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