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 void mark_object(struct object *object)
  60 {
  61         last_idnum++;
  62         add_decoration(&idnums, object, ((uint32_t *)NULL) + last_idnum);
  63 }
  64
  65 static int get_object_mark(struct object *object)
  66 {
  67         void *decoration = lookup_decoration(&idnums, object);
  68         if (!decoration)
  69                 return 0;
  70         return (uint32_t *)decoration - (uint32_t *)NULL;
  71 }
  72
  73 static void show_progress(void)
  74 {
  75         static int counter = 0;
  76         if (!progress)
  77                 return;
  78         if ((++counter % progress) == 0)
  79                 printf("progress %d objects\n", counter);
  80 }
  81
  82 static void handle_object(const unsigned char *sha1)
  83 {
  84         unsigned long size;
  85         enum object_type type;
  86         char *buf;
  87         struct object *object;
  88
  89         if (is_null_sha1(sha1))
  90                 return;
  91
  92         object = parse_object(sha1);
  93         if (!object)
  94                 die ("Could not read blob %s", sha1_to_hex(sha1));
  95
  96         if (object->flags & SHOWN)
  97                 return;
  98
  99         buf = read_sha1_file(sha1, &type, &size);
 100         if (!buf)
 101                 die ("Could not read blob %s", sha1_to_hex(sha1));
 102
 103         mark_object(object);
 104
 105         printf("blob\nmark :%d\ndata %lu\n", last_idnum, size);
 106         if (size && fwrite(buf, size, 1, stdout) != 1)
 107                 die ("Could not write blob %s", sha1_to_hex(sha1));
 108         printf("\n");
 109
 110         show_progress();
 111
 112         object->flags |= SHOWN;
 113         free(buf);
 114 }
 115
 116 static void show_filemodify(struct diff_queue_struct *q,
 117                             struct diff_options *options, void *data)
 118 {
 119         int i;
 120         for (i = 0; i < q->nr; i++) {
 121                 struct diff_filespec *spec = q->queue[i]->two;
 122                 if (is_null_sha1(spec->sha1))
 123                         printf("D %s\n", spec->path);
 124                 else {
 125                         struct object *object = lookup_object(spec->sha1);
 126                         printf("M %06o :%d %s\n", spec->mode,
 127                                get_object_mark(object), spec->path);
 128                 }
 129         }
 130 }
 131
 132 static const char *find_encoding(const char *begin, const char *end)
 133 {
 134         const char *needle = "\nencoding ";
 135         char *bol, *eol;
 136
 137         bol = memmem(begin, end ? end - begin : strlen(begin),
 138                      needle, strlen(needle));
 139         if (!bol)
 140                 return git_commit_encoding;
 141         bol += strlen(needle);
 142         eol = strchrnul(bol, '\n');
 143         *eol = '\0';
 144         return bol;
 145 }
 146
 147 static void handle_commit(struct commit *commit, struct rev_info *rev)
 148 {
 149         int saved_output_format = rev->diffopt.output_format;
 150         const char *author, *author_end, *committer, *committer_end;
 151         const char *encoding, *message;
 152         char *reencoded = NULL;
 153         struct commit_list *p;
 154         int i;
 155
 156         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
 157
 158         parse_commit(commit);
 159         author = strstr(commit->buffer, "\nauthor ");
 160         if (!author)
 161                 die ("Could not find author in commit %s",
 162                      sha1_to_hex(commit->object.sha1));
 163         author++;
 164         author_end = strchrnul(author, '\n');
 165         committer = strstr(author_end, "\ncommitter ");
 166         if (!committer)
 167                 die ("Could not find committer in commit %s",
 168                      sha1_to_hex(commit->object.sha1));
 169         committer++;
 170         committer_end = strchrnul(committer, '\n');
 171         message = strstr(committer_end, "\n\n");
 172         encoding = find_encoding(committer_end, message);
 173         if (message)
 174                 message += 2;
 175
 176         if (commit->parents) {
 177                 parse_commit(commit->parents->item);
 178                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
 179                                commit->tree->object.sha1, "", &rev->diffopt);
 180         }
 181         else
 182                 diff_root_tree_sha1(commit->tree->object.sha1,
 183                                     "", &rev->diffopt);
 184
 185         for (i = 0; i < diff_queued_diff.nr; i++)
 186                 handle_object(diff_queued_diff.queue[i]->two->sha1);
 187
 188         mark_object(&commit->object);
 189         if (!is_encoding_utf8(encoding))
 190                 reencoded = reencode_string(message, "UTF-8", encoding);
 191         printf("commit %s\nmark :%d\n%.*s\n%.*s\ndata %u\n%s",
 192                (const char *)commit->util, last_idnum,
 193                (int)(author_end - author), author,
 194                (int)(committer_end - committer), committer,
 195                (unsigned)(reencoded
 196                           ? strlen(reencoded) : message
 197                           ? strlen(message) : 0),
 198                reencoded ? reencoded : message ? message : "");
 199         free(reencoded);
 200
 201         for (i = 0, p = commit->parents; p; p = p->next) {
 202                 int mark = get_object_mark(&p->item->object);
 203                 if (!mark)
 204                         continue;
 205                 if (i == 0)
 206                         printf("from :%d\n", mark);
 207                 else
 208                         printf("merge :%d\n", mark);
 209                 i++;
 210         }
 211
 212         log_tree_diff_flush(rev);
 213         rev->diffopt.output_format = saved_output_format;
 214
 215         printf("\n");
 216
 217         show_progress();
 218 }
 219
 220 static void handle_tail(struct object_array *commits, struct rev_info *revs)
 221 {
 222         struct commit *commit;
 223         while (commits->nr) {
 224                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
 225                 if (has_unshown_parent(commit))
 226                         return;
 227                 handle_commit(commit, revs);
 228                 commits->nr--;
 229         }
 230 }
 231
 232 static void handle_tag(const char *name, struct tag *tag)
 233 {
 234         unsigned long size;
 235         enum object_type type;
 236         char *buf;
 237         const char *tagger, *tagger_end, *message;
 238         size_t message_size = 0;
 239
 240         buf = read_sha1_file(tag->object.sha1, &type, &size);
 241         if (!buf)
 242                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
 243         message = memmem(buf, size, "\n\n", 2);
 244         if (message) {
 245                 message += 2;
 246                 message_size = strlen(message);
 247         }
 248         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
 249         if (!tagger)
 250                 die ("No tagger for tag %s", sha1_to_hex(tag->object.sha1));
 251         tagger++;
 252         tagger_end = strchrnul(tagger, '\n');
 253
 254         /* handle signed tags */
 255         if (message) {
 256                 const char *signature = strstr(message,
 257                                                "\n-----BEGIN PGP SIGNATURE-----\n");
 258                 if (signature)
 259                         switch(signed_tag_mode) {
 260                         case ABORT:
 261                                 die ("Encountered signed tag %s; use "
 262                                      "--signed-tag=<mode> to handle it.",
 263                                      sha1_to_hex(tag->object.sha1));
 264                         case WARN:
 265                                 warning ("Exporting signed tag %s",
 266                                          sha1_to_hex(tag->object.sha1));
 267                                 /* fallthru */
 268                         case VERBATIM:
 269                                 break;
 270                         case STRIP:
 271                                 message_size = signature + 1 - message;
 272                                 break;
 273                         }
 274         }
 275
 276         if (!prefixcmp(name, "refs/tags/"))
 277                 name += 10;
 278         printf("tag %s\nfrom :%d\n%.*s\ndata %d\n%.*s\n",
 279                name, get_object_mark(tag->tagged),
 280                (int)(tagger_end - tagger), tagger,
 281                (int)message_size, (int)message_size, message ? message : "");
 282 }
 283
 284 static void get_tags_and_duplicates(struct object_array *pending,
 285                                     struct path_list *extra_refs)
 286 {
 287         struct tag *tag;
 288         int i;
 289
 290         for (i = 0; i < pending->nr; i++) {
 291                 struct object_array_entry *e = pending->objects + i;
 292                 unsigned char sha1[20];
 293                 struct commit *commit = commit;
 294                 char *full_name;
 295
 296                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
 297                         continue;
 298
 299                 switch (e->item->type) {
 300                 case OBJ_COMMIT:
 301                         commit = (struct commit *)e->item;
 302                         break;
 303                 case OBJ_TAG:
 304                         tag = (struct tag *)e->item;
 305                         while (tag && tag->object.type == OBJ_TAG) {
 306                                 path_list_insert(full_name, extra_refs)->util = tag;
 307                                 tag = (struct tag *)tag->tagged;
 308                         }
 309                         if (!tag)
 310                                 die ("Tag %s points nowhere?", e->name);
 311                         switch(tag->object.type) {
 312                         case OBJ_COMMIT:
 313                                 commit = (struct commit *)tag;
 314                                 break;
 315                         case OBJ_BLOB:
 316                                 handle_object(tag->object.sha1);
 317                                 continue;
 318                         }
 319                         break;
 320                 default:
 321                         die ("Unexpected object of type %s",
 322                              typename(e->item->type));
 323                 }
 324                 if (commit->util)
 325                         /* more than one name for the same object */
 326                         path_list_insert(full_name, extra_refs)->util = commit;
 327                 else
 328                         commit->util = full_name;
 329         }
 330 }
 331
 332 static void handle_tags_and_duplicates(struct path_list *extra_refs)
 333 {
 334         struct commit *commit;
 335         int i;
 336
 337         for (i = extra_refs->nr - 1; i >= 0; i--) {
 338                 const char *name = extra_refs->items[i].path;
 339                 struct object *object = extra_refs->items[i].util;
 340                 switch (object->type) {
 341                 case OBJ_TAG:
 342                         handle_tag(name, (struct tag *)object);
 343                         break;
 344                 case OBJ_COMMIT:
 345                         /* create refs pointing to already seen commits */
 346                         commit = (struct commit *)object;
 347                         printf("reset %s\nfrom :%d\n\n", name,
 348                                get_object_mark(&commit->object));
 349                         show_progress();
 350                         break;
 351                 }
 352         }
 353 }
 354
 355 int cmd_fast_export(int argc, const char **argv, const char *prefix)
 356 {
 357         struct rev_info revs;
 358         struct object_array commits = { 0, 0, NULL };
 359         struct path_list extra_refs = { NULL, 0, 0, 0 };
 360         struct commit *commit;
 361         struct option options[] = {
 362                 OPT_INTEGER(0, "progress", &progress,
 363                             "show progress after <n> objects"),
 364                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
 365                              "select handling of signed tags",
 366                              parse_opt_signed_tag_mode),
 367                 OPT_END()
 368         };
 369
 370         /* we handle encodings */
 371         git_config(git_default_config, NULL);
 372
 373         init_revisions(&revs, prefix);
 374         argc = setup_revisions(argc, argv, &revs, NULL);
 375         argc = parse_options(argc, argv, options, fast_export_usage, 0);
 376         if (argc > 1)
 377                 usage_with_options (fast_export_usage, options);
 378
 379         get_tags_and_duplicates(&revs.pending, &extra_refs);
 380
 381         if (prepare_revision_walk(&revs))
 382                 die("revision walk setup failed");
 383         revs.diffopt.format_callback = show_filemodify;
 384         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
 385         while ((commit = get_revision(&revs))) {
 386                 if (has_unshown_parent(commit)) {
 387                         struct commit_list *parent = commit->parents;
 388                         add_object_array(&commit->object, NULL, &commits);
 389                         for (; parent; parent = parent->next)
 390                                 if (!parent->item->util)
 391                                         parent->item->util = commit->util;
 392                 }
 393                 else {
 394                         handle_commit(commit, &revs);
 395                         handle_tail(&commits, &revs);
 396                 }
 397         }
 398
 399         handle_tags_and_duplicates(&extra_refs);
 400
 401         return 0;
 402 }