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         cmp = len_b - len_a;
 171         if (cmp)
 172                 return cmp;
 173         /*
 174          * Move 'R'ename entries last so that all references of the file
 175          * appear in the output before it is renamed (e.g., when a file
 176          * was copied and renamed in the same commit).
 177          */
 178         return (a->status == 'R') - (b->status == 'R');
 179 }
 180
 181 static void show_filemodify(struct diff_queue_struct *q,
 182                             struct diff_options *options, void *data)
 183 {
 184         int i;
 185
 186         /*
 187          * Handle files below a directory first, in case they are all deleted
 188          * and the directory changes to a file or symlink.
 189          */
 190         qsort(q->queue, q->nr, sizeof(q->queue[0]), depth_first);
 191
 192         for (i = 0; i < q->nr; i++) {
 193                 struct diff_filespec *ospec = q->queue[i]->one;
 194                 struct diff_filespec *spec = q->queue[i]->two;
 195
 196                 switch (q->queue[i]->status) {
 197                 case DIFF_STATUS_DELETED:
 198                         printf("D %s\n", spec->path);
 199                         break;
 200
 201                 case DIFF_STATUS_COPIED:
 202                 case DIFF_STATUS_RENAMED:
 203                         printf("%c \"%s\" \"%s\"\n", q->queue[i]->status,
 204                                ospec->path, spec->path);
 205
 206                         if (!hashcmp(ospec->sha1, spec->sha1) &&
 207                             ospec->mode == spec->mode)
 208                                 break;
 209                         /* fallthrough */
 210
 211                 case DIFF_STATUS_TYPE_CHANGED:
 212                 case DIFF_STATUS_MODIFIED:
 213                 case DIFF_STATUS_ADDED:
 214                         /*
 215                          * Links refer to objects in another repositories;
 216                          * output the SHA-1 verbatim.
 217                          */
 218                         if (no_data || S_ISGITLINK(spec->mode))
 219                                 printf("M %06o %s %s\n", spec->mode,
 220                                        sha1_to_hex(spec->sha1), spec->path);
 221                         else {
 222                                 struct object *object = lookup_object(spec->sha1);
 223                                 printf("M %06o :%d %s\n", spec->mode,
 224                                        get_object_mark(object), spec->path);
 225                         }
 226                         break;
 227
 228                 default:
 229                         die("Unexpected comparison status '%c' for %s, %s",
 230                                 q->queue[i]->status,
 231                                 ospec->path ? ospec->path : "none",
 232                                 spec->path ? spec->path : "none");
 233                 }
 234         }
 235 }
 236
 237 static const char *find_encoding(const char *begin, const char *end)
 238 {
 239         const char *needle = "\nencoding ";
 240         char *bol, *eol;
 241
 242         bol = memmem(begin, end ? end - begin : strlen(begin),
 243                      needle, strlen(needle));
 244         if (!bol)
 245                 return git_commit_encoding;
 246         bol += strlen(needle);
 247         eol = strchrnul(bol, '\n');
 248         *eol = '\0';
 249         return bol;
 250 }
 251
 252 static void handle_commit(struct commit *commit, struct rev_info *rev)
 253 {
 254         int saved_output_format = rev->diffopt.output_format;
 255         const char *author, *author_end, *committer, *committer_end;
 256         const char *encoding, *message;
 257         char *reencoded = NULL;
 258         struct commit_list *p;
 259         int i;
 260
 261         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
 262
 263         parse_commit(commit);
 264         author = strstr(commit->buffer, "\nauthor ");
 265         if (!author)
 266                 die ("Could not find author in commit %s",
 267                      sha1_to_hex(commit->object.sha1));
 268         author++;
 269         author_end = strchrnul(author, '\n');
 270         committer = strstr(author_end, "\ncommitter ");
 271         if (!committer)
 272                 die ("Could not find committer in commit %s",
 273                      sha1_to_hex(commit->object.sha1));
 274         committer++;
 275         committer_end = strchrnul(committer, '\n');
 276         message = strstr(committer_end, "\n\n");
 277         encoding = find_encoding(committer_end, message);
 278         if (message)
 279                 message += 2;
 280
 281         if (commit->parents &&
 282             get_object_mark(&commit->parents->item->object) != 0 &&
 283             !full_tree) {
 284                 parse_commit(commit->parents->item);
 285                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
 286                                commit->tree->object.sha1, "", &rev->diffopt);
 287         }
 288         else
 289                 diff_root_tree_sha1(commit->tree->object.sha1,
 290                                     "", &rev->diffopt);
 291
 292         /* Export the referenced blobs, and remember the marks. */
 293         for (i = 0; i < diff_queued_diff.nr; i++)
 294                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
 295                         handle_object(diff_queued_diff.queue[i]->two->sha1);
 296
 297         mark_next_object(&commit->object);
 298         if (!is_encoding_utf8(encoding))
 299                 reencoded = reencode_string(message, "UTF-8", encoding);
 300         if (!commit->parents)
 301                 printf("reset %s\n", (const char*)commit->util);
 302         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
 303                (const char *)commit->util, last_idnum,
 304                (int)(author_end - author), author,
 305                (int)(committer_end - committer), committer,
 306                (unsigned)(reencoded
 307                           ? strlen(reencoded) : message
 308                           ? strlen(message) : 0),
 309                reencoded ? reencoded : message ? message : "");
 310         free(reencoded);
 311
 312         for (i = 0, p = commit->parents; p; p = p->next) {
 313                 int mark = get_object_mark(&p->item->object);
 314                 if (!mark)
 315                         continue;
 316                 if (i == 0)
 317                         printf("from :%d\n", mark);
 318                 else
 319                         printf("merge :%d\n", mark);
 320                 i++;
 321         }
 322
 323         if (full_tree)
 324                 printf("deleteall\n");
 325         log_tree_diff_flush(rev);
 326         rev->diffopt.output_format = saved_output_format;
 327
 328         printf("\n");
 329
 330         show_progress();
 331 }
 332
 333 static void handle_tail(struct object_array *commits, struct rev_info *revs)
 334 {
 335         struct commit *commit;
 336         while (commits->nr) {
 337                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
 338                 if (has_unshown_parent(commit))
 339                         return;
 340                 handle_commit(commit, revs);
 341                 commits->nr--;
 342         }
 343 }
 344
 345 static void handle_tag(const char *name, struct tag *tag)
 346 {
 347         unsigned long size;
 348         enum object_type type;
 349         char *buf;
 350         const char *tagger, *tagger_end, *message;
 351         size_t message_size = 0;
 352         struct object *tagged;
 353         int tagged_mark;
 354         struct commit *p;
 355
 356         /* Trees have no identifer in fast-export output, thus we have no way
 357          * to output tags of trees, tags of tags of trees, etc.  Simply omit
 358          * such tags.
 359          */
 360         tagged = tag->tagged;
 361         while (tagged->type == OBJ_TAG) {
 362                 tagged = ((struct tag *)tagged)->tagged;
 363         }
 364         if (tagged->type == OBJ_TREE) {
 365                 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
 366                         sha1_to_hex(tag->object.sha1));
 367                 return;
 368         }
 369
 370         buf = read_sha1_file(tag->object.sha1, &type, &size);
 371         if (!buf)
 372                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
 373         message = memmem(buf, size, "\n\n", 2);
 374         if (message) {
 375                 message += 2;
 376                 message_size = strlen(message);
 377         }
 378         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
 379         if (!tagger) {
 380                 if (fake_missing_tagger)
 381                         tagger = "tagger Unspecified Tagger "
 382                                 "<unspecified-tagger> 0 +0000";
 383                 else
 384                         tagger = "";
 385                 tagger_end = tagger + strlen(tagger);
 386         } else {
 387                 tagger++;
 388                 tagger_end = strchrnul(tagger, '\n');
 389         }
 390
 391         /* handle signed tags */
 392         if (message) {
 393                 const char *signature = strstr(message,
 394                                                "\n-----BEGIN PGP SIGNATURE-----\n");
 395                 if (signature)
 396                         switch(signed_tag_mode) {
 397                         case ABORT:
 398                                 die ("Encountered signed tag %s; use "
 399                                      "--signed-tag=<mode> to handle it.",
 400                                      sha1_to_hex(tag->object.sha1));
 401                         case WARN:
 402                                 warning ("Exporting signed tag %s",
 403                                          sha1_to_hex(tag->object.sha1));
 404                                 /* fallthru */
 405                         case VERBATIM:
 406                                 break;
 407                         case STRIP:
 408                                 message_size = signature + 1 - message;
 409                                 break;
 410                         }
 411         }
 412
 413         /* handle tag->tagged having been filtered out due to paths specified */
 414         tagged = tag->tagged;
 415         tagged_mark = get_object_mark(tagged);
 416         if (!tagged_mark) {
 417                 switch(tag_of_filtered_mode) {
 418                 case ABORT:
 419                         die ("Tag %s tags unexported object; use "
 420                              "--tag-of-filtered-object=<mode> to handle it.",
 421                              sha1_to_hex(tag->object.sha1));
 422                 case DROP:
 423                         /* Ignore this tag altogether */
 424                         return;
 425                 case REWRITE:
 426                         if (tagged->type != OBJ_COMMIT) {
 427                                 die ("Tag %s tags unexported %s!",
 428                                      sha1_to_hex(tag->object.sha1),
 429                                      typename(tagged->type));
 430                         }
 431                         p = (struct commit *)tagged;
 432                         for (;;) {
 433                                 if (p->parents && p->parents->next)
 434                                         break;
 435                                 if (p->object.flags & UNINTERESTING)
 436                                         break;
 437                                 if (!(p->object.flags & TREESAME))
 438                                         break;
 439                                 if (!p->parents)
 440                                         die ("Can't find replacement commit for tag %s\n",
 441                                              sha1_to_hex(tag->object.sha1));
 442                                 p = p->parents->item;
 443                         }
 444                         tagged_mark = get_object_mark(&p->object);
 445                 }
 446         }
 447
 448         if (!prefixcmp(name, "refs/tags/"))
 449                 name += 10;
 450         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
 451                name, tagged_mark,
 452                (int)(tagger_end - tagger), tagger,
 453                tagger == tagger_end ? "" : "\n",
 454                (int)message_size, (int)message_size, message ? message : "");
 455 }
 456
 457 static void get_tags_and_duplicates(struct object_array *pending,
 458                                     struct string_list *extra_refs)
 459 {
 460         struct tag *tag;
 461         int i;
 462
 463         for (i = 0; i < pending->nr; i++) {
 464                 struct object_array_entry *e = pending->objects + i;
 465                 unsigned char sha1[20];
 466                 struct commit *commit = commit;
 467                 char *full_name;
 468
 469                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
 470                         continue;
 471
 472                 switch (e->item->type) {
 473                 case OBJ_COMMIT:
 474                         commit = (struct commit *)e->item;
 475                         break;
 476                 case OBJ_TAG:
 477                         tag = (struct tag *)e->item;
 478
 479                         /* handle nested tags */
 480                         while (tag && tag->object.type == OBJ_TAG) {
 481                                 parse_object(tag->object.sha1);
 482                                 string_list_append(extra_refs, full_name)->util = tag;
 483                                 tag = (struct tag *)tag->tagged;
 484                         }
 485                         if (!tag)
 486                                 die ("Tag %s points nowhere?", e->name);
 487                         switch(tag->object.type) {
 488                         case OBJ_COMMIT:
 489                                 commit = (struct commit *)tag;
 490                                 break;
 491                         case OBJ_BLOB:
 492                                 handle_object(tag->object.sha1);
 493                                 continue;
 494                         default: /* OBJ_TAG (nested tags) is already handled */
 495                                 warning("Tag points to object of unexpected type %s, skipping.",
 496                                         typename(tag->object.type));
 497                                 continue;
 498                         }
 499                         break;
 500                 default:
 501                         warning("%s: Unexpected object of type %s, skipping.",
 502                                 e->name,
 503                                 typename(e->item->type));
 504                         continue;
 505                 }
 506                 if (commit->util)
 507                         /* more than one name for the same object */
 508                         string_list_append(extra_refs, full_name)->util = commit;
 509                 else
 510                         commit->util = full_name;
 511         }
 512 }
 513
 514 static void handle_tags_and_duplicates(struct string_list *extra_refs)
 515 {
 516         struct commit *commit;
 517         int i;
 518
 519         for (i = extra_refs->nr - 1; i >= 0; i--) {
 520                 const char *name = extra_refs->items[i].string;
 521                 struct object *object = extra_refs->items[i].util;
 522                 switch (object->type) {
 523                 case OBJ_TAG:
 524                         handle_tag(name, (struct tag *)object);
 525                         break;
 526                 case OBJ_COMMIT:
 527                         /* create refs pointing to already seen commits */
 528                         commit = (struct commit *)object;
 529                         printf("reset %s\nfrom :%d\n\n", name,
 530                                get_object_mark(&commit->object));
 531                         show_progress();
 532                         break;
 533                 }
 534         }
 535 }
 536
 537 static void export_marks(char *file)
 538 {
 539         unsigned int i;
 540         uint32_t mark;
 541         struct object_decoration *deco = idnums.hash;
 542         FILE *f;
 543         int e = 0;
 544
 545         f = fopen(file, "w");
 546         if (!f)
 547                 die_errno("Unable to open marks file %s for writing.", file);
 548
 549         for (i = 0; i < idnums.size; i++) {
 550                 if (deco->base && deco->base->type == 1) {
 551                         mark = ptr_to_mark(deco->decoration);
 552                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
 553                                 sha1_to_hex(deco->base->sha1)) < 0) {
 554                             e = 1;
 555                             break;
 556                         }
 557                 }
 558                 deco++;
 559         }
 560
 561         e |= ferror(f);
 562         e |= fclose(f);
 563         if (e)
 564                 error("Unable to write marks file %s.", file);
 565 }
 566
 567 static void import_marks(char *input_file)
 568 {
 569         char line[512];
 570         FILE *f = fopen(input_file, "r");
 571         if (!f)
 572                 die_errno("cannot read '%s'", input_file);
 573
 574         while (fgets(line, sizeof(line), f)) {
 575                 uint32_t mark;
 576                 char *line_end, *mark_end;
 577                 unsigned char sha1[20];
 578                 struct object *object;
 579
 580                 line_end = strchr(line, '\n');
 581                 if (line[0] != ':' || !line_end)
 582                         die("corrupt mark line: %s", line);
 583                 *line_end = '\0';
 584
 585                 mark = strtoumax(line + 1, &mark_end, 10);
 586                 if (!mark || mark_end == line + 1
 587                         || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
 588                         die("corrupt mark line: %s", line);
 589
 590                 object = parse_object(sha1);
 591                 if (!object)
 592                         die ("Could not read blob %s", sha1_to_hex(sha1));
 593
 594                 if (object->flags & SHOWN)
 595                         error("Object %s already has a mark", sha1);
 596
 597                 mark_object(object, mark);
 598                 if (last_idnum < mark)
 599                         last_idnum = mark;
 600
 601                 object->flags |= SHOWN;
 602         }
 603         fclose(f);
 604 }
 605
 606 int cmd_fast_export(int argc, const char **argv, const char *prefix)
 607 {
 608         struct rev_info revs;
 609         struct object_array commits = OBJECT_ARRAY_INIT;
 610         struct string_list extra_refs = STRING_LIST_INIT_NODUP;
 611         struct commit *commit;
 612         char *export_filename = NULL, *import_filename = NULL;
 613         struct option options[] = {
 614                 OPT_INTEGER(0, "progress", &progress,
 615                             "show progress after <n> objects"),
 616                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
 617                              "select handling of signed tags",
 618                              parse_opt_signed_tag_mode),
 619                 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, "mode",
 620                              "select handling of tags that tag filtered objects",
 621                              parse_opt_tag_of_filtered_mode),
 622                 OPT_STRING(0, "export-marks", &export_filename, "FILE",
 623                              "Dump marks to this file"),
 624                 OPT_STRING(0, "import-marks", &import_filename, "FILE",
 625                              "Import marks from this file"),
 626                 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
 627                              "Fake a tagger when tags lack one"),
 628                 OPT_BOOLEAN(0, "full-tree", &full_tree,
 629                              "Output full tree for each commit"),
 630                 { OPTION_NEGBIT, 0, "data", &no_data, NULL,
 631                         "Skip output of blob data",
 632                         PARSE_OPT_NOARG | PARSE_OPT_NEGHELP, NULL, 1 },
 633                 OPT_END()
 634         };
 635
 636         if (argc == 1)
 637                 usage_with_options (fast_export_usage, options);
 638
 639         /* we handle encodings */
 640         git_config(git_default_config, NULL);
 641
 642         init_revisions(&revs, prefix);
 643         revs.topo_order = 1;
 644         revs.show_source = 1;
 645         revs.rewrite_parents = 1;
 646         argc = setup_revisions(argc, argv, &revs, NULL);
 647         argc = parse_options(argc, argv, prefix, options, fast_export_usage, 0);
 648         if (argc > 1)
 649                 usage_with_options (fast_export_usage, options);
 650
 651         if (import_filename)
 652                 import_marks(import_filename);
 653
 654         if (import_filename && revs.prune_data.nr)
 655                 full_tree = 1;
 656
 657         get_tags_and_duplicates(&revs.pending, &extra_refs);
 658
 659         if (prepare_revision_walk(&revs))
 660                 die("revision walk setup failed");
 661         revs.diffopt.format_callback = show_filemodify;
 662         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
 663         while ((commit = get_revision(&revs))) {
 664                 if (has_unshown_parent(commit)) {
 665                         add_object_array(&commit->object, NULL, &commits);
 666                 }
 667                 else {
 668                         handle_commit(commit, &revs);
 669                         handle_tail(&commits, &revs);
 670                 }
 671         }
 672
 673         handle_tags_and_duplicates(&extra_refs);
 674
 675         if (export_filename)
 676                 export_marks(export_filename);
 677
 678         return 0;
 679 }