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 "commit.h"
   9 #include "object.h"
  10 #include "tag.h"
  11 #include "diff.h"
  12 #include "diffcore.h"
  13 #include "log-tree.h"
  14 #include "revision.h"
  15 #include "decorate.h"
  16 #include "string-list.h"
  17 #include "utf8.h"
  18 #include "parse-options.h"
  19 #include "quote.h"
  20 #include "remote.h"
  21 #include "blob.h"
  22
  23 static const char *fast_export_usage[] = {
  24         N_("git fast-export [rev-list-opts]"),
  25         NULL
  26 };
  27
  28 static int progress;
  29 static enum { ABORT, VERBATIM, WARN, WARN_STRIP, STRIP } signed_tag_mode = ABORT;
  30 static enum { ERROR, DROP, REWRITE } tag_of_filtered_mode = ERROR;
  31 static int fake_missing_tagger;
  32 static int use_done_feature;
  33 static int no_data;
  34 static int full_tree;
  35 static struct string_list extra_refs = STRING_LIST_INIT_NODUP;
  36 static struct refspec *refspecs;
  37 static int refspecs_nr;
  38 static int anonymize;
  39
  40 static int parse_opt_signed_tag_mode(const struct option *opt,
  41                                      const char *arg, int unset)
  42 {
  43         if (unset || !strcmp(arg, "abort"))
  44                 signed_tag_mode = ABORT;
  45         else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
  46                 signed_tag_mode = VERBATIM;
  47         else if (!strcmp(arg, "warn"))
  48                 signed_tag_mode = WARN;
  49         else if (!strcmp(arg, "warn-strip"))
  50                 signed_tag_mode = WARN_STRIP;
  51         else if (!strcmp(arg, "strip"))
  52                 signed_tag_mode = STRIP;
  53         else
  54                 return error("Unknown signed-tags mode: %s", arg);
  55         return 0;
  56 }
  57
  58 static int parse_opt_tag_of_filtered_mode(const struct option *opt,
  59                                           const char *arg, int unset)
  60 {
  61         if (unset || !strcmp(arg, "abort"))
  62                 tag_of_filtered_mode = ERROR;
  63         else if (!strcmp(arg, "drop"))
  64                 tag_of_filtered_mode = DROP;
  65         else if (!strcmp(arg, "rewrite"))
  66                 tag_of_filtered_mode = REWRITE;
  67         else
  68                 return error("Unknown tag-of-filtered mode: %s", arg);
  69         return 0;
  70 }
  71
  72 static struct decoration idnums;
  73 static uint32_t last_idnum;
  74
  75 static int has_unshown_parent(struct commit *commit)
  76 {
  77         struct commit_list *parent;
  78
  79         for (parent = commit->parents; parent; parent = parent->next)
  80                 if (!(parent->item->object.flags & SHOWN) &&
  81                     !(parent->item->object.flags & UNINTERESTING))
  82                         return 1;
  83         return 0;
  84 }
  85
  86 struct anonymized_entry {
  87         struct hashmap_entry hash;
  88         const char *orig;
  89         size_t orig_len;
  90         const char *anon;
  91         size_t anon_len;
  92 };
  93
  94 static int anonymized_entry_cmp(const void *va, const void *vb,
  95                                 const void *data)
  96 {
  97         const struct anonymized_entry *a = va, *b = vb;
  98         return a->orig_len != b->orig_len ||
  99                 memcmp(a->orig, b->orig, a->orig_len);
 100 }
 101
 102 /*
 103  * Basically keep a cache of X->Y so that we can repeatedly replace
 104  * the same anonymized string with another. The actual generation
 105  * is farmed out to the generate function.
 106  */
 107 static const void *anonymize_mem(struct hashmap *map,
 108                                  void *(*generate)(const void *, size_t *),
 109                                  const void *orig, size_t *len)
 110 {
 111         struct anonymized_entry key, *ret;
 112
 113         if (!map->cmpfn)
 114                 hashmap_init(map, anonymized_entry_cmp, 0);
 115
 116         hashmap_entry_init(&key, memhash(orig, *len));
 117         key.orig = orig;
 118         key.orig_len = *len;
 119         ret = hashmap_get(map, &key, NULL);
 120
 121         if (!ret) {
 122                 ret = xmalloc(sizeof(*ret));
 123                 hashmap_entry_init(&ret->hash, key.hash.hash);
 124                 ret->orig = xstrdup(orig);
 125                 ret->orig_len = *len;
 126                 ret->anon = generate(orig, len);
 127                 ret->anon_len = *len;
 128                 hashmap_put(map, ret);
 129         }
 130
 131         *len = ret->anon_len;
 132         return ret->anon;
 133 }
 134
 135 /*
 136  * We anonymize each component of a path individually,
 137  * so that paths a/b and a/c will share a common root.
 138  * The paths are cached via anonymize_mem so that repeated
 139  * lookups for "a" will yield the same value.
 140  */
 141 static void anonymize_path(struct strbuf *out, const char *path,
 142                            struct hashmap *map,
 143                            void *(*generate)(const void *, size_t *))
 144 {
 145         while (*path) {
 146                 const char *end_of_component = strchrnul(path, '/');
 147                 size_t len = end_of_component - path;
 148                 const char *c = anonymize_mem(map, generate, path, &len);
 149                 strbuf_add(out, c, len);
 150                 path = end_of_component;
 151                 if (*path)
 152                         strbuf_addch(out, *path++);
 153         }
 154 }
 155
 156 /* Since intptr_t is C99, we do not use it here */
 157 static inline uint32_t *mark_to_ptr(uint32_t mark)
 158 {
 159         return ((uint32_t *)NULL) + mark;
 160 }
 161
 162 static inline uint32_t ptr_to_mark(void * mark)
 163 {
 164         return (uint32_t *)mark - (uint32_t *)NULL;
 165 }
 166
 167 static inline void mark_object(struct object *object, uint32_t mark)
 168 {
 169         add_decoration(&idnums, object, mark_to_ptr(mark));
 170 }
 171
 172 static inline void mark_next_object(struct object *object)
 173 {
 174         mark_object(object, ++last_idnum);
 175 }
 176
 177 static int get_object_mark(struct object *object)
 178 {
 179         void *decoration = lookup_decoration(&idnums, object);
 180         if (!decoration)
 181                 return 0;
 182         return ptr_to_mark(decoration);
 183 }
 184
 185 static void show_progress(void)
 186 {
 187         static int counter = 0;
 188         if (!progress)
 189                 return;
 190         if ((++counter % progress) == 0)
 191                 printf("progress %d objects\n", counter);
 192 }
 193
 194 /*
 195  * Ideally we would want some transformation of the blob data here
 196  * that is unreversible, but would still be the same size and have
 197  * the same data relationship to other blobs (so that we get the same
 198  * delta and packing behavior as the original). But the first and last
 199  * requirements there are probably mutually exclusive, so let's take
 200  * the easy way out for now, and just generate arbitrary content.
 201  *
 202  * There's no need to cache this result with anonymize_mem, since
 203  * we already handle blob content caching with marks.
 204  */
 205 static char *anonymize_blob(unsigned long *size)
 206 {
 207         static int counter;
 208         struct strbuf out = STRBUF_INIT;
 209         strbuf_addf(&out, "anonymous blob %d", counter++);
 210         *size = out.len;
 211         return strbuf_detach(&out, NULL);
 212 }
 213
 214 static void export_blob(const unsigned char *sha1)
 215 {
 216         unsigned long size;
 217         enum object_type type;
 218         char *buf;
 219         struct object *object;
 220         int eaten;
 221
 222         if (no_data)
 223                 return;
 224
 225         if (is_null_sha1(sha1))
 226                 return;
 227
 228         object = lookup_object(sha1);
 229         if (object && object->flags & SHOWN)
 230                 return;
 231
 232         if (anonymize) {
 233                 buf = anonymize_blob(&size);
 234                 object = (struct object *)lookup_blob(sha1);
 235                 eaten = 0;
 236         } else {
 237                 buf = read_sha1_file(sha1, &type, &size);
 238                 if (!buf)
 239                         die ("Could not read blob %s", sha1_to_hex(sha1));
 240                 if (check_sha1_signature(sha1, buf, size, typename(type)) < 0)
 241                         die("sha1 mismatch in blob %s", sha1_to_hex(sha1));
 242                 object = parse_object_buffer(sha1, type, size, buf, &eaten);
 243         }
 244
 245         if (!object)
 246                 die("Could not read blob %s", sha1_to_hex(sha1));
 247
 248         mark_next_object(object);
 249
 250         printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
 251         if (size && fwrite(buf, size, 1, stdout) != 1)
 252                 die_errno ("Could not write blob '%s'", sha1_to_hex(sha1));
 253         printf("\n");
 254
 255         show_progress();
 256
 257         object->flags |= SHOWN;
 258         if (!eaten)
 259                 free(buf);
 260 }
 261
 262 static int depth_first(const void *a_, const void *b_)
 263 {
 264         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
 265         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
 266         const char *name_a, *name_b;
 267         int len_a, len_b, len;
 268         int cmp;
 269
 270         name_a = a->one ? a->one->path : a->two->path;
 271         name_b = b->one ? b->one->path : b->two->path;
 272
 273         len_a = strlen(name_a);
 274         len_b = strlen(name_b);
 275         len = (len_a < len_b) ? len_a : len_b;
 276
 277         /* strcmp will sort 'd' before 'd/e', we want 'd/e' before 'd' */
 278         cmp = memcmp(name_a, name_b, len);
 279         if (cmp)
 280                 return cmp;
 281         cmp = len_b - len_a;
 282         if (cmp)
 283                 return cmp;
 284         /*
 285          * Move 'R'ename entries last so that all references of the file
 286          * appear in the output before it is renamed (e.g., when a file
 287          * was copied and renamed in the same commit).
 288          */
 289         return (a->status == 'R') - (b->status == 'R');
 290 }
 291
 292 static void print_path_1(const char *path)
 293 {
 294         int need_quote = quote_c_style(path, NULL, NULL, 0);
 295         if (need_quote)
 296                 quote_c_style(path, NULL, stdout, 0);
 297         else if (strchr(path, ' '))
 298                 printf("\"%s\"", path);
 299         else
 300                 printf("%s", path);
 301 }
 302
 303 static void *anonymize_path_component(const void *path, size_t *len)
 304 {
 305         static int counter;
 306         struct strbuf out = STRBUF_INIT;
 307         strbuf_addf(&out, "path%d", counter++);
 308         return strbuf_detach(&out, len);
 309 }
 310
 311 static void print_path(const char *path)
 312 {
 313         if (!anonymize)
 314                 print_path_1(path);
 315         else {
 316                 static struct hashmap paths;
 317                 static struct strbuf anon = STRBUF_INIT;
 318
 319                 anonymize_path(&anon, path, &paths, anonymize_path_component);
 320                 print_path_1(anon.buf);
 321                 strbuf_reset(&anon);
 322         }
 323 }
 324
 325 static void *generate_fake_sha1(const void *old, size_t *len)
 326 {
 327         static uint32_t counter = 1; /* avoid null sha1 */
 328         unsigned char *out = xcalloc(20, 1);
 329         put_be32(out + 16, counter++);
 330         return out;
 331 }
 332
 333 static const unsigned char *anonymize_sha1(const unsigned char *sha1)
 334 {
 335         static struct hashmap sha1s;
 336         size_t len = 20;
 337         return anonymize_mem(&sha1s, generate_fake_sha1, sha1, &len);
 338 }
 339
 340 static void show_filemodify(struct diff_queue_struct *q,
 341                             struct diff_options *options, void *data)
 342 {
 343         int i;
 344
 345         /*
 346          * Handle files below a directory first, in case they are all deleted
 347          * and the directory changes to a file or symlink.
 348          */
 349         qsort(q->queue, q->nr, sizeof(q->queue[0]), depth_first);
 350
 351         for (i = 0; i < q->nr; i++) {
 352                 struct diff_filespec *ospec = q->queue[i]->one;
 353                 struct diff_filespec *spec = q->queue[i]->two;
 354
 355                 switch (q->queue[i]->status) {
 356                 case DIFF_STATUS_DELETED:
 357                         printf("D ");
 358                         print_path(spec->path);
 359                         putchar('\n');
 360                         break;
 361
 362                 case DIFF_STATUS_COPIED:
 363                 case DIFF_STATUS_RENAMED:
 364                         printf("%c ", q->queue[i]->status);
 365                         print_path(ospec->path);
 366                         putchar(' ');
 367                         print_path(spec->path);
 368                         putchar('\n');
 369
 370                         if (!hashcmp(ospec->sha1, spec->sha1) &&
 371                             ospec->mode == spec->mode)
 372                                 break;
 373                         /* fallthrough */
 374
 375                 case DIFF_STATUS_TYPE_CHANGED:
 376                 case DIFF_STATUS_MODIFIED:
 377                 case DIFF_STATUS_ADDED:
 378                         /*
 379                          * Links refer to objects in another repositories;
 380                          * output the SHA-1 verbatim.
 381                          */
 382                         if (no_data || S_ISGITLINK(spec->mode))
 383                                 printf("M %06o %s ", spec->mode,
 384                                        sha1_to_hex(anonymize ?
 385                                                    anonymize_sha1(spec->sha1) :
 386                                                    spec->sha1));
 387                         else {
 388                                 struct object *object = lookup_object(spec->sha1);
 389                                 printf("M %06o :%d ", spec->mode,
 390                                        get_object_mark(object));
 391                         }
 392                         print_path(spec->path);
 393                         putchar('\n');
 394                         break;
 395
 396                 default:
 397                         die("Unexpected comparison status '%c' for %s, %s",
 398                                 q->queue[i]->status,
 399                                 ospec->path ? ospec->path : "none",
 400                                 spec->path ? spec->path : "none");
 401                 }
 402         }
 403 }
 404
 405 static const char *find_encoding(const char *begin, const char *end)
 406 {
 407         const char *needle = "\nencoding ";
 408         char *bol, *eol;
 409
 410         bol = memmem(begin, end ? end - begin : strlen(begin),
 411                      needle, strlen(needle));
 412         if (!bol)
 413                 return git_commit_encoding;
 414         bol += strlen(needle);
 415         eol = strchrnul(bol, '\n');
 416         *eol = '\0';
 417         return bol;
 418 }
 419
 420 static void *anonymize_ref_component(const void *old, size_t *len)
 421 {
 422         static int counter;
 423         struct strbuf out = STRBUF_INIT;
 424         strbuf_addf(&out, "ref%d", counter++);
 425         return strbuf_detach(&out, len);
 426 }
 427
 428 static const char *anonymize_refname(const char *refname)
 429 {
 430         /*
 431          * If any of these prefixes is found, we will leave it intact
 432          * so that tags remain tags and so forth.
 433          */
 434         static const char *prefixes[] = {
 435                 "refs/heads/",
 436                 "refs/tags/",
 437                 "refs/remotes/",
 438                 "refs/"
 439         };
 440         static struct hashmap refs;
 441         static struct strbuf anon = STRBUF_INIT;
 442         int i;
 443
 444         /*
 445          * We also leave "master" as a special case, since it does not reveal
 446          * anything interesting.
 447          */
 448         if (!strcmp(refname, "refs/heads/master"))
 449                 return refname;
 450
 451         strbuf_reset(&anon);
 452         for (i = 0; i < ARRAY_SIZE(prefixes); i++) {
 453                 if (skip_prefix(refname, prefixes[i], &refname)) {
 454                         strbuf_addstr(&anon, prefixes[i]);
 455                         break;
 456                 }
 457         }
 458
 459         anonymize_path(&anon, refname, &refs, anonymize_ref_component);
 460         return anon.buf;
 461 }
 462
 463 /*
 464  * We do not even bother to cache commit messages, as they are unlikely
 465  * to be repeated verbatim, and it is not that interesting when they are.
 466  */
 467 static char *anonymize_commit_message(const char *old)
 468 {
 469         static int counter;
 470         return xstrfmt("subject %d\n\nbody\n", counter++);
 471 }
 472
 473 static struct hashmap idents;
 474 static void *anonymize_ident(const void *old, size_t *len)
 475 {
 476         static int counter;
 477         struct strbuf out = STRBUF_INIT;
 478         strbuf_addf(&out, "User %d <user%d@example.com>", counter, counter);
 479         counter++;
 480         return strbuf_detach(&out, len);
 481 }
 482
 483 /*
 484  * Our strategy here is to anonymize the names and email addresses,
 485  * but keep timestamps intact, as they influence things like traversal
 486  * order (and by themselves should not be too revealing).
 487  */
 488 static void anonymize_ident_line(const char **beg, const char **end)
 489 {
 490         static struct strbuf buffers[] = { STRBUF_INIT, STRBUF_INIT };
 491         static unsigned which_buffer;
 492
 493         struct strbuf *out;
 494         struct ident_split split;
 495         const char *end_of_header;
 496
 497         out = &buffers[which_buffer++];
 498         which_buffer %= ARRAY_SIZE(buffers);
 499         strbuf_reset(out);
 500
 501         /* skip "committer", "author", "tagger", etc */
 502         end_of_header = strchr(*beg, ' ');
 503         if (!end_of_header)
 504                 die("BUG: malformed line fed to anonymize_ident_line: %.*s",
 505                     (int)(*end - *beg), *beg);
 506         end_of_header++;
 507         strbuf_add(out, *beg, end_of_header - *beg);
 508
 509         if (!split_ident_line(&split, end_of_header, *end - end_of_header) &&
 510             split.date_begin) {
 511                 const char *ident;
 512                 size_t len;
 513
 514                 len = split.mail_end - split.name_begin;
 515                 ident = anonymize_mem(&idents, anonymize_ident,
 516                                       split.name_begin, &len);
 517                 strbuf_add(out, ident, len);
 518                 strbuf_addch(out, ' ');
 519                 strbuf_add(out, split.date_begin, split.tz_end - split.date_begin);
 520         } else {
 521                 strbuf_addstr(out, "Malformed Ident <malformed@example.com> 0 -0000");
 522         }
 523
 524         *beg = out->buf;
 525         *end = out->buf + out->len;
 526 }
 527
 528 static void handle_commit(struct commit *commit, struct rev_info *rev)
 529 {
 530         int saved_output_format = rev->diffopt.output_format;
 531         const char *commit_buffer;
 532         const char *author, *author_end, *committer, *committer_end;
 533         const char *encoding, *message;
 534         char *reencoded = NULL;
 535         struct commit_list *p;
 536         const char *refname;
 537         int i;
 538
 539         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
 540
 541         parse_commit_or_die(commit);
 542         commit_buffer = get_commit_buffer(commit, NULL);
 543         author = strstr(commit_buffer, "\nauthor ");
 544         if (!author)
 545                 die ("Could not find author in commit %s",
 546                      sha1_to_hex(commit->object.sha1));
 547         author++;
 548         author_end = strchrnul(author, '\n');
 549         committer = strstr(author_end, "\ncommitter ");
 550         if (!committer)
 551                 die ("Could not find committer in commit %s",
 552                      sha1_to_hex(commit->object.sha1));
 553         committer++;
 554         committer_end = strchrnul(committer, '\n');
 555         message = strstr(committer_end, "\n\n");
 556         encoding = find_encoding(committer_end, message);
 557         if (message)
 558                 message += 2;
 559
 560         if (commit->parents &&
 561             get_object_mark(&commit->parents->item->object) != 0 &&
 562             !full_tree) {
 563                 parse_commit_or_die(commit->parents->item);
 564                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
 565                                commit->tree->object.sha1, "", &rev->diffopt);
 566         }
 567         else
 568                 diff_root_tree_sha1(commit->tree->object.sha1,
 569                                     "", &rev->diffopt);
 570
 571         /* Export the referenced blobs, and remember the marks. */
 572         for (i = 0; i < diff_queued_diff.nr; i++)
 573                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
 574                         export_blob(diff_queued_diff.queue[i]->two->sha1);
 575
 576         refname = commit->util;
 577         if (anonymize) {
 578                 refname = anonymize_refname(refname);
 579                 anonymize_ident_line(&committer, &committer_end);
 580                 anonymize_ident_line(&author, &author_end);
 581         }
 582
 583         mark_next_object(&commit->object);
 584         if (anonymize)
 585                 reencoded = anonymize_commit_message(message);
 586         else if (!is_encoding_utf8(encoding))
 587                 reencoded = reencode_string(message, "UTF-8", encoding);
 588         if (!commit->parents)
 589                 printf("reset %s\n", refname);
 590         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
 591                refname, last_idnum,
 592                (int)(author_end - author), author,
 593                (int)(committer_end - committer), committer,
 594                (unsigned)(reencoded
 595                           ? strlen(reencoded) : message
 596                           ? strlen(message) : 0),
 597                reencoded ? reencoded : message ? message : "");
 598         free(reencoded);
 599         unuse_commit_buffer(commit, commit_buffer);
 600
 601         for (i = 0, p = commit->parents; p; p = p->next) {
 602                 int mark = get_object_mark(&p->item->object);
 603                 if (!mark)
 604                         continue;
 605                 if (i == 0)
 606                         printf("from :%d\n", mark);
 607                 else
 608                         printf("merge :%d\n", mark);
 609                 i++;
 610         }
 611
 612         if (full_tree)
 613                 printf("deleteall\n");
 614         log_tree_diff_flush(rev);
 615         rev->diffopt.output_format = saved_output_format;
 616
 617         printf("\n");
 618
 619         show_progress();
 620 }
 621
 622 static void *anonymize_tag(const void *old, size_t *len)
 623 {
 624         static int counter;
 625         struct strbuf out = STRBUF_INIT;
 626         strbuf_addf(&out, "tag message %d", counter++);
 627         return strbuf_detach(&out, len);
 628 }
 629
 630 static void handle_tail(struct object_array *commits, struct rev_info *revs)
 631 {
 632         struct commit *commit;
 633         while (commits->nr) {
 634                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
 635                 if (has_unshown_parent(commit))
 636                         return;
 637                 handle_commit(commit, revs);
 638                 commits->nr--;
 639         }
 640 }
 641
 642 static void handle_tag(const char *name, struct tag *tag)
 643 {
 644         unsigned long size;
 645         enum object_type type;
 646         char *buf;
 647         const char *tagger, *tagger_end, *message;
 648         size_t message_size = 0;
 649         struct object *tagged;
 650         int tagged_mark;
 651         struct commit *p;
 652
 653         /* Trees have no identifier in fast-export output, thus we have no way
 654          * to output tags of trees, tags of tags of trees, etc.  Simply omit
 655          * such tags.
 656          */
 657         tagged = tag->tagged;
 658         while (tagged->type == OBJ_TAG) {
 659                 tagged = ((struct tag *)tagged)->tagged;
 660         }
 661         if (tagged->type == OBJ_TREE) {
 662                 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
 663                         sha1_to_hex(tag->object.sha1));
 664                 return;
 665         }
 666
 667         buf = read_sha1_file(tag->object.sha1, &type, &size);
 668         if (!buf)
 669                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
 670         message = memmem(buf, size, "\n\n", 2);
 671         if (message) {
 672                 message += 2;
 673                 message_size = strlen(message);
 674         }
 675         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
 676         if (!tagger) {
 677                 if (fake_missing_tagger)
 678                         tagger = "tagger Unspecified Tagger "
 679                                 "<unspecified-tagger> 0 +0000";
 680                 else
 681                         tagger = "";
 682                 tagger_end = tagger + strlen(tagger);
 683         } else {
 684                 tagger++;
 685                 tagger_end = strchrnul(tagger, '\n');
 686                 if (anonymize)
 687                         anonymize_ident_line(&tagger, &tagger_end);
 688         }
 689
 690         if (anonymize) {
 691                 name = anonymize_refname(name);
 692                 if (message) {
 693                         static struct hashmap tags;
 694                         message = anonymize_mem(&tags, anonymize_tag,
 695                                                 message, &message_size);
 696                 }
 697         }
 698
 699         /* handle signed tags */
 700         if (message) {
 701                 const char *signature = strstr(message,
 702                                                "\n-----BEGIN PGP SIGNATURE-----\n");
 703                 if (signature)
 704                         switch(signed_tag_mode) {
 705                         case ABORT:
 706                                 die ("Encountered signed tag %s; use "
 707                                      "--signed-tags=<mode> to handle it.",
 708                                      sha1_to_hex(tag->object.sha1));
 709                         case WARN:
 710                                 warning ("Exporting signed tag %s",
 711                                          sha1_to_hex(tag->object.sha1));
 712                                 /* fallthru */
 713                         case VERBATIM:
 714                                 break;
 715                         case WARN_STRIP:
 716                                 warning ("Stripping signature from tag %s",
 717                                          sha1_to_hex(tag->object.sha1));
 718                                 /* fallthru */
 719                         case STRIP:
 720                                 message_size = signature + 1 - message;
 721                                 break;
 722                         }
 723         }
 724
 725         /* handle tag->tagged having been filtered out due to paths specified */
 726         tagged = tag->tagged;
 727         tagged_mark = get_object_mark(tagged);
 728         if (!tagged_mark) {
 729                 switch(tag_of_filtered_mode) {
 730                 case ABORT:
 731                         die ("Tag %s tags unexported object; use "
 732                              "--tag-of-filtered-object=<mode> to handle it.",
 733                              sha1_to_hex(tag->object.sha1));
 734                 case DROP:
 735                         /* Ignore this tag altogether */
 736                         return;
 737                 case REWRITE:
 738                         if (tagged->type != OBJ_COMMIT) {
 739                                 die ("Tag %s tags unexported %s!",
 740                                      sha1_to_hex(tag->object.sha1),
 741                                      typename(tagged->type));
 742                         }
 743                         p = (struct commit *)tagged;
 744                         for (;;) {
 745                                 if (p->parents && p->parents->next)
 746                                         break;
 747                                 if (p->object.flags & UNINTERESTING)
 748                                         break;
 749                                 if (!(p->object.flags & TREESAME))
 750                                         break;
 751                                 if (!p->parents)
 752                                         die ("Can't find replacement commit for tag %s\n",
 753                                              sha1_to_hex(tag->object.sha1));
 754                                 p = p->parents->item;
 755                         }
 756                         tagged_mark = get_object_mark(&p->object);
 757                 }
 758         }
 759
 760         if (starts_with(name, "refs/tags/"))
 761                 name += 10;
 762         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
 763                name, tagged_mark,
 764                (int)(tagger_end - tagger), tagger,
 765                tagger == tagger_end ? "" : "\n",
 766                (int)message_size, (int)message_size, message ? message : "");
 767 }
 768
 769 static struct commit *get_commit(struct rev_cmdline_entry *e, char *full_name)
 770 {
 771         switch (e->item->type) {
 772         case OBJ_COMMIT:
 773                 return (struct commit *)e->item;
 774         case OBJ_TAG: {
 775                 struct tag *tag = (struct tag *)e->item;
 776
 777                 /* handle nested tags */
 778                 while (tag && tag->object.type == OBJ_TAG) {
 779                         parse_object(tag->object.sha1);
 780                         string_list_append(&extra_refs, full_name)->util = tag;
 781                         tag = (struct tag *)tag->tagged;
 782                 }
 783                 if (!tag)
 784                         die("Tag %s points nowhere?", e->name);
 785                 return (struct commit *)tag;
 786                 break;
 787         }
 788         default:
 789                 return NULL;
 790         }
 791 }
 792
 793 static void get_tags_and_duplicates(struct rev_cmdline_info *info)
 794 {
 795         int i;
 796
 797         for (i = 0; i < info->nr; i++) {
 798                 struct rev_cmdline_entry *e = info->rev + i;
 799                 unsigned char sha1[20];
 800                 struct commit *commit;
 801                 char *full_name;
 802
 803                 if (e->flags & UNINTERESTING)
 804                         continue;
 805
 806                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
 807                         continue;
 808
 809                 if (refspecs) {
 810                         char *private;
 811                         private = apply_refspecs(refspecs, refspecs_nr, full_name);
 812                         if (private) {
 813                                 free(full_name);
 814                                 full_name = private;
 815                         }
 816                 }
 817
 818                 commit = get_commit(e, full_name);
 819                 if (!commit) {
 820                         warning("%s: Unexpected object of type %s, skipping.",
 821                                 e->name,
 822                                 typename(e->item->type));
 823                         continue;
 824                 }
 825
 826                 switch(commit->object.type) {
 827                 case OBJ_COMMIT:
 828                         break;
 829                 case OBJ_BLOB:
 830                         export_blob(commit->object.sha1);
 831                         continue;
 832                 default: /* OBJ_TAG (nested tags) is already handled */
 833                         warning("Tag points to object of unexpected type %s, skipping.",
 834                                 typename(commit->object.type));
 835                         continue;
 836                 }
 837
 838                 /*
 839                  * This ref will not be updated through a commit, lets make
 840                  * sure it gets properly updated eventually.
 841                  */
 842                 if (commit->util || commit->object.flags & SHOWN)
 843                         string_list_append(&extra_refs, full_name)->util = commit;
 844                 if (!commit->util)
 845                         commit->util = full_name;
 846         }
 847 }
 848
 849 static void handle_tags_and_duplicates(void)
 850 {
 851         struct commit *commit;
 852         int i;
 853
 854         for (i = extra_refs.nr - 1; i >= 0; i--) {
 855                 const char *name = extra_refs.items[i].string;
 856                 struct object *object = extra_refs.items[i].util;
 857                 switch (object->type) {
 858                 case OBJ_TAG:
 859                         handle_tag(name, (struct tag *)object);
 860                         break;
 861                 case OBJ_COMMIT:
 862                         if (anonymize)
 863                                 name = anonymize_refname(name);
 864                         /* create refs pointing to already seen commits */
 865                         commit = (struct commit *)object;
 866                         printf("reset %s\nfrom :%d\n\n", name,
 867                                get_object_mark(&commit->object));
 868                         show_progress();
 869                         break;
 870                 }
 871         }
 872 }
 873
 874 static void export_marks(char *file)
 875 {
 876         unsigned int i;
 877         uint32_t mark;
 878         struct object_decoration *deco = idnums.hash;
 879         FILE *f;
 880         int e = 0;
 881
 882         f = fopen(file, "w");
 883         if (!f)
 884                 die_errno("Unable to open marks file %s for writing.", file);
 885
 886         for (i = 0; i < idnums.size; i++) {
 887                 if (deco->base && deco->base->type == 1) {
 888                         mark = ptr_to_mark(deco->decoration);
 889                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
 890                                 sha1_to_hex(deco->base->sha1)) < 0) {
 891                             e = 1;
 892                             break;
 893                         }
 894                 }
 895                 deco++;
 896         }
 897
 898         e |= ferror(f);
 899         e |= fclose(f);
 900         if (e)
 901                 error("Unable to write marks file %s.", file);
 902 }
 903
 904 static void import_marks(char *input_file)
 905 {
 906         char line[512];
 907         FILE *f = fopen(input_file, "r");
 908         if (!f)
 909                 die_errno("cannot read '%s'", input_file);
 910
 911         while (fgets(line, sizeof(line), f)) {
 912                 uint32_t mark;
 913                 char *line_end, *mark_end;
 914                 unsigned char sha1[20];
 915                 struct object *object;
 916                 struct commit *commit;
 917                 enum object_type type;
 918
 919                 line_end = strchr(line, '\n');
 920                 if (line[0] != ':' || !line_end)
 921                         die("corrupt mark line: %s", line);
 922                 *line_end = '\0';
 923
 924                 mark = strtoumax(line + 1, &mark_end, 10);
 925                 if (!mark || mark_end == line + 1
 926                         || *mark_end != ' ' || get_sha1_hex(mark_end + 1, sha1))
 927                         die("corrupt mark line: %s", line);
 928
 929                 if (last_idnum < mark)
 930                         last_idnum = mark;
 931
 932                 type = sha1_object_info(sha1, NULL);
 933                 if (type < 0)
 934                         die("object not found: %s", sha1_to_hex(sha1));
 935
 936                 if (type != OBJ_COMMIT)
 937                         /* only commits */
 938                         continue;
 939
 940                 commit = lookup_commit(sha1);
 941                 if (!commit)
 942                         die("not a commit? can't happen: %s", sha1_to_hex(sha1));
 943
 944                 object = &commit->object;
 945
 946                 if (object->flags & SHOWN)
 947                         error("Object %s already has a mark", sha1_to_hex(sha1));
 948
 949                 mark_object(object, mark);
 950
 951                 object->flags |= SHOWN;
 952         }
 953         fclose(f);
 954 }
 955
 956 static void handle_deletes(void)
 957 {
 958         int i;
 959         for (i = 0; i < refspecs_nr; i++) {
 960                 struct refspec *refspec = &refspecs[i];
 961                 if (*refspec->src)
 962                         continue;
 963
 964                 printf("reset %s\nfrom %s\n\n",
 965                                 refspec->dst, sha1_to_hex(null_sha1));
 966         }
 967 }
 968
 969 int cmd_fast_export(int argc, const char **argv, const char *prefix)
 970 {
 971         struct rev_info revs;
 972         struct object_array commits = OBJECT_ARRAY_INIT;
 973         struct commit *commit;
 974         char *export_filename = NULL, *import_filename = NULL;
 975         uint32_t lastimportid;
 976         struct string_list refspecs_list = STRING_LIST_INIT_NODUP;
 977         struct option options[] = {
 978                 OPT_INTEGER(0, "progress", &progress,
 979                             N_("show progress after <n> objects")),
 980                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, N_("mode"),
 981                              N_("select handling of signed tags"),
 982                              parse_opt_signed_tag_mode),
 983                 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, N_("mode"),
 984                              N_("select handling of tags that tag filtered objects"),
 985                              parse_opt_tag_of_filtered_mode),
 986                 OPT_STRING(0, "export-marks", &export_filename, N_("file"),
 987                              N_("Dump marks to this file")),
 988                 OPT_STRING(0, "import-marks", &import_filename, N_("file"),
 989                              N_("Import marks from this file")),
 990                 OPT_BOOL(0, "fake-missing-tagger", &fake_missing_tagger,
 991                          N_("Fake a tagger when tags lack one")),
 992                 OPT_BOOL(0, "full-tree", &full_tree,
 993                          N_("Output full tree for each commit")),
 994                 OPT_BOOL(0, "use-done-feature", &use_done_feature,
 995                              N_("Use the done feature to terminate the stream")),
 996                 OPT_BOOL(0, "no-data", &no_data, N_("Skip output of blob data")),
 997                 OPT_STRING_LIST(0, "refspec", &refspecs_list, N_("refspec"),
 998                              N_("Apply refspec to exported refs")),
 999                 OPT_BOOL(0, "anonymize", &anonymize, N_("anonymize output")),
1000                 OPT_END()
1001         };
1002
1003         if (argc == 1)
1004                 usage_with_options (fast_export_usage, options);
1005
1006         /* we handle encodings */
1007         git_config(git_default_config, NULL);
1008
1009         init_revisions(&revs, prefix);
1010         revs.topo_order = 1;
1011         revs.show_source = 1;
1012         revs.rewrite_parents = 1;
1013         argc = parse_options(argc, argv, prefix, options, fast_export_usage,
1014                         PARSE_OPT_KEEP_ARGV0 | PARSE_OPT_KEEP_UNKNOWN);
1015         argc = setup_revisions(argc, argv, &revs, NULL);
1016         if (argc > 1)
1017                 usage_with_options (fast_export_usage, options);
1018
1019         if (refspecs_list.nr) {
1020                 const char **refspecs_str;
1021                 int i;
1022
1023                 refspecs_str = xmalloc(sizeof(*refspecs_str) * refspecs_list.nr);
1024                 for (i = 0; i < refspecs_list.nr; i++)
1025                         refspecs_str[i] = refspecs_list.items[i].string;
1026
1027                 refspecs_nr = refspecs_list.nr;
1028                 refspecs = parse_fetch_refspec(refspecs_nr, refspecs_str);
1029
1030                 string_list_clear(&refspecs_list, 1);
1031                 free(refspecs_str);
1032         }
1033
1034         if (use_done_feature)
1035                 printf("feature done\n");
1036
1037         if (import_filename)
1038                 import_marks(import_filename);
1039         lastimportid = last_idnum;
1040
1041         if (import_filename && revs.prune_data.nr)
1042                 full_tree = 1;
1043
1044         get_tags_and_duplicates(&revs.cmdline);
1045
1046         if (prepare_revision_walk(&revs))
1047                 die("revision walk setup failed");
1048         revs.diffopt.format_callback = show_filemodify;
1049         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
1050         while ((commit = get_revision(&revs))) {
1051                 if (has_unshown_parent(commit)) {
1052                         add_object_array(&commit->object, NULL, &commits);
1053                 }
1054                 else {
1055                         handle_commit(commit, &revs);
1056                         handle_tail(&commits, &revs);
1057                 }
1058         }
1059
1060         handle_tags_and_duplicates();
1061         handle_deletes();
1062
1063         if (export_filename && lastimportid != last_idnum)
1064                 export_marks(export_filename);
1065
1066         if (use_done_feature)
1067                 printf("done\n");
1068
1069         free_refspec(refspecs_nr, refspecs);
1070
1071         return 0;
1072 }