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