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