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 { VERBATIM, WARN, STRIP, ABORT } signed_tag_mode = ABORT;
  27 static int fake_missing_tagger;
  28
  29 static int parse_opt_signed_tag_mode(const struct option *opt,
  30                                      const char *arg, int unset)
  31 {
  32         if (unset || !strcmp(arg, "abort"))
  33                 signed_tag_mode = ABORT;
  34         else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
  35                 signed_tag_mode = VERBATIM;
  36         else if (!strcmp(arg, "warn"))
  37                 signed_tag_mode = WARN;
  38         else if (!strcmp(arg, "strip"))
  39                 signed_tag_mode = STRIP;
  40         else
  41                 return error("Unknown signed-tag mode: %s", arg);
  42         return 0;
  43 }
  44
  45 static struct decoration idnums;
  46 static uint32_t last_idnum;
  47
  48 static int has_unshown_parent(struct commit *commit)
  49 {
  50         struct commit_list *parent;
  51
  52         for (parent = commit->parents; parent; parent = parent->next)
  53                 if (!(parent->item->object.flags & SHOWN) &&
  54                     !(parent->item->object.flags & UNINTERESTING))
  55                         return 1;
  56         return 0;
  57 }
  58
  59 /* Since intptr_t is C99, we do not use it here */
  60 static inline uint32_t *mark_to_ptr(uint32_t mark)
  61 {
  62         return ((uint32_t *)NULL) + mark;
  63 }
  64
  65 static inline uint32_t ptr_to_mark(void * mark)
  66 {
  67         return (uint32_t *)mark - (uint32_t *)NULL;
  68 }
  69
  70 static inline void mark_object(struct object *object, uint32_t mark)
  71 {
  72         add_decoration(&idnums, object, mark_to_ptr(mark));
  73 }
  74
  75 static inline void mark_next_object(struct object *object)
  76 {
  77         mark_object(object, ++last_idnum);
  78 }
  79
  80 static int get_object_mark(struct object *object)
  81 {
  82         void *decoration = lookup_decoration(&idnums, object);
  83         if (!decoration)
  84                 return 0;
  85         return ptr_to_mark(decoration);
  86 }
  87
  88 static void show_progress(void)
  89 {
  90         static int counter = 0;
  91         if (!progress)
  92                 return;
  93         if ((++counter % progress) == 0)
  94                 printf("progress %d objects\n", counter);
  95 }
  96
  97 static void handle_object(const unsigned char *sha1)
  98 {
  99         unsigned long size;
 100         enum object_type type;
 101         char *buf;
 102         struct object *object;
 103
 104         if (is_null_sha1(sha1))
 105                 return;
 106
 107         object = parse_object(sha1);
 108         if (!object)
 109                 die ("Could not read blob %s", sha1_to_hex(sha1));
 110
 111         if (object->flags & SHOWN)
 112                 return;
 113
 114         buf = read_sha1_file(sha1, &type, &size);
 115         if (!buf)
 116                 die ("Could not read blob %s", sha1_to_hex(sha1));
 117
 118         mark_next_object(object);
 119
 120         printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
 121         if (size && fwrite(buf, size, 1, stdout) != 1)
 122                 die_errno ("Could not write blob '%s'", sha1_to_hex(sha1));
 123         printf("\n");
 124
 125         show_progress();
 126
 127         object->flags |= SHOWN;
 128         free(buf);
 129 }
 130
 131 static void show_filemodify(struct diff_queue_struct *q,
 132                             struct diff_options *options, void *data)
 133 {
 134         int i;
 135         for (i = 0; i < q->nr; i++) {
 136                 struct diff_filespec *ospec = q->queue[i]->one;
 137                 struct diff_filespec *spec = q->queue[i]->two;
 138
 139                 switch (q->queue[i]->status) {
 140                 case DIFF_STATUS_DELETED:
 141                         printf("D %s\n", spec->path);
 142                         break;
 143
 144                 case DIFF_STATUS_COPIED:
 145                 case DIFF_STATUS_RENAMED:
 146                         printf("%c \"%s\" \"%s\"\n", q->queue[i]->status,
 147                                ospec->path, spec->path);
 148
 149                         if (!hashcmp(ospec->sha1, spec->sha1) &&
 150                             ospec->mode == spec->mode)
 151                                 break;
 152                         /* fallthrough */
 153
 154                 case DIFF_STATUS_TYPE_CHANGED:
 155                 case DIFF_STATUS_MODIFIED:
 156                 case DIFF_STATUS_ADDED:
 157                         /*
 158                          * Links refer to objects in another repositories;
 159                          * output the SHA-1 verbatim.
 160                          */
 161                         if (S_ISGITLINK(spec->mode))
 162                                 printf("M %06o %s %s\n", spec->mode,
 163                                        sha1_to_hex(spec->sha1), spec->path);
 164                         else {
 165                                 struct object *object = lookup_object(spec->sha1);
 166                                 printf("M %06o :%d %s\n", spec->mode,
 167                                        get_object_mark(object), spec->path);
 168                         }
 169                         break;
 170
 171                 default:
 172                         die("Unexpected comparison status '%c' for %s, %s",
 173                                 q->queue[i]->status,
 174                                 ospec->path ? ospec->path : "none",
 175                                 spec->path ? spec->path : "none");
 176                 }
 177         }
 178 }
 179
 180 static const char *find_encoding(const char *begin, const char *end)
 181 {
 182         const char *needle = "\nencoding ";
 183         char *bol, *eol;
 184
 185         bol = memmem(begin, end ? end - begin : strlen(begin),
 186                      needle, strlen(needle));
 187         if (!bol)
 188                 return git_commit_encoding;
 189         bol += strlen(needle);
 190         eol = strchrnul(bol, '\n');
 191         *eol = '\0';
 192         return bol;
 193 }
 194
 195 static void handle_commit(struct commit *commit, struct rev_info *rev)
 196 {
 197         int saved_output_format = rev->diffopt.output_format;
 198         const char *author, *author_end, *committer, *committer_end;
 199         const char *encoding, *message;
 200         char *reencoded = NULL;
 201         struct commit_list *p;
 202         int i;
 203
 204         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
 205
 206         parse_commit(commit);
 207         author = strstr(commit->buffer, "\nauthor ");
 208         if (!author)
 209                 die ("Could not find author in commit %s",
 210                      sha1_to_hex(commit->object.sha1));
 211         author++;
 212         author_end = strchrnul(author, '\n');
 213         committer = strstr(author_end, "\ncommitter ");
 214         if (!committer)
 215                 die ("Could not find committer in commit %s",
 216                      sha1_to_hex(commit->object.sha1));
 217         committer++;
 218         committer_end = strchrnul(committer, '\n');
 219         message = strstr(committer_end, "\n\n");
 220         encoding = find_encoding(committer_end, message);
 221         if (message)
 222                 message += 2;
 223
 224         if (commit->parents &&
 225             get_object_mark(&commit->parents->item->object) != 0) {
 226                 parse_commit(commit->parents->item);
 227                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
 228                                commit->tree->object.sha1, "", &rev->diffopt);
 229         }
 230         else
 231                 diff_root_tree_sha1(commit->tree->object.sha1,
 232                                     "", &rev->diffopt);
 233
 234         /* Export the referenced blobs, and remember the marks. */
 235         for (i = 0; i < diff_queued_diff.nr; i++)
 236                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
 237                         handle_object(diff_queued_diff.queue[i]->two->sha1);
 238
 239         mark_next_object(&commit->object);
 240         if (!is_encoding_utf8(encoding))
 241                 reencoded = reencode_string(message, "UTF-8", encoding);
 242         if (!commit->parents)
 243                 printf("reset %s\n", (const char*)commit->util);
 244         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
 245                (const char *)commit->util, last_idnum,
 246                (int)(author_end - author), author,
 247                (int)(committer_end - committer), committer,
 248                (unsigned)(reencoded
 249                           ? strlen(reencoded) : message
 250                           ? strlen(message) : 0),
 251                reencoded ? reencoded : message ? message : "");
 252         free(reencoded);
 253
 254         for (i = 0, p = commit->parents; p; p = p->next) {
 255                 int mark = get_object_mark(&p->item->object);
 256                 if (!mark)
 257                         continue;
 258                 if (i == 0)
 259                         printf("from :%d\n", mark);
 260                 else
 261                         printf("merge :%d\n", mark);
 262                 i++;
 263         }
 264
 265         log_tree_diff_flush(rev);
 266         rev->diffopt.output_format = saved_output_format;
 267
 268         printf("\n");
 269
 270         show_progress();
 271 }
 272
 273 static void handle_tail(struct object_array *commits, struct rev_info *revs)
 274 {
 275         struct commit *commit;
 276         while (commits->nr) {
 277                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
 278                 if (has_unshown_parent(commit))
 279                         return;
 280                 handle_commit(commit, revs);
 281                 commits->nr--;
 282         }
 283 }
 284
 285 static void handle_tag(const char *name, struct tag *tag)
 286 {
 287         unsigned long size;
 288         enum object_type type;
 289         char *buf;
 290         const char *tagger, *tagger_end, *message;
 291         size_t message_size = 0;
 292
 293         buf = read_sha1_file(tag->object.sha1, &type, &size);
 294         if (!buf)
 295                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
 296         message = memmem(buf, size, "\n\n", 2);
 297         if (message) {
 298                 message += 2;
 299                 message_size = strlen(message);
 300         }
 301         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
 302         if (!tagger) {
 303                 if (fake_missing_tagger)
 304                         tagger = "tagger Unspecified Tagger "
 305                                 "<unspecified-tagger> 0 +0000";
 306                 else
 307                         tagger = "";
 308                 tagger_end = tagger + strlen(tagger);
 309         } else {
 310                 tagger++;
 311                 tagger_end = strchrnul(tagger, '\n');
 312         }
 313
 314         /* handle signed tags */
 315         if (message) {
 316                 const char *signature = strstr(message,
 317                                                "\n-----BEGIN PGP SIGNATURE-----\n");
 318                 if (signature)
 319                         switch(signed_tag_mode) {
 320                         case ABORT:
 321                                 die ("Encountered signed tag %s; use "
 322                                      "--signed-tag=<mode> to handle it.",
 323                                      sha1_to_hex(tag->object.sha1));
 324                         case WARN:
 325                                 warning ("Exporting signed tag %s",
 326                                          sha1_to_hex(tag->object.sha1));
 327                                 /* fallthru */
 328                         case VERBATIM:
 329                                 break;
 330                         case STRIP:
 331                                 message_size = signature + 1 - message;
 332                                 break;
 333                         }
 334         }
 335
 336         if (!prefixcmp(name, "refs/tags/"))
 337                 name += 10;
 338         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
 339                name, get_object_mark(tag->tagged),
 340                (int)(tagger_end - tagger), tagger,
 341                tagger == tagger_end ? "" : "\n",
 342                (int)message_size, (int)message_size, message ? message : "");
 343 }
 344
 345 static void get_tags_and_duplicates(struct object_array *pending,
 346                                     struct string_list *extra_refs)
 347 {
 348         struct tag *tag;
 349         int i;
 350
 351         for (i = 0; i < pending->nr; i++) {
 352                 struct object_array_entry *e = pending->objects + i;
 353                 unsigned char sha1[20];
 354                 struct commit *commit = commit;
 355                 char *full_name;
 356
 357                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
 358                         continue;
 359
 360                 switch (e->item->type) {
 361                 case OBJ_COMMIT:
 362                         commit = (struct commit *)e->item;
 363                         break;
 364                 case OBJ_TAG:
 365                         tag = (struct tag *)e->item;
 366
 367                         /* handle nested tags */
 368                         while (tag && tag->object.type == OBJ_TAG) {
 369                                 parse_object(tag->object.sha1);
 370                                 string_list_append(full_name, extra_refs)->util = tag;
 371                                 tag = (struct tag *)tag->tagged;
 372                         }
 373                         if (!tag)
 374                                 die ("Tag %s points nowhere?", e->name);
 375                         switch(tag->object.type) {
 376                         case OBJ_COMMIT:
 377                                 commit = (struct commit *)tag;
 378                                 break;
 379                         case OBJ_BLOB:
 380                                 handle_object(tag->object.sha1);
 381                                 continue;
 382                         default: /* OBJ_TAG (nested tags) is already handled */
 383                                 warning("Tag points to object of unexpected type %s, skipping.",
 384                                         typename(tag->object.type));
 385                                 continue;
 386                         }
 387                         break;
 388                 default:
 389                         warning("%s: Unexpected object of type %s, skipping.",
 390                                 e->name,
 391                                 typename(e->item->type));
 392                         continue;
 393                 }
 394                 if (commit->util)
 395                         /* more than one name for the same object */
 396                         string_list_append(full_name, extra_refs)->util = commit;
 397                 else
 398                         commit->util = full_name;
 399         }
 400 }
 401
 402 static void handle_tags_and_duplicates(struct string_list *extra_refs)
 403 {
 404         struct commit *commit;
 405         int i;
 406
 407         for (i = extra_refs->nr - 1; i >= 0; i--) {
 408                 const char *name = extra_refs->items[i].string;
 409                 struct object *object = extra_refs->items[i].util;
 410                 switch (object->type) {
 411                 case OBJ_TAG:
 412                         handle_tag(name, (struct tag *)object);
 413                         break;
 414                 case OBJ_COMMIT:
 415                         /* create refs pointing to already seen commits */
 416                         commit = (struct commit *)object;
 417                         printf("reset %s\nfrom :%d\n\n", name,
 418                                get_object_mark(&commit->object));
 419                         show_progress();
 420                         break;
 421                 }
 422         }
 423 }
 424
 425 static void export_marks(char *file)
 426 {
 427         unsigned int i;
 428         uint32_t mark;
 429         struct object_decoration *deco = idnums.hash;
 430         FILE *f;
 431         int e = 0;
 432
 433         f = fopen(file, "w");
 434         if (!f)
 435                 error("Unable to open marks file %s for writing.", file);
 436
 437         for (i = 0; i < idnums.size; i++) {
 438                 if (deco->base && deco->base->type == 1) {
 439                         mark = ptr_to_mark(deco->decoration);
 440                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
 441                                 sha1_to_hex(deco->base->sha1)) < 0) {
 442                             e = 1;
 443                             break;
 444                         }
 445                 }
 446                 deco++;
 447         }
 448
 449         e |= ferror(f);
 450         e |= fclose(f);
 451         if (e)
 452                 error("Unable to write marks file %s.", file);
 453 }
 454
 455 static void import_marks(char *input_file)
 456 {
 457         char line[512];
 458         FILE *f = fopen(input_file, "r");
 459         if (!f)
 460                 die_errno("cannot read '%s'", input_file);
 461
 462         while (fgets(line, sizeof(line), f)) {
 463                 uint32_t mark;
 464                 char *line_end, *mark_end;
 465                 unsigned char sha1[20];
 466                 struct object *object;
 467
 468                 line_end = strchr(line, '\n');
 469                 if (line[0] != ':' || !line_end)
 470                         die("corrupt mark line: %s", line);
 471                 *line_end = '\0';
 472
 473                 mark = strtoumax(line + 1, &mark_end, 10);
 474                 if (!mark || mark_end == line + 1
 475                         || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
 476                         die("corrupt mark line: %s", line);
 477
 478                 object = parse_object(sha1);
 479                 if (!object)
 480                         die ("Could not read blob %s", sha1_to_hex(sha1));
 481
 482                 if (object->flags & SHOWN)
 483                         error("Object %s already has a mark", sha1);
 484
 485                 mark_object(object, mark);
 486                 if (last_idnum < mark)
 487                         last_idnum = mark;
 488
 489                 object->flags |= SHOWN;
 490         }
 491         fclose(f);
 492 }
 493
 494 int cmd_fast_export(int argc, const char **argv, const char *prefix)
 495 {
 496         struct rev_info revs;
 497         struct object_array commits = { 0, 0, NULL };
 498         struct string_list extra_refs = { NULL, 0, 0, 0 };
 499         struct commit *commit;
 500         char *export_filename = NULL, *import_filename = NULL;
 501         struct option options[] = {
 502                 OPT_INTEGER(0, "progress", &progress,
 503                             "show progress after <n> objects"),
 504                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
 505                              "select handling of signed tags",
 506                              parse_opt_signed_tag_mode),
 507                 OPT_STRING(0, "export-marks", &export_filename, "FILE",
 508                              "Dump marks to this file"),
 509                 OPT_STRING(0, "import-marks", &import_filename, "FILE",
 510                              "Import marks from this file"),
 511                 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
 512                              "Fake a tagger when tags lack one"),
 513                 OPT_END()
 514         };
 515
 516         if (argc == 1)
 517                 usage_with_options (fast_export_usage, options);
 518
 519         /* we handle encodings */
 520         git_config(git_default_config, NULL);
 521
 522         init_revisions(&revs, prefix);
 523         argc = setup_revisions(argc, argv, &revs, NULL);
 524         argc = parse_options(argc, argv, prefix, options, fast_export_usage, 0);
 525         if (argc > 1)
 526                 usage_with_options (fast_export_usage, options);
 527
 528         if (import_filename)
 529                 import_marks(import_filename);
 530
 531         get_tags_and_duplicates(&revs.pending, &extra_refs);
 532
 533         revs.topo_order = 1;
 534         if (prepare_revision_walk(&revs))
 535                 die("revision walk setup failed");
 536         revs.diffopt.format_callback = show_filemodify;
 537         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
 538         while ((commit = get_revision(&revs))) {
 539                 if (has_unshown_parent(commit)) {
 540                         struct commit_list *parent = commit->parents;
 541                         add_object_array(&commit->object, NULL, &commits);
 542                         for (; parent; parent = parent->next)
 543                                 if (!parent->item->util)
 544                                         parent->item->util = commit->util;
 545                 }
 546                 else {
 547                         handle_commit(commit, &revs);
 548                         handle_tail(&commits, &revs);
 549                 }
 550         }
 551
 552         handle_tags_and_duplicates(&extra_refs);
 553
 554         if (export_filename)
 555                 export_marks(export_filename);
 556
 557         return 0;
 558 }