Merge branch 'jn/fast-import-blob-access'
[git] / archive.c
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
9 static 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
17 #define USES_ZLIB_COMPRESSION 1
18
19 static const struct archiver {
20         const char *name;
21         write_archive_fn_t write_archive;
22         unsigned int flags;
23 } archivers[] = {
24         { "tar", write_tar_archive },
25         { "zip", write_zip_archive, USES_ZLIB_COMPRESSION },
26 };
27
28 static void format_subst(const struct commit *commit,
29                          const char *src, size_t len,
30                          struct strbuf *buf)
31 {
32         char *to_free = NULL;
33         struct strbuf fmt = STRBUF_INIT;
34         struct pretty_print_context ctx = {0};
35         ctx.date_mode = DATE_NORMAL;
36         ctx.abbrev = DEFAULT_ABBREV;
37
38         if (src == buf->buf)
39                 to_free = strbuf_detach(buf, NULL);
40         for (;;) {
41                 const char *b, *c;
42
43                 b = memmem(src, len, "$Format:", 8);
44                 if (!b)
45                         break;
46                 c = memchr(b + 8, '$', (src + len) - b - 8);
47                 if (!c)
48                         break;
49
50                 strbuf_reset(&fmt);
51                 strbuf_add(&fmt, b + 8, c - b - 8);
52
53                 strbuf_add(buf, src, b - src);
54                 format_commit_message(commit, fmt.buf, buf, &ctx);
55                 len -= c + 1 - src;
56                 src  = c + 1;
57         }
58         strbuf_add(buf, src, len);
59         strbuf_release(&fmt);
60         free(to_free);
61 }
62
63 static void *sha1_file_to_archive(const char *path, const unsigned char *sha1,
64                 unsigned int mode, enum object_type *type,
65                 unsigned long *sizep, const struct commit *commit)
66 {
67         void *buffer;
68
69         buffer = read_sha1_file(sha1, type, sizep);
70         if (buffer && S_ISREG(mode)) {
71                 struct strbuf buf = STRBUF_INIT;
72                 size_t size = 0;
73
74                 strbuf_attach(&buf, buffer, *sizep, *sizep + 1);
75                 convert_to_working_tree(path, buf.buf, buf.len, &buf);
76                 if (commit)
77                         format_subst(commit, buf.buf, buf.len, &buf);
78                 buffer = strbuf_detach(&buf, &size);
79                 *sizep = size;
80         }
81
82         return buffer;
83 }
84
85 static void setup_archive_check(struct git_attr_check *check)
86 {
87         static struct git_attr *attr_export_ignore;
88         static struct git_attr *attr_export_subst;
89
90         if (!attr_export_ignore) {
91                 attr_export_ignore = git_attr("export-ignore");
92                 attr_export_subst = git_attr("export-subst");
93         }
94         check[0].attr = attr_export_ignore;
95         check[1].attr = attr_export_subst;
96 }
97
98 struct archiver_context {
99         struct archiver_args *args;
100         write_archive_entry_fn_t write_entry;
101 };
102
103 static int write_archive_entry(const unsigned char *sha1, const char *base,
104                 int baselen, const char *filename, unsigned mode, int stage,
105                 void *context)
106 {
107         static struct strbuf path = STRBUF_INIT;
108         struct archiver_context *c = context;
109         struct archiver_args *args = c->args;
110         write_archive_entry_fn_t write_entry = c->write_entry;
111         struct git_attr_check check[2];
112         const char *path_without_prefix;
113         int convert = 0;
114         int err;
115         enum object_type type;
116         unsigned long size;
117         void *buffer;
118
119         strbuf_reset(&path);
120         strbuf_grow(&path, PATH_MAX);
121         strbuf_add(&path, args->base, args->baselen);
122         strbuf_add(&path, base, baselen);
123         strbuf_addstr(&path, filename);
124         path_without_prefix = path.buf + args->baselen;
125
126         setup_archive_check(check);
127         if (!git_checkattr(path_without_prefix, ARRAY_SIZE(check), check)) {
128                 if (ATTR_TRUE(check[0].value))
129                         return 0;
130                 convert = ATTR_TRUE(check[1].value);
131         }
132
133         if (S_ISDIR(mode) || S_ISGITLINK(mode)) {
134                 strbuf_addch(&path, '/');
135                 if (args->verbose)
136                         fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
137                 err = write_entry(args, sha1, path.buf, path.len, mode, NULL, 0);
138                 if (err)
139                         return err;
140                 return (S_ISDIR(mode) ? READ_TREE_RECURSIVE : 0);
141         }
142
143         buffer = sha1_file_to_archive(path_without_prefix, sha1, mode,
144                         &type, &size, convert ? args->commit : NULL);
145         if (!buffer)
146                 return error("cannot read %s", sha1_to_hex(sha1));
147         if (args->verbose)
148                 fprintf(stderr, "%.*s\n", (int)path.len, path.buf);
149         err = write_entry(args, sha1, path.buf, path.len, mode, buffer, size);
150         free(buffer);
151         return err;
152 }
153
154 int write_archive_entries(struct archiver_args *args,
155                 write_archive_entry_fn_t write_entry)
156 {
157         struct archiver_context context;
158         struct unpack_trees_options opts;
159         struct tree_desc t;
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         err = read_tree_recursive(args->tree, "", 0, 0, args->pathspec,
195                                   write_archive_entry, &context);
196         if (err == READ_TREE_RECURSIVE)
197                 err = 0;
198         return err;
199 }
200
201 static const struct archiver *lookup_archiver(const char *name)
202 {
203         int i;
204
205         if (!name)
206                 return NULL;
207
208         for (i = 0; i < ARRAY_SIZE(archivers); i++) {
209                 if (!strcmp(name, archivers[i].name))
210                         return &archivers[i];
211         }
212         return NULL;
213 }
214
215 static int reject_entry(const unsigned char *sha1, const char *base,
216                         int baselen, const char *filename, unsigned mode,
217                         int stage, void *context)
218 {
219         return -1;
220 }
221
222 static int path_exists(struct tree *tree, const char *path)
223 {
224         const char *pathspec[] = { path, NULL };
225
226         if (read_tree_recursive(tree, "", 0, 0, pathspec, reject_entry, NULL))
227                 return 1;
228         return 0;
229 }
230
231 static void parse_pathspec_arg(const char **pathspec,
232                 struct archiver_args *ar_args)
233 {
234         ar_args->pathspec = pathspec = get_pathspec("", pathspec);
235         if (pathspec) {
236                 while (*pathspec) {
237                         if (!path_exists(ar_args->tree, *pathspec))
238                                 die("path not found: %s", *pathspec);
239                         pathspec++;
240                 }
241         }
242 }
243
244 static void parse_treeish_arg(const char **argv,
245                 struct archiver_args *ar_args, const char *prefix)
246 {
247         const char *name = argv[0];
248         const unsigned char *commit_sha1;
249         time_t archive_time;
250         struct tree *tree;
251         const struct commit *commit;
252         unsigned char sha1[20];
253
254         if (get_sha1(name, sha1))
255                 die("Not a valid object name");
256
257         commit = lookup_commit_reference_gently(sha1, 1);
258         if (commit) {
259                 commit_sha1 = commit->object.sha1;
260                 archive_time = commit->date;
261         } else {
262                 commit_sha1 = NULL;
263                 archive_time = time(NULL);
264         }
265
266         tree = parse_tree_indirect(sha1);
267         if (tree == NULL)
268                 die("not a tree object");
269
270         if (prefix) {
271                 unsigned char tree_sha1[20];
272                 unsigned int mode;
273                 int err;
274
275                 err = get_tree_entry(tree->object.sha1, prefix,
276                                      tree_sha1, &mode);
277                 if (err || !S_ISDIR(mode))
278                         die("current working directory is untracked");
279
280                 tree = parse_tree_indirect(tree_sha1);
281         }
282         ar_args->tree = tree;
283         ar_args->commit_sha1 = commit_sha1;
284         ar_args->commit = commit;
285         ar_args->time = archive_time;
286 }
287
288 #define OPT__COMPR(s, v, h, p) \
289         { OPTION_SET_INT, (s), NULL, (v), NULL, (h), \
290           PARSE_OPT_NOARG | PARSE_OPT_NONEG, NULL, (p) }
291 #define OPT__COMPR_HIDDEN(s, v, p) \
292         { OPTION_SET_INT, (s), NULL, (v), NULL, "", \
293           PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
294
295 static int parse_archive_args(int argc, const char **argv,
296                 const struct archiver **ar, struct archiver_args *args)
297 {
298         const char *format = "tar";
299         const char *base = NULL;
300         const char *remote = NULL;
301         const char *exec = NULL;
302         const char *output = NULL;
303         int compression_level = -1;
304         int verbose = 0;
305         int i;
306         int list = 0;
307         int worktree_attributes = 0;
308         struct option opts[] = {
309                 OPT_GROUP(""),
310                 OPT_STRING(0, "format", &format, "fmt", "archive format"),
311                 OPT_STRING(0, "prefix", &base, "prefix",
312                         "prepend prefix to each pathname in the archive"),
313                 OPT_STRING('o', "output", &output, "file",
314                         "write the archive to this file"),
315                 OPT_BOOLEAN(0, "worktree-attributes", &worktree_attributes,
316                         "read .gitattributes in working directory"),
317                 OPT__VERBOSE(&verbose, "report archived files on stderr"),
318                 OPT__COMPR('0', &compression_level, "store only", 0),
319                 OPT__COMPR('1', &compression_level, "compress faster", 1),
320                 OPT__COMPR_HIDDEN('2', &compression_level, 2),
321                 OPT__COMPR_HIDDEN('3', &compression_level, 3),
322                 OPT__COMPR_HIDDEN('4', &compression_level, 4),
323                 OPT__COMPR_HIDDEN('5', &compression_level, 5),
324                 OPT__COMPR_HIDDEN('6', &compression_level, 6),
325                 OPT__COMPR_HIDDEN('7', &compression_level, 7),
326                 OPT__COMPR_HIDDEN('8', &compression_level, 8),
327                 OPT__COMPR('9', &compression_level, "compress better", 9),
328                 OPT_GROUP(""),
329                 OPT_BOOLEAN('l', "list", &list,
330                         "list supported archive formats"),
331                 OPT_GROUP(""),
332                 OPT_STRING(0, "remote", &remote, "repo",
333                         "retrieve the archive from remote repository <repo>"),
334                 OPT_STRING(0, "exec", &exec, "cmd",
335                         "path to the remote git-upload-archive command"),
336                 OPT_END()
337         };
338
339         argc = parse_options(argc, argv, NULL, opts, archive_usage, 0);
340
341         if (remote)
342                 die("Unexpected option --remote");
343         if (exec)
344                 die("Option --exec can only be used together with --remote");
345         if (output)
346                 die("Unexpected option --output");
347
348         if (!base)
349                 base = "";
350
351         if (list) {
352                 for (i = 0; i < ARRAY_SIZE(archivers); i++)
353                         printf("%s\n", archivers[i].name);
354                 exit(0);
355         }
356
357         /* We need at least one parameter -- tree-ish */
358         if (argc < 1)
359                 usage_with_options(archive_usage, opts);
360         *ar = lookup_archiver(format);
361         if (!*ar)
362                 die("Unknown archive format '%s'", format);
363
364         args->compression_level = Z_DEFAULT_COMPRESSION;
365         if (compression_level != -1) {
366                 if ((*ar)->flags & USES_ZLIB_COMPRESSION)
367                         args->compression_level = compression_level;
368                 else {
369                         die("Argument not supported for format '%s': -%d",
370                                         format, compression_level);
371                 }
372         }
373         args->verbose = verbose;
374         args->base = base;
375         args->baselen = strlen(base);
376         args->worktree_attributes = worktree_attributes;
377
378         return argc;
379 }
380
381 int write_archive(int argc, const char **argv, const char *prefix,
382                 int setup_prefix)
383 {
384         const struct archiver *ar = NULL;
385         struct archiver_args args;
386
387         argc = parse_archive_args(argc, argv, &ar, &args);
388         if (setup_prefix && prefix == NULL)
389                 prefix = setup_git_directory();
390
391         parse_treeish_arg(argv, &args, prefix);
392         parse_pathspec_arg(argv + 1, &args);
393
394         git_config(git_default_config, NULL);
395
396         return ar->write_archive(&args);
397 }