec <cmd>] [options] <tree-ish> [<path>...]",
13 "git archive --remote <repo> [--exec <cmd>] --list",
14 NULL
15};
16
17static const struct archiver **archivers;
18static int nr_archivers;
19static int alloc_archivers;
20
21void register_archiver(struct archiver *ar)
22{
23 ALLOC_GROW(archivers, nr_archivers + 1, alloc_archivers);
24 archivers[nr_archivers++] = ar;
25}
26
27static void format_subst(const struct commit *commit,
28 const char *src, size_t len,
29 struct strbuf *buf)
30{
31 char *to_free = NULL;
32 struct strbuf fmt = STRBUF_INIT;
33 struct pretty_print_context ctx = {0};
34 ctx.date_mode = DATE_NORMAL;
35 ctx.abbrev = DEFAULT_ABBREV;
36
37 if (src == buf->buf)
38 to_free = strbuf_detach(buf, NULL);
39 for (;;) {
40 const char *b, *c;
41
42 b = memmem(src, len, "$Format:", 8);
43 if (!b)
44 break;
45 c = memchr(b + 8, '$', (src + len) - b - 8);
46 if (!c)
47 break;
48
49 strbuf_reset(&fmt);
50 strbuf_add(&fmt, b + 8, c - b - 8);
51
52 strbuf_add(buf, src, b - src);
53 format_commit_message(commit, fmt.buf, buf, &ctx);
54 len -= c + 1 - src;
55 src = c + 1;
56 }
57 strbuf_add(buf, src, len);
58 strbuf_release(&fmt);
59 free(to_free);
60}
61
62static void *sha1_file_to_archive(const char *path, const unsigned char *sha1,
63 unsigned int mode, enum object_type *type,
64 unsigned long *sizep, const struct commit *commit)
65{
66 void *buffer;
67
68 buffer = read_sha1_file(sha1, type, sizep);
69 if (buffer && S_ISREG(mode)) {
70 struct strbuf buf = STRBUF_INIT;
71 size_t size = 0;
72
73 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
74 convert_to_working_tree(path, buf.buf, buf.len, &buf);
75 if (commit)
76 format_subst(commit, buf.buf, buf.len, &buf);
77 buffer = strbuf_detach(&buf, &size);
78 *sizep = size;
79 }
80
81 return buffer;
82}
83
84static void setup_archive_check(struct git_attr_check *check)
85{
86 static struct git_attr *attr_export_ignore;
87 static struct git_attr *attr_export_subst;
88
89 if (!attr_export_ignore) {
90 attr_export_ignore = git_attr("export-ignore");
91 attr_export_subst = git_attr("export-subst");
92 }
93 check[0].attr = attr_export_ignore;
94 check[1].attr = attr_export_subst;
95}
96
97struct archiver_context {
98 struct archiver_args *args;
99 write_archive_entry_fn_t write_entry;
100};
101
102static int write_archive_entry(const unsigned char *sha1, const char *base,
103 int baselen, const char *filename, unsigned mode, int stage,
104 void *context)
105{
106 static struct strbuf path = STRBUF_INIT;
107 struct archiver_context *c = context;
108 struct archiver_args *args = c->args;
109 write_archive_entry_fn_t write_entry = c->write_entry;
110 struct git_attr_check check[2];
111 const char *path_without_prefix;
112 int convert = 0;
113 int err;
114 enum object_type type;
115 unsigned long size;
116 void *buffer;
117
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 path_without_prefix = path.buf + args->baselen;
124
125 setup_archive_check(check);
126 if (!git_check_attr(path_without_prefix, ARRAY_SIZE(check), check)) {
127 if (ATTR_TRUE(check[0].value))
128 return 0;
129 convert = ATTR_TRUE(check[1].value);
130 }
131
132 if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
133 strbuf_addch(&path, '/');
134 if (args->verbose)
135 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
136 err = write_entry(args, sha1, path.buf, path.len, mode, NULL, 0);
137 if (err)
138 return err;
139 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
140 }
141
142 buffer = sha1_file_to_archive(path_without_prefix, sha1, mode,
143 &type, &size, convert ? args->commit : NULL);
144 if (!buffer)
145 return error("cannot read %s", sha1_to_hex(sha1));
146 if (args->verbose)
147 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
148 err = write_entry(args, sha1, path.buf, path.len, mode, buffer, size);
149 free(buffer);
150 return err;
151}
152
153int write_archive_entries(struct archiver_args *args,
154 write_archive_entry_fn_t write_entry)
155{
156 struct archiver_context context;
157 struct unpack_trees_options opts;
158 struct tree_desc t;
159 struct pathspec pathspec;
160 int err;
161
162 if (args->baselen > 0 && args->base[args->baselen - 1] == '/') {
163 size_t len = args->baselen;
164
165 while (len > 1 && args->base[len - 2] == '/')
166 len--;
167 if (args->verbose)
168 fprintf(stderr, "%.*s\n", (int)len, args->base);
169 err = write_entry(args, args->tree->object.sha1, args->base,
170 len, 040777, NULL, 0);
171 if (err)
172 return err;
173 }
174
175 context.args = args;
176 context.write_entry = write_entry;
177
178 /*
179 * Setup index and instruct attr to read index only
180 */
181 if (!args->worktree_attributes) {
182 memset(&opts, 0, sizeof(opts));
183 opts.index_only = 1;
184 opts.head_idx = -1;
185 opts.src_index = &the_index;
186 opts.dst_index = &the_index;
187 opts.fn = oneway_merge;
188 init_tree_desc(&t, args->tree->buffer, args->tree->size);
189 if (unpack_trees(1, &t, &opts))
190 return -1;
191 git_attr_set_direction(GIT_ATTR_INDEX, &the_index);
192 }
193
194 init_pathspec(&pathspec, args->pathspec);
195 err = read_tree_recursive(args->tree, "", 0, 0, &pathspec,
196 write_archive_entry, &context);
197 free_pathspec(&pathspec);
198 if (err == READ_TREE_RECURSIVE)
199 err = 0;
200 return err;
201}
202
203static const struct archiver *lookup_archiver(const char *name)
204{
205 int i;
206
207 if (!name)
208 return NULL;
209
210 for (i = 0; i < nr_archivers; i++) {
211 if (!strcmp(name, archivers[i]->name))
212 return archivers[i];
213 }
214 return NULL;
215}
216
217static int reject_entry(const unsigned char *sha1, const char *base,
218 int baselen, const char *filename, unsigned mode,
219 int stage, void *context)
220{
221 return -1;
222}
223
224static int path_exists(struct tree *tree, const char *path)
225{
226 const char *paths[] = { path, NULL };
227 struct pathspec pathspec;
228 int ret;
229
230 init_pathspec(&pathspec, paths);
231 ret = read_tree_recursive(tree, "", 0, 0, &pathspec, reject_entry, NULL);
232 free_pathspec(&pathspec);
233 return ret != 0;
234}
235
236static void parse_pathspec_arg(const char **pathspec,
237 struct archiver_args *ar_args)
238{
239 ar_args->pathspec = pathspec = get_pathspec("", pathspec);
240 if (pathspec) {
241 while (*pathspec) {
242 if (!path_exists(ar_args->tree, *pathspec))
243 die("path not found: %s", *pathspec);
244 pathspec++;
245 }
246 }
247}
248
249static void parse_treeish_arg(const char **argv,
250 struct archiver_args *ar_args, const char *prefix,
251 int remote)
252{
253 const char *name = argv[0];
254 const unsigned char *commit_sha1;
255 time_t archive_time;
256 struct tree *tree;
257 const struct commit *commit;
258 unsigned char sha1[20];
259
260 /* Remotes are only allowed to fetch actual refs */
261 if (remote) {
262 char *ref = NULL;
263 if (!dwim_ref(name, strlen(name), sha1, &ref))
264 die("no such ref: %s", name);
265 free(ref);
266 }
267 else {
268 if (get_sha1(name, sha1))
269 die("Not a valid object name");
270 }
271
272 commit = lookup_commit_reference_gently(sha1, 1);
273 if (commit) {
274 commit_sha1 = commit->object.sha1;
275 archive_time = commit->date;
276 } else {
277 commit_sha1 = NULL;
278 archive_time = time(NULL);
279 }
280
281 tree = parse_tree_indirect(sha1);
282 if (tree == NULL)
283 die("not a tree object");
284
285 if (prefix) {
286 unsigned char tree_sha1[20];
287 unsigned int mode;
288 int err;
289
290 err = get_tree_entry(tree->object.sha1, prefix,
291 tree_sha1, &mode);
292 if (err || !S_ISDIR(mode))
293 die("current working directory is untracked");
294
295 tree = parse_tree_indirect(tree_sha1);
296 }
297 ar_args->tree = tree;
298 ar_args->commit_sha1 = commit_sha1;
299 ar_args->commit = commit;
300 ar_args->time = archive_time;
301}
302
303#define OPT__COMPR(s, v, h, p) \
304 { OPTION_SET_INT, (s), NULL, (v), NULL, (h), \
305 PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, (p) }
306#define OPT__COMPR_HIDDEN(s, v, p) \
307 { OPTION_SET_INT, (s), NULL, (v), NULL, "", \
308 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
309
310static int parse_archive_args(int argc, const char **argv,
311 const struct archiver **ar, struct archiver_args *args,
312 const char *name_hint, int is_remote)
313{
314 const char *format = NULL;
315 const char *base = NULL;
316 const char *remote = NULL;
317 const char *exec = NULL;
318 const char *output = NULL;
319 int compression_level = -1;
320 int verbose = 0;
321 int i;
322 int list = 0;
323 int worktree_attributes = 0;
324 struct option opts[] = {
325 OPT_GROUP(""),
326 OPT_STRING(0, "format", &format, "fmt", "archive format"),
327 OPT_STRING(0, "prefix", &base, "prefix",
328 "prepend prefix to each pathname in the archive"),
329 OPT_STRING('o', "output", &output, "file",
330 "write the archive to this file"),
331 OPT_BOOL(0, "worktree-attributes", &worktree_attributes,
332 "read .gitattributes in working directory"),
333 OPT__VERBOSE(&verbose, "report archived files on stderr"),
334 OPT__COMPR('0', &compression_level, "store only", 0),
335 OPT__COMPR('1', &compression_level, "compress faster", 1),
336 OPT__COMPR_HIDDEN('2', &compression_level, 2),
337 OPT__COMPR_HIDDEN('3', &compression_level, 3),
338 OPT__COMPR_HIDDEN('4', &compression_level, 4),
339 OPT__COMPR_HIDDEN('5', &compression_level, 5),
340 OPT__COMPR_HIDDEN('6', &compression_level, 6),
341 OPT__COMPR_HIDDEN('7', &compression_level, 7),
342 OPT__COMPR_HIDDEN('8', &compression_level, 8),
343 OPT__COMPR('9', &compression_level, "compress better", 9),
344 OPT_GROUP(""),
345 OPT_BOOL('l', "list", &list,
346 "list supported archive formats"),
347 OPT_GROUP(""),
348 OPT_STRING(0, "remote", &remote, "repo",
349 "retrieve the archive from remote repository <repo>"),
350 OPT_STRING(0, "exec", &exec, "cmd",
351 "path to the remote git-upload-archive command"),
352 OPT_END()
353 };
354
355 argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
356
357 if (remote)
358 die("Unexpected option --remote");
359 if (exec)
360 die("Option --exec can only be used together with --remote");
361 if (output)
362 die("Unexpected option --output");
363
364 if (!base)
365 base = "";
366
367 if (list) {
368 for (i = 0; i < nr_archivers; i++)
369 if (!is_remote || archivers[i]->flags & ARCHIVER_REMOTE)
370 printf("%s\n", archivers[i]->name);
371 exit(0);
372 }
373
374 if (!format && name_hint)
375 format = archive_format_from_filename(name_hint);
376 if (!format)
377 format = "tar";
378
379 /* We need at least one parameter -- tree-ish */
380 if (argc < 1)
381 usage_with_options(archive_usage, opts);
382 *ar = lookup_archiver(format);
383 if (!*ar || (is_remote && !((*ar)->flags & ARCHIVER_REMOTE)))
384 die("Unknown archive format '%s'", format);
385
386 args->compression_level = Z_DEFAULT_COMPRESSION;
387 if (compression_level != -1) {
388 if ((*ar)->flags & ARCHIVER_WANT_COMPRESSION_LEVELS)
389 args->compression_level = compression_level;
390 else {
391 die("Argument not supported for format '%s': -%d",
392 format, compression_level);
393 }
394 }
395 args->verbose = verbose;
396 args->base = base;
397 args->baselen = strlen(base);
398 args->worktree_attributes = worktree_attributes;
399
400 return argc;
401}
402
403int write_archive(int argc, const char **argv, const char *prefix,
404 int setup_prefix, const char *name_hint, int remote)
405{
406 int nongit = 0;
407 const struct archiver *ar = NULL;
408 struct archiver_args args;
409
410 if (setup_prefix && prefix == NULL)
411 prefix = setup_git_directory_gently(&nongit);
412
413 git_config(git_default_config, NULL);
414 init_tar_archiver();
415 init_zip_archiver();
416
417 argc = parse_archive_args(argc, argv, &ar, &args, name_hint, remote);
418 if (nongit) {
419 /*
420 * We know this will die() with an error, so we could just
421 * die ourselves; but its error message will be more specific
422 * than what we could write here.
423 */
424 setup_git_directory();
425 }
426
427 parse_treeish_arg(argv, &args, prefix, remote);
428 parse_pathspec_arg(argv + 1, &args);
429
430 return ar->write_archive(ar, &args);
431}
432
433static int match_extension(const char *filename, const char *ext)
434{
435 int prefixlen = strlen(filename) - strlen(ext);
436
437 /*
438 * We need 1 character for the '.', and 1 character to ensure that the
439 * prefix is non-empty (k.e., we don't match .tar.gz with no actual
440 * filename).
441 */
442 if (prefixlen < 2 || filename[prefixlen-1] != '.')
443 return 0;
444 return !strcmp(filename + prefixlen, ext);
445}
446
447const char *archive_format_from_filename(const char *filename)
448{
449 int i;
450
451 for (i = 0; i < nr_archivers; i++)
452 if (match_extension(filename, archivers[i]->name))
453 return archivers[i]->name;
454 return NULL;
455}