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