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