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 static int full_tree;
  31
  32 static int parse_opt_signed_tag_mode(const struct option *opt,
  33                                      const char *arg, int unset)
  34 {
  35         if (unset || !strcmp(arg, "abort"))
  36                 signed_tag_mode = ABORT;
  37         else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
  38                 signed_tag_mode = VERBATIM;
  39         else if (!strcmp(arg, "warn"))
  40                 signed_tag_mode = WARN;
  41         else if (!strcmp(arg, "strip"))
  42                 signed_tag_mode = STRIP;
  43         else
  44                 return error("Unknown signed-tag mode: %s", arg);
  45         return 0;
  46 }
  47
  48 static int parse_opt_tag_of_filtered_mode(const struct option *opt,
  49                                           const char *arg, int unset)
  50 {
  51         if (unset || !strcmp(arg, "abort"))
  52                 tag_of_filtered_mode = ABORT;
  53         else if (!strcmp(arg, "drop"))
  54                 tag_of_filtered_mode = DROP;
  55         else if (!strcmp(arg, "rewrite"))
  56                 tag_of_filtered_mode = REWRITE;
  57         else
  58                 return error("Unknown tag-of-filtered mode: %s", arg);
  59         return 0;
  60 }
  61
  62 static struct decoration idnums;
  63 static uint32_t last_idnum;
  64
  65 static int has_unshown_parent(struct commit *commit)
  66 {
  67         struct commit_list *parent;
  68
  69         for (parent = commit->parents; parent; parent = parent->next)
  70                 if (!(parent->item->object.flags & SHOWN) &&
  71                     !(parent->item->object.flags & UNINTERESTING))
  72                         return 1;
  73         return 0;
  74 }
  75
  76 /* Since intptr_t is C99, we do not use it here */
  77 static inline uint32_t *mark_to_ptr(uint32_t mark)
  78 {
  79         return ((uint32_t *)NULL) + mark;
  80 }
  81
  82 static inline uint32_t ptr_to_mark(void * mark)
  83 {
  84         return (uint32_t *)mark - (uint32_t *)NULL;
  85 }
  86
  87 static inline void mark_object(struct object *object, uint32_t mark)
  88 {
  89         add_decoration(&idnums, object, mark_to_ptr(mark));
  90 }
  91
  92 static inline void mark_next_object(struct object *object)
  93 {
  94         mark_object(object, ++last_idnum);
  95 }
  96
  97 static int get_object_mark(struct object *object)
  98 {
  99         void *decoration = lookup_decoration(&idnums, object);
 100         if (!decoration)
 101                 return 0;
 102         return ptr_to_mark(decoration);
 103 }
 104
 105 static void show_progress(void)
 106 {
 107         static int counter = 0;
 108         if (!progress)
 109                 return;
 110         if ((++counter % progress) == 0)
 111                 printf("progress %d objects\n", counter);
 112 }
 113
 114 static void handle_object(const unsigned char *sha1)
 115 {
 116         unsigned long size;
 117         enum object_type type;
 118         char *buf;
 119         struct object *object;
 120
 121         if (no_data)
 122                 return;
 123
 124         if (is_null_sha1(sha1))
 125                 return;
 126
 127         object = parse_object(sha1);
 128         if (!object)
 129                 die ("Could not read blob %s", sha1_to_hex(sha1));
 130
 131         if (object->flags & SHOWN)
 132                 return;
 133
 134         buf = read_sha1_file(sha1, &type, &size);
 135         if (!buf)
 136                 die ("Could not read blob %s", sha1_to_hex(sha1));
 137
 138         mark_next_object(object);
 139
 140         printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
 141         if (size && fwrite(buf, size, 1, stdout) != 1)
 142                 die_errno ("Could not write blob '%s'", sha1_to_hex(sha1));
 143         printf("\n");
 144
 145         show_progress();
 146
 147         object->flags |= SHOWN;
 148         free(buf);
 149 }
 150
 151 static int depth_first(const void *a_, const void *b_)
 152 {
 153         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
 154         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
 155         const char *name_a, *name_b;
 156         int len_a, len_b, len;
 157         int cmp;
 158
 159         name_a = a->one ? a->one->path : a->two->path;
 160         name_b = b->one ? b->one->path : b->two->path;
 161
 162         len_a = strlen(name_a);
 163         len_b = strlen(name_b);
 164         len = (len_a < len_b) ? len_a : len_b;
 165
 166         /* strcmp will sort 'd' before 'd/e', we want 'd/e' before 'd' */
 167         cmp = memcmp(name_a, name_b, len);
 168         if (cmp)
 169                 return cmp;
 170         return (len_b - len_a);
 171 }
 172
 173 static void show_filemodify(struct diff_queue_struct *q,
 174                             struct diff_options *options, void *data)
 175 {
 176         int i;
 177
 178         /*
 179          * Handle files below a directory first, in case they are all deleted
 180          * and the directory changes to a file or symlink.
 181          */
 182         qsort(q->queue, q->nr, sizeof(q->queue[0]), depth_first);
 183
 184         for (i = 0; i < q->nr; i++) {
 185                 struct diff_filespec *ospec = q->queue[i]->one;
 186                 struct diff_filespec *spec = q->queue[i]->two;
 187
 188                 switch (q->queue[i]->status) {
 189                 case DIFF_STATUS_DELETED:
 190                         printf("D %s\n", spec->path);
 191                         break;
 192
 193                 case DIFF_STATUS_COPIED:
 194                 case DIFF_STATUS_RENAMED:
 195                         printf("%c \"%s\" \"%s\"\n", q->queue[i]->status,
 196                                ospec->path, spec->path);
 197
 198                         if (!hashcmp(ospec->sha1, spec->sha1) &&
 199                             ospec->mode == spec->mode)
 200                                 break;
 201                         /* fallthrough */
 202
 203                 case DIFF_STATUS_TYPE_CHANGED:
 204                 case DIFF_STATUS_MODIFIED:
 205                 case DIFF_STATUS_ADDED:
 206                         /*
 207                          * Links refer to objects in another repositories;
 208                          * output the SHA-1 verbatim.
 209                          */
 210                         if (no_data || S_ISGITLINK(spec->mode))
 211                                 printf("M %06o %s %s\n", spec->mode,
 212                                        sha1_to_hex(spec->sha1), spec->path);
 213                         else {
 214                                 struct object *object = lookup_object(spec->sha1);
 215                                 printf("M %06o :%d %s\n", spec->mode,
 216                                        get_object_mark(object), spec->path);
 217                         }
 218                         break;
 219
 220                 default:
 221                         die("Unexpected comparison status '%c' for %s, %s",
 222                                 q->queue[i]->status,
 223                                 ospec->path ? ospec->path : "none",
 224                                 spec->path ? spec->path : "none");
 225                 }
 226         }
 227 }
 228
 229 static const char *find_encoding(const char *begin, const char *end)
 230 {
 231         const char *needle = "\nencoding ";
 232         char *bol, *eol;
 233
 234         bol = memmem(begin, end ? end - begin : strlen(begin),
 235                      needle, strlen(needle));
 236         if (!bol)
 237                 return git_commit_encoding;
 238         bol += strlen(needle);
 239         eol = strchrnul(bol, '\n');
 240         *eol = '\0';
 241         return bol;
 242 }
 243
 244 static void handle_commit(struct commit *commit, struct rev_info *rev)
 245 {
 246         int saved_output_format = rev->diffopt.output_format;
 247         const char *author, *author_end, *committer, *committer_end;
 248         const char *encoding, *message;
 249         char *reencoded = NULL;
 250         struct commit_list *p;
 251         int i;
 252
 253         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
 254
 255         parse_commit(commit);
 256         author = strstr(commit->buffer, "\nauthor ");
 257         if (!author)
 258                 die ("Could not find author in commit %s",
 259                      sha1_to_hex(commit->object.sha1));
 260         author++;
 261         author_end = strchrnul(author, '\n');
 262         committer = strstr(author_end, "\ncommitter ");
 263         if (!committer)
 264                 die ("Could not find committer in commit %s",
 265                      sha1_to_hex(commit->object.sha1));
 266         committer++;
 267         committer_end = strchrnul(committer, '\n');
 268         message = strstr(committer_end, "\n\n");
 269         encoding = find_encoding(committer_end, message);
 270         if (message)
 271                 message += 2;
 272
 273         if (commit->parents &&
 274             get_object_mark(&commit->parents->item->object) != 0 &&
 275             !full_tree) {
 276                 parse_commit(commit->parents->item);
 277                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
 278                                commit->tree->object.sha1, "", &rev->diffopt);
 279         }
 280         else
 281                 diff_root_tree_sha1(commit->tree->object.sha1,
 282                                     "", &rev->diffopt);
 283
 284         /* Export the referenced blobs, and remember the marks. */
 285         for (i = 0; i < diff_queued_diff.nr; i++)
 286                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
 287                         handle_object(diff_queued_diff.queue[i]->two->sha1);
 288
 289         mark_next_object(&commit->object);
 290         if (!is_encoding_utf8(encoding))
 291                 reencoded = reencode_string(message, "UTF-8", encoding);
 292         if (!commit->parents)
 293                 printf("reset %s\n", (const char*)commit->util);
 294         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
 295                (const char *)commit->util, last_idnum,
 296                (int)(author_end - author), author,
 297                (int)(committer_end - committer), committer,
 298                (unsigned)(reencoded
 299                           ? strlen(reencoded) : message
 300                           ? strlen(message) : 0),
 301                reencoded ? reencoded : message ? message : "");
 302         free(reencoded);
 303
 304         for (i = 0, p = commit->parents; p; p = p->next) {
 305                 int mark = get_object_mark(&p->item->object);
 306                 if (!mark)
 307                         continue;
 308                 if (i == 0)
 309                         printf("from :%d\n", mark);
 310                 else
 311                         printf("merge :%d\n", mark);
 312                 i++;
 313         }
 314
 315         if (full_tree)
 316                 printf("deleteall\n");
 317         log_tree_diff_flush(rev);
 318         rev->diffopt.output_format = saved_output_format;
 319
 320         printf("\n");
 321
 322         show_progress();
 323 }
 324
 325 static void handle_tail(struct object_array *commits, struct rev_info *revs)
 326 {
 327         struct commit *commit;
 328         while (commits->nr) {
 329                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
 330                 if (has_unshown_parent(commit))
 331                         return;
 332                 handle_commit(commit, revs);
 333                 commits->nr--;
 334         }
 335 }
 336
 337 static void handle_tag(const char *name, struct tag *tag)
 338 {
 339         unsigned long size;
 340         enum object_type type;
 341         char *buf;
 342         const char *tagger, *tagger_end, *message;
 343         size_t message_size = 0;
 344         struct object *tagged;
 345         int tagged_mark;
 346         struct commit *p;
 347
 348         /* Trees have no identifer in fast-export output, thus we have no way
 349          * to output tags of trees, tags of tags of trees, etc.  Simply omit
 350          * such tags.
 351          */
 352         tagged = tag->tagged;
 353         while (tagged->type == OBJ_TAG) {
 354                 tagged = ((struct tag *)tagged)->tagged;
 355         }
 356         if (tagged->type == OBJ_TREE) {
 357                 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
 358                         sha1_to_hex(tag->object.sha1));
 359                 return;
 360         }
 361
 362         buf = read_sha1_file(tag->object.sha1, &type, &size);
 363         if (!buf)
 364                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
 365         message = memmem(buf, size, "\n\n", 2);
 366         if (message) {
 367                 message += 2;
 368                 message_size = strlen(message);
 369         }
 370         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
 371         if (!tagger) {
 372                 if (fake_missing_tagger)
 373                         tagger = "tagger Unspecified Tagger "
 374                                 "<unspecified-tagger> 0 +0000";
 375                 else
 376                         tagger = "";
 377                 tagger_end = tagger + strlen(tagger);
 378         } else {
 379                 tagger++;
 380                 tagger_end = strchrnul(tagger, '\n');
 381         }
 382
 383         /* handle signed tags */
 384         if (message) {
 385                 const char *signature = strstr(message,
 386                                                "\n-----BEGIN PGP SIGNATURE-----\n");
 387                 if (signature)
 388                         switch(signed_tag_mode) {
 389                         case ABORT:
 390                                 die ("Encountered signed tag %s; use "
 391                                      "--signed-tag=<mode> to handle it.",
 392                                      sha1_to_hex(tag->object.sha1));
 393                         case WARN:
 394                                 warning ("Exporting signed tag %s",
 395                                          sha1_to_hex(tag->object.sha1));
 396                                 /* fallthru */
 397                         case VERBATIM:
 398                                 break;
 399                         case STRIP:
 400                                 message_size = signature + 1 - message;
 401                                 break;
 402                         }
 403         }
 404
 405         /* handle tag->tagged having been filtered out due to paths specified */
 406         tagged = tag->tagged;
 407         tagged_mark = get_object_mark(tagged);
 408         if (!tagged_mark) {
 409                 switch(tag_of_filtered_mode) {
 410                 case ABORT:
 411                         die ("Tag %s tags unexported object; use "
 412                              "--tag-of-filtered-object=<mode> to handle it.",
 413                              sha1_to_hex(tag->object.sha1));
 414                 case DROP:
 415                         /* Ignore this tag altogether */
 416                         return;
 417                 case REWRITE:
 418                         if (tagged->type != OBJ_COMMIT) {
 419                                 die ("Tag %s tags unexported %s!",
 420                                      sha1_to_hex(tag->object.sha1),
 421                                      typename(tagged->type));
 422                         }
 423                         p = (struct commit *)tagged;
 424                         for (;;) {
 425                                 if (p->parents && p->parents->next)
 426                                         break;
 427                                 if (p->object.flags & UNINTERESTING)
 428                                         break;
 429                                 if (!(p->object.flags & TREESAME))
 430                                         break;
 431                                 if (!p->parents)
 432                                         die ("Can't find replacement commit for tag %s\n",
 433                                              sha1_to_hex(tag->object.sha1));
 434                                 p = p->parents->item;
 435                         }
 436                         tagged_mark = get_object_mark(&p->object);
 437                 }
 438         }
 439
 440         if (!prefixcmp(name, "refs/tags/"))
 441                 name += 10;
 442         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
 443                name, tagged_mark,
 444                (int)(tagger_end - tagger), tagger,
 445                tagger == tagger_end ? "" : "\n",
 446                (int)message_size, (int)message_size, message ? message : "");
 447 }
 448
 449 static void get_tags_and_duplicates(struct object_array *pending,
 450                                     struct string_list *extra_refs)
 451 {
 452         struct tag *tag;
 453         int i;
 454
 455         for (i = 0; i < pending->nr; i++) {
 456                 struct object_array_entry *e = pending->objects + i;
 457                 unsigned char sha1[20];
 458                 struct commit *commit = commit;
 459                 char *full_name;
 460
 461                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
 462                         continue;
 463
 464                 switch (e->item->type) {
 465                 case OBJ_COMMIT:
 466                         commit = (struct commit *)e->item;
 467                         break;
 468                 case OBJ_TAG:
 469                         tag = (struct tag *)e->item;
 470
 471                         /* handle nested tags */
 472                         while (tag && tag->object.type == OBJ_TAG) {
 473                                 parse_object(tag->object.sha1);
 474                                 string_list_append(extra_refs, full_name)->util = tag;
 475                                 tag = (struct tag *)tag->tagged;
 476                         }
 477                         if (!tag)
 478                                 die ("Tag %s points nowhere?", e->name);
 479                         switch(tag->object.type) {
 480                         case OBJ_COMMIT:
 481                                 commit = (struct commit *)tag;
 482                                 break;
 483                         case OBJ_BLOB:
 484                                 handle_object(tag->object.sha1);
 485                                 continue;
 486                         default: /* OBJ_TAG (nested tags) is already handled */
 487                                 warning("Tag points to object of unexpected type %s, skipping.",
 488                                         typename(tag->object.type));
 489                                 continue;
 490                         }
 491                         break;
 492                 default:
 493                         warning("%s: Unexpected object of type %s, skipping.",
 494                                 e->name,
 495                                 typename(e->item->type));
 496                         continue;
 497                 }
 498                 if (commit->util)
 499                         /* more than one name for the same object */
 500                         string_list_append(extra_refs, full_name)->util = commit;
 501                 else
 502                         commit->util = full_name;
 503         }
 504 }
 505
 506 static void handle_tags_and_duplicates(struct string_list *extra_refs)
 507 {
 508         struct commit *commit;
 509         int i;
 510
 511         for (i = extra_refs->nr - 1; i >= 0; i--) {
 512                 const char *name = extra_refs->items[i].string;
 513                 struct object *object = extra_refs->items[i].util;
 514                 switch (object->type) {
 515                 case OBJ_TAG:
 516                         handle_tag(name, (struct tag *)object);
 517                         break;
 518                 case OBJ_COMMIT:
 519                         /* create refs pointing to already seen commits */
 520                         commit = (struct commit *)object;
 521                         printf("reset %s\nfrom :%d\n\n", name,
 522                                get_object_mark(&commit->object));
 523                         show_progress();
 524                         break;
 525                 }
 526         }
 527 }
 528
 529 static void export_marks(char *file)
 530 {
 531         unsigned int i;
 532         uint32_t mark;
 533         struct object_decoration *deco = idnums.hash;
 534         FILE *f;
 535         int e = 0;
 536
 537         f = fopen(file, "w");
 538         if (!f)
 539                 die_errno("Unable to open marks file %s for writing.", file);
 540
 541         for (i = 0; i < idnums.size; i++) {
 542                 if (deco->base && deco->base->type == 1) {
 543                         mark = ptr_to_mark(deco->decoration);
 544                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
 545                                 sha1_to_hex(deco->base->sha1)) < 0) {
 546                             e = 1;
 547                             break;
 548                         }
 549                 }
 550                 deco++;
 551         }
 552
 553         e |= ferror(f);
 554         e |= fclose(f);
 555         if (e)
 556                 error("Unable to write marks file %s.", file);
 557 }
 558
 559 static void import_marks(char *input_file)
 560 {
 561         char line[512];
 562         FILE *f = fopen(input_file, "r");
 563         if (!f)
 564                 die_errno("cannot read '%s'", input_file);
 565
 566         while (fgets(line, sizeof(line), f)) {
 567                 uint32_t mark;
 568                 char *line_end, *mark_end;
 569                 unsigned char sha1[20];
 570                 struct object *object;
 571
 572                 line_end = strchr(line, '\n');
 573                 if (line[0] != ':' || !line_end)
 574                         die("corrupt mark line: %s", line);
 575                 *line_end = '\0';
 576
 577                 mark = strtoumax(line + 1, &mark_end, 10);
 578                 if (!mark || mark_end == line + 1
 579                         || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
 580                         die("corrupt mark line: %s", line);
 581
 582                 object = parse_object(sha1);
 583                 if (!object)
 584                         die ("Could not read blob %s", sha1_to_hex(sha1));
 585
 586                 if (object->flags & SHOWN)
 587                         error("Object %s already has a mark", sha1);
 588
 589                 mark_object(object, mark);
 590                 if (last_idnum < mark)
 591                         last_idnum = mark;
 592
 593                 object->flags |= SHOWN;
 594         }
 595         fclose(f);
 596 }
 597
 598 int cmd_fast_export(int argc, const char **argv, const char *prefix)
 599 {
 600         struct rev_info revs;
 601         struct object_array commits = OBJECT_ARRAY_INIT;
 602         struct string_list extra_refs = STRING_LIST_INIT_NODUP;
 603         struct commit *commit;
 604         char *export_filename = NULL, *import_filename = NULL;
 605         struct option options[] = {
 606                 OPT_INTEGER(0, "progress", &progress,
 607                             "show progress after <n> objects"),
 608                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
 609                              "select handling of signed tags",
 610                              parse_opt_signed_tag_mode),
 611                 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, "mode",
 612                              "select handling of tags that tag filtered objects",
 613                              parse_opt_tag_of_filtered_mode),
 614                 OPT_STRING(0, "export-marks", &export_filename, "FILE",
 615                              "Dump marks to this file"),
 616                 OPT_STRING(0, "import-marks", &import_filename, "FILE",
 617                              "Import marks from this file"),
 618                 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
 619                              "Fake a tagger when tags lack one"),
 620                 OPT_BOOLEAN(0, "full-tree", &full_tree,
 621                              "Output full tree for each commit"),
 622                 { OPTION_NEGBIT, 0, "data", &no_data, NULL,
 623                         "Skip output of blob data",
 624                         PARSE_OPT_NOARG | PARSE_OPT_NEGHELP, NULL, 1 },
 625                 OPT_END()
 626         };
 627
 628         if (argc == 1)
 629                 usage_with_options (fast_export_usage, options);
 630
 631         /* we handle encodings */
 632         git_config(git_default_config, NULL);
 633
 634         init_revisions(&revs, prefix);
 635         revs.topo_order = 1;
 636         revs.show_source = 1;
 637         revs.rewrite_parents = 1;
 638         argc = setup_revisions(argc, argv, &revs, NULL);
 639         argc = parse_options(argc, argv, prefix, options, fast_export_usage, 0);
 640         if (argc > 1)
 641                 usage_with_options (fast_export_usage, options);
 642
 643         if (import_filename)
 644                 import_marks(import_filename);
 645
 646         if (import_filename && revs.prune_data)
 647                 full_tree = 1;
 648
 649         get_tags_and_duplicates(&revs.pending, &extra_refs);
 650
 651         if (prepare_revision_walk(&revs))
 652                 die("revision walk setup failed");
 653         revs.diffopt.format_callback = show_filemodify;
 654         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
 655         while ((commit = get_revision(&revs))) {
 656                 if (has_unshown_parent(commit)) {
 657                         add_object_array(&commit->object, NULL, &commits);
 658                 }
 659                 else {
 660                         handle_commit(commit, &revs);
 661                         handle_tail(&commits, &revs);
 662                 }
 663         }
 664
 665         handle_tags_and_duplicates(&extra_refs);
 666
 667         if (export_filename)
 668                 export_marks(export_filename);
 669
 670         return 0;
 671 }