diff: undocument the compaction heuristic knobs for experimentation
[git] / bundle.c
1 #include "cache.h"
2 #include "lockfile.h"
3 #include "bundle.h"
4 #include "object.h"
5 #include "commit.h"
6 #include "diff.h"
7 #include "revision.h"
8 #include "list-objects.h"
9 #include "run-command.h"
10 #include "refs.h"
11 #include "argv-array.h"
12
13 static const char bundle_signature[] = "# v2 git bundle\n";
14
15 static void add_to_ref_list(const unsigned char *sha1, const char *name,
16                 struct ref_list *list)
17 {
18         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
19         hashcpy(list->list[list->nr].sha1, sha1);
20         list->list[list->nr].name = xstrdup(name);
21         list->nr++;
22 }
23
24 static int parse_bundle_header(int fd, struct bundle_header *header,
25                                const char *report_path)
26 {
27         struct strbuf buf = STRBUF_INIT;
28         int status = 0;
29
30         /* The bundle header begins with the signature */
31         if (strbuf_getwholeline_fd(&buf, fd, '\n') ||
32             strcmp(buf.buf, bundle_signature)) {
33                 if (report_path)
34                         error(_("'%s' does not look like a v2 bundle file"),
35                               report_path);
36                 status = -1;
37                 goto abort;
38         }
39
40         /* The bundle header ends with an empty line */
41         while (!strbuf_getwholeline_fd(&buf, fd, '\n') &&
42                buf.len && buf.buf[0] != '\n') {
43                 unsigned char sha1[20];
44                 int is_prereq = 0;
45
46                 if (*buf.buf == '-') {
47                         is_prereq = 1;
48                         strbuf_remove(&buf, 0, 1);
49                 }
50                 strbuf_rtrim(&buf);
51
52                 /*
53                  * Tip lines have object name, SP, and refname.
54                  * Prerequisites have object name that is optionally
55                  * followed by SP and subject line.
56                  */
57                 if (get_sha1_hex(buf.buf, sha1) ||
58                     (buf.len > 40 && !isspace(buf.buf[40])) ||
59                     (!is_prereq && buf.len <= 40)) {
60                         if (report_path)
61                                 error(_("unrecognized header: %s%s (%d)"),
62                                       (is_prereq ? "-" : ""), buf.buf, (int)buf.len);
63                         status = -1;
64                         break;
65                 } else {
66                         if (is_prereq)
67                                 add_to_ref_list(sha1, "", &header->prerequisites);
68                         else
69                                 add_to_ref_list(sha1, buf.buf + 41, &header->references);
70                 }
71         }
72
73  abort:
74         if (status) {
75                 close(fd);
76                 fd = -1;
77         }
78         strbuf_release(&buf);
79         return fd;
80 }
81
82 int read_bundle_header(const char *path, struct bundle_header *header)
83 {
84         int fd = open(path, O_RDONLY);
85
86         if (fd < 0)
87                 return error(_("could not open '%s'"), path);
88         return parse_bundle_header(fd, header, path);
89 }
90
91 int is_bundle(const char *path, int quiet)
92 {
93         struct bundle_header header;
94         int fd = open(path, O_RDONLY);
95
96         if (fd < 0)
97                 return 0;
98         memset(&header, 0, sizeof(header));
99         fd = parse_bundle_header(fd, &header, quiet ? NULL : path);
100         if (fd >= 0)
101                 close(fd);
102         return (fd >= 0);
103 }
104
105 static int list_refs(struct ref_list *r, int argc, const char **argv)
106 {
107         int i;
108
109         for (i = 0; i < r->nr; i++) {
110                 if (argc > 1) {
111                         int j;
112                         for (j = 1; j < argc; j++)
113                                 if (!strcmp(r->list[i].name, argv[j]))
114                                         break;
115                         if (j == argc)
116                                 continue;
117                 }
118                 printf("%s %s\n", sha1_to_hex(r->list[i].sha1),
119                                 r->list[i].name);
120         }
121         return 0;
122 }
123
124 /* Remember to update object flag allocation in object.h */
125 #define PREREQ_MARK (1u<<16)
126
127 int verify_bundle(struct bundle_header *header, int verbose)
128 {
129         /*
130          * Do fast check, then if any prereqs are missing then go line by line
131          * to be verbose about the errors
132          */
133         struct ref_list *p = &header->prerequisites;
134         struct rev_info revs;
135         const char *argv[] = {NULL, "--all", NULL};
136         struct object_array refs;
137         struct commit *commit;
138         int i, ret = 0, req_nr;
139         const char *message = _("Repository lacks these prerequisite commits:");
140
141         init_revisions(&revs, NULL);
142         for (i = 0; i < p->nr; i++) {
143                 struct ref_list_entry *e = p->list + i;
144                 struct object *o = parse_object(e->sha1);
145                 if (o) {
146                         o->flags |= PREREQ_MARK;
147                         add_pending_object(&revs, o, e->name);
148                         continue;
149                 }
150                 if (++ret == 1)
151                         error("%s", message);
152                 error("%s %s", sha1_to_hex(e->sha1), e->name);
153         }
154         if (revs.pending.nr != p->nr)
155                 return ret;
156         req_nr = revs.pending.nr;
157         setup_revisions(2, argv, &revs, NULL);
158
159         refs = revs.pending;
160         revs.leak_pending = 1;
161
162         if (prepare_revision_walk(&revs))
163                 die(_("revision walk setup failed"));
164
165         i = req_nr;
166         while (i && (commit = get_revision(&revs)))
167                 if (commit->object.flags & PREREQ_MARK)
168                         i--;
169
170         for (i = 0; i < req_nr; i++)
171                 if (!(refs.objects[i].item->flags & SHOWN)) {
172                         if (++ret == 1)
173                                 error("%s", message);
174                         error("%s %s", sha1_to_hex(refs.objects[i].item->sha1),
175                                 refs.objects[i].name);
176                 }
177
178         clear_commit_marks_for_object_array(&refs, ALL_REV_FLAGS);
179         free(refs.objects);
180
181         if (verbose) {
182                 struct ref_list *r;
183
184                 r = &header->references;
185                 printf_ln(Q_("The bundle contains this ref:",
186                              "The bundle contains these %d refs:",
187                              r->nr),
188                           r->nr);
189                 list_refs(r, 0, NULL);
190                 r = &header->prerequisites;
191                 if (!r->nr) {
192                         printf_ln(_("The bundle records a complete history."));
193                 } else {
194                         printf_ln(Q_("The bundle requires this ref:",
195                                      "The bundle requires these %d refs:",
196                                      r->nr),
197                                   r->nr);
198                         list_refs(r, 0, NULL);
199                 }
200         }
201         return ret;
202 }
203
204 int list_bundle_refs(struct bundle_header *header, int argc, const char **argv)
205 {
206         return list_refs(&header->references, argc, argv);
207 }
208
209 static int is_tag_in_date_range(struct object *tag, struct rev_info *revs)
210 {
211         unsigned long size;
212         enum object_type type;
213         char *buf = NULL, *line, *lineend;
214         unsigned long date;
215         int result = 1;
216
217         if (revs->max_age == -1 && revs->min_age == -1)
218                 goto out;
219
220         buf = read_sha1_file(tag->sha1, &type, &size);
221         if (!buf)
222                 goto out;
223         line = memmem(buf, size, "\ntagger ", 8);
224         if (!line++)
225                 goto out;
226         lineend = memchr(line, '\n', buf + size - line);
227         line = memchr(line, '>', lineend ? lineend - line : buf + size - line);
228         if (!line++)
229                 goto out;
230         date = strtoul(line, NULL, 10);
231         result = (revs->max_age == -1 || revs->max_age < date) &&
232                 (revs->min_age == -1 || revs->min_age > date);
233 out:
234         free(buf);
235         return result;
236 }
237
238 static int write_pack_data(int bundle_fd, struct lock_file *lock, struct rev_info *revs)
239 {
240         struct child_process pack_objects = CHILD_PROCESS_INIT;
241         int i;
242
243         argv_array_pushl(&pack_objects.args,
244                          "pack-objects", "--all-progress-implied",
245                          "--stdout", "--thin", "--delta-base-offset",
246                          NULL);
247         pack_objects.in = -1;
248         pack_objects.out = bundle_fd;
249         pack_objects.git_cmd = 1;
250         if (start_command(&pack_objects))
251                 return error(_("Could not spawn pack-objects"));
252
253         /*
254          * start_command closed bundle_fd if it was > 1
255          * so set the lock fd to -1 so commit_lock_file()
256          * won't fail trying to close it.
257          */
258         lock->fd = -1;
259
260         for (i = 0; i < revs->pending.nr; i++) {
261                 struct object *object = revs->pending.objects[i].item;
262                 if (object->flags & UNINTERESTING)
263                         write_or_die(pack_objects.in, "^", 1);
264                 write_or_die(pack_objects.in, sha1_to_hex(object->sha1), 40);
265                 write_or_die(pack_objects.in, "\n", 1);
266         }
267         close(pack_objects.in);
268         if (finish_command(&pack_objects))
269                 return error(_("pack-objects died"));
270         return 0;
271 }
272
273 static int compute_and_write_prerequisites(int bundle_fd,
274                                            struct rev_info *revs,
275                                            int argc, const char **argv)
276 {
277         struct child_process rls = CHILD_PROCESS_INIT;
278         struct strbuf buf = STRBUF_INIT;
279         FILE *rls_fout;
280         int i;
281
282         argv_array_pushl(&rls.args,
283                          "rev-list", "--boundary", "--pretty=oneline",
284                          NULL);
285         for (i = 1; i < argc; i++)
286                 argv_array_push(&rls.args, argv[i]);
287         rls.out = -1;
288         rls.git_cmd = 1;
289         if (start_command(&rls))
290                 return -1;
291         rls_fout = xfdopen(rls.out, "r");
292         while (strbuf_getwholeline(&buf, rls_fout, '\n') != EOF) {
293                 unsigned char sha1[20];
294                 if (buf.len > 0 && buf.buf[0] == '-') {
295                         write_or_die(bundle_fd, buf.buf, buf.len);
296                         if (!get_sha1_hex(buf.buf + 1, sha1)) {
297                                 struct object *object = parse_object_or_die(sha1, buf.buf);
298                                 object->flags |= UNINTERESTING;
299                                 add_pending_object(revs, object, buf.buf);
300                         }
301                 } else if (!get_sha1_hex(buf.buf, sha1)) {
302                         struct object *object = parse_object_or_die(sha1, buf.buf);
303                         object->flags |= SHOWN;
304                 }
305         }
306         strbuf_release(&buf);
307         fclose(rls_fout);
308         if (finish_command(&rls))
309                 return error(_("rev-list died"));
310         return 0;
311 }
312
313 /*
314  * Write out bundle refs based on the tips already
315  * parsed into revs.pending. As a side effect, may
316  * manipulate revs.pending to include additional
317  * necessary objects (like tags).
318  *
319  * Returns the number of refs written, or negative
320  * on error.
321  */
322 static int write_bundle_refs(int bundle_fd, struct rev_info *revs)
323 {
324         int i;
325         int ref_count = 0;
326
327         for (i = 0; i < revs->pending.nr; i++) {
328                 struct object_array_entry *e = revs->pending.objects + i;
329                 unsigned char sha1[20];
330                 char *ref;
331                 const char *display_ref;
332                 int flag;
333
334                 if (e->item->flags & UNINTERESTING)
335                         continue;
336                 if (dwim_ref(e->name, strlen(e->name), sha1, &ref) != 1)
337                         goto skip_write_ref;
338                 if (read_ref_full(e->name, RESOLVE_REF_READING, sha1, &flag))
339                         flag = 0;
340                 display_ref = (flag & REF_ISSYMREF) ? e->name : ref;
341
342                 if (e->item->type == OBJ_TAG &&
343                                 !is_tag_in_date_range(e->item, revs)) {
344                         e->item->flags |= UNINTERESTING;
345                         goto skip_write_ref;
346                 }
347
348                 /*
349                  * Make sure the refs we wrote out is correct; --max-count and
350                  * other limiting options could have prevented all the tips
351                  * from getting output.
352                  *
353                  * Non commit objects such as tags and blobs do not have
354                  * this issue as they are not affected by those extra
355                  * constraints.
356                  */
357                 if (!(e->item->flags & SHOWN) && e->item->type == OBJ_COMMIT) {
358                         warning(_("ref '%s' is excluded by the rev-list options"),
359                                 e->name);
360                         goto skip_write_ref;
361                 }
362                 /*
363                  * If you run "git bundle create bndl v1.0..v2.0", the
364                  * name of the positive ref is "v2.0" but that is the
365                  * commit that is referenced by the tag, and not the tag
366                  * itself.
367                  */
368                 if (hashcmp(sha1, e->item->sha1)) {
369                         /*
370                          * Is this the positive end of a range expressed
371                          * in terms of a tag (e.g. v2.0 from the range
372                          * "v1.0..v2.0")?
373                          */
374                         struct commit *one = lookup_commit_reference(sha1);
375                         struct object *obj;
376
377                         if (e->item == &(one->object)) {
378                                 /*
379                                  * Need to include e->name as an
380                                  * independent ref to the pack-objects
381                                  * input, so that the tag is included
382                                  * in the output; otherwise we would
383                                  * end up triggering "empty bundle"
384                                  * error.
385                                  */
386                                 obj = parse_object_or_die(sha1, e->name);
387                                 obj->flags |= SHOWN;
388                                 add_pending_object(revs, obj, e->name);
389                         }
390                         goto skip_write_ref;
391                 }
392
393                 ref_count++;
394                 write_or_die(bundle_fd, sha1_to_hex(e->item->sha1), 40);
395                 write_or_die(bundle_fd, " ", 1);
396                 write_or_die(bundle_fd, display_ref, strlen(display_ref));
397                 write_or_die(bundle_fd, "\n", 1);
398  skip_write_ref:
399                 free(ref);
400         }
401
402         /* end header */
403         write_or_die(bundle_fd, "\n", 1);
404         return ref_count;
405 }
406
407 int create_bundle(struct bundle_header *header, const char *path,
408                   int argc, const char **argv)
409 {
410         static struct lock_file lock;
411         int bundle_fd = -1;
412         int bundle_to_stdout;
413         int ref_count = 0;
414         struct rev_info revs;
415
416         bundle_to_stdout = !strcmp(path, "-");
417         if (bundle_to_stdout)
418                 bundle_fd = 1;
419         else
420                 bundle_fd = hold_lock_file_for_update(&lock, path,
421                                                       LOCK_DIE_ON_ERROR);
422
423         /* write signature */
424         write_or_die(bundle_fd, bundle_signature, strlen(bundle_signature));
425
426         /* init revs to list objects for pack-objects later */
427         save_commit_buffer = 0;
428         init_revisions(&revs, NULL);
429
430         /* write prerequisites */
431         if (compute_and_write_prerequisites(bundle_fd, &revs, argc, argv))
432                 return -1;
433
434         argc = setup_revisions(argc, argv, &revs, NULL);
435
436         if (argc > 1)
437                 return error(_("unrecognized argument: %s"), argv[1]);
438
439         object_array_remove_duplicates(&revs.pending);
440
441         ref_count = write_bundle_refs(bundle_fd, &revs);
442         if (!ref_count)
443                 die(_("Refusing to create empty bundle."));
444         else if (ref_count < 0)
445                 return -1;
446
447         /* write pack */
448         if (write_pack_data(bundle_fd, &lock, &revs))
449                 return -1;
450
451         if (!bundle_to_stdout) {
452                 if (commit_lock_file(&lock))
453                         die_errno(_("cannot create '%s'"), path);
454         }
455         return 0;
456 }
457
458 int unbundle(struct bundle_header *header, int bundle_fd, int flags)
459 {
460         const char *argv_index_pack[] = {"index-pack",
461                                          "--fix-thin", "--stdin", NULL, NULL};
462         struct child_process ip = CHILD_PROCESS_INIT;
463
464         if (flags & BUNDLE_VERBOSE)
465                 argv_index_pack[3] = "-v";
466
467         if (verify_bundle(header, 0))
468                 return -1;
469         ip.argv = argv_index_pack;
470         ip.in = bundle_fd;
471         ip.no_stdout = 1;
472         ip.git_cmd = 1;
473         if (run_command(&ip))
474                 return error(_("index-pack died"));
475         return 0;
476 }