git.oblomov.eu Git - git/blob - 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
 348         /*
 349          * Handle files below a directory first, in case they are all deleted
 350          * and the directory changes to a file or symlink.
 351          */
 352         QSORT(q->queue, q->nr, depth_first);
 353
 354         for (i = 0; i < q->nr; i++) {
 355                 struct diff_filespec *ospec = q->queue[i]->one;
 356                 struct diff_filespec *spec = q->queue[i]->two;
 357
 358                 switch (q->queue[i]->status) {
 359                 case DIFF_STATUS_DELETED:
 360                         printf("D ");
 361                         print_path(spec->path);
 362                         putchar('\n');
 363                         break;
 364
 365                 case DIFF_STATUS_COPIED:
 366                 case DIFF_STATUS_RENAMED:
 367                         printf("%c ", q->queue[i]->status);
 368                         print_path(ospec->path);
 369                         putchar(' ');
 370                         print_path(spec->path);
 371                         putchar('\n');
 372
 373                         if (!oidcmp(&ospec->oid, &spec->oid) &&
 374                             ospec->mode == spec->mode)
 375                                 break;
 376                         /* fallthrough */
 377
 378                 case DIFF_STATUS_TYPE_CHANGED:
 379                 case DIFF_STATUS_MODIFIED:
 380                 case DIFF_STATUS_ADDED:
 381                         /*
 382                          * Links refer to objects in another repositories;
 383                          * output the SHA-1 verbatim.
 384                          */
 385                         if (no_data || S_ISGITLINK(spec->mode))
 386                                 printf("M %06o %s ", spec->mode,
 387                                        sha1_to_hex(anonymize ?
 388                                                    anonymize_sha1(&spec->oid) :
 389                                                    spec->oid.hash));
 390                         else {
 391                                 struct object *object = lookup_object(spec->oid.hash);
 392                                 printf("M %06o :%d ", spec->mode,
 393                                        get_object_mark(object));
 394                         }
 395                         print_path(spec->path);
 396                         putchar('\n');
 397                         break;
 398
 399                 default:
 400                         die("Unexpected comparison status '%c' for %s, %s",
 401                                 q->queue[i]->status,
 402                                 ospec->path ? ospec->path : "none",
 403                                 spec->path ? spec->path : "none");
 404                 }
 405         }
 406 }
 407
 408 static const char *find_encoding(const char *begin, const char *end)
 409 {
 410         const char *needle = "\nencoding ";
 411         char *bol, *eol;
 412
 413         bol = memmem(begin, end ? end - begin : strlen(begin),
 414                      needle, strlen(needle));
 415         if (!bol)
 416                 return git_commit_encoding;
 417         bol += strlen(needle);
 418         eol = strchrnul(bol, '\n');
 419         *eol = '\0';
 420         return bol;
 421 }
 422
 423 static void *anonymize_ref_component(const void *old, size_t *len)
 424 {
 425         static int counter;
 426         struct strbuf out = STRBUF_INIT;
 427         strbuf_addf(&out, "ref%d", counter++);
 428         return strbuf_detach(&out, len);
 429 }
 430
 431 static const char *anonymize_refname(const char *refname)
 432 {
 433         /*
 434          * If any of these prefixes is found, we will leave it intact
 435          * so that tags remain tags and so forth.
 436          */
 437         static const char *prefixes[] = {
 438                 "refs/heads/",
 439                 "refs/tags/",
 440                 "refs/remotes/",
 441                 "refs/"
 442         };
 443         static struct hashmap refs;
 444         static struct strbuf anon = STRBUF_INIT;
 445         int i;
 446
 447         /*
 448          * We also leave "master" as a special case, since it does not reveal
 449          * anything interesting.
 450          */
 451         if (!strcmp(refname, "refs/heads/master"))
 452                 return refname;
 453
 454         strbuf_reset(&anon);
 455         for (i = 0; i < ARRAY_SIZE(prefixes); i++) {
 456                 if (skip_prefix(refname, prefixes[i], &refname)) {
 457                         strbuf_addstr(&anon, prefixes[i]);
 458                         break;
 459                 }
 460         }
 461
 462         anonymize_path(&anon, refname, &refs, anonymize_ref_component);
 463         return anon.buf;
 464 }
 465
 466 /*
 467  * We do not even bother to cache commit messages, as they are unlikely
 468  * to be repeated verbatim, and it is not that interesting when they are.
 469  */
 470 static char *anonymize_commit_message(const char *old)
 471 {
 472         static int counter;
 473         return xstrfmt("subject %d\n\nbody\n", counter++);
 474 }
 475
 476 static struct hashmap idents;
 477 static void *anonymize_ident(const void *old, size_t *len)
 478 {
 479         static int counter;
 480         struct strbuf out = STRBUF_INIT;
 481         strbuf_addf(&out, "User %d <user%d@example.com>", counter, counter);
 482         counter++;
 483         return strbuf_detach(&out, len);
 484 }
 485
 486 /*
 487  * Our strategy here is to anonymize the names and email addresses,
 488  * but keep timestamps intact, as they influence things like traversal
 489  * order (and by themselves should not be too revealing).
 490  */
 491 static void anonymize_ident_line(const char **beg, const char **end)
 492 {
 493         static struct strbuf buffers[] = { STRBUF_INIT, STRBUF_INIT };
 494         static unsigned which_buffer;
 495
 496         struct strbuf *out;
 497         struct ident_split split;
 498         const char *end_of_header;
 499
 500         out = &buffers[which_buffer++];
 501         which_buffer %= ARRAY_SIZE(buffers);
 502         strbuf_reset(out);
 503
 504         /* skip "committer", "author", "tagger", etc */
 505         end_of_header = strchr(*beg, ' ');
 506         if (!end_of_header)
 507                 die("BUG: malformed line fed to anonymize_ident_line: %.*s",
 508                     (int)(*end - *beg), *beg);
 509         end_of_header++;
 510         strbuf_add(out, *beg, end_of_header - *beg);
 511
 512         if (!split_ident_line(&split, end_of_header, *end - end_of_header) &&
 513             split.date_begin) {
 514                 const char *ident;
 515                 size_t len;
 516
 517                 len = split.mail_end - split.name_begin;
 518                 ident = anonymize_mem(&idents, anonymize_ident,
 519                                       split.name_begin, &len);
 520                 strbuf_add(out, ident, len);
 521                 strbuf_addch(out, ' ');
 522                 strbuf_add(out, split.date_begin, split.tz_end - split.date_begin);
 523         } else {
 524                 strbuf_addstr(out, "Malformed Ident <malformed@example.com> 0 -0000");
 525         }
 526
 527         *beg = out->buf;
 528         *end = out->buf + out->len;
 529 }
 530
 531 static void handle_commit(struct commit *commit, struct rev_info *rev)
 532 {
 533         int saved_output_format = rev->diffopt.output_format;
 534         const char *commit_buffer;
 535         const char *author, *author_end, *committer, *committer_end;
 536         const char *encoding, *message;
 537         char *reencoded = NULL;
 538         struct commit_list *p;
 539         const char *refname;
 540         int i;
 541
 542         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
 543
 544         parse_commit_or_die(commit);
 545         commit_buffer = get_commit_buffer(commit, NULL);
 546         author = strstr(commit_buffer, "\nauthor ");
 547         if (!author)
 548                 die ("Could not find author in commit %s",
 549                      oid_to_hex(&commit->object.oid));
 550         author++;
 551         author_end = strchrnul(author, '\n');
 552         committer = strstr(author_end, "\ncommitter ");
 553         if (!committer)
 554                 die ("Could not find committer in commit %s",
 555                      oid_to_hex(&commit->object.oid));
 556         committer++;
 557         committer_end = strchrnul(committer, '\n');
 558         message = strstr(committer_end, "\n\n");
 559         encoding = find_encoding(committer_end, message);
 560         if (message)
 561                 message += 2;
 562
 563         if (commit->parents &&
 564             get_object_mark(&commit->parents->item->object) != 0 &&
 565             !full_tree) {
 566                 parse_commit_or_die(commit->parents->item);
 567                 diff_tree_oid(&commit->parents->item->tree->object.oid,
 568                               &commit->tree->object.oid, "", &rev->diffopt);
 569         }
 570         else
 571                 diff_root_tree_oid(&commit->tree->object.oid,
 572                                    "", &rev->diffopt);
 573
 574         /* Export the referenced blobs, and remember the marks. */
 575         for (i = 0; i < diff_queued_diff.nr; i++)
 576                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
 577                         export_blob(&diff_queued_diff.queue[i]->two->oid);
 578
 579         refname = commit->util;
 580         if (anonymize) {
 581                 refname = anonymize_refname(refname);
 582                 anonymize_ident_line(&committer, &committer_end);
 583                 anonymize_ident_line(&author, &author_end);
 584         }
 585
 586         mark_next_object(&commit->object);
 587         if (anonymize)
 588                 reencoded = anonymize_commit_message(message);
 589         else if (!is_encoding_utf8(encoding))
 590                 reencoded = reencode_string(message, "UTF-8", encoding);
 591         if (!commit->parents)
 592                 printf("reset %s\n", refname);
 593         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
 594                refname, last_idnum,
 595                (int)(author_end - author), author,
 596                (int)(committer_end - committer), committer,
 597                (unsigned)(reencoded
 598                           ? strlen(reencoded) : message
 599                           ? strlen(message) : 0),
 600                reencoded ? reencoded : message ? message : "");
 601         free(reencoded);
 602         unuse_commit_buffer(commit, commit_buffer);
 603
 604         for (i = 0, p = commit->parents; p; p = p->next) {
 605                 int mark = get_object_mark(&p->item->object);
 606                 if (!mark)
 607                         continue;
 608                 if (i == 0)
 609                         printf("from :%d\n", mark);
 610                 else
 611                         printf("merge :%d\n", mark);
 612                 i++;
 613         }
 614
 615         if (full_tree)
 616                 printf("deleteall\n");
 617         log_tree_diff_flush(rev);
 618         rev->diffopt.output_format = saved_output_format;
 619
 620         printf("\n");
 621
 622         show_progress();
 623 }
 624
 625 static void *anonymize_tag(const void *old, size_t *len)
 626 {
 627         static int counter;
 628         struct strbuf out = STRBUF_INIT;
 629         strbuf_addf(&out, "tag message %d", counter++);
 630         return strbuf_detach(&out, len);
 631 }
 632
 633 static void handle_tail(struct object_array *commits, struct rev_info *revs)
 634 {
 635         struct commit *commit;
 636         while (commits->nr) {
 637                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
 638                 if (has_unshown_parent(commit))
 639                         return;
 640                 handle_commit(commit, revs);
 641                 commits->nr--;
 642         }
 643 }
 644
 645 static void handle_tag(const char *name, struct tag *tag)
 646 {
 647         unsigned long size;
 648         enum object_type type;
 649         char *buf;
 650         const char *tagger, *tagger_end, *message;
 651         size_t message_size = 0;
 652         struct object *tagged;
 653         int tagged_mark;
 654         struct commit *p;
 655
 656         /* Trees have no identifier in fast-export output, thus we have no way
 657          * to output tags of trees, tags of tags of trees, etc.  Simply omit
 658          * such tags.
 659          */
 660         tagged = tag->tagged;
 661         while (tagged->type == OBJ_TAG) {
 662                 tagged = ((struct tag *)tagged)->tagged;
 663         }
 664         if (tagged->type == OBJ_TREE) {
 665                 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
 666                         oid_to_hex(&tag->object.oid));
 667                 return;
 668         }
 669
 670         buf = read_sha1_file(tag->object.oid.hash, &type, &size);
 671         if (!buf)
 672                 die ("Could not read tag %s", oid_to_hex(&tag->object.oid));
 673         message = memmem(buf, size, "\n\n", 2);
 674         if (message) {
 675                 message += 2;
 676                 message_size = strlen(message);
 677         }
 678         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
 679         if (!tagger) {
 680                 if (fake_missing_tagger)
 681                         tagger = "tagger Unspecified Tagger "
 682                                 "<unspecified-tagger> 0 +0000";
 683                 else
 684                         tagger = "";
 685                 tagger_end = tagger + strlen(tagger);
 686         } else {
 687                 tagger++;
 688                 tagger_end = strchrnul(tagger, '\n');
 689                 if (anonymize)
 690                         anonymize_ident_line(&tagger, &tagger_end);
 691         }
 692
 693         if (anonymize) {
 694                 name = anonymize_refname(name);
 695                 if (message) {
 696                         static struct hashmap tags;
 697                         message = anonymize_mem(&tags, anonymize_tag,
 698                                                 message, &message_size);
 699                 }
 700         }
 701
 702         /* handle signed tags */
 703         if (message) {
 704                 const char *signature = strstr(message,
 705                                                "\n-----BEGIN PGP SIGNATURE-----\n");
 706                 if (signature)
 707                         switch(signed_tag_mode) {
 708                         case ABORT:
 709                                 die ("Encountered signed tag %s; use "
 710                                      "--signed-tags=<mode> to handle it.",
 711                                      oid_to_hex(&tag->object.oid));
 712                         case WARN:
 713                                 warning ("Exporting signed tag %s",
 714                                          oid_to_hex(&tag->object.oid));
 715                                 /* fallthru */
 716                         case VERBATIM:
 717                                 break;
 718                         case WARN_STRIP:
 719                                 warning ("Stripping signature from tag %s",
 720                                          oid_to_hex(&tag->object.oid));
 721                                 /* fallthru */
 722                         case STRIP:
 723                                 message_size = signature + 1 - message;
 724                                 break;
 725                         }
 726         }
 727
 728         /* handle tag->tagged having been filtered out due to paths specified */
 729         tagged = tag->tagged;
 730         tagged_mark = get_object_mark(tagged);
 731         if (!tagged_mark) {
 732                 switch(tag_of_filtered_mode) {
 733                 case ABORT:
 734                         die ("Tag %s tags unexported object; use "
 735                              "--tag-of-filtered-object=<mode> to handle it.",
 736                              oid_to_hex(&tag->object.oid));
 737                 case DROP:
 738                         /* Ignore this tag altogether */
 739                         free(buf);
 740                         return;
 741                 case REWRITE:
 742                         if (tagged->type != OBJ_COMMIT) {
 743                                 die ("Tag %s tags unexported %s!",
 744                                      oid_to_hex(&tag->object.oid),
 745                                      typename(tagged->type));
 746                         }
 747                         p = (struct commit *)tagged;
 748                         for (;;) {
 749                                 if (p->parents && p->parents->next)
 750                                         break;
 751                                 if (p->object.flags & UNINTERESTING)
 752                                         break;
 753                                 if (!(p->object.flags & TREESAME))
 754                                         break;
 755                                 if (!p->parents)
 756                                         die ("Can't find replacement commit for tag %s\n",
 757                                              oid_to_hex(&tag->object.oid));
 758                                 p = p->parents->item;
 759                         }
 760                         tagged_mark = get_object_mark(&p->object);
 761                 }
 762         }
 763
 764         if (starts_with(name, "refs/tags/"))
 765                 name += 10;
 766         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
 767                name, tagged_mark,
 768                (int)(tagger_end - tagger), tagger,
 769                tagger == tagger_end ? "" : "\n",
 770                (int)message_size, (int)message_size, message ? message : "");
 771         free(buf);
 772 }
 773
 774 static struct commit *get_commit(struct rev_cmdline_entry *e, char *full_name)
 775 {
 776         switch (e->item->type) {
 777         case OBJ_COMMIT:
 778                 return (struct commit *)e->item;
 779         case OBJ_TAG: {
 780                 struct tag *tag = (struct tag *)e->item;
 781
 782                 /* handle nested tags */
 783                 while (tag && tag->object.type == OBJ_TAG) {
 784                         parse_object(&tag->object.oid);
 785                         string_list_append(&extra_refs, full_name)->util = tag;
 786                         tag = (struct tag *)tag->tagged;
 787                 }
 788                 if (!tag)
 789                         die("Tag %s points nowhere?", e->name);
 790                 return (struct commit *)tag;
 791                 break;
 792         }
 793         default:
 794                 return NULL;
 795         }
 796 }
 797
 798 static void get_tags_and_duplicates(struct rev_cmdline_info *info)
 799 {
 800         int i;
 801
 802         for (i = 0; i < info->nr; i++) {
 803                 struct rev_cmdline_entry *e = info->rev + i;
 804                 struct object_id oid;
 805                 struct commit *commit;
 806                 char *full_name;
 807
 808                 if (e->flags & UNINTERESTING)
 809                         continue;
 810
 811                 if (dwim_ref(e->name, strlen(e->name), oid.hash, &full_name) != 1)
 812                         continue;
 813
 814                 if (refspecs) {
 815                         char *private;
 816                         private = apply_refspecs(refspecs, refspecs_nr, full_name);
 817                         if (private) {
 818                                 free(full_name);
 819                                 full_name = private;
 820                         }
 821                 }
 822
 823                 commit = get_commit(e, full_name);
 824                 if (!commit) {
 825                         warning("%s: Unexpected object of type %s, skipping.",
 826                                 e->name,
 827                                 typename(e->item->type));
 828                         continue;
 829                 }
 830
 831                 switch(commit->object.type) {
 832                 case OBJ_COMMIT:
 833                         break;
 834                 case OBJ_BLOB:
 835                         export_blob(&commit->object.oid);
 836                         continue;
 837                 default: /* OBJ_TAG (nested tags) is already handled */
 838                         warning("Tag points to object of unexpected type %s, skipping.",
 839                                 typename(commit->object.type));
 840                         continue;
 841                 }
 842
 843                 /*
 844                  * This ref will not be updated through a commit, lets make
 845                  * sure it gets properly updated eventually.
 846                  */
 847                 if (commit->util || commit->object.flags & SHOWN)
 848                         string_list_append(&extra_refs, full_name)->util = commit;
 849                 if (!commit->util)
 850                         commit->util = full_name;
 851         }
 852 }
 853
 854 static void handle_tags_and_duplicates(void)
 855 {
 856         struct commit *commit;
 857         int i;
 858
 859         for (i = extra_refs.nr - 1; i >= 0; i--) {
 860                 const char *name = extra_refs.items[i].string;
 861                 struct object *object = extra_refs.items[i].util;
 862                 switch (object->type) {
 863                 case OBJ_TAG:
 864                         handle_tag(name, (struct tag *)object);
 865                         break;
 866                 case OBJ_COMMIT:
 867                         if (anonymize)
 868                                 name = anonymize_refname(name);
 869                         /* create refs pointing to already seen commits */
 870                         commit = (struct commit *)object;
 871                         printf("reset %s\nfrom :%d\n\n", name,
 872                                get_object_mark(&commit->object));
 873                         show_progress();
 874                         break;
 875                 }
 876         }
 877 }
 878
 879 static void export_marks(char *file)
 880 {
 881         unsigned int i;
 882         uint32_t mark;
 883         struct object_decoration *deco = idnums.hash;
 884         FILE *f;
 885         int e = 0;
 886
 887         f = fopen_for_writing(file);
 888         if (!f)
 889                 die_errno("Unable to open marks file %s for writing.", file);
 890
 891         for (i = 0; i < idnums.size; i++) {
 892                 if (deco->base && deco->base->type == 1) {
 893                         mark = ptr_to_mark(deco->decoration);
 894                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
 895                                 oid_to_hex(&deco->base->oid)) < 0) {
 896                             e = 1;
 897                             break;
 898                         }
 899                 }
 900                 deco++;
 901         }
 902
 903         e |= ferror(f);
 904         e |= fclose(f);
 905         if (e)
 906                 error("Unable to write marks file %s.", file);
 907 }
 908
 909 static void import_marks(char *input_file)
 910 {
 911         char line[512];
 912         FILE *f = xfopen(input_file, "r");
 913
 914         while (fgets(line, sizeof(line), f)) {
 915                 uint32_t mark;
 916                 char *line_end, *mark_end;
 917                 struct object_id oid;
 918                 struct object *object;
 919                 struct commit *commit;
 920                 enum object_type type;
 921
 922                 line_end = strchr(line, '\n');
 923                 if (line[0] != ':' || !line_end)
 924                         die("corrupt mark line: %s", line);
 925                 *line_end = '\0';
 926
 927                 mark = strtoumax(line + 1, &mark_end, 10);
 928                 if (!mark || mark_end == line + 1
 929                         || *mark_end != ' ' || get_oid_hex(mark_end + 1, &oid))
 930                         die("corrupt mark line: %s", line);
 931
 932                 if (last_idnum < mark)
 933                         last_idnum = mark;
 934
 935                 type = sha1_object_info(oid.hash, NULL);
 936                 if (type < 0)
 937                         die("object not found: %s", oid_to_hex(&oid));
 938
 939                 if (type != OBJ_COMMIT)
 940                         /* only commits */
 941                         continue;
 942
 943                 commit = lookup_commit(&oid);
 944                 if (!commit)
 945                         die("not a commit? can't happen: %s", oid_to_hex(&oid));
 946
 947                 object = &commit->object;
 948
 949                 if (object->flags & SHOWN)
 950                         error("Object %s already has a mark", oid_to_hex(&oid));
 951
 952                 mark_object(object, mark);
 953
 954                 object->flags |= SHOWN;
 955         }
 956         fclose(f);
 957 }
 958
 959 static void handle_deletes(void)
 960 {
 961         int i;
 962         for (i = 0; i < refspecs_nr; i++) {
 963                 struct refspec *refspec = &refspecs[i];
 964                 if (*refspec->src)
 965                         continue;
 966
 967                 printf("reset %s\nfrom %s\n\n",
 968                                 refspec->dst, sha1_to_hex(null_sha1));
 969         }
 970 }
 971
 972 int cmd_fast_export(int argc, const char **argv, const char *prefix)
 973 {
 974         struct rev_info revs;
 975         struct object_array commits = OBJECT_ARRAY_INIT;
 976         struct commit *commit;
 977         char *export_filename = NULL, *import_filename = NULL;
 978         uint32_t lastimportid;
 979         struct string_list refspecs_list = STRING_LIST_INIT_NODUP;
 980         struct option options[] = {
 981                 OPT_INTEGER(0, "progress", &progress,
 982                             N_("show progress after <n> objects")),
 983                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, N_("mode"),
 984                              N_("select handling of signed tags"),
 985                              parse_opt_signed_tag_mode),
 986                 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, N_("mode"),
 987                              N_("select handling of tags that tag filtered objects"),
 988                              parse_opt_tag_of_filtered_mode),
 989                 OPT_STRING(0, "export-marks", &export_filename, N_("file"),
 990                              N_("Dump marks to this file")),
 991                 OPT_STRING(0, "import-marks", &import_filename, N_("file"),
 992                              N_("Import marks from this file")),
 993                 OPT_BOOL(0, "fake-missing-tagger", &fake_missing_tagger,
 994                          N_("Fake a tagger when tags lack one")),
 995                 OPT_BOOL(0, "full-tree", &full_tree,
 996                          N_("Output full tree for each commit")),
 997                 OPT_BOOL(0, "use-done-feature", &use_done_feature,
 998                              N_("Use the done feature to terminate the stream")),
 999                 OPT_BOOL(0, "no-data", &no_data, N_("Skip output of blob data")),
1000                 OPT_STRING_LIST(0, "refspec", &refspecs_list, N_("refspec"),
1001                              N_("Apply refspec to exported refs")),
1002                 OPT_BOOL(0, "anonymize", &anonymize, N_("anonymize output")),
1003                 OPT_END()
1004         };
1005
1006         if (argc == 1)
1007                 usage_with_options (fast_export_usage, options);
1008
1009         /* we handle encodings */
1010         git_config(git_default_config, NULL);
1011
1012         init_revisions(&revs, prefix);
1013         revs.topo_order = 1;
1014         revs.show_source = 1;
1015         revs.rewrite_parents = 1;
1016         argc = parse_options(argc, argv, prefix, options, fast_export_usage,
1017                         PARSE_OPT_KEEP_ARGV0 | PARSE_OPT_KEEP_UNKNOWN);
1018         argc = setup_revisions(argc, argv, &revs, NULL);
1019         if (argc > 1)
1020                 usage_with_options (fast_export_usage, options);
1021
1022         if (refspecs_list.nr) {
1023                 const char **refspecs_str;
1024                 int i;
1025
1026                 ALLOC_ARRAY(refspecs_str, refspecs_list.nr);
1027                 for (i = 0; i < refspecs_list.nr; i++)
1028                         refspecs_str[i] = refspecs_list.items[i].string;
1029
1030                 refspecs_nr = refspecs_list.nr;
1031                 refspecs = parse_fetch_refspec(refspecs_nr, refspecs_str);
1032
1033                 string_list_clear(&refspecs_list, 1);
1034                 free(refspecs_str);
1035         }
1036
1037         if (use_done_feature)
1038                 printf("feature done\n");
1039
1040         if (import_filename)
1041                 import_marks(import_filename);
1042         lastimportid = last_idnum;
1043
1044         if (import_filename && revs.prune_data.nr)
1045                 full_tree = 1;
1046
1047         get_tags_and_duplicates(&revs.cmdline);
1048
1049         if (prepare_revision_walk(&revs))
1050                 die("revision walk setup failed");
1051         revs.diffopt.format_callback = show_filemodify;
1052         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
1053         while ((commit = get_revision(&revs))) {
1054                 if (has_unshown_parent(commit)) {
1055                         add_object_array(&commit->object, NULL, &commits);
1056                 }
1057                 else {
1058                         handle_commit(commit, &revs);
1059                         handle_tail(&commits, &revs);
1060                 }
1061         }
1062
1063         handle_tags_and_duplicates();
1064         handle_deletes();
1065
1066         if (export_filename && lastimportid != last_idnum)
1067                 export_marks(export_filename);
1068
1069         if (use_done_feature)
1070                 printf("done\n");
1071
1072         free_refspec(refspecs_nr, refspecs);
1073
1074         return 0;
1075 }