builtin/fast-export.c

   1 /*
   2  * "git fast-export" builtin command
   3  *
   4  * Copyright (C) 2007 Johannes E. Schindelin
   5  */
   6 #include "builtin.h"
   7 #include "cache.h"
   8 #include "config.h"
   9 #include "refs.h"
  10 #include "commit.h"
  11 #include "object.h"
  12 #include "tag.h"
  13 #include "diff.h"
  14 #include "diffcore.h"
  15 #include "log-tree.h"
  16 #include "revision.h"
  17 #include "decorate.h"
  18 #include "string-list.h"
  19 #include "utf8.h"
  20 #include "parse-options.h"
  21 #include "quote.h"
  22 #include "remote.h"
  23 #include "blob.h"
  24
  25 static const char *fast_export_usage[] = {
  26         N_("git fast-export [rev-list-opts]"),
  27         NULL
  28 };
  29
  30 static int progress;
  31 static enum { ABORT, VERBATIM, WARN, WARN_STRIP, STRIP } signed_tag_mode = ABORT;
  32 static enum { ERROR, DROP, REWRITE } tag_of_filtered_mode = ERROR;
  33 static int fake_missing_tagger;
  34 static int use_done_feature;
  35 static int no_data;
  36 static int full_tree;
  37 static struct string_list extra_refs = STRING_LIST_INIT_NODUP;
  38 static struct refspec *refspecs;
  39 static int refspecs_nr;
  40 static int anonymize;
  41
  42 static int parse_opt_signed_tag_mode(const struct option *opt,
  43                                      const char *arg, int unset)
  44 {
  45         if (unset || !strcmp(arg, "abort"))
  46                 signed_tag_mode = ABORT;
  47         else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
  48                 signed_tag_mode = VERBATIM;
  49         else if (!strcmp(arg, "warn"))
  50                 signed_tag_mode = WARN;
  51         else if (!strcmp(arg, "warn-strip"))
  52                 signed_tag_mode = WARN_STRIP;
  53         else if (!strcmp(arg, "strip"))
  54                 signed_tag_mode = STRIP;
  55         else
  56                 return error("Unknown signed-tags mode: %s", arg);
  57         return 0;
  58 }
  59
  60 static int parse_opt_tag_of_filtered_mode(const struct option *opt,
  61                                           const char *arg, int unset)
  62 {
  63         if (unset || !strcmp(arg, "abort"))
  64                 tag_of_filtered_mode = ERROR;
  65         else if (!strcmp(arg, "drop"))
  66                 tag_of_filtered_mode = DROP;
  67         else if (!strcmp(arg, "rewrite"))
  68                 tag_of_filtered_mode = REWRITE;
  69         else
  70                 return error("Unknown tag-of-filtered mode: %s", arg);
  71         return 0;
  72 }
  73
  74 static struct decoration idnums;
  75 static uint32_t last_idnum;
  76
  77 static int has_unshown_parent(struct commit *commit)
  78 {
  79         struct commit_list *parent;
  80
  81         for (parent = commit->parents; parent; parent = parent->next)
  82                 if (!(parent->item->object.flags & SHOWN) &&
  83                     !(parent->item->object.flags & UNINTERESTING))
  84                         return 1;
  85         return 0;
  86 }
  87
  88 struct anonymized_entry {
  89         struct hashmap_entry hash;
  90         const char *orig;
  91         size_t orig_len;
  92         const char *anon;
  93         size_t anon_len;
  94 };
  95
  96 static int anonymized_entry_cmp(const void *unused_cmp_data,
  97                                 const void *va, const void *vb,
  98                                 const void *unused_keydata)
  99 {
 100         const struct anonymized_entry *a = va, *b = vb;
 101         return a->orig_len != b->orig_len ||
 102                 memcmp(a->orig, b->orig, a->orig_len);
 103 }
 104
 105 /*
 106  * Basically keep a cache of X->Y so that we can repeatedly replace
 107  * the same anonymized string with another. The actual generation
 108  * is farmed out to the generate function.
 109  */
 110 static const void *anonymize_mem(struct hashmap *map,
 111                                  void *(*generate)(const void *, size_t *),
 112                                  const void *orig, size_t *len)
 113 {
 114         struct anonymized_entry key, *ret;
 115
 116         if (!map->cmpfn)
 117                 hashmap_init(map, anonymized_entry_cmp, NULL, 0);
 118
 119         hashmap_entry_init(&key, memhash(orig, *len));
 120         key.orig = orig;
 121         key.orig_len = *len;
 122         ret = hashmap_get(map, &key, NULL);
 123
 124         if (!ret) {
 125                 ret = xmalloc(sizeof(*ret));
 126                 hashmap_entry_init(&ret->hash, key.hash.hash);
 127                 ret->orig = xstrdup(orig);
 128                 ret->orig_len = *len;
 129                 ret->anon = generate(orig, len);
 130                 ret->anon_len = *len;
 131                 hashmap_put(map, ret);
 132         }
 133
 134         *len = ret->anon_len;
 135         return ret->anon;
 136 }
 137
 138 /*
 139  * We anonymize each component of a path individually,
 140  * so that paths a/b and a/c will share a common root.
 141  * The paths are cached via anonymize_mem so that repeated
 142  * lookups for "a" will yield the same value.
 143  */
 144 static void anonymize_path(struct strbuf *out, const char *path,
 145                            struct hashmap *map,
 146                            void *(*generate)(const void *, size_t *))
 147 {
 148         while (*path) {
 149                 const char *end_of_component = strchrnul(path, '/');
 150                 size_t len = end_of_component - path;
 151                 const char *c = anonymize_mem(map, generate, path, &len);
 152                 strbuf_add(out, c, len);
 153                 path = end_of_component;
 154                 if (*path)
 155                         strbuf_addch(out, *path++);
 156         }
 157 }
 158
 159 /* Since intptr_t is C99, we do not use it here */
 160 static inline uint32_t *mark_to_ptr(uint32_t mark)
 161 {
 162         return ((uint32_t *)NULL) + mark;
 163 }
 164
 165 static inline uint32_t ptr_to_mark(void * mark)
 166 {
 167         return (uint32_t *)mark - (uint32_t *)NULL;
 168 }
 169
 170 static inline void mark_object(struct object *object, uint32_t mark)
 171 {
 172         add_decoration(&idnums, object, mark_to_ptr(mark));
 173 }
 174
 175 static inline void mark_next_object(struct object *object)
 176 {
 177         mark_object(object, ++last_idnum);
 178 }
 179
 180 static int get_object_mark(struct object *object)
 181 {
 182         void *decoration = lookup_decoration(&idnums, object);
 183         if (!decoration)
 184                 return 0;
 185         return ptr_to_mark(decoration);
 186 }
 187
 188 static void show_progress(void)
 189 {
 190         static int counter = 0;
 191         if (!progress)
 192                 return;
 193         if ((++counter % progress) == 0)
 194                 printf("progress %d objects\n", counter);
 195 }
 196
 197 /*
 198  * Ideally we would want some transformation of the blob data here
 199  * that is unreversible, but would still be the same size and have
 200  * the same data relationship to other blobs (so that we get the same
 201  * delta and packing behavior as the original). But the first and last
 202  * requirements there are probably mutually exclusive, so let's take
 203  * the easy way out for now, and just generate arbitrary content.
 204  *
 205  * There's no need to cache this result with anonymize_mem, since
 206  * we already handle blob content caching with marks.
 207  */
 208 static char *anonymize_blob(unsigned long *size)
 209 {
 210         static int counter;
 211         struct strbuf out = STRBUF_INIT;
 212         strbuf_addf(&out, "anonymous blob %d", counter++);
 213         *size = out.len;
 214         return strbuf_detach(&out, NULL);
 215 }
 216
 217 static void export_blob(const struct object_id *oid)
 218 {
 219         unsigned long size;
 220         enum object_type type;
 221         char *buf;
 222         struct object *object;
 223         int eaten;
 224
 225         if (no_data)
 226                 return;
 227
 228         if (is_null_oid(oid))
 229                 return;
 230
 231         object = lookup_object(oid->hash);
 232         if (object && object->flags & SHOWN)
 233                 return;
 234
 235         if (anonymize) {
 236                 buf = anonymize_blob(&size);
 237                 object = (struct object *)lookup_blob(oid);
 238                 eaten = 0;
 239         } else {
 240                 buf = read_sha1_file(oid->hash, &type, &size);
 241                 if (!buf)
 242                         die ("Could not read blob %s", oid_to_hex(oid));
 243                 if (check_sha1_signature(oid->hash, buf, size, typename(type)) < 0)
 244                         die("sha1 mismatch in blob %s", oid_to_hex(oid));
 245                 object = parse_object_buffer(oid, type, size, buf, &eaten);
 246         }
 247
 248         if (!object)
 249                 die("Could not read blob %s", oid_to_hex(oid));
 250
 251         mark_next_object(object);
 252
 253         printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
 254         if (size && fwrite(buf, size, 1, stdout) != 1)
 255                 die_errno ("Could not write blob '%s'", oid_to_hex(oid));
 256         printf("\n");
 257
 258         show_progress();
 259
 260         object->flags |= SHOWN;
 261         if (!eaten)
 262                 free(buf);
 263 }
 264
 265 static int depth_first(const void *a_, const void *b_)
 266 {
 267         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
 268         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
 269         const char *name_a, *name_b;
 270         int len_a, len_b, len;
 271         int cmp;
 272
 273         name_a = a->one ? a->one->path : a->two->path;
 274         name_b = b->one ? b->one->path : b->two->path;
 275
 276         len_a = strlen(name_a);
 277         len_b = strlen(name_b);
 278         len = (len_a < len_b) ? len_a : len_b;
 279
 280         /* strcmp will sort 'd' before 'd/e', we want 'd/e' before 'd' */
 281         cmp = memcmp(name_a, name_b, len);
 282         if (cmp)
 283                 return cmp;
 284         cmp = len_b - len_a;
 285         if (cmp)
 286                 return cmp;
 287         /*
 288          * Move 'R'ename entries last so that all references of the file
 289          * appear in the output before it is renamed (e.g., when a file
 290          * was copied and renamed in the same commit).
 291          */
 292         return (a->status == 'R') - (b->status == 'R');
 293 }
 294
 295 static void print_path_1(const char *path)
 296 {
 297         int need_quote = quote_c_style(path, NULL, NULL, 0);
 298         if (need_quote)
 299                 quote_c_style(path, NULL, stdout, 0);
 300         else if (strchr(path, ' '))
 301                 printf("\"%s\"", path);
 302         else
 303                 printf("%s", path);
 304 }
 305
 306 static void *anonymize_path_component(const void *path, size_t *len)
 307 {
 308         static int counter;
 309         struct strbuf out = STRBUF_INIT;
 310         strbuf_addf(&out, "path%d", counter++);
 311         return strbuf_detach(&out, len);
 312 }
 313
 314 static void print_path(const char *path)
 315 {
 316         if (!anonymize)
 317                 print_path_1(path);
 318         else {
 319                 static struct hashmap paths;
 320                 static struct strbuf anon = STRBUF_INIT;
 321
 322                 anonymize_path(&anon, path, &paths, anonymize_path_component);
 323                 print_path_1(anon.buf);
 324                 strbuf_reset(&anon);
 325         }
 326 }
 327
 328 static void *generate_fake_oid(const void *old, size_t *len)
 329 {
 330         static uint32_t counter = 1; /* avoid null sha1 */
 331         unsigned char *out = xcalloc(GIT_SHA1_RAWSZ, 1);
 332         put_be32(out + GIT_SHA1_RAWSZ - 4, counter++);
 333         return out;
 334 }
 335
 336 static const unsigned char *anonymize_sha1(const struct object_id *oid)
 337 {
 338         static struct hashmap sha1s;
 339         size_t len = GIT_SHA1_RAWSZ;
 340         return anonymize_mem(&sha1s, generate_fake_oid, oid, &len);
 341 }
 342
 343 static void show_filemodify(struct diff_queue_struct *q,
 344                             struct diff_options *options, void *data)
 345 {
 346         int i;
 347         struct string_list *changed = data;
 348
 349         /*
 350          * Handle files below a directory first, in case they are all deleted
 351          * and the directory changes to a file or symlink.
 352          */
 353         QSORT(q->queue, q->nr, depth_first);
 354
 355         for (i = 0; i < q->nr; i++) {
 356                 struct diff_filespec *ospec = q->queue[i]->one;
 357                 struct diff_filespec *spec = q->queue[i]->two;
 358
 359                 switch (q->queue[i]->status) {
 360                 case DIFF_STATUS_DELETED:
 361                         printf("D ");
 362                         print_path(spec->path);
 363                         string_list_insert(changed, spec->path);
 364                         putchar('\n');
 365                         break;
 366
 367                 case DIFF_STATUS_COPIED:
 368                 case DIFF_STATUS_RENAMED:
 369                         /*
 370                          * If a change in the file corresponding to ospec->path
 371                          * has been observed, we cannot trust its contents
 372                          * because the diff is calculated based on the prior
 373                          * contents, not the current contents.  So, declare a
 374                          * copy or rename only if there was no change observed.
 375                          */
 376                         if (!string_list_has_string(changed, ospec->path)) {
 377                                 printf("%c ", q->queue[i]->status);
 378                                 print_path(ospec->path);
 379                                 putchar(' ');
 380                                 print_path(spec->path);
 381                                 string_list_insert(changed, spec->path);
 382                                 putchar('\n');
 383
 384                                 if (!oidcmp(&ospec->oid, &spec->oid) &&
 385                                     ospec->mode == spec->mode)
 386                                         break;
 387                         }
 388                         /* fallthrough */
 389
 390                 case DIFF_STATUS_TYPE_CHANGED:
 391                 case DIFF_STATUS_MODIFIED:
 392                 case DIFF_STATUS_ADDED:
 393                         /*
 394                          * Links refer to objects in another repositories;
 395                          * output the SHA-1 verbatim.
 396                          */
 397                         if (no_data || S_ISGITLINK(spec->mode))
 398                                 printf("M %06o %s ", spec->mode,
 399                                        sha1_to_hex(anonymize ?
 400                                                    anonymize_sha1(&spec->oid) :
 401                                                    spec->oid.hash));
 402                         else {
 403                                 struct object *object = lookup_object(spec->oid.hash);
 404                                 printf("M %06o :%d ", spec->mode,
 405                                        get_object_mark(object));
 406                         }
 407                         print_path(spec->path);
 408                         string_list_insert(changed, spec->path);
 409                         putchar('\n');
 410                         break;
 411
 412                 default:
 413                         die("Unexpected comparison status '%c' for %s, %s",
 414                                 q->queue[i]->status,
 415                                 ospec->path ? ospec->path : "none",
 416                                 spec->path ? spec->path : "none");
 417                 }
 418         }
 419 }
 420
 421 static const char *find_encoding(const char *begin, const char *end)
 422 {
 423         const char *needle = "\nencoding ";
 424         char *bol, *eol;
 425
 426         bol = memmem(begin, end ? end - begin : strlen(begin),
 427                      needle, strlen(needle));
 428         if (!bol)
 429                 return git_commit_encoding;
 430         bol += strlen(needle);
 431         eol = strchrnul(bol, '\n');
 432         *eol = '\0';
 433         return bol;
 434 }
 435
 436 static void *anonymize_ref_component(const void *old, size_t *len)
 437 {
 438         static int counter;
 439         struct strbuf out = STRBUF_INIT;
 440         strbuf_addf(&out, "ref%d", counter++);
 441         return strbuf_detach(&out, len);
 442 }
 443
 444 static const char *anonymize_refname(const char *refname)
 445 {
 446         /*
 447          * If any of these prefixes is found, we will leave it intact
 448          * so that tags remain tags and so forth.
 449          */
 450         static const char *prefixes[] = {
 451                 "refs/heads/",
 452                 "refs/tags/",
 453                 "refs/remotes/",
 454                 "refs/"
 455         };
 456         static struct hashmap refs;
 457         static struct strbuf anon = STRBUF_INIT;
 458         int i;
 459
 460         /*
 461          * We also leave "master" as a special case, since it does not reveal
 462          * anything interesting.
 463          */
 464         if (!strcmp(refname, "refs/heads/master"))
 465                 return refname;
 466
 467         strbuf_reset(&anon);
 468         for (i = 0; i < ARRAY_SIZE(prefixes); i++) {
 469                 if (skip_prefix(refname, prefixes[i], &refname)) {
 470                         strbuf_addstr(&anon, prefixes[i]);
 471                         break;
 472                 }
 473         }
 474
 475         anonymize_path(&anon, refname, &refs, anonymize_ref_component);
 476         return anon.buf;
 477 }
 478
 479 /*
 480  * We do not even bother to cache commit messages, as they are unlikely
 481  * to be repeated verbatim, and it is not that interesting when they are.
 482  */
 483 static char *anonymize_commit_message(const char *old)
 484 {
 485         static int counter;
 486         return xstrfmt("subject %d\n\nbody\n", counter++);
 487 }
 488
 489 static struct hashmap idents;
 490 static void *anonymize_ident(const void *old, size_t *len)
 491 {
 492         static int counter;
 493         struct strbuf out = STRBUF_INIT;
 494         strbuf_addf(&out, "User %d <user%d@example.com>", counter, counter);
 495         counter++;
 496         return strbuf_detach(&out, len);
 497 }
 498
 499 /*
 500  * Our strategy here is to anonymize the names and email addresses,
 501  * but keep timestamps intact, as they influence things like traversal
 502  * order (and by themselves should not be too revealing).
 503  */
 504 static void anonymize_ident_line(const char **beg, const char **end)
 505 {
 506         static struct strbuf buffers[] = { STRBUF_INIT, STRBUF_INIT };
 507         static unsigned which_buffer;
 508
 509         struct strbuf *out;
 510         struct ident_split split;
 511         const char *end_of_header;
 512
 513         out = &buffers[which_buffer++];
 514         which_buffer %= ARRAY_SIZE(buffers);
 515         strbuf_reset(out);
 516
 517         /* skip "committer", "author", "tagger", etc */
 518         end_of_header = strchr(*beg, ' ');
 519         if (!end_of_header)
 520                 die("BUG: malformed line fed to anonymize_ident_line: %.*s",
 521                     (int)(*end - *beg), *beg);
 522         end_of_header++;
 523         strbuf_add(out, *beg, end_of_header - *beg);
 524
 525         if (!split_ident_line(&split, end_of_header, *end - end_of_header) &&
 526             split.date_begin) {
 527                 const char *ident;
 528                 size_t len;
 529
 530                 len = split.mail_end - split.name_begin;
 531                 ident = anonymize_mem(&idents, anonymize_ident,
 532                                       split.name_begin, &len);
 533                 strbuf_add(out, ident, len);
 534                 strbuf_addch(out, ' ');
 535                 strbuf_add(out, split.date_begin, split.tz_end - split.date_begin);
 536         } else {
 537                 strbuf_addstr(out, "Malformed Ident <malformed@example.com> 0 -0000");
 538         }
 539
 540         *beg = out->buf;
 541         *end = out->buf + out->len;
 542 }
 543
 544 static void handle_commit(struct commit *commit, struct rev_info *rev,
 545                           struct string_list *paths_of_changed_objects)
 546 {
 547         int saved_output_format = rev->diffopt.output_format;
 548         const char *commit_buffer;
 549         const char *author, *author_end, *committer, *committer_end;
 550         const char *encoding, *message;
 551         char *reencoded = NULL;
 552         struct commit_list *p;
 553         const char *refname;
 554         int i;
 555
 556         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
 557
 558         parse_commit_or_die(commit);
 559         commit_buffer = get_commit_buffer(commit, NULL);
 560         author = strstr(commit_buffer, "\nauthor ");
 561         if (!author)
 562                 die ("Could not find author in commit %s",
 563                      oid_to_hex(&commit->object.oid));
 564         author++;
 565         author_end = strchrnul(author, '\n');
 566         committer = strstr(author_end, "\ncommitter ");
 567         if (!committer)
 568                 die ("Could not find committer in commit %s",
 569                      oid_to_hex(&commit->object.oid));
 570         committer++;
 571         committer_end = strchrnul(committer, '\n');
 572         message = strstr(committer_end, "\n\n");
 573         encoding = find_encoding(committer_end, message);
 574         if (message)
 575                 message += 2;
 576
 577         if (commit->parents &&
 578             get_object_mark(&commit->parents->item->object) != 0 &&
 579             !full_tree) {
 580                 parse_commit_or_die(commit->parents->item);
 581                 diff_tree_oid(&commit->parents->item->tree->object.oid,
 582                               &commit->tree->object.oid, "", &rev->diffopt);
 583         }
 584         else
 585                 diff_root_tree_oid(&commit->tree->object.oid,
 586                                    "", &rev->diffopt);
 587
 588         /* Export the referenced blobs, and remember the marks. */
 589         for (i = 0; i < diff_queued_diff.nr; i++)
 590                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
 591                         export_blob(&diff_queued_diff.queue[i]->two->oid);
 592
 593         refname = commit->util;
 594         if (anonymize) {
 595                 refname = anonymize_refname(refname);
 596                 anonymize_ident_line(&committer, &committer_end);
 597                 anonymize_ident_line(&author, &author_end);
 598         }
 599
 600         mark_next_object(&commit->object);
 601         if (anonymize)
 602                 reencoded = anonymize_commit_message(message);
 603         else if (!is_encoding_utf8(encoding))
 604                 reencoded = reencode_string(message, "UTF-8", encoding);
 605         if (!commit->parents)
 606                 printf("reset %s\n", refname);
 607         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
 608                refname, last_idnum,
 609                (int)(author_end - author), author,
 610                (int)(committer_end - committer), committer,
 611                (unsigned)(reencoded
 612                           ? strlen(reencoded) : message
 613                           ? strlen(message) : 0),
 614                reencoded ? reencoded : message ? message : "");
 615         free(reencoded);
 616         unuse_commit_buffer(commit, commit_buffer);
 617
 618         for (i = 0, p = commit->parents; p; p = p->next) {
 619                 int mark = get_object_mark(&p->item->object);
 620                 if (!mark)
 621                         continue;
 622                 if (i == 0)
 623                         printf("from :%d\n", mark);
 624                 else
 625                         printf("merge :%d\n", mark);
 626                 i++;
 627         }
 628
 629         if (full_tree)
 630                 printf("deleteall\n");
 631         log_tree_diff_flush(rev);
 632         string_list_clear(paths_of_changed_objects, 0);
 633         rev->diffopt.output_format = saved_output_format;
 634
 635         printf("\n");
 636
 637         show_progress();
 638 }
 639
 640 static void *anonymize_tag(const void *old, size_t *len)
 641 {
 642         static int counter;
 643         struct strbuf out = STRBUF_INIT;
 644         strbuf_addf(&out, "tag message %d", counter++);
 645         return strbuf_detach(&out, len);
 646 }
 647
 648 static void handle_tail(struct object_array *commits, struct rev_info *revs,
 649                         struct string_list *paths_of_changed_objects)
 650 {
 651         struct commit *commit;
 652         while (commits->nr) {
 653                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
 654                 if (has_unshown_parent(commit))
 655                         return;
 656                 handle_commit(commit, revs, paths_of_changed_objects);
 657                 commits->nr--;
 658         }
 659 }
 660
 661 static void handle_tag(const char *name, struct tag *tag)
 662 {
 663         unsigned long size;
 664         enum object_type type;
 665         char *buf;
 666         const char *tagger, *tagger_end, *message;
 667         size_t message_size = 0;
 668         struct object *tagged;
 669         int tagged_mark;
 670         struct commit *p;
 671
 672         /* Trees have no identifier in fast-export output, thus we have no way
 673          * to output tags of trees, tags of tags of trees, etc.  Simply omit
 674          * such tags.
 675          */
 676         tagged = tag->tagged;
 677         while (tagged->type == OBJ_TAG) {
 678                 tagged = ((struct tag *)tagged)->tagged;
 679         }
 680         if (tagged->type == OBJ_TREE) {
 681                 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
 682                         oid_to_hex(&tag->object.oid));
 683                 return;
 684         }
 685
 686         buf = read_sha1_file(tag->object.oid.hash, &type, &size);
 687         if (!buf)
 688                 die ("Could not read tag %s", oid_to_hex(&tag->object.oid));
 689         message = memmem(buf, size, "\n\n", 2);
 690         if (message) {
 691                 message += 2;
 692                 message_size = strlen(message);
 693         }
 694         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
 695         if (!tagger) {
 696                 if (fake_missing_tagger)
 697                         tagger = "tagger Unspecified Tagger "
 698                                 "<unspecified-tagger> 0 +0000";
 699                 else
 700                         tagger = "";
 701                 tagger_end = tagger + strlen(tagger);
 702         } else {
 703                 tagger++;
 704                 tagger_end = strchrnul(tagger, '\n');
 705                 if (anonymize)
 706                         anonymize_ident_line(&tagger, &tagger_end);
 707         }
 708
 709         if (anonymize) {
 710                 name = anonymize_refname(name);
 711                 if (message) {
 712                         static struct hashmap tags;
 713                         message = anonymize_mem(&tags, anonymize_tag,
 714                                                 message, &message_size);
 715                 }
 716         }
 717
 718         /* handle signed tags */
 719         if (message) {
 720                 const char *signature = strstr(message,
 721                                                "\n-----BEGIN PGP SIGNATURE-----\n");
 722                 if (signature)
 723                         switch(signed_tag_mode) {
 724                         case ABORT:
 725                                 die ("Encountered signed tag %s; use "
 726                                      "--signed-tags=<mode> to handle it.",
 727                                      oid_to_hex(&tag->object.oid));
 728                         case WARN:
 729                                 warning ("Exporting signed tag %s",
 730                                          oid_to_hex(&tag->object.oid));
 731                                 /* fallthru */
 732                         case VERBATIM:
 733                                 break;
 734                         case WARN_STRIP:
 735                                 warning ("Stripping signature from tag %s",
 736                                          oid_to_hex(&tag->object.oid));
 737                                 /* fallthru */
 738                         case STRIP:
 739                                 message_size = signature + 1 - message;
 740                                 break;
 741                         }
 742         }
 743
 744         /* handle tag->tagged having been filtered out due to paths specified */
 745         tagged = tag->tagged;
 746         tagged_mark = get_object_mark(tagged);
 747         if (!tagged_mark) {
 748                 switch(tag_of_filtered_mode) {
 749                 case ABORT:
 750                         die ("Tag %s tags unexported object; use "
 751                              "--tag-of-filtered-object=<mode> to handle it.",
 752                              oid_to_hex(&tag->object.oid));
 753                 case DROP:
 754                         /* Ignore this tag altogether */
 755                         free(buf);
 756                         return;
 757                 case REWRITE:
 758                         if (tagged->type != OBJ_COMMIT) {
 759                                 die ("Tag %s tags unexported %s!",
 760                                      oid_to_hex(&tag->object.oid),
 761                                      typename(tagged->type));
 762                         }
 763                         p = (struct commit *)tagged;
 764                         for (;;) {
 765                                 if (p->parents && p->parents->next)
 766                                         break;
 767                                 if (p->object.flags & UNINTERESTING)
 768                                         break;
 769                                 if (!(p->object.flags & TREESAME))
 770                                         break;
 771                                 if (!p->parents)
 772                                         die ("Can't find replacement commit for tag %s\n",
 773                                              oid_to_hex(&tag->object.oid));
 774                                 p = p->parents->item;
 775                         }
 776                         tagged_mark = get_object_mark(&p->object);
 777                 }
 778         }
 779
 780         if (starts_with(name, "refs/tags/"))
 781                 name += 10;
 782         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
 783                name, tagged_mark,
 784                (int)(tagger_end - tagger), tagger,
 785                tagger == tagger_end ? "" : "\n",
 786                (int)message_size, (int)message_size, message ? message : "");
 787         free(buf);
 788 }
 789
 790 static struct commit *get_commit(struct rev_cmdline_entry *e, char *full_name)
 791 {
 792         switch (e->item->type) {
 793         case OBJ_COMMIT:
 794                 return (struct commit *)e->item;
 795         case OBJ_TAG: {
 796                 struct tag *tag = (struct tag *)e->item;
 797
 798                 /* handle nested tags */
 799                 while (tag && tag->object.type == OBJ_TAG) {
 800                         parse_object(&tag->object.oid);
 801                         string_list_append(&extra_refs, full_name)->util = tag;
 802                         tag = (struct tag *)tag->tagged;
 803                 }
 804                 if (!tag)
 805                         die("Tag %s points nowhere?", e->name);
 806                 return (struct commit *)tag;
 807                 break;
 808         }
 809         default:
 810                 return NULL;
 811         }
 812 }
 813
 814 static void get_tags_and_duplicates(struct rev_cmdline_info *info)
 815 {
 816         int i;
 817
 818         for (i = 0; i < info->nr; i++) {
 819                 struct rev_cmdline_entry *e = info->rev + i;
 820                 struct object_id oid;
 821                 struct commit *commit;
 822                 char *full_name;
 823
 824                 if (e->flags & UNINTERESTING)
 825                         continue;
 826
 827                 if (dwim_ref(e->name, strlen(e->name), oid.hash, &full_name) != 1)
 828                         continue;
 829
 830                 if (refspecs) {
 831                         char *private;
 832                         private = apply_refspecs(refspecs, refspecs_nr, full_name);
 833                         if (private) {
 834                                 free(full_name);
 835                                 full_name = private;
 836                         }
 837                 }
 838
 839                 commit = get_commit(e, full_name);
 840                 if (!commit) {
 841                         warning("%s: Unexpected object of type %s, skipping.",
 842                                 e->name,
 843                                 typename(e->item->type));
 844                         continue;
 845                 }
 846
 847                 switch(commit->object.type) {
 848                 case OBJ_COMMIT:
 849                         break;
 850                 case OBJ_BLOB:
 851                         export_blob(&commit->object.oid);
 852                         continue;
 853                 default: /* OBJ_TAG (nested tags) is already handled */
 854                         warning("Tag points to object of unexpected type %s, skipping.",
 855                                 typename(commit->object.type));
 856                         continue;
 857                 }
 858
 859                 /*
 860                  * This ref will not be updated through a commit, lets make
 861                  * sure it gets properly updated eventually.
 862                  */
 863                 if (commit->util || commit->object.flags & SHOWN)
 864                         string_list_append(&extra_refs, full_name)->util = commit;
 865                 if (!commit->util)
 866                         commit->util = full_name;
 867         }
 868 }
 869
 870 static void handle_tags_and_duplicates(void)
 871 {
 872         struct commit *commit;
 873         int i;
 874
 875         for (i = extra_refs.nr - 1; i >= 0; i--) {
 876                 const char *name = extra_refs.items[i].string;
 877                 struct object *object = extra_refs.items[i].util;
 878                 switch (object->type) {
 879                 case OBJ_TAG:
 880                         handle_tag(name, (struct tag *)object);
 881                         break;
 882                 case OBJ_COMMIT:
 883                         if (anonymize)
 884                                 name = anonymize_refname(name);
 885                         /* create refs pointing to already seen commits */
 886                         commit = (struct commit *)object;
 887                         printf("reset %s\nfrom :%d\n\n", name,
 888                                get_object_mark(&commit->object));
 889                         show_progress();
 890                         break;
 891                 }
 892         }
 893 }
 894
 895 static void export_marks(char *file)
 896 {
 897         unsigned int i;
 898         uint32_t mark;
 899         struct object_decoration *deco = idnums.hash;
 900         FILE *f;
 901         int e = 0;
 902
 903         f = fopen_for_writing(file);
 904         if (!f)
 905                 die_errno("Unable to open marks file %s for writing.", file);
 906
 907         for (i = 0; i < idnums.size; i++) {
 908                 if (deco->base && deco->base->type == 1) {
 909                         mark = ptr_to_mark(deco->decoration);
 910                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
 911                                 oid_to_hex(&deco->base->oid)) < 0) {
 912                             e = 1;
 913                             break;
 914                         }
 915                 }
 916                 deco++;
 917         }
 918
 919         e |= ferror(f);
 920         e |= fclose(f);
 921         if (e)
 922                 error("Unable to write marks file %s.", file);
 923 }
 924
 925 static void import_marks(char *input_file)
 926 {
 927         char line[512];
 928         FILE *f = xfopen(input_file, "r");
 929
 930         while (fgets(line, sizeof(line), f)) {
 931                 uint32_t mark;
 932                 char *line_end, *mark_end;
 933                 struct object_id oid;
 934                 struct object *object;
 935                 struct commit *commit;
 936                 enum object_type type;
 937
 938                 line_end = strchr(line, '\n');
 939                 if (line[0] != ':' || !line_end)
 940                         die("corrupt mark line: %s", line);
 941                 *line_end = '\0';
 942
 943                 mark = strtoumax(line + 1, &mark_end, 10);
 944                 if (!mark || mark_end == line + 1
 945                         || *mark_end != ' ' || get_oid_hex(mark_end + 1, &oid))
 946                         die("corrupt mark line: %s", line);
 947
 948                 if (last_idnum < mark)
 949                         last_idnum = mark;
 950
 951                 type = sha1_object_info(oid.hash, NULL);
 952                 if (type < 0)
 953                         die("object not found: %s", oid_to_hex(&oid));
 954
 955                 if (type != OBJ_COMMIT)
 956                         /* only commits */
 957                         continue;
 958
 959                 commit = lookup_commit(&oid);
 960                 if (!commit)
 961                         die("not a commit? can't happen: %s", oid_to_hex(&oid));
 962
 963                 object = &commit->object;
 964
 965                 if (object->flags & SHOWN)
 966                         error("Object %s already has a mark", oid_to_hex(&oid));
 967
 968                 mark_object(object, mark);
 969
 970                 object->flags |= SHOWN;
 971         }
 972         fclose(f);
 973 }
 974
 975 static void handle_deletes(void)
 976 {
 977         int i;
 978         for (i = 0; i < refspecs_nr; i++) {
 979                 struct refspec *refspec = &refspecs[i];
 980                 if (*refspec->src)
 981                         continue;
 982
 983                 printf("reset %s\nfrom %s\n\n",
 984                                 refspec->dst, sha1_to_hex(null_sha1));
 985         }
 986 }
 987
 988 int cmd_fast_export(int argc, const char **argv, const char *prefix)
 989 {
 990         struct rev_info revs;
 991         struct object_array commits = OBJECT_ARRAY_INIT;
 992         struct commit *commit;
 993         char *export_filename = NULL, *import_filename = NULL;
 994         uint32_t lastimportid;
 995         struct string_list refspecs_list = STRING_LIST_INIT_NODUP;
 996         struct string_list paths_of_changed_objects = STRING_LIST_INIT_DUP;
 997         struct option options[] = {
 998                 OPT_INTEGER(0, "progress", &progress,
 999                             N_("show progress after <n> objects")),
1000                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, N_("mode"),
1001                              N_("select handling of signed tags"),
1002                              parse_opt_signed_tag_mode),
1003                 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, N_("mode"),
1004                              N_("select handling of tags that tag filtered objects"),
1005                              parse_opt_tag_of_filtered_mode),
1006                 OPT_STRING(0, "export-marks", &export_filename, N_("file"),
1007                              N_("Dump marks to this file")),
1008                 OPT_STRING(0, "import-marks", &import_filename, N_("file"),
1009                              N_("Import marks from this file")),
1010                 OPT_BOOL(0, "fake-missing-tagger", &fake_missing_tagger,
1011                          N_("Fake a tagger when tags lack one")),
1012                 OPT_BOOL(0, "full-tree", &full_tree,
1013                          N_("Output full tree for each commit")),
1014                 OPT_BOOL(0, "use-done-feature", &use_done_feature,
1015                              N_("Use the done feature to terminate the stream")),
1016                 OPT_BOOL(0, "no-data", &no_data, N_("Skip output of blob data")),
1017                 OPT_STRING_LIST(0, "refspec", &refspecs_list, N_("refspec"),
1018                              N_("Apply refspec to exported refs")),
1019                 OPT_BOOL(0, "anonymize", &anonymize, N_("anonymize output")),
1020                 OPT_END()
1021         };
1022
1023         if (argc == 1)
1024                 usage_with_options (fast_export_usage, options);
1025
1026         /* we handle encodings */
1027         git_config(git_default_config, NULL);
1028
1029         init_revisions(&revs, prefix);
1030         revs.topo_order = 1;
1031         revs.show_source = 1;
1032         revs.rewrite_parents = 1;
1033         argc = parse_options(argc, argv, prefix, options, fast_export_usage,
1034                         PARSE_OPT_KEEP_ARGV0 | PARSE_OPT_KEEP_UNKNOWN);
1035         argc = setup_revisions(argc, argv, &revs, NULL);
1036         if (argc > 1)
1037                 usage_with_options (fast_export_usage, options);
1038
1039         if (refspecs_list.nr) {
1040                 const char **refspecs_str;
1041                 int i;
1042
1043                 ALLOC_ARRAY(refspecs_str, refspecs_list.nr);
1044                 for (i = 0; i < refspecs_list.nr; i++)
1045                         refspecs_str[i] = refspecs_list.items[i].string;
1046
1047                 refspecs_nr = refspecs_list.nr;
1048                 refspecs = parse_fetch_refspec(refspecs_nr, refspecs_str);
1049
1050                 string_list_clear(&refspecs_list, 1);
1051                 free(refspecs_str);
1052         }
1053
1054         if (use_done_feature)
1055                 printf("feature done\n");
1056
1057         if (import_filename)
1058                 import_marks(import_filename);
1059         lastimportid = last_idnum;
1060
1061         if (import_filename && revs.prune_data.nr)
1062                 full_tree = 1;
1063
1064         get_tags_and_duplicates(&revs.cmdline);
1065
1066         if (prepare_revision_walk(&revs))
1067                 die("revision walk setup failed");
1068         revs.diffopt.format_callback = show_filemodify;
1069         revs.diffopt.format_callback_data = &paths_of_changed_objects;
1070         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
1071         while ((commit = get_revision(&revs))) {
1072                 if (has_unshown_parent(commit)) {
1073                         add_object_array(&commit->object, NULL, &commits);
1074                 }
1075                 else {
1076                         handle_commit(commit, &revs, &paths_of_changed_objects);
1077                         handle_tail(&commits, &revs, &paths_of_changed_objects);
1078                 }
1079         }
1080
1081         handle_tags_and_duplicates();
1082         handle_deletes();
1083
1084         if (export_filename && lastimportid != last_idnum)
1085                 export_marks(export_filename);
1086
1087         if (use_done_feature)
1088                 printf("done\n");
1089
1090         free_refspec(refspecs_nr, refspecs);
1091
1092         return 0;
1093 }