refspec: rename struct refspec to struct refspec_item
[git/git.git] / builtin / fast-export.c
CommitLineData
f2dc849e
JS
1/*
2 * "git fast-export" builtin command
3 *
4 * Copyright (C) 2007 Johannes E. Schindelin
5 */
6#include "builtin.h"
7#include "cache.h"
b2141fc1 8#include "config.h"
fb58c8d5 9#include "refs.h"
ec0cb496 10#include "refspec.h"
f2dc849e
JS
11#include "commit.h"
12#include "object.h"
13#include "tag.h"
14#include "diff.h"
15#include "diffcore.h"
16#include "log-tree.h"
17#include "revision.h"
18#include "decorate.h"
c455c87c 19#include "string-list.h"
f2dc849e
JS
20#include "utf8.h"
21#include "parse-options.h"
6280dfdc 22#include "quote.h"
03e9010c 23#include "remote.h"
a8722750 24#include "blob.h"
f2dc849e
JS
25
26static const char *fast_export_usage[] = {
3b787b96 27 N_("git fast-export [rev-list-opts]"),
f2dc849e
JS
28 NULL
29};
30
31static int progress;
cd16c59b 32static enum { ABORT, VERBATIM, WARN, WARN_STRIP, STRIP } signed_tag_mode = ABORT;
d7a10c31 33static enum { ERROR, DROP, REWRITE } tag_of_filtered_mode = ERROR;
4e46a8d6 34static int fake_missing_tagger;
82670a5c 35static int use_done_feature;
79559f27 36static int no_data;
7f40ab09 37static int full_tree;
1d844ee7 38static struct string_list extra_refs = STRING_LIST_INIT_NODUP;
0ad4a5ff 39static struct refspec_item *refspecs;
03e9010c 40static int refspecs_nr;
a8722750 41static int anonymize;
f2dc849e
JS
42
43static int parse_opt_signed_tag_mode(const struct option *opt,
44 const char *arg, int unset)
45{
46 if (unset || !strcmp(arg, "abort"))
47 signed_tag_mode = ABORT;
ee4bc371
JS
48 else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
49 signed_tag_mode = VERBATIM;
f2dc849e
JS
50 else if (!strcmp(arg, "warn"))
51 signed_tag_mode = WARN;
cd16c59b
JK
52 else if (!strcmp(arg, "warn-strip"))
53 signed_tag_mode = WARN_STRIP;
f2dc849e
JS
54 else if (!strcmp(arg, "strip"))
55 signed_tag_mode = STRIP;
56 else
04a74b6c 57 return error("Unknown signed-tags mode: %s", arg);
f2dc849e
JS
58 return 0;
59}
60
2d8ad469
EN
61static int parse_opt_tag_of_filtered_mode(const struct option *opt,
62 const char *arg, int unset)
63{
64 if (unset || !strcmp(arg, "abort"))
d7a10c31 65 tag_of_filtered_mode = ERROR;
2d8ad469
EN
66 else if (!strcmp(arg, "drop"))
67 tag_of_filtered_mode = DROP;
68 else if (!strcmp(arg, "rewrite"))
69 tag_of_filtered_mode = REWRITE;
70 else
71 return error("Unknown tag-of-filtered mode: %s", arg);
72 return 0;
73}
74
f2dc849e
JS
75static struct decoration idnums;
76static uint32_t last_idnum;
77
78static int has_unshown_parent(struct commit *commit)
79{
80 struct commit_list *parent;
81
82 for (parent = commit->parents; parent; parent = parent->next)
83 if (!(parent->item->object.flags & SHOWN) &&
84 !(parent->item->object.flags & UNINTERESTING))
85 return 1;
86 return 0;
87}
88
a8722750
JK
89struct anonymized_entry {
90 struct hashmap_entry hash;
91 const char *orig;
92 size_t orig_len;
93 const char *anon;
94 size_t anon_len;
95};
96
7663cdc8
SB
97static int anonymized_entry_cmp(const void *unused_cmp_data,
98 const void *va, const void *vb,
99 const void *unused_keydata)
a8722750
JK
100{
101 const struct anonymized_entry *a = va, *b = vb;
102 return a->orig_len != b->orig_len ||
103 memcmp(a->orig, b->orig, a->orig_len);
104}
105
106/*
107 * Basically keep a cache of X->Y so that we can repeatedly replace
108 * the same anonymized string with another. The actual generation
109 * is farmed out to the generate function.
110 */
111static const void *anonymize_mem(struct hashmap *map,
112 void *(*generate)(const void *, size_t *),
113 const void *orig, size_t *len)
114{
115 struct anonymized_entry key, *ret;
116
117 if (!map->cmpfn)
7663cdc8 118 hashmap_init(map, anonymized_entry_cmp, NULL, 0);
a8722750
JK
119
120 hashmap_entry_init(&key, memhash(orig, *len));
121 key.orig = orig;
122 key.orig_len = *len;
123 ret = hashmap_get(map, &key, NULL);
124
125 if (!ret) {
126 ret = xmalloc(sizeof(*ret));
127 hashmap_entry_init(&ret->hash, key.hash.hash);
128 ret->orig = xstrdup(orig);
129 ret->orig_len = *len;
130 ret->anon = generate(orig, len);
131 ret->anon_len = *len;
132 hashmap_put(map, ret);
133 }
134
135 *len = ret->anon_len;
136 return ret->anon;
137}
138
139/*
140 * We anonymize each component of a path individually,
141 * so that paths a/b and a/c will share a common root.
142 * The paths are cached via anonymize_mem so that repeated
143 * lookups for "a" will yield the same value.
144 */
145static void anonymize_path(struct strbuf *out, const char *path,
146 struct hashmap *map,
147 void *(*generate)(const void *, size_t *))
148{
149 while (*path) {
150 const char *end_of_component = strchrnul(path, '/');
151 size_t len = end_of_component - path;
152 const char *c = anonymize_mem(map, generate, path, &len);
153 strbuf_add(out, c, len);
154 path = end_of_component;
155 if (*path)
156 strbuf_addch(out, *path++);
157 }
158}
159
f2dc849e 160/* Since intptr_t is C99, we do not use it here */
df6a7ff7 161static inline uint32_t *mark_to_ptr(uint32_t mark)
f2dc849e 162{
df6a7ff7
PB
163 return ((uint32_t *)NULL) + mark;
164}
165
166static inline uint32_t ptr_to_mark(void * mark)
167{
168 return (uint32_t *)mark - (uint32_t *)NULL;
169}
170
171static inline void mark_object(struct object *object, uint32_t mark)
172{
173 add_decoration(&idnums, object, mark_to_ptr(mark));
174}
175
176static inline void mark_next_object(struct object *object)
177{
178 mark_object(object, ++last_idnum);
f2dc849e
JS
179}
180
181static int get_object_mark(struct object *object)
182{
183 void *decoration = lookup_decoration(&idnums, object);
184 if (!decoration)
185 return 0;
df6a7ff7 186 return ptr_to_mark(decoration);
f2dc849e
JS
187}
188
189static void show_progress(void)
190{
191 static int counter = 0;
192 if (!progress)
193 return;
194 if ((++counter % progress) == 0)
195 printf("progress %d objects\n", counter);
196}
197
a8722750
JK
198/*
199 * Ideally we would want some transformation of the blob data here
200 * that is unreversible, but would still be the same size and have
201 * the same data relationship to other blobs (so that we get the same
202 * delta and packing behavior as the original). But the first and last
203 * requirements there are probably mutually exclusive, so let's take
204 * the easy way out for now, and just generate arbitrary content.
205 *
206 * There's no need to cache this result with anonymize_mem, since
207 * we already handle blob content caching with marks.
208 */
209static char *anonymize_blob(unsigned long *size)
210{
211 static int counter;
212 struct strbuf out = STRBUF_INIT;
213 strbuf_addf(&out, "anonymous blob %d", counter++);
214 *size = out.len;
215 return strbuf_detach(&out, NULL);
216}
217
273f8ee8 218static void export_blob(const struct object_id *oid)
f2dc849e
JS
219{
220 unsigned long size;
221 enum object_type type;
222 char *buf;
223 struct object *object;
30b939c3 224 int eaten;
f2dc849e 225
79559f27
GI
226 if (no_data)
227 return;
228
273f8ee8 229 if (is_null_oid(oid))
f2dc849e
JS
230 return;
231
273f8ee8 232 object = lookup_object(oid->hash);
30b939c3 233 if (object && object->flags & SHOWN)
f2dc849e
JS
234 return;
235
a8722750
JK
236 if (anonymize) {
237 buf = anonymize_blob(&size);
3aca1fc6 238 object = (struct object *)lookup_blob(oid);
a8722750
JK
239 eaten = 0;
240 } else {
b4f5aca4 241 buf = read_object_file(oid, &type, &size);
a8722750 242 if (!buf)
273f8ee8 243 die ("Could not read blob %s", oid_to_hex(oid));
17e65451 244 if (check_object_signature(oid, buf, size, type_name(type)) < 0)
273f8ee8 245 die("sha1 mismatch in blob %s", oid_to_hex(oid));
c251c83d 246 object = parse_object_buffer(oid, type, size, buf, &eaten);
a8722750
JK
247 }
248
30b939c3 249 if (!object)
273f8ee8 250 die("Could not read blob %s", oid_to_hex(oid));
f2dc849e 251
df6a7ff7 252 mark_next_object(object);
f2dc849e 253
6e1c2344 254 printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
b0fe0d72 255 if (size && fwrite(buf, size, 1, stdout) != 1)
273f8ee8 256 die_errno ("Could not write blob '%s'", oid_to_hex(oid));
f2dc849e
JS
257 printf("\n");
258
259 show_progress();
260
261 object->flags |= SHOWN;
30b939c3
JK
262 if (!eaten)
263 free(buf);
f2dc849e
JS
264}
265
060df624
EN
266static int depth_first(const void *a_, const void *b_)
267{
268 const struct diff_filepair *a = *((const struct diff_filepair **)a_);
269 const struct diff_filepair *b = *((const struct diff_filepair **)b_);
270 const char *name_a, *name_b;
271 int len_a, len_b, len;
272 int cmp;
273
274 name_a = a->one ? a->one->path : a->two->path;
275 name_b = b->one ? b->one->path : b->two->path;
276
277 len_a = strlen(name_a);
278 len_b = strlen(name_b);
279 len = (len_a < len_b) ? len_a : len_b;
280
281 /* strcmp will sort 'd' before 'd/e', we want 'd/e' before 'd' */
282 cmp = memcmp(name_a, name_b, len);
283 if (cmp)
284 return cmp;
4ce6fb80
JS
285 cmp = len_b - len_a;
286 if (cmp)
287 return cmp;
288 /*
289 * Move 'R'ename entries last so that all references of the file
290 * appear in the output before it is renamed (e.g., when a file
291 * was copied and renamed in the same commit).
292 */
293 return (a->status == 'R') - (b->status == 'R');
060df624
EN
294}
295
a8722750 296static void print_path_1(const char *path)
6280dfdc
JK
297{
298 int need_quote = quote_c_style(path, NULL, NULL, 0);
299 if (need_quote)
300 quote_c_style(path, NULL, stdout, 0);
ff59f6da
JS
301 else if (strchr(path, ' '))
302 printf("\"%s\"", path);
6280dfdc
JK
303 else
304 printf("%s", path);
305}
306
a8722750
JK
307static void *anonymize_path_component(const void *path, size_t *len)
308{
309 static int counter;
310 struct strbuf out = STRBUF_INIT;
311 strbuf_addf(&out, "path%d", counter++);
312 return strbuf_detach(&out, len);
313}
314
315static void print_path(const char *path)
316{
317 if (!anonymize)
318 print_path_1(path);
319 else {
320 static struct hashmap paths;
321 static struct strbuf anon = STRBUF_INIT;
322
323 anonymize_path(&anon, path, &paths, anonymize_path_component);
324 print_path_1(anon.buf);
325 strbuf_reset(&anon);
326 }
327}
328
273f8ee8 329static void *generate_fake_oid(const void *old, size_t *len)
a8722750
JK
330{
331 static uint32_t counter = 1; /* avoid null sha1 */
273f8ee8 332 unsigned char *out = xcalloc(GIT_SHA1_RAWSZ, 1);
333 put_be32(out + GIT_SHA1_RAWSZ - 4, counter++);
a8722750
JK
334 return out;
335}
336
273f8ee8 337static const unsigned char *anonymize_sha1(const struct object_id *oid)
a8722750
JK
338{
339 static struct hashmap sha1s;
273f8ee8 340 size_t len = GIT_SHA1_RAWSZ;
341 return anonymize_mem(&sha1s, generate_fake_oid, oid, &len);
a8722750
JK
342}
343
f2dc849e
JS
344static void show_filemodify(struct diff_queue_struct *q,
345 struct diff_options *options, void *data)
346{
347 int i;
b3e8ca89 348 struct string_list *changed = data;
060df624
EN
349
350 /*
351 * Handle files below a directory first, in case they are all deleted
352 * and the directory changes to a file or symlink.
353 */
9ed0d8d6 354 QSORT(q->queue, q->nr, depth_first);
060df624 355
f2dc849e 356 for (i = 0; i < q->nr; i++) {
ae7c5dce 357 struct diff_filespec *ospec = q->queue[i]->one;
f2dc849e 358 struct diff_filespec *spec = q->queue[i]->two;
ae7c5dce
AG
359
360 switch (q->queue[i]->status) {
361 case DIFF_STATUS_DELETED:
6280dfdc
JK
362 printf("D ");
363 print_path(spec->path);
b3e8ca89 364 string_list_insert(changed, spec->path);
6280dfdc 365 putchar('\n');
ae7c5dce
AG
366 break;
367
368 case DIFF_STATUS_COPIED:
369 case DIFF_STATUS_RENAMED:
b3e8ca89
JT
370 /*
371 * If a change in the file corresponding to ospec->path
372 * has been observed, we cannot trust its contents
373 * because the diff is calculated based on the prior
374 * contents, not the current contents. So, declare a
375 * copy or rename only if there was no change observed.
376 */
377 if (!string_list_has_string(changed, ospec->path)) {
378 printf("%c ", q->queue[i]->status);
379 print_path(ospec->path);
380 putchar(' ');
381 print_path(spec->path);
382 string_list_insert(changed, spec->path);
383 putchar('\n');
384
385 if (!oidcmp(&ospec->oid, &spec->oid) &&
386 ospec->mode == spec->mode)
387 break;
388 }
ae7c5dce
AG
389 /* fallthrough */
390
391 case DIFF_STATUS_TYPE_CHANGED:
392 case DIFF_STATUS_MODIFIED:
393 case DIFF_STATUS_ADDED:
03db4525
AG
394 /*
395 * Links refer to objects in another repositories;
396 * output the SHA-1 verbatim.
397 */
79559f27 398 if (no_data || S_ISGITLINK(spec->mode))
6280dfdc 399 printf("M %06o %s ", spec->mode,
a8722750 400 sha1_to_hex(anonymize ?
273f8ee8 401 anonymize_sha1(&spec->oid) :
a0d12c44 402 spec->oid.hash));
03db4525 403 else {
a0d12c44 404 struct object *object = lookup_object(spec->oid.hash);
6280dfdc
JK
405 printf("M %06o :%d ", spec->mode,
406 get_object_mark(object));
03db4525 407 }
6280dfdc 408 print_path(spec->path);
b3e8ca89 409 string_list_insert(changed, spec->path);
6280dfdc 410 putchar('\n');
ae7c5dce
AG
411 break;
412
413 default:
414 die("Unexpected comparison status '%c' for %s, %s",
415 q->queue[i]->status,
416 ospec->path ? ospec->path : "none",
417 spec->path ? spec->path : "none");
f2dc849e
JS
418 }
419 }
420}
421
422static const char *find_encoding(const char *begin, const char *end)
423{
424 const char *needle = "\nencoding ";
425 char *bol, *eol;
426
427 bol = memmem(begin, end ? end - begin : strlen(begin),
428 needle, strlen(needle));
429 if (!bol)
430 return git_commit_encoding;
431 bol += strlen(needle);
432 eol = strchrnul(bol, '\n');
433 *eol = '\0';
434 return bol;
435}
436
a8722750
JK
437static void *anonymize_ref_component(const void *old, size_t *len)
438{
439 static int counter;
440 struct strbuf out = STRBUF_INIT;
441 strbuf_addf(&out, "ref%d", counter++);
442 return strbuf_detach(&out, len);
443}
444
445static const char *anonymize_refname(const char *refname)
446{
447 /*
448 * If any of these prefixes is found, we will leave it intact
449 * so that tags remain tags and so forth.
450 */
451 static const char *prefixes[] = {
452 "refs/heads/",
453 "refs/tags/",
454 "refs/remotes/",
455 "refs/"
456 };
457 static struct hashmap refs;
458 static struct strbuf anon = STRBUF_INIT;
459 int i;
460
461 /*
462 * We also leave "master" as a special case, since it does not reveal
463 * anything interesting.
464 */
465 if (!strcmp(refname, "refs/heads/master"))
466 return refname;
467
468 strbuf_reset(&anon);
469 for (i = 0; i < ARRAY_SIZE(prefixes); i++) {
470 if (skip_prefix(refname, prefixes[i], &refname)) {
471 strbuf_addstr(&anon, prefixes[i]);
472 break;
473 }
474 }
475
476 anonymize_path(&anon, refname, &refs, anonymize_ref_component);
477 return anon.buf;
478}
479
480/*
481 * We do not even bother to cache commit messages, as they are unlikely
482 * to be repeated verbatim, and it is not that interesting when they are.
483 */
484static char *anonymize_commit_message(const char *old)
485{
486 static int counter;
487 return xstrfmt("subject %d\n\nbody\n", counter++);
488}
489
490static struct hashmap idents;
491static void *anonymize_ident(const void *old, size_t *len)
492{
493 static int counter;
494 struct strbuf out = STRBUF_INIT;
495 strbuf_addf(&out, "User %d <user%d@example.com>", counter, counter);
496 counter++;
497 return strbuf_detach(&out, len);
498}
499
500/*
501 * Our strategy here is to anonymize the names and email addresses,
502 * but keep timestamps intact, as they influence things like traversal
503 * order (and by themselves should not be too revealing).
504 */
505static void anonymize_ident_line(const char **beg, const char **end)
506{
507 static struct strbuf buffers[] = { STRBUF_INIT, STRBUF_INIT };
508 static unsigned which_buffer;
509
510 struct strbuf *out;
511 struct ident_split split;
512 const char *end_of_header;
513
514 out = &buffers[which_buffer++];
515 which_buffer %= ARRAY_SIZE(buffers);
516 strbuf_reset(out);
517
518 /* skip "committer", "author", "tagger", etc */
519 end_of_header = strchr(*beg, ' ');
520 if (!end_of_header)
521 die("BUG: malformed line fed to anonymize_ident_line: %.*s",
522 (int)(*end - *beg), *beg);
523 end_of_header++;
524 strbuf_add(out, *beg, end_of_header - *beg);
525
526 if (!split_ident_line(&split, end_of_header, *end - end_of_header) &&
527 split.date_begin) {
528 const char *ident;
529 size_t len;
530
531 len = split.mail_end - split.name_begin;
532 ident = anonymize_mem(&idents, anonymize_ident,
533 split.name_begin, &len);
534 strbuf_add(out, ident, len);
535 strbuf_addch(out, ' ');
536 strbuf_add(out, split.date_begin, split.tz_end - split.date_begin);
537 } else {
538 strbuf_addstr(out, "Malformed Ident <malformed@example.com> 0 -0000");
539 }
540
541 *beg = out->buf;
542 *end = out->buf + out->len;
543}
544
b3e8ca89
JT
545static void handle_commit(struct commit *commit, struct rev_info *rev,
546 struct string_list *paths_of_changed_objects)
f2dc849e
JS
547{
548 int saved_output_format = rev->diffopt.output_format;
bc6b8fc1 549 const char *commit_buffer;
f2dc849e
JS
550 const char *author, *author_end, *committer, *committer_end;
551 const char *encoding, *message;
552 char *reencoded = NULL;
553 struct commit_list *p;
a8722750 554 const char *refname;
f2dc849e
JS
555 int i;
556
557 rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
558
683ff884 559 parse_commit_or_die(commit);
8597ea3a 560 commit_buffer = get_commit_buffer(commit, NULL);
bc6b8fc1 561 author = strstr(commit_buffer, "\nauthor ");
f2dc849e
JS
562 if (!author)
563 die ("Could not find author in commit %s",
f2fd0760 564 oid_to_hex(&commit->object.oid));
f2dc849e
JS
565 author++;
566 author_end = strchrnul(author, '\n');
567 committer = strstr(author_end, "\ncommitter ");
568 if (!committer)
569 die ("Could not find committer in commit %s",
f2fd0760 570 oid_to_hex(&commit->object.oid));
f2dc849e
JS
571 committer++;
572 committer_end = strchrnul(committer, '\n');
573 message = strstr(committer_end, "\n\n");
574 encoding = find_encoding(committer_end, message);
575 if (message)
576 message += 2;
577
ebeec7db 578 if (commit->parents &&
4087a02e
EN
579 get_object_mark(&commit->parents->item->object) != 0 &&
580 !full_tree) {
683ff884 581 parse_commit_or_die(commit->parents->item);
66f414f8
BW
582 diff_tree_oid(&commit->parents->item->tree->object.oid,
583 &commit->tree->object.oid, "", &rev->diffopt);
f2dc849e
JS
584 }
585 else
7b8dea0c
BW
586 diff_root_tree_oid(&commit->tree->object.oid,
587 "", &rev->diffopt);
f2dc849e 588
03db4525 589 /* Export the referenced blobs, and remember the marks. */
f2dc849e 590 for (i = 0; i < diff_queued_diff.nr; i++)
03db4525 591 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
273f8ee8 592 export_blob(&diff_queued_diff.queue[i]->two->oid);
f2dc849e 593
a8722750
JK
594 refname = commit->util;
595 if (anonymize) {
596 refname = anonymize_refname(refname);
597 anonymize_ident_line(&committer, &committer_end);
598 anonymize_ident_line(&author, &author_end);
599 }
600
df6a7ff7 601 mark_next_object(&commit->object);
a8722750
JK
602 if (anonymize)
603 reencoded = anonymize_commit_message(message);
604 else if (!is_encoding_utf8(encoding))
f2dc849e 605 reencoded = reencode_string(message, "UTF-8", encoding);
d8933f01 606 if (!commit->parents)
a8722750 607 printf("reset %s\n", refname);
6e1c2344 608 printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
a8722750 609 refname, last_idnum,
f2dc849e
JS
610 (int)(author_end - author), author,
611 (int)(committer_end - committer), committer,
612 (unsigned)(reencoded
613 ? strlen(reencoded) : message
614 ? strlen(message) : 0),
615 reencoded ? reencoded : message ? message : "");
8e0f7003 616 free(reencoded);
bc6b8fc1 617 unuse_commit_buffer(commit, commit_buffer);
f2dc849e
JS
618
619 for (i = 0, p = commit->parents; p; p = p->next) {
620 int mark = get_object_mark(&p->item->object);
621 if (!mark)
622 continue;
623 if (i == 0)
624 printf("from :%d\n", mark);
f2dc849e 625 else
5070b49e 626 printf("merge :%d\n", mark);
f2dc849e
JS
627 i++;
628 }
f2dc849e 629
4087a02e
EN
630 if (full_tree)
631 printf("deleteall\n");
f2dc849e 632 log_tree_diff_flush(rev);
b3e8ca89 633 string_list_clear(paths_of_changed_objects, 0);
f2dc849e
JS
634 rev->diffopt.output_format = saved_output_format;
635
636 printf("\n");
637
638 show_progress();
639}
640
a8722750
JK
641static void *anonymize_tag(const void *old, size_t *len)
642{
643 static int counter;
644 struct strbuf out = STRBUF_INIT;
645 strbuf_addf(&out, "tag message %d", counter++);
646 return strbuf_detach(&out, len);
647}
648
b3e8ca89
JT
649static void handle_tail(struct object_array *commits, struct rev_info *revs,
650 struct string_list *paths_of_changed_objects)
f2dc849e
JS
651{
652 struct commit *commit;
653 while (commits->nr) {
71992039 654 commit = (struct commit *)object_array_pop(commits);
be011bbe
655 if (has_unshown_parent(commit)) {
656 /* Queue again, to be handled later */
657 add_object_array(&commit->object, NULL, commits);
f2dc849e 658 return;
be011bbe 659 }
b3e8ca89 660 handle_commit(commit, revs, paths_of_changed_objects);
f2dc849e
JS
661 }
662}
663
664static void handle_tag(const char *name, struct tag *tag)
665{
666 unsigned long size;
667 enum object_type type;
668 char *buf;
669 const char *tagger, *tagger_end, *message;
670 size_t message_size = 0;
02c48cd6 671 struct object *tagged;
2d8ad469
EN
672 int tagged_mark;
673 struct commit *p;
02c48cd6 674
98e023de 675 /* Trees have no identifier in fast-export output, thus we have no way
02c48cd6
EN
676 * to output tags of trees, tags of tags of trees, etc. Simply omit
677 * such tags.
678 */
679 tagged = tag->tagged;
680 while (tagged->type == OBJ_TAG) {
681 tagged = ((struct tag *)tagged)->tagged;
682 }
683 if (tagged->type == OBJ_TREE) {
684 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
f2fd0760 685 oid_to_hex(&tag->object.oid));
02c48cd6
EN
686 return;
687 }
f2dc849e 688
b4f5aca4 689 buf = read_object_file(&tag->object.oid, &type, &size);
f2dc849e 690 if (!buf)
f2fd0760 691 die ("Could not read tag %s", oid_to_hex(&tag->object.oid));
f2dc849e
JS
692 message = memmem(buf, size, "\n\n", 2);
693 if (message) {
694 message += 2;
695 message_size = strlen(message);
696 }
697 tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
4e46a8d6
JS
698 if (!tagger) {
699 if (fake_missing_tagger)
700 tagger = "tagger Unspecified Tagger "
701 "<unspecified-tagger> 0 +0000";
702 else
703 tagger = "";
704 tagger_end = tagger + strlen(tagger);
705 } else {
706 tagger++;
707 tagger_end = strchrnul(tagger, '\n');
a8722750
JK
708 if (anonymize)
709 anonymize_ident_line(&tagger, &tagger_end);
710 }
711
712 if (anonymize) {
713 name = anonymize_refname(name);
714 if (message) {
715 static struct hashmap tags;
716 message = anonymize_mem(&tags, anonymize_tag,
717 message, &message_size);
718 }
4e46a8d6 719 }
f2dc849e
JS
720
721 /* handle signed tags */
722 if (message) {
723 const char *signature = strstr(message,
724 "\n-----BEGIN PGP SIGNATURE-----\n");
725 if (signature)
726 switch(signed_tag_mode) {
727 case ABORT:
728 die ("Encountered signed tag %s; use "
04a74b6c 729 "--signed-tags=<mode> to handle it.",
f2fd0760 730 oid_to_hex(&tag->object.oid));
f2dc849e
JS
731 case WARN:
732 warning ("Exporting signed tag %s",
f2fd0760 733 oid_to_hex(&tag->object.oid));
f2dc849e 734 /* fallthru */
ee4bc371 735 case VERBATIM:
f2dc849e 736 break;
cd16c59b
JK
737 case WARN_STRIP:
738 warning ("Stripping signature from tag %s",
f2fd0760 739 oid_to_hex(&tag->object.oid));
cd16c59b 740 /* fallthru */
f2dc849e
JS
741 case STRIP:
742 message_size = signature + 1 - message;
743 break;
744 }
745 }
746
2d8ad469
EN
747 /* handle tag->tagged having been filtered out due to paths specified */
748 tagged = tag->tagged;
749 tagged_mark = get_object_mark(tagged);
750 if (!tagged_mark) {
751 switch(tag_of_filtered_mode) {
752 case ABORT:
753 die ("Tag %s tags unexported object; use "
754 "--tag-of-filtered-object=<mode> to handle it.",
f2fd0760 755 oid_to_hex(&tag->object.oid));
2d8ad469
EN
756 case DROP:
757 /* Ignore this tag altogether */
1efb1e9a 758 free(buf);
2d8ad469
EN
759 return;
760 case REWRITE:
761 if (tagged->type != OBJ_COMMIT) {
762 die ("Tag %s tags unexported %s!",
f2fd0760 763 oid_to_hex(&tag->object.oid),
debca9d2 764 type_name(tagged->type));
2d8ad469
EN
765 }
766 p = (struct commit *)tagged;
767 for (;;) {
768 if (p->parents && p->parents->next)
769 break;
770 if (p->object.flags & UNINTERESTING)
771 break;
772 if (!(p->object.flags & TREESAME))
773 break;
774 if (!p->parents)
775 die ("Can't find replacement commit for tag %s\n",
f2fd0760 776 oid_to_hex(&tag->object.oid));
2d8ad469
EN
777 p = p->parents->item;
778 }
779 tagged_mark = get_object_mark(&p->object);
780 }
781 }
782
59556548 783 if (starts_with(name, "refs/tags/"))
f2dc849e 784 name += 10;
4e46a8d6 785 printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
2d8ad469 786 name, tagged_mark,
f2dc849e 787 (int)(tagger_end - tagger), tagger,
4e46a8d6 788 tagger == tagger_end ? "" : "\n",
f2dc849e 789 (int)message_size, (int)message_size, message ? message : "");
1efb1e9a 790 free(buf);
f2dc849e
JS
791}
792
3e9b9cb1
FC
793static struct commit *get_commit(struct rev_cmdline_entry *e, char *full_name)
794{
795 switch (e->item->type) {
796 case OBJ_COMMIT:
797 return (struct commit *)e->item;
798 case OBJ_TAG: {
799 struct tag *tag = (struct tag *)e->item;
800
801 /* handle nested tags */
802 while (tag && tag->object.type == OBJ_TAG) {
c251c83d 803 parse_object(&tag->object.oid);
3e9b9cb1
FC
804 string_list_append(&extra_refs, full_name)->util = tag;
805 tag = (struct tag *)tag->tagged;
806 }
807 if (!tag)
808 die("Tag %s points nowhere?", e->name);
809 return (struct commit *)tag;
810 break;
811 }
812 default:
813 return NULL;
814 }
815}
816
1d844ee7 817static void get_tags_and_duplicates(struct rev_cmdline_info *info)
f2dc849e 818{
f2dc849e
JS
819 int i;
820
49266e8a
FC
821 for (i = 0; i < info->nr; i++) {
822 struct rev_cmdline_entry *e = info->rev + i;
273f8ee8 823 struct object_id oid;
2d242de4 824 struct commit *commit;
f2dc849e
JS
825 char *full_name;
826
49266e8a
FC
827 if (e->flags & UNINTERESTING)
828 continue;
829
cca5fa64 830 if (dwim_ref(e->name, strlen(e->name), &oid, &full_name) != 1)
f2dc849e
JS
831 continue;
832
03e9010c
FC
833 if (refspecs) {
834 char *private;
835 private = apply_refspecs(refspecs, refspecs_nr, full_name);
836 if (private) {
837 free(full_name);
838 full_name = private;
839 }
840 }
841
3e9b9cb1
FC
842 commit = get_commit(e, full_name);
843 if (!commit) {
2d07f6d4
EFL
844 warning("%s: Unexpected object of type %s, skipping.",
845 e->name,
debca9d2 846 type_name(e->item->type));
2d07f6d4 847 continue;
f2dc849e 848 }
f28e7c90 849
3e9b9cb1
FC
850 switch(commit->object.type) {
851 case OBJ_COMMIT:
852 break;
853 case OBJ_BLOB:
273f8ee8 854 export_blob(&commit->object.oid);
3e9b9cb1
FC
855 continue;
856 default: /* OBJ_TAG (nested tags) is already handled */
857 warning("Tag points to object of unexpected type %s, skipping.",
debca9d2 858 type_name(commit->object.type));
3e9b9cb1
FC
859 continue;
860 }
861
f28e7c90
FC
862 /*
863 * This ref will not be updated through a commit, lets make
864 * sure it gets properly updated eventually.
865 */
866 if (commit->util || commit->object.flags & SHOWN)
1d844ee7 867 string_list_append(&extra_refs, full_name)->util = commit;
f28e7c90 868 if (!commit->util)
f2dc849e
JS
869 commit->util = full_name;
870 }
871}
872
1d844ee7 873static void handle_tags_and_duplicates(void)
f2dc849e
JS
874{
875 struct commit *commit;
876 int i;
877
1d844ee7
FC
878 for (i = extra_refs.nr - 1; i >= 0; i--) {
879 const char *name = extra_refs.items[i].string;
880 struct object *object = extra_refs.items[i].util;
f2dc849e
JS
881 switch (object->type) {
882 case OBJ_TAG:
883 handle_tag(name, (struct tag *)object);
884 break;
885 case OBJ_COMMIT:
a8722750
JK
886 if (anonymize)
887 name = anonymize_refname(name);
f2dc849e
JS
888 /* create refs pointing to already seen commits */
889 commit = (struct commit *)object;
890 printf("reset %s\nfrom :%d\n\n", name,
891 get_object_mark(&commit->object));
892 show_progress();
893 break;
894 }
895 }
896}
897
df6a7ff7
PB
898static void export_marks(char *file)
899{
900 unsigned int i;
901 uint32_t mark;
ddd3e312 902 struct decoration_entry *deco = idnums.entries;
df6a7ff7 903 FILE *f;
96d69b55 904 int e = 0;
df6a7ff7 905
ea56518d 906 f = fopen_for_writing(file);
df6a7ff7 907 if (!f)
bb6ad28c 908 die_errno("Unable to open marks file %s for writing.", file);
df6a7ff7 909
69913575
JH
910 for (i = 0; i < idnums.size; i++) {
911 if (deco->base && deco->base->type == 1) {
df6a7ff7 912 mark = ptr_to_mark(deco->decoration);
96d69b55 913 if (fprintf(f, ":%"PRIu32" %s\n", mark,
f2fd0760 914 oid_to_hex(&deco->base->oid)) < 0) {
96d69b55
MA
915 e = 1;
916 break;
917 }
df6a7ff7 918 }
69913575 919 deco++;
df6a7ff7
PB
920 }
921
96d69b55
MA
922 e |= ferror(f);
923 e |= fclose(f);
924 if (e)
df6a7ff7
PB
925 error("Unable to write marks file %s.", file);
926}
927
69913575 928static void import_marks(char *input_file)
df6a7ff7
PB
929{
930 char line[512];
23a9e071 931 FILE *f = xfopen(input_file, "r");
df6a7ff7
PB
932
933 while (fgets(line, sizeof(line), f)) {
934 uint32_t mark;
935 char *line_end, *mark_end;
273f8ee8 936 struct object_id oid;
df6a7ff7 937 struct object *object;
47bd9bf8 938 struct commit *commit;
e6812cfa 939 enum object_type type;
df6a7ff7
PB
940
941 line_end = strchr(line, '\n');
942 if (line[0] != ':' || !line_end)
943 die("corrupt mark line: %s", line);
69913575 944 *line_end = '\0';
df6a7ff7
PB
945
946 mark = strtoumax(line + 1, &mark_end, 10);
947 if (!mark || mark_end == line + 1
273f8ee8 948 || *mark_end != ' ' || get_oid_hex(mark_end + 1, &oid))
df6a7ff7
PB
949 die("corrupt mark line: %s", line);
950
c4458ecd
AP
951 if (last_idnum < mark)
952 last_idnum = mark;
953
abef9020 954 type = oid_object_info(&oid, NULL);
e6812cfa 955 if (type < 0)
273f8ee8 956 die("object not found: %s", oid_to_hex(&oid));
e6812cfa
FC
957
958 if (type != OBJ_COMMIT)
959 /* only commits */
c4458ecd 960 continue;
df6a7ff7 961
bc83266a 962 commit = lookup_commit(&oid);
47bd9bf8 963 if (!commit)
273f8ee8 964 die("not a commit? can't happen: %s", oid_to_hex(&oid));
47bd9bf8
FC
965
966 object = &commit->object;
e6812cfa 967
df6a7ff7 968 if (object->flags & SHOWN)
273f8ee8 969 error("Object %s already has a mark", oid_to_hex(&oid));
df6a7ff7
PB
970
971 mark_object(object, mark);
df6a7ff7
PB
972
973 object->flags |= SHOWN;
974 }
975 fclose(f);
976}
977
60ed2643
FC
978static void handle_deletes(void)
979{
980 int i;
981 for (i = 0; i < refspecs_nr; i++) {
0ad4a5ff 982 struct refspec_item *refspec = &refspecs[i];
60ed2643
FC
983 if (*refspec->src)
984 continue;
985
986 printf("reset %s\nfrom %s\n\n",
987 refspec->dst, sha1_to_hex(null_sha1));
988 }
989}
990
f2dc849e
JS
991int cmd_fast_export(int argc, const char **argv, const char *prefix)
992{
993 struct rev_info revs;
3cd47459 994 struct object_array commits = OBJECT_ARRAY_INIT;
f2dc849e 995 struct commit *commit;
df6a7ff7 996 char *export_filename = NULL, *import_filename = NULL;
c4458ecd 997 uint32_t lastimportid;
03e9010c 998 struct string_list refspecs_list = STRING_LIST_INIT_NODUP;
b3e8ca89 999 struct string_list paths_of_changed_objects = STRING_LIST_INIT_DUP;
f2dc849e
JS
1000 struct option options[] = {
1001 OPT_INTEGER(0, "progress", &progress,
3b787b96
NTND
1002 N_("show progress after <n> objects")),
1003 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, N_("mode"),
1004 N_("select handling of signed tags"),
f2dc849e 1005 parse_opt_signed_tag_mode),
3b787b96
NTND
1006 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, N_("mode"),
1007 N_("select handling of tags that tag filtered objects"),
2d8ad469 1008 parse_opt_tag_of_filtered_mode),
3b787b96
NTND
1009 OPT_STRING(0, "export-marks", &export_filename, N_("file"),
1010 N_("Dump marks to this file")),
1011 OPT_STRING(0, "import-marks", &import_filename, N_("file"),
1012 N_("Import marks from this file")),
d5d09d47
SB
1013 OPT_BOOL(0, "fake-missing-tagger", &fake_missing_tagger,
1014 N_("Fake a tagger when tags lack one")),
1015 OPT_BOOL(0, "full-tree", &full_tree,
1016 N_("Output full tree for each commit")),
1017 OPT_BOOL(0, "use-done-feature", &use_done_feature,
3b787b96
NTND
1018 N_("Use the done feature to terminate the stream")),
1019 OPT_BOOL(0, "no-data", &no_data, N_("Skip output of blob data")),
03e9010c
FC
1020 OPT_STRING_LIST(0, "refspec", &refspecs_list, N_("refspec"),
1021 N_("Apply refspec to exported refs")),
a8722750 1022 OPT_BOOL(0, "anonymize", &anonymize, N_("anonymize output")),
f2dc849e
JS
1023 OPT_END()
1024 };
1025
dcfdbdf0
MV
1026 if (argc == 1)
1027 usage_with_options (fast_export_usage, options);
1028
f2dc849e 1029 /* we handle encodings */
ef90d6d4 1030 git_config(git_default_config, NULL);
f2dc849e
JS
1031
1032 init_revisions(&revs, prefix);
668f3aa7 1033 revs.topo_order = 1;
2374502c 1034 revs.show_source = 1;
32164131 1035 revs.rewrite_parents = 1;
8b2f86a7
FC
1036 argc = parse_options(argc, argv, prefix, options, fast_export_usage,
1037 PARSE_OPT_KEEP_ARGV0 | PARSE_OPT_KEEP_UNKNOWN);
f2dc849e 1038 argc = setup_revisions(argc, argv, &revs, NULL);
f2dc849e
JS
1039 if (argc > 1)
1040 usage_with_options (fast_export_usage, options);
1041
03e9010c
FC
1042 if (refspecs_list.nr) {
1043 const char **refspecs_str;
1044 int i;
1045
b32fa95f 1046 ALLOC_ARRAY(refspecs_str, refspecs_list.nr);
03e9010c
FC
1047 for (i = 0; i < refspecs_list.nr; i++)
1048 refspecs_str[i] = refspecs_list.items[i].string;
1049
1050 refspecs_nr = refspecs_list.nr;
1051 refspecs = parse_fetch_refspec(refspecs_nr, refspecs_str);
1052
1053 string_list_clear(&refspecs_list, 1);
1054 free(refspecs_str);
1055 }
1056
82670a5c
SR
1057 if (use_done_feature)
1058 printf("feature done\n");
1059
df6a7ff7
PB
1060 if (import_filename)
1061 import_marks(import_filename);
c4458ecd 1062 lastimportid = last_idnum;
df6a7ff7 1063
afe069d1 1064 if (import_filename && revs.prune_data.nr)
4087a02e
EN
1065 full_tree = 1;
1066
1d844ee7 1067 get_tags_and_duplicates(&revs.cmdline);
f2dc849e 1068
3d51e1b5
MK
1069 if (prepare_revision_walk(&revs))
1070 die("revision walk setup failed");
f2dc849e 1071 revs.diffopt.format_callback = show_filemodify;
b3e8ca89 1072 revs.diffopt.format_callback_data = &paths_of_changed_objects;
0d1e0e78 1073 revs.diffopt.flags.recursive = 1;
f2dc849e
JS
1074 while ((commit = get_revision(&revs))) {
1075 if (has_unshown_parent(commit)) {
f2dc849e 1076 add_object_array(&commit->object, NULL, &commits);
f2dc849e
JS
1077 }
1078 else {
b3e8ca89
JT
1079 handle_commit(commit, &revs, &paths_of_changed_objects);
1080 handle_tail(&commits, &revs, &paths_of_changed_objects);
f2dc849e
JS
1081 }
1082 }
1083
1d844ee7 1084 handle_tags_and_duplicates();
60ed2643 1085 handle_deletes();
f2dc849e 1086
c4458ecd 1087 if (export_filename && lastimportid != last_idnum)
df6a7ff7
PB
1088 export_marks(export_filename);
1089
82670a5c
SR
1090 if (use_done_feature)
1091 printf("done\n");
1092
03e9010c
FC
1093 free_refspec(refspecs_nr, refspecs);
1094
f2dc849e
JS
1095 return 0;
1096}