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
  20 static const char *fast_export_usage[] = {
  21         "git fast-export [rev-list-opts]",
  22         NULL
  23 };
  24
  25 static int progress;
  26 static enum { ABORT, VERBATIM, WARN, STRIP } signed_tag_mode = ABORT;
  27 static enum { ERROR, DROP, REWRITE } tag_of_filtered_mode = ABORT;
  28 static int fake_missing_tagger;
  29 static int no_data;
  30
  31 static int parse_opt_signed_tag_mode(const struct option *opt,
  32                                      const char *arg, int unset)
  33 {
  34         if (unset || !strcmp(arg, "abort"))
  35                 signed_tag_mode = ABORT;
  36         else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
  37                 signed_tag_mode = VERBATIM;
  38         else if (!strcmp(arg, "warn"))
  39                 signed_tag_mode = WARN;
  40         else if (!strcmp(arg, "strip"))
  41                 signed_tag_mode = STRIP;
  42         else
  43                 return error("Unknown signed-tag mode: %s", arg);
  44         return 0;
  45 }
  46
  47 static int parse_opt_tag_of_filtered_mode(const struct option *opt,
  48                                           const char *arg, int unset)
  49 {
  50         if (unset || !strcmp(arg, "abort"))
  51                 tag_of_filtered_mode = ABORT;
  52         else if (!strcmp(arg, "drop"))
  53                 tag_of_filtered_mode = DROP;
  54         else if (!strcmp(arg, "rewrite"))
  55                 tag_of_filtered_mode = REWRITE;
  56         else
  57                 return error("Unknown tag-of-filtered mode: %s", arg);
  58         return 0;
  59 }
  60
  61 static struct decoration idnums;
  62 static uint32_t last_idnum;
  63
  64 static int has_unshown_parent(struct commit *commit)
  65 {
  66         struct commit_list *parent;
  67
  68         for (parent = commit->parents; parent; parent = parent->next)
  69                 if (!(parent->item->object.flags & SHOWN) &&
  70                     !(parent->item->object.flags & UNINTERESTING))
  71                         return 1;
  72         return 0;
  73 }
  74
  75 /* Since intptr_t is C99, we do not use it here */
  76 static inline uint32_t *mark_to_ptr(uint32_t mark)
  77 {
  78         return ((uint32_t *)NULL) + mark;
  79 }
  80
  81 static inline uint32_t ptr_to_mark(void * mark)
  82 {
  83         return (uint32_t *)mark - (uint32_t *)NULL;
  84 }
  85
  86 static inline void mark_object(struct object *object, uint32_t mark)
  87 {
  88         add_decoration(&idnums, object, mark_to_ptr(mark));
  89 }
  90
  91 static inline void mark_next_object(struct object *object)
  92 {
  93         mark_object(object, ++last_idnum);
  94 }
  95
  96 static int get_object_mark(struct object *object)
  97 {
  98         void *decoration = lookup_decoration(&idnums, object);
  99         if (!decoration)
 100                 return 0;
 101         return ptr_to_mark(decoration);
 102 }
 103
 104 static void show_progress(void)
 105 {
 106         static int counter = 0;
 107         if (!progress)
 108                 return;
 109         if ((++counter % progress) == 0)
 110                 printf("progress %d objects\n", counter);
 111 }
 112
 113 static void handle_object(const unsigned char *sha1)
 114 {
 115         unsigned long size;
 116         enum object_type type;
 117         char *buf;
 118         struct object *object;
 119
 120         if (no_data)
 121                 return;
 122
 123         if (is_null_sha1(sha1))
 124                 return;
 125
 126         object = parse_object(sha1);
 127         if (!object)
 128                 die ("Could not read blob %s", sha1_to_hex(sha1));
 129
 130         if (object->flags & SHOWN)
 131                 return;
 132
 133         buf = read_sha1_file(sha1, &type, &size);
 134         if (!buf)
 135                 die ("Could not read blob %s", sha1_to_hex(sha1));
 136
 137         mark_next_object(object);
 138
 139         printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
 140         if (size && fwrite(buf, size, 1, stdout) != 1)
 141                 die_errno ("Could not write blob '%s'", sha1_to_hex(sha1));
 142         printf("\n");
 143
 144         show_progress();
 145
 146         object->flags |= SHOWN;
 147         free(buf);
 148 }
 149
 150 static int depth_first(const void *a_, const void *b_)
 151 {
 152         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
 153         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
 154         const char *name_a, *name_b;
 155         int len_a, len_b, len;
 156         int cmp;
 157
 158         name_a = a->one ? a->one->path : a->two->path;
 159         name_b = b->one ? b->one->path : b->two->path;
 160
 161         len_a = strlen(name_a);
 162         len_b = strlen(name_b);
 163         len = (len_a < len_b) ? len_a : len_b;
 164
 165         /* strcmp will sort 'd' before 'd/e', we want 'd/e' before 'd' */
 166         cmp = memcmp(name_a, name_b, len);
 167         if (cmp)
 168                 return cmp;
 169         cmp = len_b - len_a;
 170         if (cmp)
 171                 return cmp;
 172         /*
 173          * Move 'R'ename entries last so that all references of the file
 174          * appear in the output before it is renamed (e.g., when a file
 175          * was copied and renamed in the same commit).
 176          */
 177         return (a->status == 'R') - (b->status == 'R');
 178 }
 179
 180 static void show_filemodify(struct diff_queue_struct *q,
 181                             struct diff_options *options, void *data)
 182 {
 183         int i;
 184
 185         /*
 186          * Handle files below a directory first, in case they are all deleted
 187          * and the directory changes to a file or symlink.
 188          */
 189         qsort(q->queue, q->nr, sizeof(q->queue[0]), depth_first);
 190
 191         for (i = 0; i < q->nr; i++) {
 192                 struct diff_filespec *ospec = q->queue[i]->one;
 193                 struct diff_filespec *spec = q->queue[i]->two;
 194
 195                 switch (q->queue[i]->status) {
 196                 case DIFF_STATUS_DELETED:
 197                         printf("D %s\n", spec->path);
 198                         break;
 199
 200                 case DIFF_STATUS_COPIED:
 201                 case DIFF_STATUS_RENAMED:
 202                         printf("%c \"%s\" \"%s\"\n", q->queue[i]->status,
 203                                ospec->path, spec->path);
 204
 205                         if (!hashcmp(ospec->sha1, spec->sha1) &&
 206                             ospec->mode == spec->mode)
 207                                 break;
 208                         /* fallthrough */
 209
 210                 case DIFF_STATUS_TYPE_CHANGED:
 211                 case DIFF_STATUS_MODIFIED:
 212                 case DIFF_STATUS_ADDED:
 213                         /*
 214                          * Links refer to objects in another repositories;
 215                          * output the SHA-1 verbatim.
 216                          */
 217                         if (no_data || S_ISGITLINK(spec->mode))
 218                                 printf("M %06o %s %s\n", spec->mode,
 219                                        sha1_to_hex(spec->sha1), spec->path);
 220                         else {
 221                                 struct object *object = lookup_object(spec->sha1);
 222                                 printf("M %06o :%d %s\n", spec->mode,
 223                                        get_object_mark(object), spec->path);
 224                         }
 225                         break;
 226
 227                 default:
 228                         die("Unexpected comparison status '%c' for %s, %s",
 229                                 q->queue[i]->status,
 230                                 ospec->path ? ospec->path : "none",
 231                                 spec->path ? spec->path : "none");
 232                 }
 233         }
 234 }
 235
 236 static const char *find_encoding(const char *begin, const char *end)
 237 {
 238         const char *needle = "\nencoding ";
 239         char *bol, *eol;
 240
 241         bol = memmem(begin, end ? end - begin : strlen(begin),
 242                      needle, strlen(needle));
 243         if (!bol)
 244                 return git_commit_encoding;
 245         bol += strlen(needle);
 246         eol = strchrnul(bol, '\n');
 247         *eol = '\0';
 248         return bol;
 249 }
 250
 251 static void handle_commit(struct commit *commit, struct rev_info *rev)
 252 {
 253         int saved_output_format = rev->diffopt.output_format;
 254         const char *author, *author_end, *committer, *committer_end;
 255         const char *encoding, *message;
 256         char *reencoded = NULL;
 257         struct commit_list *p;
 258         int i;
 259
 260         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
 261
 262         parse_commit(commit);
 263         author = strstr(commit->buffer, "\nauthor ");
 264         if (!author)
 265                 die ("Could not find author in commit %s",
 266                      sha1_to_hex(commit->object.sha1));
 267         author++;
 268         author_end = strchrnul(author, '\n');
 269         committer = strstr(author_end, "\ncommitter ");
 270         if (!committer)
 271                 die ("Could not find committer in commit %s",
 272                      sha1_to_hex(commit->object.sha1));
 273         committer++;
 274         committer_end = strchrnul(committer, '\n');
 275         message = strstr(committer_end, "\n\n");
 276         encoding = find_encoding(committer_end, message);
 277         if (message)
 278                 message += 2;
 279
 280         if (commit->parents &&
 281             get_object_mark(&commit->parents->item->object) != 0) {
 282                 parse_commit(commit->parents->item);
 283                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
 284                                commit->tree->object.sha1, "", &rev->diffopt);
 285         }
 286         else
 287                 diff_root_tree_sha1(commit->tree->object.sha1,
 288                                     "", &rev->diffopt);
 289
 290         /* Export the referenced blobs, and remember the marks. */
 291         for (i = 0; i < diff_queued_diff.nr; i++)
 292                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
 293                         handle_object(diff_queued_diff.queue[i]->two->sha1);
 294
 295         mark_next_object(&commit->object);
 296         if (!is_encoding_utf8(encoding))
 297                 reencoded = reencode_string(message, "UTF-8", encoding);
 298         if (!commit->parents)
 299                 printf("reset %s\n", (const char*)commit->util);
 300         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
 301                (const char *)commit->util, last_idnum,
 302                (int)(author_end - author), author,
 303                (int)(committer_end - committer), committer,
 304                (unsigned)(reencoded
 305                           ? strlen(reencoded) : message
 306                           ? strlen(message) : 0),
 307                reencoded ? reencoded : message ? message : "");
 308         free(reencoded);
 309
 310         for (i = 0, p = commit->parents; p; p = p->next) {
 311                 int mark = get_object_mark(&p->item->object);
 312                 if (!mark)
 313                         continue;
 314                 if (i == 0)
 315                         printf("from :%d\n", mark);
 316                 else
 317                         printf("merge :%d\n", mark);
 318                 i++;
 319         }
 320
 321         log_tree_diff_flush(rev);
 322         rev->diffopt.output_format = saved_output_format;
 323
 324         printf("\n");
 325
 326         show_progress();
 327 }
 328
 329 static void handle_tail(struct object_array *commits, struct rev_info *revs)
 330 {
 331         struct commit *commit;
 332         while (commits->nr) {
 333                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
 334                 if (has_unshown_parent(commit))
 335                         return;
 336                 handle_commit(commit, revs);
 337                 commits->nr--;
 338         }
 339 }
 340
 341 static void handle_tag(const char *name, struct tag *tag)
 342 {
 343         unsigned long size;
 344         enum object_type type;
 345         char *buf;
 346         const char *tagger, *tagger_end, *message;
 347         size_t message_size = 0;
 348         struct object *tagged;
 349         int tagged_mark;
 350         struct commit *p;
 351
 352         /* Trees have no identifer in fast-export output, thus we have no way
 353          * to output tags of trees, tags of tags of trees, etc.  Simply omit
 354          * such tags.
 355          */
 356         tagged = tag->tagged;
 357         while (tagged->type == OBJ_TAG) {
 358                 tagged = ((struct tag *)tagged)->tagged;
 359         }
 360         if (tagged->type == OBJ_TREE) {
 361                 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
 362                         sha1_to_hex(tag->object.sha1));
 363                 return;
 364         }
 365
 366         buf = read_sha1_file(tag->object.sha1, &type, &size);
 367         if (!buf)
 368                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
 369         message = memmem(buf, size, "\n\n", 2);
 370         if (message) {
 371                 message += 2;
 372                 message_size = strlen(message);
 373         }
 374         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
 375         if (!tagger) {
 376                 if (fake_missing_tagger)
 377                         tagger = "tagger Unspecified Tagger "
 378                                 "<unspecified-tagger> 0 +0000";
 379                 else
 380                         tagger = "";
 381                 tagger_end = tagger + strlen(tagger);
 382         } else {
 383                 tagger++;
 384                 tagger_end = strchrnul(tagger, '\n');
 385         }
 386
 387         /* handle signed tags */
 388         if (message) {
 389                 const char *signature = strstr(message,
 390                                                "\n-----BEGIN PGP SIGNATURE-----\n");
 391                 if (signature)
 392                         switch(signed_tag_mode) {
 393                         case ABORT:
 394                                 die ("Encountered signed tag %s; use "
 395                                      "--signed-tag=<mode> to handle it.",
 396                                      sha1_to_hex(tag->object.sha1));
 397                         case WARN:
 398                                 warning ("Exporting signed tag %s",
 399                                          sha1_to_hex(tag->object.sha1));
 400                                 /* fallthru */
 401                         case VERBATIM:
 402                                 break;
 403                         case STRIP:
 404                                 message_size = signature + 1 - message;
 405                                 break;
 406                         }
 407         }
 408
 409         /* handle tag->tagged having been filtered out due to paths specified */
 410         tagged = tag->tagged;
 411         tagged_mark = get_object_mark(tagged);
 412         if (!tagged_mark) {
 413                 switch(tag_of_filtered_mode) {
 414                 case ABORT:
 415                         die ("Tag %s tags unexported object; use "
 416                              "--tag-of-filtered-object=<mode> to handle it.",
 417                              sha1_to_hex(tag->object.sha1));
 418                 case DROP:
 419                         /* Ignore this tag altogether */
 420                         return;
 421                 case REWRITE:
 422                         if (tagged->type != OBJ_COMMIT) {
 423                                 die ("Tag %s tags unexported %s!",
 424                                      sha1_to_hex(tag->object.sha1),
 425                                      typename(tagged->type));
 426                         }
 427                         p = (struct commit *)tagged;
 428                         for (;;) {
 429                                 if (p->parents && p->parents->next)
 430                                         break;
 431                                 if (p->object.flags & UNINTERESTING)
 432                                         break;
 433                                 if (!(p->object.flags & TREESAME))
 434                                         break;
 435                                 if (!p->parents)
 436                                         die ("Can't find replacement commit for tag %s\n",
 437                                              sha1_to_hex(tag->object.sha1));
 438                                 p = p->parents->item;
 439                         }
 440                         tagged_mark = get_object_mark(&p->object);
 441                 }
 442         }
 443
 444         if (!prefixcmp(name, "refs/tags/"))
 445                 name += 10;
 446         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
 447                name, tagged_mark,
 448                (int)(tagger_end - tagger), tagger,
 449                tagger == tagger_end ? "" : "\n",
 450                (int)message_size, (int)message_size, message ? message : "");
 451 }
 452
 453 static void get_tags_and_duplicates(struct object_array *pending,
 454                                     struct string_list *extra_refs)
 455 {
 456         struct tag *tag;
 457         int i;
 458
 459         for (i = 0; i < pending->nr; i++) {
 460                 struct object_array_entry *e = pending->objects + i;
 461                 unsigned char sha1[20];
 462                 struct commit *commit = commit;
 463                 char *full_name;
 464
 465                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
 466                         continue;
 467
 468                 switch (e->item->type) {
 469                 case OBJ_COMMIT:
 470                         commit = (struct commit *)e->item;
 471                         break;
 472                 case OBJ_TAG:
 473                         tag = (struct tag *)e->item;
 474
 475                         /* handle nested tags */
 476                         while (tag && tag->object.type == OBJ_TAG) {
 477                                 parse_object(tag->object.sha1);
 478                                 string_list_append(full_name, extra_refs)->util = tag;
 479                                 tag = (struct tag *)tag->tagged;
 480                         }
 481                         if (!tag)
 482                                 die ("Tag %s points nowhere?", e->name);
 483                         switch(tag->object.type) {
 484                         case OBJ_COMMIT:
 485                                 commit = (struct commit *)tag;
 486                                 break;
 487                         case OBJ_BLOB:
 488                                 handle_object(tag->object.sha1);
 489                                 continue;
 490                         default: /* OBJ_TAG (nested tags) is already handled */
 491                                 warning("Tag points to object of unexpected type %s, skipping.",
 492                                         typename(tag->object.type));
 493                                 continue;
 494                         }
 495                         break;
 496                 default:
 497                         warning("%s: Unexpected object of type %s, skipping.",
 498                                 e->name,
 499                                 typename(e->item->type));
 500                         continue;
 501                 }
 502                 if (commit->util)
 503                         /* more than one name for the same object */
 504                         string_list_append(full_name, extra_refs)->util = commit;
 505                 else
 506                         commit->util = full_name;
 507         }
 508 }
 509
 510 static void handle_tags_and_duplicates(struct string_list *extra_refs)
 511 {
 512         struct commit *commit;
 513         int i;
 514
 515         for (i = extra_refs->nr - 1; i >= 0; i--) {
 516                 const char *name = extra_refs->items[i].string;
 517                 struct object *object = extra_refs->items[i].util;
 518                 switch (object->type) {
 519                 case OBJ_TAG:
 520                         handle_tag(name, (struct tag *)object);
 521                         break;
 522                 case OBJ_COMMIT:
 523                         /* create refs pointing to already seen commits */
 524                         commit = (struct commit *)object;
 525                         printf("reset %s\nfrom :%d\n\n", name,
 526                                get_object_mark(&commit->object));
 527                         show_progress();
 528                         break;
 529                 }
 530         }
 531 }
 532
 533 static void export_marks(char *file)
 534 {
 535         unsigned int i;
 536         uint32_t mark;
 537         struct object_decoration *deco = idnums.hash;
 538         FILE *f;
 539         int e = 0;
 540
 541         f = fopen(file, "w");
 542         if (!f)
 543                 die_errno("Unable to open marks file %s for writing.", file);
 544
 545         for (i = 0; i < idnums.size; i++) {
 546                 if (deco->base && deco->base->type == 1) {
 547                         mark = ptr_to_mark(deco->decoration);
 548                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
 549                                 sha1_to_hex(deco->base->sha1)) < 0) {
 550                             e = 1;
 551                             break;
 552                         }
 553                 }
 554                 deco++;
 555         }
 556
 557         e |= ferror(f);
 558         e |= fclose(f);
 559         if (e)
 560                 error("Unable to write marks file %s.", file);
 561 }
 562
 563 static void import_marks(char *input_file)
 564 {
 565         char line[512];
 566         FILE *f = fopen(input_file, "r");
 567         if (!f)
 568                 die_errno("cannot read '%s'", input_file);
 569
 570         while (fgets(line, sizeof(line), f)) {
 571                 uint32_t mark;
 572                 char *line_end, *mark_end;
 573                 unsigned char sha1[20];
 574                 struct object *object;
 575
 576                 line_end = strchr(line, '\n');
 577                 if (line[0] != ':' || !line_end)
 578                         die("corrupt mark line: %s", line);
 579                 *line_end = '\0';
 580
 581                 mark = strtoumax(line + 1, &mark_end, 10);
 582                 if (!mark || mark_end == line + 1
 583                         || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
 584                         die("corrupt mark line: %s", line);
 585
 586                 object = parse_object(sha1);
 587                 if (!object)
 588                         die ("Could not read blob %s", sha1_to_hex(sha1));
 589
 590                 if (object->flags & SHOWN)
 591                         error("Object %s already has a mark", sha1);
 592
 593                 mark_object(object, mark);
 594                 if (last_idnum < mark)
 595                         last_idnum = mark;
 596
 597                 object->flags |= SHOWN;
 598         }
 599         fclose(f);
 600 }
 601
 602 int cmd_fast_export(int argc, const char **argv, const char *prefix)
 603 {
 604         struct rev_info revs;
 605         struct object_array commits = { 0, 0, NULL };
 606         struct string_list extra_refs = { NULL, 0, 0, 0 };
 607         struct commit *commit;
 608         char *export_filename = NULL, *import_filename = NULL;
 609         struct option options[] = {
 610                 OPT_INTEGER(0, "progress", &progress,
 611                             "show progress after <n> objects"),
 612                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
 613                              "select handling of signed tags",
 614                              parse_opt_signed_tag_mode),
 615                 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, "mode",
 616                              "select handling of tags that tag filtered objects",
 617                              parse_opt_tag_of_filtered_mode),
 618                 OPT_STRING(0, "export-marks", &export_filename, "FILE",
 619                              "Dump marks to this file"),
 620                 OPT_STRING(0, "import-marks", &import_filename, "FILE",
 621                              "Import marks from this file"),
 622                 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
 623                              "Fake a tagger when tags lack one"),
 624                 { OPTION_NEGBIT, 0, "data", &no_data, NULL,
 625                         "Skip output of blob data",
 626                         PARSE_OPT_NOARG | PARSE_OPT_NEGHELP, NULL, 1 },
 627                 OPT_END()
 628         };
 629
 630         if (argc == 1)
 631                 usage_with_options (fast_export_usage, options);
 632
 633         /* we handle encodings */
 634         git_config(git_default_config, NULL);
 635
 636         init_revisions(&revs, prefix);
 637         revs.topo_order = 1;
 638         revs.show_source = 1;
 639         revs.rewrite_parents = 1;
 640         argc = setup_revisions(argc, argv, &revs, NULL);
 641         argc = parse_options(argc, argv, prefix, options, fast_export_usage, 0);
 642         if (argc > 1)
 643                 usage_with_options (fast_export_usage, options);
 644
 645         if (import_filename)
 646                 import_marks(import_filename);
 647
 648         get_tags_and_duplicates(&revs.pending, &extra_refs);
 649
 650         if (prepare_revision_walk(&revs))
 651                 die("revision walk setup failed");
 652         revs.diffopt.format_callback = show_filemodify;
 653         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
 654         while ((commit = get_revision(&revs))) {
 655                 if (has_unshown_parent(commit)) {
 656                         add_object_array(&commit->object, NULL, &commits);
 657                 }
 658                 else {
 659                         handle_commit(commit, &revs);
 660                         handle_tail(&commits, &revs);
 661                 }
 662         }
 663
 664         handle_tags_and_duplicates(&extra_refs);
 665
 666         if (export_filename)
 667                 export_marks(export_filename);
 668
 669         return 0;
 670 }