builtin/cat-file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  */
   6 #define USE_THE_INDEX_VARIABLE
   7 #include "cache.h"
   8 #include "config.h"
   9 #include "builtin.h"
  10 #include "diff.h"
  11 #include "parse-options.h"
  12 #include "userdiff.h"
  13 #include "streaming.h"
  14 #include "tree-walk.h"
  15 #include "oid-array.h"
  16 #include "packfile.h"
  17 #include "object-store.h"
  18 #include "promisor-remote.h"
  19 #include "mailmap.h"
  20
  21 enum batch_mode {
  22         BATCH_MODE_CONTENTS,
  23         BATCH_MODE_INFO,
  24         BATCH_MODE_QUEUE_AND_DISPATCH,
  25 };
  26
  27 struct batch_options {
  28         int enabled;
  29         int follow_symlinks;
  30         enum batch_mode batch_mode;
  31         int buffer_output;
  32         int all_objects;
  33         int unordered;
  34         int transform_mode; /* may be 'w' or 'c' for --filters or --textconv */
  35         int nul_terminated;
  36         const char *format;
  37 };
  38
  39 static const char *force_path;
  40
  41 static struct string_list mailmap = STRING_LIST_INIT_NODUP;
  42 static int use_mailmap;
  43
  44 static char *replace_idents_using_mailmap(char *, size_t *);
  45
  46 static char *replace_idents_using_mailmap(char *object_buf, size_t *size)
  47 {
  48         struct strbuf sb = STRBUF_INIT;
  49         const char *headers[] = { "author ", "committer ", "tagger ", NULL };
  50
  51         strbuf_attach(&sb, object_buf, *size, *size + 1);
  52         apply_mailmap_to_header(&sb, headers, &mailmap);
  53         *size = sb.len;
  54         return strbuf_detach(&sb, NULL);
  55 }
  56
  57 static int filter_object(const char *path, unsigned mode,
  58                          const struct object_id *oid,
  59                          char **buf, unsigned long *size)
  60 {
  61         enum object_type type;
  62
  63         *buf = read_object_file(oid, &type, size);
  64         if (!*buf)
  65                 return error(_("cannot read object %s '%s'"),
  66                              oid_to_hex(oid), path);
  67         if ((type == OBJ_BLOB) && S_ISREG(mode)) {
  68                 struct strbuf strbuf = STRBUF_INIT;
  69                 struct checkout_metadata meta;
  70
  71                 init_checkout_metadata(&meta, NULL, NULL, oid);
  72                 if (convert_to_working_tree(&the_index, path, *buf, *size, &strbuf, &meta)) {
  73                         free(*buf);
  74                         *size = strbuf.len;
  75                         *buf = strbuf_detach(&strbuf, NULL);
  76                 }
  77         }
  78
  79         return 0;
  80 }
  81
  82 static int stream_blob(const struct object_id *oid)
  83 {
  84         if (stream_blob_to_fd(1, oid, NULL, 0))
  85                 die("unable to stream %s to stdout", oid_to_hex(oid));
  86         return 0;
  87 }
  88
  89 static int cat_one_file(int opt, const char *exp_type, const char *obj_name,
  90                         int unknown_type)
  91 {
  92         int ret;
  93         struct object_id oid;
  94         enum object_type type;
  95         char *buf;
  96         unsigned long size;
  97         struct object_context obj_context;
  98         struct object_info oi = OBJECT_INFO_INIT;
  99         struct strbuf sb = STRBUF_INIT;
 100         unsigned flags = OBJECT_INFO_LOOKUP_REPLACE;
 101         unsigned get_oid_flags = GET_OID_RECORD_PATH | GET_OID_ONLY_TO_DIE;
 102         const char *path = force_path;
 103         const int opt_cw = (opt == 'c' || opt == 'w');
 104         if (!path && opt_cw)
 105                 get_oid_flags |= GET_OID_REQUIRE_PATH;
 106
 107         if (unknown_type)
 108                 flags |= OBJECT_INFO_ALLOW_UNKNOWN_TYPE;
 109
 110         if (get_oid_with_context(the_repository, obj_name, get_oid_flags, &oid,
 111                                  &obj_context))
 112                 die("Not a valid object name %s", obj_name);
 113
 114         if (!path)
 115                 path = obj_context.path;
 116         if (obj_context.mode == S_IFINVALID)
 117                 obj_context.mode = 0100644;
 118
 119         buf = NULL;
 120         switch (opt) {
 121         case 't':
 122                 oi.type_name = &sb;
 123                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 124                         die("git cat-file: could not get object info");
 125                 if (sb.len) {
 126                         printf("%s\n", sb.buf);
 127                         strbuf_release(&sb);
 128                         ret = 0;
 129                         goto cleanup;
 130                 }
 131                 break;
 132
 133         case 's':
 134                 oi.sizep = &size;
 135                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 136                         die("git cat-file: could not get object info");
 137                 printf("%"PRIuMAX"\n", (uintmax_t)size);
 138                 ret = 0;
 139                 goto cleanup;
 140
 141         case 'e':
 142                 return !has_object_file(&oid);
 143
 144         case 'w':
 145
 146                 if (filter_object(path, obj_context.mode,
 147                                   &oid, &buf, &size)) {
 148                         ret = -1;
 149                         goto cleanup;
 150                 }
 151                 break;
 152
 153         case 'c':
 154                 if (textconv_object(the_repository, path, obj_context.mode,
 155                                     &oid, 1, &buf, &size))
 156                         break;
 157                 /* else fallthrough */
 158
 159         case 'p':
 160                 type = oid_object_info(the_repository, &oid, NULL);
 161                 if (type < 0)
 162                         die("Not a valid object name %s", obj_name);
 163
 164                 /* custom pretty-print here */
 165                 if (type == OBJ_TREE) {
 166                         const char *ls_args[3] = { NULL };
 167                         ls_args[0] =  "ls-tree";
 168                         ls_args[1] =  obj_name;
 169                         ret = cmd_ls_tree(2, ls_args, NULL);
 170                         goto cleanup;
 171                 }
 172
 173                 if (type == OBJ_BLOB) {
 174                         ret = stream_blob(&oid);
 175                         goto cleanup;
 176                 }
 177                 buf = read_object_file(&oid, &type, &size);
 178                 if (!buf)
 179                         die("Cannot read object %s", obj_name);
 180
 181                 if (use_mailmap) {
 182                         size_t s = size;
 183                         buf = replace_idents_using_mailmap(buf, &s);
 184                         size = cast_size_t_to_ulong(s);
 185                 }
 186
 187                 /* otherwise just spit out the data */
 188                 break;
 189
 190         case 0:
 191         {
 192                 enum object_type exp_type_id = type_from_string(exp_type);
 193
 194                 if (exp_type_id == OBJ_BLOB) {
 195                         struct object_id blob_oid;
 196                         if (oid_object_info(the_repository, &oid, NULL) == OBJ_TAG) {
 197                                 char *buffer = read_object_file(&oid, &type,
 198                                                                 &size);
 199                                 const char *target;
 200                                 if (!skip_prefix(buffer, "object ", &target) ||
 201                                     get_oid_hex(target, &blob_oid))
 202                                         die("%s not a valid tag", oid_to_hex(&oid));
 203                                 free(buffer);
 204                         } else
 205                                 oidcpy(&blob_oid, &oid);
 206
 207                         if (oid_object_info(the_repository, &blob_oid, NULL) == OBJ_BLOB) {
 208                                 ret = stream_blob(&blob_oid);
 209                                 goto cleanup;
 210                         }
 211                         /*
 212                          * we attempted to dereference a tag to a blob
 213                          * and failed; there may be new dereference
 214                          * mechanisms this code is not aware of.
 215                          * fall-back to the usual case.
 216                          */
 217                 }
 218                 buf = read_object_with_reference(the_repository, &oid,
 219                                                  exp_type_id, &size, NULL);
 220
 221                 if (use_mailmap) {
 222                         size_t s = size;
 223                         buf = replace_idents_using_mailmap(buf, &s);
 224                         size = cast_size_t_to_ulong(s);
 225                 }
 226                 break;
 227         }
 228         default:
 229                 die("git cat-file: unknown option: %s", exp_type);
 230         }
 231
 232         if (!buf)
 233                 die("git cat-file %s: bad file", obj_name);
 234
 235         write_or_die(1, buf, size);
 236         ret = 0;
 237 cleanup:
 238         free(buf);
 239         free(obj_context.path);
 240         return ret;
 241 }
 242
 243 struct expand_data {
 244         struct object_id oid;
 245         enum object_type type;
 246         unsigned long size;
 247         off_t disk_size;
 248         const char *rest;
 249         struct object_id delta_base_oid;
 250
 251         /*
 252          * If mark_query is true, we do not expand anything, but rather
 253          * just mark the object_info with items we wish to query.
 254          */
 255         int mark_query;
 256
 257         /*
 258          * Whether to split the input on whitespace before feeding it to
 259          * get_sha1; this is decided during the mark_query phase based on
 260          * whether we have a %(rest) token in our format.
 261          */
 262         int split_on_whitespace;
 263
 264         /*
 265          * After a mark_query run, this object_info is set up to be
 266          * passed to oid_object_info_extended. It will point to the data
 267          * elements above, so you can retrieve the response from there.
 268          */
 269         struct object_info info;
 270
 271         /*
 272          * This flag will be true if the requested batch format and options
 273          * don't require us to call oid_object_info, which can then be
 274          * optimized out.
 275          */
 276         unsigned skip_object_info : 1;
 277 };
 278
 279 static int is_atom(const char *atom, const char *s, int slen)
 280 {
 281         int alen = strlen(atom);
 282         return alen == slen && !memcmp(atom, s, alen);
 283 }
 284
 285 static void expand_atom(struct strbuf *sb, const char *atom, int len,
 286                         void *vdata)
 287 {
 288         struct expand_data *data = vdata;
 289
 290         if (is_atom("objectname", atom, len)) {
 291                 if (!data->mark_query)
 292                         strbuf_addstr(sb, oid_to_hex(&data->oid));
 293         } else if (is_atom("objecttype", atom, len)) {
 294                 if (data->mark_query)
 295                         data->info.typep = &data->type;
 296                 else
 297                         strbuf_addstr(sb, type_name(data->type));
 298         } else if (is_atom("objectsize", atom, len)) {
 299                 if (data->mark_query)
 300                         data->info.sizep = &data->size;
 301                 else
 302                         strbuf_addf(sb, "%"PRIuMAX , (uintmax_t)data->size);
 303         } else if (is_atom("objectsize:disk", atom, len)) {
 304                 if (data->mark_query)
 305                         data->info.disk_sizep = &data->disk_size;
 306                 else
 307                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 308         } else if (is_atom("rest", atom, len)) {
 309                 if (data->mark_query)
 310                         data->split_on_whitespace = 1;
 311                 else if (data->rest)
 312                         strbuf_addstr(sb, data->rest);
 313         } else if (is_atom("deltabase", atom, len)) {
 314                 if (data->mark_query)
 315                         data->info.delta_base_oid = &data->delta_base_oid;
 316                 else
 317                         strbuf_addstr(sb,
 318                                       oid_to_hex(&data->delta_base_oid));
 319         } else
 320                 die("unknown format element: %.*s", len, atom);
 321 }
 322
 323 static size_t expand_format(struct strbuf *sb, const char *start, void *data)
 324 {
 325         const char *end;
 326
 327         if (*start != '(')
 328                 return 0;
 329         end = strchr(start + 1, ')');
 330         if (!end)
 331                 die("format element '%s' does not end in ')'", start);
 332
 333         expand_atom(sb, start + 1, end - start - 1, data);
 334
 335         return end - start + 1;
 336 }
 337
 338 static void batch_write(struct batch_options *opt, const void *data, int len)
 339 {
 340         if (opt->buffer_output) {
 341                 if (fwrite(data, 1, len, stdout) != len)
 342                         die_errno("unable to write to stdout");
 343         } else
 344                 write_or_die(1, data, len);
 345 }
 346
 347 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
 348 {
 349         const struct object_id *oid = &data->oid;
 350
 351         assert(data->info.typep);
 352
 353         if (data->type == OBJ_BLOB) {
 354                 if (opt->buffer_output)
 355                         fflush(stdout);
 356                 if (opt->transform_mode) {
 357                         char *contents;
 358                         unsigned long size;
 359
 360                         if (!data->rest)
 361                                 die("missing path for '%s'", oid_to_hex(oid));
 362
 363                         if (opt->transform_mode == 'w') {
 364                                 if (filter_object(data->rest, 0100644, oid,
 365                                                   &contents, &size))
 366                                         die("could not convert '%s' %s",
 367                                             oid_to_hex(oid), data->rest);
 368                         } else if (opt->transform_mode == 'c') {
 369                                 enum object_type type;
 370                                 if (!textconv_object(the_repository,
 371                                                      data->rest, 0100644, oid,
 372                                                      1, &contents, &size))
 373                                         contents = read_object_file(oid,
 374                                                                     &type,
 375                                                                     &size);
 376                                 if (!contents)
 377                                         die("could not convert '%s' %s",
 378                                             oid_to_hex(oid), data->rest);
 379                         } else
 380                                 BUG("invalid transform_mode: %c", opt->transform_mode);
 381                         batch_write(opt, contents, size);
 382                         free(contents);
 383                 } else {
 384                         stream_blob(oid);
 385                 }
 386         }
 387         else {
 388                 enum object_type type;
 389                 unsigned long size;
 390                 void *contents;
 391
 392                 contents = read_object_file(oid, &type, &size);
 393
 394                 if (use_mailmap) {
 395                         size_t s = size;
 396                         contents = replace_idents_using_mailmap(contents, &s);
 397                         size = cast_size_t_to_ulong(s);
 398                 }
 399
 400                 if (!contents)
 401                         die("object %s disappeared", oid_to_hex(oid));
 402                 if (type != data->type)
 403                         die("object %s changed type!?", oid_to_hex(oid));
 404                 if (data->info.sizep && size != data->size && !use_mailmap)
 405                         die("object %s changed size!?", oid_to_hex(oid));
 406
 407                 batch_write(opt, contents, size);
 408                 free(contents);
 409         }
 410 }
 411
 412 static void print_default_format(struct strbuf *scratch, struct expand_data *data)
 413 {
 414         strbuf_addf(scratch, "%s %s %"PRIuMAX"\n", oid_to_hex(&data->oid),
 415                     type_name(data->type),
 416                     (uintmax_t)data->size);
 417 }
 418
 419 /*
 420  * If "pack" is non-NULL, then "offset" is the byte offset within the pack from
 421  * which the object may be accessed (though note that we may also rely on
 422  * data->oid, too). If "pack" is NULL, then offset is ignored.
 423  */
 424 static void batch_object_write(const char *obj_name,
 425                                struct strbuf *scratch,
 426                                struct batch_options *opt,
 427                                struct expand_data *data,
 428                                struct packed_git *pack,
 429                                off_t offset)
 430 {
 431         if (!data->skip_object_info) {
 432                 int ret;
 433
 434                 if (pack)
 435                         ret = packed_object_info(the_repository, pack, offset,
 436                                                  &data->info);
 437                 else
 438                         ret = oid_object_info_extended(the_repository,
 439                                                        &data->oid, &data->info,
 440                                                        OBJECT_INFO_LOOKUP_REPLACE);
 441                 if (ret < 0) {
 442                         printf("%s missing\n",
 443                                obj_name ? obj_name : oid_to_hex(&data->oid));
 444                         fflush(stdout);
 445                         return;
 446                 }
 447         }
 448
 449         strbuf_reset(scratch);
 450
 451         if (!opt->format) {
 452                 print_default_format(scratch, data);
 453         } else {
 454                 strbuf_expand(scratch, opt->format, expand_format, data);
 455                 strbuf_addch(scratch, '\n');
 456         }
 457
 458         batch_write(opt, scratch->buf, scratch->len);
 459
 460         if (opt->batch_mode == BATCH_MODE_CONTENTS) {
 461                 print_object_or_die(opt, data);
 462                 batch_write(opt, "\n", 1);
 463         }
 464 }
 465
 466 static void batch_one_object(const char *obj_name,
 467                              struct strbuf *scratch,
 468                              struct batch_options *opt,
 469                              struct expand_data *data)
 470 {
 471         struct object_context ctx;
 472         int flags = opt->follow_symlinks ? GET_OID_FOLLOW_SYMLINKS : 0;
 473         enum get_oid_result result;
 474
 475         result = get_oid_with_context(the_repository, obj_name,
 476                                       flags, &data->oid, &ctx);
 477         if (result != FOUND) {
 478                 switch (result) {
 479                 case MISSING_OBJECT:
 480                         printf("%s missing\n", obj_name);
 481                         break;
 482                 case SHORT_NAME_AMBIGUOUS:
 483                         printf("%s ambiguous\n", obj_name);
 484                         break;
 485                 case DANGLING_SYMLINK:
 486                         printf("dangling %"PRIuMAX"\n%s\n",
 487                                (uintmax_t)strlen(obj_name), obj_name);
 488                         break;
 489                 case SYMLINK_LOOP:
 490                         printf("loop %"PRIuMAX"\n%s\n",
 491                                (uintmax_t)strlen(obj_name), obj_name);
 492                         break;
 493                 case NOT_DIR:
 494                         printf("notdir %"PRIuMAX"\n%s\n",
 495                                (uintmax_t)strlen(obj_name), obj_name);
 496                         break;
 497                 default:
 498                         BUG("unknown get_sha1_with_context result %d\n",
 499                                result);
 500                         break;
 501                 }
 502                 fflush(stdout);
 503                 return;
 504         }
 505
 506         if (ctx.mode == 0) {
 507                 printf("symlink %"PRIuMAX"\n%s\n",
 508                        (uintmax_t)ctx.symlink_path.len,
 509                        ctx.symlink_path.buf);
 510                 fflush(stdout);
 511                 return;
 512         }
 513
 514         batch_object_write(obj_name, scratch, opt, data, NULL, 0);
 515 }
 516
 517 struct object_cb_data {
 518         struct batch_options *opt;
 519         struct expand_data *expand;
 520         struct oidset *seen;
 521         struct strbuf *scratch;
 522 };
 523
 524 static int batch_object_cb(const struct object_id *oid, void *vdata)
 525 {
 526         struct object_cb_data *data = vdata;
 527         oidcpy(&data->expand->oid, oid);
 528         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 529                            NULL, 0);
 530         return 0;
 531 }
 532
 533 static int collect_loose_object(const struct object_id *oid,
 534                                 const char *path,
 535                                 void *data)
 536 {
 537         oid_array_append(data, oid);
 538         return 0;
 539 }
 540
 541 static int collect_packed_object(const struct object_id *oid,
 542                                  struct packed_git *pack,
 543                                  uint32_t pos,
 544                                  void *data)
 545 {
 546         oid_array_append(data, oid);
 547         return 0;
 548 }
 549
 550 static int batch_unordered_object(const struct object_id *oid,
 551                                   struct packed_git *pack, off_t offset,
 552                                   void *vdata)
 553 {
 554         struct object_cb_data *data = vdata;
 555
 556         if (oidset_insert(data->seen, oid))
 557                 return 0;
 558
 559         oidcpy(&data->expand->oid, oid);
 560         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 561                            pack, offset);
 562         return 0;
 563 }
 564
 565 static int batch_unordered_loose(const struct object_id *oid,
 566                                  const char *path,
 567                                  void *data)
 568 {
 569         return batch_unordered_object(oid, NULL, 0, data);
 570 }
 571
 572 static int batch_unordered_packed(const struct object_id *oid,
 573                                   struct packed_git *pack,
 574                                   uint32_t pos,
 575                                   void *data)
 576 {
 577         return batch_unordered_object(oid, pack,
 578                                       nth_packed_object_offset(pack, pos),
 579                                       data);
 580 }
 581
 582 typedef void (*parse_cmd_fn_t)(struct batch_options *, const char *,
 583                                struct strbuf *, struct expand_data *);
 584
 585 struct queued_cmd {
 586         parse_cmd_fn_t fn;
 587         char *line;
 588 };
 589
 590 static void parse_cmd_contents(struct batch_options *opt,
 591                              const char *line,
 592                              struct strbuf *output,
 593                              struct expand_data *data)
 594 {
 595         opt->batch_mode = BATCH_MODE_CONTENTS;
 596         batch_one_object(line, output, opt, data);
 597 }
 598
 599 static void parse_cmd_info(struct batch_options *opt,
 600                            const char *line,
 601                            struct strbuf *output,
 602                            struct expand_data *data)
 603 {
 604         opt->batch_mode = BATCH_MODE_INFO;
 605         batch_one_object(line, output, opt, data);
 606 }
 607
 608 static void dispatch_calls(struct batch_options *opt,
 609                 struct strbuf *output,
 610                 struct expand_data *data,
 611                 struct queued_cmd *cmd,
 612                 int nr)
 613 {
 614         int i;
 615
 616         if (!opt->buffer_output)
 617                 die(_("flush is only for --buffer mode"));
 618
 619         for (i = 0; i < nr; i++)
 620                 cmd[i].fn(opt, cmd[i].line, output, data);
 621
 622         fflush(stdout);
 623 }
 624
 625 static void free_cmds(struct queued_cmd *cmd, size_t *nr)
 626 {
 627         size_t i;
 628
 629         for (i = 0; i < *nr; i++)
 630                 FREE_AND_NULL(cmd[i].line);
 631
 632         *nr = 0;
 633 }
 634
 635
 636 static const struct parse_cmd {
 637         const char *name;
 638         parse_cmd_fn_t fn;
 639         unsigned takes_args;
 640 } commands[] = {
 641         { "contents", parse_cmd_contents, 1},
 642         { "info", parse_cmd_info, 1},
 643         { "flush", NULL, 0},
 644 };
 645
 646 static void batch_objects_command(struct batch_options *opt,
 647                                     struct strbuf *output,
 648                                     struct expand_data *data)
 649 {
 650         struct strbuf input = STRBUF_INIT;
 651         struct queued_cmd *queued_cmd = NULL;
 652         size_t alloc = 0, nr = 0;
 653
 654         while (1) {
 655                 int i, ret;
 656                 const struct parse_cmd *cmd = NULL;
 657                 const char *p = NULL, *cmd_end;
 658                 struct queued_cmd call = {0};
 659
 660                 if (opt->nul_terminated)
 661                         ret = strbuf_getline_nul(&input, stdin);
 662                 else
 663                         ret = strbuf_getline(&input, stdin);
 664
 665                 if (ret)
 666                         break;
 667
 668                 if (!input.len)
 669                         die(_("empty command in input"));
 670                 if (isspace(*input.buf))
 671                         die(_("whitespace before command: '%s'"), input.buf);
 672
 673                 for (i = 0; i < ARRAY_SIZE(commands); i++) {
 674                         if (!skip_prefix(input.buf, commands[i].name, &cmd_end))
 675                                 continue;
 676
 677                         cmd = &commands[i];
 678                         if (cmd->takes_args) {
 679                                 if (*cmd_end != ' ')
 680                                         die(_("%s requires arguments"),
 681                                             commands[i].name);
 682
 683                                 p = cmd_end + 1;
 684                         } else if (*cmd_end) {
 685                                 die(_("%s takes no arguments"),
 686                                     commands[i].name);
 687                         }
 688
 689                         break;
 690                 }
 691
 692                 if (!cmd)
 693                         die(_("unknown command: '%s'"), input.buf);
 694
 695                 if (!strcmp(cmd->name, "flush")) {
 696                         dispatch_calls(opt, output, data, queued_cmd, nr);
 697                         free_cmds(queued_cmd, &nr);
 698                 } else if (!opt->buffer_output) {
 699                         cmd->fn(opt, p, output, data);
 700                 } else {
 701                         ALLOC_GROW(queued_cmd, nr + 1, alloc);
 702                         call.fn = cmd->fn;
 703                         call.line = xstrdup_or_null(p);
 704                         queued_cmd[nr++] = call;
 705                 }
 706         }
 707
 708         if (opt->buffer_output &&
 709             nr &&
 710             !git_env_bool("GIT_TEST_CAT_FILE_NO_FLUSH_ON_EXIT", 0)) {
 711                 dispatch_calls(opt, output, data, queued_cmd, nr);
 712                 free_cmds(queued_cmd, &nr);
 713         }
 714
 715         free_cmds(queued_cmd, &nr);
 716         free(queued_cmd);
 717         strbuf_release(&input);
 718 }
 719
 720 #define DEFAULT_FORMAT "%(objectname) %(objecttype) %(objectsize)"
 721
 722 static int batch_objects(struct batch_options *opt)
 723 {
 724         struct strbuf input = STRBUF_INIT;
 725         struct strbuf output = STRBUF_INIT;
 726         struct expand_data data;
 727         int save_warning;
 728         int retval = 0;
 729
 730         /*
 731          * Expand once with our special mark_query flag, which will prime the
 732          * object_info to be handed to oid_object_info_extended for each
 733          * object.
 734          */
 735         memset(&data, 0, sizeof(data));
 736         data.mark_query = 1;
 737         strbuf_expand(&output,
 738                       opt->format ? opt->format : DEFAULT_FORMAT,
 739                       expand_format,
 740                       &data);
 741         data.mark_query = 0;
 742         strbuf_release(&output);
 743         if (opt->transform_mode)
 744                 data.split_on_whitespace = 1;
 745
 746         if (opt->format && !strcmp(opt->format, DEFAULT_FORMAT))
 747                 opt->format = NULL;
 748         /*
 749          * If we are printing out the object, then always fill in the type,
 750          * since we will want to decide whether or not to stream.
 751          */
 752         if (opt->batch_mode == BATCH_MODE_CONTENTS)
 753                 data.info.typep = &data.type;
 754
 755         if (opt->all_objects) {
 756                 struct object_cb_data cb;
 757                 struct object_info empty = OBJECT_INFO_INIT;
 758
 759                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 760                         data.skip_object_info = 1;
 761
 762                 if (has_promisor_remote())
 763                         warning("This repository uses promisor remotes. Some objects may not be loaded.");
 764
 765                 read_replace_refs = 0;
 766
 767                 cb.opt = opt;
 768                 cb.expand = &data;
 769                 cb.scratch = &output;
 770
 771                 if (opt->unordered) {
 772                         struct oidset seen = OIDSET_INIT;
 773
 774                         cb.seen = &seen;
 775
 776                         for_each_loose_object(batch_unordered_loose, &cb, 0);
 777                         for_each_packed_object(batch_unordered_packed, &cb,
 778                                                FOR_EACH_OBJECT_PACK_ORDER);
 779
 780                         oidset_clear(&seen);
 781                 } else {
 782                         struct oid_array sa = OID_ARRAY_INIT;
 783
 784                         for_each_loose_object(collect_loose_object, &sa, 0);
 785                         for_each_packed_object(collect_packed_object, &sa, 0);
 786
 787                         oid_array_for_each_unique(&sa, batch_object_cb, &cb);
 788
 789                         oid_array_clear(&sa);
 790                 }
 791
 792                 strbuf_release(&output);
 793                 return 0;
 794         }
 795
 796         /*
 797          * We are going to call get_sha1 on a potentially very large number of
 798          * objects. In most large cases, these will be actual object sha1s. The
 799          * cost to double-check that each one is not also a ref (just so we can
 800          * warn) ends up dwarfing the actual cost of the object lookups
 801          * themselves. We can work around it by just turning off the warning.
 802          */
 803         save_warning = warn_on_object_refname_ambiguity;
 804         warn_on_object_refname_ambiguity = 0;
 805
 806         if (opt->batch_mode == BATCH_MODE_QUEUE_AND_DISPATCH) {
 807                 batch_objects_command(opt, &output, &data);
 808                 goto cleanup;
 809         }
 810
 811         while (1) {
 812                 int ret;
 813                 if (opt->nul_terminated)
 814                         ret = strbuf_getline_nul(&input, stdin);
 815                 else
 816                         ret = strbuf_getline(&input, stdin);
 817
 818                 if (ret == EOF)
 819                         break;
 820
 821                 if (data.split_on_whitespace) {
 822                         /*
 823                          * Split at first whitespace, tying off the beginning
 824                          * of the string and saving the remainder (or NULL) in
 825                          * data.rest.
 826                          */
 827                         char *p = strpbrk(input.buf, " \t");
 828                         if (p) {
 829                                 while (*p && strchr(" \t", *p))
 830                                         *p++ = '\0';
 831                         }
 832                         data.rest = p;
 833                 }
 834
 835                 batch_one_object(input.buf, &output, opt, &data);
 836         }
 837
 838  cleanup:
 839         strbuf_release(&input);
 840         strbuf_release(&output);
 841         warn_on_object_refname_ambiguity = save_warning;
 842         return retval;
 843 }
 844
 845 static int git_cat_file_config(const char *var, const char *value, void *cb)
 846 {
 847         if (userdiff_config(var, value) < 0)
 848                 return -1;
 849
 850         return git_default_config(var, value, cb);
 851 }
 852
 853 static int batch_option_callback(const struct option *opt,
 854                                  const char *arg,
 855                                  int unset)
 856 {
 857         struct batch_options *bo = opt->value;
 858
 859         BUG_ON_OPT_NEG(unset);
 860
 861         if (bo->enabled) {
 862                 return error(_("only one batch option may be specified"));
 863         }
 864
 865         bo->enabled = 1;
 866
 867         if (!strcmp(opt->long_name, "batch"))
 868                 bo->batch_mode = BATCH_MODE_CONTENTS;
 869         else if (!strcmp(opt->long_name, "batch-check"))
 870                 bo->batch_mode = BATCH_MODE_INFO;
 871         else if (!strcmp(opt->long_name, "batch-command"))
 872                 bo->batch_mode = BATCH_MODE_QUEUE_AND_DISPATCH;
 873         else
 874                 BUG("%s given to batch-option-callback", opt->long_name);
 875
 876         bo->format = arg;
 877
 878         return 0;
 879 }
 880
 881 int cmd_cat_file(int argc, const char **argv, const char *prefix)
 882 {
 883         int opt = 0;
 884         int opt_cw = 0;
 885         int opt_epts = 0;
 886         const char *exp_type = NULL, *obj_name = NULL;
 887         struct batch_options batch = {0};
 888         int unknown_type = 0;
 889
 890         const char * const usage[] = {
 891                 N_("git cat-file <type> <object>"),
 892                 N_("git cat-file (-e | -p) <object>"),
 893                 N_("git cat-file (-t | -s) [--allow-unknown-type] <object>"),
 894                 N_("git cat-file (--batch | --batch-check | --batch-command) [--batch-all-objects]\n"
 895                    "             [--buffer] [--follow-symlinks] [--unordered]\n"
 896                    "             [--textconv | --filters] [-z]"),
 897                 N_("git cat-file (--textconv | --filters)\n"
 898                    "             [<rev>:<path|tree-ish> | --path=<path|tree-ish> <rev>]"),
 899                 NULL
 900         };
 901         const struct option options[] = {
 902                 /* Simple queries */
 903                 OPT_GROUP(N_("Check object existence or emit object contents")),
 904                 OPT_CMDMODE('e', NULL, &opt,
 905                             N_("check if <object> exists"), 'e'),
 906                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print <object> content"), 'p'),
 907
 908                 OPT_GROUP(N_("Emit [broken] object attributes")),
 909                 OPT_CMDMODE('t', NULL, &opt, N_("show object type (one of 'blob', 'tree', 'commit', 'tag', ...)"), 't'),
 910                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 911                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 912                           N_("allow -s and -t to work with broken/corrupt objects")),
 913                 OPT_BOOL(0, "use-mailmap", &use_mailmap, N_("use mail map file")),
 914                 OPT_ALIAS(0, "mailmap", "use-mailmap"),
 915                 /* Batch mode */
 916                 OPT_GROUP(N_("Batch objects requested on stdin (or --batch-all-objects)")),
 917                 OPT_CALLBACK_F(0, "batch", &batch, N_("format"),
 918                         N_("show full <object> or <rev> contents"),
 919                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 920                         batch_option_callback),
 921                 OPT_CALLBACK_F(0, "batch-check", &batch, N_("format"),
 922                         N_("like --batch, but don't emit <contents>"),
 923                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 924                         batch_option_callback),
 925                 OPT_BOOL('z', NULL, &batch.nul_terminated, N_("stdin is NUL-terminated")),
 926                 OPT_CALLBACK_F(0, "batch-command", &batch, N_("format"),
 927                         N_("read commands from stdin"),
 928                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 929                         batch_option_callback),
 930                 OPT_CMDMODE(0, "batch-all-objects", &opt,
 931                             N_("with --batch[-check]: ignores stdin, batches all known objects"), 'b'),
 932                 /* Batch-specific options */
 933                 OPT_GROUP(N_("Change or optimize batch output")),
 934                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 935                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 936                          N_("follow in-tree symlinks")),
 937                 OPT_BOOL(0, "unordered", &batch.unordered,
 938                          N_("do not order objects before emitting them")),
 939                 /* Textconv options, stand-ole*/
 940                 OPT_GROUP(N_("Emit object (blob or tree) with conversion or filter (stand-alone, or with batch)")),
 941                 OPT_CMDMODE(0, "textconv", &opt,
 942                             N_("run textconv on object's content"), 'c'),
 943                 OPT_CMDMODE(0, "filters", &opt,
 944                             N_("run filters on object's content"), 'w'),
 945                 OPT_STRING(0, "path", &force_path, N_("blob|tree"),
 946                            N_("use a <path> for (--textconv | --filters); Not with 'batch'")),
 947                 OPT_END()
 948         };
 949
 950         git_config(git_cat_file_config, NULL);
 951
 952         batch.buffer_output = -1;
 953
 954         argc = parse_options(argc, argv, prefix, options, usage, 0);
 955         opt_cw = (opt == 'c' || opt == 'w');
 956         opt_epts = (opt == 'e' || opt == 'p' || opt == 't' || opt == 's');
 957
 958         if (use_mailmap)
 959                 read_mailmap(&mailmap);
 960
 961         /* --batch-all-objects? */
 962         if (opt == 'b')
 963                 batch.all_objects = 1;
 964
 965         /* Option compatibility */
 966         if (force_path && !opt_cw)
 967                 usage_msg_optf(_("'%s=<%s>' needs '%s' or '%s'"),
 968                                usage, options,
 969                                "--path", _("path|tree-ish"), "--filters",
 970                                "--textconv");
 971
 972         /* Option compatibility with batch mode */
 973         if (batch.enabled)
 974                 ;
 975         else if (batch.follow_symlinks)
 976                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
 977                                "--follow-symlinks");
 978         else if (batch.buffer_output >= 0)
 979                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
 980                                "--buffer");
 981         else if (batch.all_objects)
 982                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
 983                                "--batch-all-objects");
 984         else if (batch.nul_terminated)
 985                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
 986                                "-z");
 987
 988         /* Batch defaults */
 989         if (batch.buffer_output < 0)
 990                 batch.buffer_output = batch.all_objects;
 991
 992         /* Return early if we're in batch mode? */
 993         if (batch.enabled) {
 994                 if (opt_cw)
 995                         batch.transform_mode = opt;
 996                 else if (opt && opt != 'b')
 997                         usage_msg_optf(_("'-%c' is incompatible with batch mode"),
 998                                        usage, options, opt);
 999                 else if (argc)
1000                         usage_msg_opt(_("batch modes take no arguments"), usage,
1001                                       options);
1002
1003                 return batch_objects(&batch);
1004         }
1005
1006         if (opt) {
1007                 if (!argc && opt == 'c')
1008                         usage_msg_optf(_("<rev> required with '%s'"),
1009                                        usage, options, "--textconv");
1010                 else if (!argc && opt == 'w')
1011                         usage_msg_optf(_("<rev> required with '%s'"),
1012                                        usage, options, "--filters");
1013                 else if (!argc && opt_epts)
1014                         usage_msg_optf(_("<object> required with '-%c'"),
1015                                        usage, options, opt);
1016                 else if (argc == 1)
1017                         obj_name = argv[0];
1018                 else
1019                         usage_msg_opt(_("too many arguments"), usage, options);
1020         } else if (!argc) {
1021                 usage_with_options(usage, options);
1022         } else if (argc != 2) {
1023                 usage_msg_optf(_("only two arguments allowed in <type> <object> mode, not %d"),
1024                               usage, options, argc);
1025         } else if (argc) {
1026                 exp_type = argv[0];
1027                 obj_name = argv[1];
1028         }
1029
1030         if (unknown_type && opt != 't' && opt != 's')
1031                 die("git cat-file --allow-unknown-type: use with -s or -t");
1032         return cat_one_file(opt, exp_type, obj_name, unknown_type);
1033 }