Merge branch 'jk/parse-object-type-mismatch'
[git/debian.git] / bundle.c
blob4ef7256aa11e8b43691f238970da6c6bf21cd4ee
1 #include "cache.h"
2 #include "lockfile.h"
3 #include "bundle.h"
4 #include "object-store.h"
5 #include "repository.h"
6 #include "object.h"
7 #include "commit.h"
8 #include "diff.h"
9 #include "revision.h"
10 #include "list-objects.h"
11 #include "run-command.h"
12 #include "refs.h"
13 #include "strvec.h"
14 #include "list-objects-filter-options.h"
16 static const char v2_bundle_signature[] = "# v2 git bundle\n";
17 static const char v3_bundle_signature[] = "# v3 git bundle\n";
18 static struct {
19 int version;
20 const char *signature;
21 } bundle_sigs[] = {
22 { 2, v2_bundle_signature },
23 { 3, v3_bundle_signature },
26 void bundle_header_init(struct bundle_header *header)
28 struct bundle_header blank = BUNDLE_HEADER_INIT;
29 memcpy(header, &blank, sizeof(*header));
32 void bundle_header_release(struct bundle_header *header)
34 string_list_clear(&header->prerequisites, 1);
35 string_list_clear(&header->references, 1);
36 list_objects_filter_release(&header->filter);
39 static int parse_capability(struct bundle_header *header, const char *capability)
41 const char *arg;
42 if (skip_prefix(capability, "object-format=", &arg)) {
43 int algo = hash_algo_by_name(arg);
44 if (algo == GIT_HASH_UNKNOWN)
45 return error(_("unrecognized bundle hash algorithm: %s"), arg);
46 header->hash_algo = &hash_algos[algo];
47 return 0;
49 if (skip_prefix(capability, "filter=", &arg)) {
50 parse_list_objects_filter(&header->filter, arg);
51 return 0;
53 return error(_("unknown capability '%s'"), capability);
56 static int parse_bundle_signature(struct bundle_header *header, const char *line)
58 int i;
60 for (i = 0; i < ARRAY_SIZE(bundle_sigs); i++) {
61 if (!strcmp(line, bundle_sigs[i].signature)) {
62 header->version = bundle_sigs[i].version;
63 return 0;
66 return -1;
69 int read_bundle_header_fd(int fd, struct bundle_header *header,
70 const char *report_path)
72 struct strbuf buf = STRBUF_INIT;
73 int status = 0;
75 /* The bundle header begins with the signature */
76 if (strbuf_getwholeline_fd(&buf, fd, '\n') ||
77 parse_bundle_signature(header, buf.buf)) {
78 if (report_path)
79 error(_("'%s' does not look like a v2 or v3 bundle file"),
80 report_path);
81 status = -1;
82 goto abort;
85 header->hash_algo = the_hash_algo;
87 /* The bundle header ends with an empty line */
88 while (!strbuf_getwholeline_fd(&buf, fd, '\n') &&
89 buf.len && buf.buf[0] != '\n') {
90 struct object_id oid;
91 int is_prereq = 0;
92 const char *p;
94 strbuf_rtrim(&buf);
96 if (header->version == 3 && *buf.buf == '@') {
97 if (parse_capability(header, buf.buf + 1)) {
98 status = -1;
99 break;
101 continue;
104 if (*buf.buf == '-') {
105 is_prereq = 1;
106 strbuf_remove(&buf, 0, 1);
110 * Tip lines have object name, SP, and refname.
111 * Prerequisites have object name that is optionally
112 * followed by SP and subject line.
114 if (parse_oid_hex_algop(buf.buf, &oid, &p, header->hash_algo) ||
115 (*p && !isspace(*p)) ||
116 (!is_prereq && !*p)) {
117 if (report_path)
118 error(_("unrecognized header: %s%s (%d)"),
119 (is_prereq ? "-" : ""), buf.buf, (int)buf.len);
120 status = -1;
121 break;
122 } else {
123 struct object_id *dup = oiddup(&oid);
124 if (is_prereq)
125 string_list_append(&header->prerequisites, "")->util = dup;
126 else
127 string_list_append(&header->references, p + 1)->util = dup;
131 abort:
132 if (status) {
133 close(fd);
134 fd = -1;
136 strbuf_release(&buf);
137 return fd;
140 int read_bundle_header(const char *path, struct bundle_header *header)
142 int fd = open(path, O_RDONLY);
144 if (fd < 0)
145 return error(_("could not open '%s'"), path);
146 return read_bundle_header_fd(fd, header, path);
149 int is_bundle(const char *path, int quiet)
151 struct bundle_header header = BUNDLE_HEADER_INIT;
152 int fd = open(path, O_RDONLY);
154 if (fd < 0)
155 return 0;
156 fd = read_bundle_header_fd(fd, &header, quiet ? NULL : path);
157 if (fd >= 0)
158 close(fd);
159 bundle_header_release(&header);
160 return (fd >= 0);
163 static int list_refs(struct string_list *r, int argc, const char **argv)
165 int i;
167 for (i = 0; i < r->nr; i++) {
168 struct object_id *oid;
169 const char *name;
171 if (argc > 1) {
172 int j;
173 for (j = 1; j < argc; j++)
174 if (!strcmp(r->items[i].string, argv[j]))
175 break;
176 if (j == argc)
177 continue;
180 oid = r->items[i].util;
181 name = r->items[i].string;
182 printf("%s %s\n", oid_to_hex(oid), name);
184 return 0;
187 /* Remember to update object flag allocation in object.h */
188 #define PREREQ_MARK (1u<<16)
190 int verify_bundle(struct repository *r,
191 struct bundle_header *header,
192 enum verify_bundle_flags flags)
195 * Do fast check, then if any prereqs are missing then go line by line
196 * to be verbose about the errors
198 struct string_list *p = &header->prerequisites;
199 struct rev_info revs = REV_INFO_INIT;
200 const char *argv[] = {NULL, "--all", NULL};
201 struct commit *commit;
202 int i, ret = 0, req_nr;
203 const char *message = _("Repository lacks these prerequisite commits:");
205 if (!r || !r->objects || !r->objects->odb)
206 return error(_("need a repository to verify a bundle"));
208 repo_init_revisions(r, &revs, NULL);
209 for (i = 0; i < p->nr; i++) {
210 struct string_list_item *e = p->items + i;
211 const char *name = e->string;
212 struct object_id *oid = e->util;
213 struct object *o = parse_object(r, oid);
214 if (o) {
215 o->flags |= PREREQ_MARK;
216 add_pending_object(&revs, o, name);
217 continue;
219 ret++;
220 if (flags & VERIFY_BUNDLE_QUIET)
221 continue;
222 if (ret == 1)
223 error("%s", message);
224 error("%s %s", oid_to_hex(oid), name);
226 if (revs.pending.nr != p->nr)
227 goto cleanup;
228 req_nr = revs.pending.nr;
229 setup_revisions(2, argv, &revs, NULL);
231 list_objects_filter_copy(&revs.filter, &header->filter);
233 if (prepare_revision_walk(&revs))
234 die(_("revision walk setup failed"));
236 i = req_nr;
237 while (i && (commit = get_revision(&revs)))
238 if (commit->object.flags & PREREQ_MARK)
239 i--;
241 for (i = 0; i < p->nr; i++) {
242 struct string_list_item *e = p->items + i;
243 const char *name = e->string;
244 const struct object_id *oid = e->util;
245 struct object *o = parse_object(r, oid);
246 assert(o); /* otherwise we'd have returned early */
247 if (o->flags & SHOWN)
248 continue;
249 ret++;
250 if (flags & VERIFY_BUNDLE_QUIET)
251 continue;
252 if (ret == 1)
253 error("%s", message);
254 error("%s %s", oid_to_hex(oid), name);
257 if (flags & VERIFY_BUNDLE_VERBOSE) {
258 struct string_list *r;
260 r = &header->references;
261 printf_ln(Q_("The bundle contains this ref:",
262 "The bundle contains these %"PRIuMAX" refs:",
263 r->nr),
264 (uintmax_t)r->nr);
265 list_refs(r, 0, NULL);
267 r = &header->prerequisites;
268 if (!r->nr) {
269 printf_ln(_("The bundle records a complete history."));
270 } else {
271 printf_ln(Q_("The bundle requires this ref:",
272 "The bundle requires these %"PRIuMAX" refs:",
273 r->nr),
274 (uintmax_t)r->nr);
275 list_refs(r, 0, NULL);
278 printf_ln("The bundle uses this hash algorithm: %s",
279 header->hash_algo->name);
280 if (header->filter.choice)
281 printf_ln("The bundle uses this filter: %s",
282 list_objects_filter_spec(&header->filter));
284 cleanup:
285 /* Clean up objects used, as they will be reused. */
286 for (i = 0; i < p->nr; i++) {
287 struct string_list_item *e = p->items + i;
288 struct object_id *oid = e->util;
289 commit = lookup_commit_reference_gently(r, oid, 1);
290 if (commit)
291 clear_commit_marks(commit, ALL_REV_FLAGS | PREREQ_MARK);
293 release_revisions(&revs);
294 return ret;
297 int list_bundle_refs(struct bundle_header *header, int argc, const char **argv)
299 return list_refs(&header->references, argc, argv);
302 static int is_tag_in_date_range(struct object *tag, struct rev_info *revs)
304 unsigned long size;
305 enum object_type type;
306 char *buf = NULL, *line, *lineend;
307 timestamp_t date;
308 int result = 1;
310 if (revs->max_age == -1 && revs->min_age == -1)
311 goto out;
313 buf = read_object_file(&tag->oid, &type, &size);
314 if (!buf)
315 goto out;
316 line = memmem(buf, size, "\ntagger ", 8);
317 if (!line++)
318 goto out;
319 lineend = memchr(line, '\n', buf + size - line);
320 line = memchr(line, '>', lineend ? lineend - line : buf + size - line);
321 if (!line++)
322 goto out;
323 date = parse_timestamp(line, NULL, 10);
324 result = (revs->max_age == -1 || revs->max_age < date) &&
325 (revs->min_age == -1 || revs->min_age > date);
326 out:
327 free(buf);
328 return result;
332 /* Write the pack data to bundle_fd */
333 static int write_pack_data(int bundle_fd, struct rev_info *revs, struct strvec *pack_options)
335 struct child_process pack_objects = CHILD_PROCESS_INIT;
336 int i;
338 strvec_pushl(&pack_objects.args,
339 "pack-objects",
340 "--stdout", "--thin", "--delta-base-offset",
341 NULL);
342 strvec_pushv(&pack_objects.args, pack_options->v);
343 if (revs->filter.choice)
344 strvec_pushf(&pack_objects.args, "--filter=%s",
345 list_objects_filter_spec(&revs->filter));
346 pack_objects.in = -1;
347 pack_objects.out = bundle_fd;
348 pack_objects.git_cmd = 1;
351 * start_command() will close our descriptor if it's >1. Duplicate it
352 * to avoid surprising the caller.
354 if (pack_objects.out > 1) {
355 pack_objects.out = dup(pack_objects.out);
356 if (pack_objects.out < 0) {
357 error_errno(_("unable to dup bundle descriptor"));
358 child_process_clear(&pack_objects);
359 return -1;
363 if (start_command(&pack_objects))
364 return error(_("Could not spawn pack-objects"));
366 for (i = 0; i < revs->pending.nr; i++) {
367 struct object *object = revs->pending.objects[i].item;
368 if (object->flags & UNINTERESTING)
369 write_or_die(pack_objects.in, "^", 1);
370 write_or_die(pack_objects.in, oid_to_hex(&object->oid), the_hash_algo->hexsz);
371 write_or_die(pack_objects.in, "\n", 1);
373 close(pack_objects.in);
374 if (finish_command(&pack_objects))
375 return error(_("pack-objects died"));
376 return 0;
380 * Write out bundle refs based on the tips already
381 * parsed into revs.pending. As a side effect, may
382 * manipulate revs.pending to include additional
383 * necessary objects (like tags).
385 * Returns the number of refs written, or negative
386 * on error.
388 static int write_bundle_refs(int bundle_fd, struct rev_info *revs)
390 int i;
391 int ref_count = 0;
393 for (i = 0; i < revs->pending.nr; i++) {
394 struct object_array_entry *e = revs->pending.objects + i;
395 struct object_id oid;
396 char *ref;
397 const char *display_ref;
398 int flag;
400 if (e->item->flags & UNINTERESTING)
401 continue;
402 if (dwim_ref(e->name, strlen(e->name), &oid, &ref, 0) != 1)
403 goto skip_write_ref;
404 if (read_ref_full(e->name, RESOLVE_REF_READING, &oid, &flag))
405 flag = 0;
406 display_ref = (flag & REF_ISSYMREF) ? e->name : ref;
408 if (e->item->type == OBJ_TAG &&
409 !is_tag_in_date_range(e->item, revs)) {
410 e->item->flags |= UNINTERESTING;
411 goto skip_write_ref;
415 * Make sure the refs we wrote out is correct; --max-count and
416 * other limiting options could have prevented all the tips
417 * from getting output.
419 * Non commit objects such as tags and blobs do not have
420 * this issue as they are not affected by those extra
421 * constraints.
423 if (!(e->item->flags & SHOWN) && e->item->type == OBJ_COMMIT) {
424 warning(_("ref '%s' is excluded by the rev-list options"),
425 e->name);
426 goto skip_write_ref;
429 * If you run "git bundle create bndl v1.0..v2.0", the
430 * name of the positive ref is "v2.0" but that is the
431 * commit that is referenced by the tag, and not the tag
432 * itself.
434 if (!oideq(&oid, &e->item->oid)) {
436 * Is this the positive end of a range expressed
437 * in terms of a tag (e.g. v2.0 from the range
438 * "v1.0..v2.0")?
440 struct commit *one = lookup_commit_reference(revs->repo, &oid);
441 struct object *obj;
443 if (e->item == &(one->object)) {
445 * Need to include e->name as an
446 * independent ref to the pack-objects
447 * input, so that the tag is included
448 * in the output; otherwise we would
449 * end up triggering "empty bundle"
450 * error.
452 obj = parse_object_or_die(&oid, e->name);
453 obj->flags |= SHOWN;
454 add_pending_object(revs, obj, e->name);
456 goto skip_write_ref;
459 ref_count++;
460 write_or_die(bundle_fd, oid_to_hex(&e->item->oid), the_hash_algo->hexsz);
461 write_or_die(bundle_fd, " ", 1);
462 write_or_die(bundle_fd, display_ref, strlen(display_ref));
463 write_or_die(bundle_fd, "\n", 1);
464 skip_write_ref:
465 free(ref);
468 /* end header */
469 write_or_die(bundle_fd, "\n", 1);
470 return ref_count;
473 struct bundle_prerequisites_info {
474 struct object_array *pending;
475 int fd;
478 static void write_bundle_prerequisites(struct commit *commit, void *data)
480 struct bundle_prerequisites_info *bpi = data;
481 struct object *object;
482 struct pretty_print_context ctx = { 0 };
483 struct strbuf buf = STRBUF_INIT;
485 if (!(commit->object.flags & BOUNDARY))
486 return;
487 strbuf_addf(&buf, "-%s ", oid_to_hex(&commit->object.oid));
488 write_or_die(bpi->fd, buf.buf, buf.len);
490 ctx.fmt = CMIT_FMT_ONELINE;
491 ctx.output_encoding = get_log_output_encoding();
492 strbuf_reset(&buf);
493 pretty_print_commit(&ctx, commit, &buf);
494 strbuf_trim(&buf);
496 object = (struct object *)commit;
497 object->flags |= UNINTERESTING;
498 add_object_array_with_path(object, buf.buf, bpi->pending, S_IFINVALID,
499 NULL);
500 strbuf_addch(&buf, '\n');
501 write_or_die(bpi->fd, buf.buf, buf.len);
502 strbuf_release(&buf);
505 int create_bundle(struct repository *r, const char *path,
506 int argc, const char **argv, struct strvec *pack_options, int version)
508 struct lock_file lock = LOCK_INIT;
509 int bundle_fd = -1;
510 int bundle_to_stdout;
511 int ref_count = 0;
512 struct rev_info revs, revs_copy;
513 int min_version = 2;
514 struct bundle_prerequisites_info bpi;
515 int i;
517 /* init revs to list objects for pack-objects later */
518 save_commit_buffer = 0;
519 repo_init_revisions(r, &revs, NULL);
522 * Pre-initialize the '--objects' flag so we can parse a
523 * --filter option successfully.
525 revs.tree_objects = revs.blob_objects = 1;
527 argc = setup_revisions(argc, argv, &revs, NULL);
530 * Reasons to require version 3:
532 * 1. @object-format is required because our hash algorithm is not
533 * SHA1.
534 * 2. @filter is required because we parsed an object filter.
536 if (the_hash_algo != &hash_algos[GIT_HASH_SHA1] || revs.filter.choice)
537 min_version = 3;
539 if (argc > 1) {
540 error(_("unrecognized argument: %s"), argv[1]);
541 goto err;
544 bundle_to_stdout = !strcmp(path, "-");
545 if (bundle_to_stdout)
546 bundle_fd = 1;
547 else
548 bundle_fd = hold_lock_file_for_update(&lock, path,
549 LOCK_DIE_ON_ERROR);
551 if (version == -1)
552 version = min_version;
554 if (version < 2 || version > 3) {
555 die(_("unsupported bundle version %d"), version);
556 } else if (version < min_version) {
557 die(_("cannot write bundle version %d with algorithm %s"), version, the_hash_algo->name);
558 } else if (version == 2) {
559 write_or_die(bundle_fd, v2_bundle_signature, strlen(v2_bundle_signature));
560 } else {
561 const char *capability = "@object-format=";
562 write_or_die(bundle_fd, v3_bundle_signature, strlen(v3_bundle_signature));
563 write_or_die(bundle_fd, capability, strlen(capability));
564 write_or_die(bundle_fd, the_hash_algo->name, strlen(the_hash_algo->name));
565 write_or_die(bundle_fd, "\n", 1);
567 if (revs.filter.choice) {
568 const char *value = expand_list_objects_filter_spec(&revs.filter);
569 capability = "@filter=";
570 write_or_die(bundle_fd, capability, strlen(capability));
571 write_or_die(bundle_fd, value, strlen(value));
572 write_or_die(bundle_fd, "\n", 1);
576 /* save revs.pending in revs_copy for later use */
577 memcpy(&revs_copy, &revs, sizeof(revs));
578 revs_copy.pending.nr = 0;
579 revs_copy.pending.alloc = 0;
580 revs_copy.pending.objects = NULL;
581 for (i = 0; i < revs.pending.nr; i++) {
582 struct object_array_entry *e = revs.pending.objects + i;
583 if (e)
584 add_object_array_with_path(e->item, e->name,
585 &revs_copy.pending,
586 e->mode, e->path);
589 /* write prerequisites */
590 revs.boundary = 1;
591 if (prepare_revision_walk(&revs))
592 die("revision walk setup failed");
593 bpi.fd = bundle_fd;
594 bpi.pending = &revs_copy.pending;
597 * Remove any object walking here. We only care about commits and
598 * tags here. The revs_copy has the right instances of these values.
600 revs.blob_objects = revs.tree_objects = 0;
601 traverse_commit_list(&revs, write_bundle_prerequisites, NULL, &bpi);
602 object_array_remove_duplicates(&revs_copy.pending);
604 /* write bundle refs */
605 ref_count = write_bundle_refs(bundle_fd, &revs_copy);
606 if (!ref_count)
607 die(_("Refusing to create empty bundle."));
608 else if (ref_count < 0)
609 goto err;
611 /* write pack */
612 if (write_pack_data(bundle_fd, &revs_copy, pack_options))
613 goto err;
615 if (!bundle_to_stdout) {
616 if (commit_lock_file(&lock))
617 die_errno(_("cannot create '%s'"), path);
619 return 0;
620 err:
621 rollback_lock_file(&lock);
622 return -1;
625 int unbundle(struct repository *r, struct bundle_header *header,
626 int bundle_fd, struct strvec *extra_index_pack_args,
627 enum verify_bundle_flags flags)
629 struct child_process ip = CHILD_PROCESS_INIT;
630 strvec_pushl(&ip.args, "index-pack", "--fix-thin", "--stdin", NULL);
632 /* If there is a filter, then we need to create the promisor pack. */
633 if (header->filter.choice)
634 strvec_push(&ip.args, "--promisor=from-bundle");
636 if (extra_index_pack_args) {
637 strvec_pushv(&ip.args, extra_index_pack_args->v);
638 strvec_clear(extra_index_pack_args);
641 if (verify_bundle(r, header, flags))
642 return -1;
643 ip.in = bundle_fd;
644 ip.no_stdout = 1;
645 ip.git_cmd = 1;
646 if (run_command(&ip))
647 return error(_("index-pack died"));
648 return 0;