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