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