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 "alloc.h"
   9 #include "config.h"
  10 #include "builtin.h"
  11 #include "diff.h"
  12 #include "environment.h"
  13 #include "gettext.h"
  14 #include "hex.h"
  15 #include "ident.h"
  16 #include "parse-options.h"
  17 #include "userdiff.h"
  18 #include "streaming.h"
  19 #include "tree-walk.h"
  20 #include "oid-array.h"
  21 #include "packfile.h"
  22 #include "object-store.h"
  23 #include "replace-object.h"
  24 #include "promisor-remote.h"
  25 #include "mailmap.h"
  26 #include "write-or-die.h"
  27
  28 enum batch_mode {
  29         BATCH_MODE_CONTENTS,
  30         BATCH_MODE_INFO,
  31         BATCH_MODE_QUEUE_AND_DISPATCH,
  32 };
  33
  34 struct batch_options {
  35         int enabled;
  36         int follow_symlinks;
  37         enum batch_mode batch_mode;
  38         int buffer_output;
  39         int all_objects;
  40         int unordered;
  41         int transform_mode; /* may be 'w' or 'c' for --filters or --textconv */
  42         int nul_terminated;
  43         const char *format;
  44 };
  45
  46 static const char *force_path;
  47
  48 static struct string_list mailmap = STRING_LIST_INIT_NODUP;
  49 static int use_mailmap;
  50
  51 static char *replace_idents_using_mailmap(char *, size_t *);
  52
  53 static char *replace_idents_using_mailmap(char *object_buf, size_t *size)
  54 {
  55         struct strbuf sb = STRBUF_INIT;
  56         const char *headers[] = { "author ", "committer ", "tagger ", NULL };
  57
  58         strbuf_attach(&sb, object_buf, *size, *size + 1);
  59         apply_mailmap_to_header(&sb, headers, &mailmap);
  60         *size = sb.len;
  61         return strbuf_detach(&sb, NULL);
  62 }
  63
  64 static int filter_object(const char *path, unsigned mode,
  65                          const struct object_id *oid,
  66                          char **buf, unsigned long *size)
  67 {
  68         enum object_type type;
  69
  70         *buf = repo_read_object_file(the_repository, oid, &type, size);
  71         if (!*buf)
  72                 return error(_("cannot read object %s '%s'"),
  73                              oid_to_hex(oid), path);
  74         if ((type == OBJ_BLOB) && S_ISREG(mode)) {
  75                 struct strbuf strbuf = STRBUF_INIT;
  76                 struct checkout_metadata meta;
  77
  78                 init_checkout_metadata(&meta, NULL, NULL, oid);
  79                 if (convert_to_working_tree(&the_index, path, *buf, *size, &strbuf, &meta)) {
  80                         free(*buf);
  81                         *size = strbuf.len;
  82                         *buf = strbuf_detach(&strbuf, NULL);
  83                 }
  84         }
  85
  86         return 0;
  87 }
  88
  89 static int stream_blob(const struct object_id *oid)
  90 {
  91         if (stream_blob_to_fd(1, oid, NULL, 0))
  92                 die("unable to stream %s to stdout", oid_to_hex(oid));
  93         return 0;
  94 }
  95
  96 static int cat_one_file(int opt, const char *exp_type, const char *obj_name,
  97                         int unknown_type)
  98 {
  99         int ret;
 100         struct object_id oid;
 101         enum object_type type;
 102         char *buf;
 103         unsigned long size;
 104         struct object_context obj_context;
 105         struct object_info oi = OBJECT_INFO_INIT;
 106         struct strbuf sb = STRBUF_INIT;
 107         unsigned flags = OBJECT_INFO_LOOKUP_REPLACE;
 108         unsigned get_oid_flags = GET_OID_RECORD_PATH | GET_OID_ONLY_TO_DIE;
 109         const char *path = force_path;
 110         const int opt_cw = (opt == 'c' || opt == 'w');
 111         if (!path && opt_cw)
 112                 get_oid_flags |= GET_OID_REQUIRE_PATH;
 113
 114         if (unknown_type)
 115                 flags |= OBJECT_INFO_ALLOW_UNKNOWN_TYPE;
 116
 117         if (get_oid_with_context(the_repository, obj_name, get_oid_flags, &oid,
 118                                  &obj_context))
 119                 die("Not a valid object name %s", obj_name);
 120
 121         if (!path)
 122                 path = obj_context.path;
 123         if (obj_context.mode == S_IFINVALID)
 124                 obj_context.mode = 0100644;
 125
 126         buf = NULL;
 127         switch (opt) {
 128         case 't':
 129                 oi.type_name = &sb;
 130                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 131                         die("git cat-file: could not get object info");
 132                 if (sb.len) {
 133                         printf("%s\n", sb.buf);
 134                         strbuf_release(&sb);
 135                         ret = 0;
 136                         goto cleanup;
 137                 }
 138                 break;
 139
 140         case 's':
 141                 oi.sizep = &size;
 142
 143                 if (use_mailmap) {
 144                         oi.typep = &type;
 145                         oi.contentp = (void**)&buf;
 146                 }
 147
 148                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 149                         die("git cat-file: could not get object info");
 150
 151                 if (use_mailmap && (type == OBJ_COMMIT || type == OBJ_TAG)) {
 152                         size_t s = size;
 153                         buf = replace_idents_using_mailmap(buf, &s);
 154                         size = cast_size_t_to_ulong(s);
 155                 }
 156
 157                 printf("%"PRIuMAX"\n", (uintmax_t)size);
 158                 ret = 0;
 159                 goto cleanup;
 160
 161         case 'e':
 162                 return !repo_has_object_file(the_repository, &oid);
 163
 164         case 'w':
 165
 166                 if (filter_object(path, obj_context.mode,
 167                                   &oid, &buf, &size)) {
 168                         ret = -1;
 169                         goto cleanup;
 170                 }
 171                 break;
 172
 173         case 'c':
 174                 if (textconv_object(the_repository, path, obj_context.mode,
 175                                     &oid, 1, &buf, &size))
 176                         break;
 177                 /* else fallthrough */
 178
 179         case 'p':
 180                 type = oid_object_info(the_repository, &oid, NULL);
 181                 if (type < 0)
 182                         die("Not a valid object name %s", obj_name);
 183
 184                 /* custom pretty-print here */
 185                 if (type == OBJ_TREE) {
 186                         const char *ls_args[3] = { NULL };
 187                         ls_args[0] =  "ls-tree";
 188                         ls_args[1] =  obj_name;
 189                         ret = cmd_ls_tree(2, ls_args, NULL);
 190                         goto cleanup;
 191                 }
 192
 193                 if (type == OBJ_BLOB) {
 194                         ret = stream_blob(&oid);
 195                         goto cleanup;
 196                 }
 197                 buf = repo_read_object_file(the_repository, &oid, &type,
 198                                             &size);
 199                 if (!buf)
 200                         die("Cannot read object %s", obj_name);
 201
 202                 if (use_mailmap) {
 203                         size_t s = size;
 204                         buf = replace_idents_using_mailmap(buf, &s);
 205                         size = cast_size_t_to_ulong(s);
 206                 }
 207
 208                 /* otherwise just spit out the data */
 209                 break;
 210
 211         case 0:
 212         {
 213                 enum object_type exp_type_id = type_from_string(exp_type);
 214
 215                 if (exp_type_id == OBJ_BLOB) {
 216                         struct object_id blob_oid;
 217                         if (oid_object_info(the_repository, &oid, NULL) == OBJ_TAG) {
 218                                 char *buffer = repo_read_object_file(the_repository,
 219                                                                      &oid,
 220                                                                      &type,
 221                                                                      &size);
 222                                 const char *target;
 223                                 if (!skip_prefix(buffer, "object ", &target) ||
 224                                     get_oid_hex(target, &blob_oid))
 225                                         die("%s not a valid tag", oid_to_hex(&oid));
 226                                 free(buffer);
 227                         } else
 228                                 oidcpy(&blob_oid, &oid);
 229
 230                         if (oid_object_info(the_repository, &blob_oid, NULL) == OBJ_BLOB) {
 231                                 ret = stream_blob(&blob_oid);
 232                                 goto cleanup;
 233                         }
 234                         /*
 235                          * we attempted to dereference a tag to a blob
 236                          * and failed; there may be new dereference
 237                          * mechanisms this code is not aware of.
 238                          * fall-back to the usual case.
 239                          */
 240                 }
 241                 buf = read_object_with_reference(the_repository, &oid,
 242                                                  exp_type_id, &size, NULL);
 243
 244                 if (use_mailmap) {
 245                         size_t s = size;
 246                         buf = replace_idents_using_mailmap(buf, &s);
 247                         size = cast_size_t_to_ulong(s);
 248                 }
 249                 break;
 250         }
 251         default:
 252                 die("git cat-file: unknown option: %s", exp_type);
 253         }
 254
 255         if (!buf)
 256                 die("git cat-file %s: bad file", obj_name);
 257
 258         write_or_die(1, buf, size);
 259         ret = 0;
 260 cleanup:
 261         free(buf);
 262         free(obj_context.path);
 263         return ret;
 264 }
 265
 266 struct expand_data {
 267         struct object_id oid;
 268         enum object_type type;
 269         unsigned long size;
 270         off_t disk_size;
 271         const char *rest;
 272         struct object_id delta_base_oid;
 273
 274         /*
 275          * If mark_query is true, we do not expand anything, but rather
 276          * just mark the object_info with items we wish to query.
 277          */
 278         int mark_query;
 279
 280         /*
 281          * Whether to split the input on whitespace before feeding it to
 282          * get_sha1; this is decided during the mark_query phase based on
 283          * whether we have a %(rest) token in our format.
 284          */
 285         int split_on_whitespace;
 286
 287         /*
 288          * After a mark_query run, this object_info is set up to be
 289          * passed to oid_object_info_extended. It will point to the data
 290          * elements above, so you can retrieve the response from there.
 291          */
 292         struct object_info info;
 293
 294         /*
 295          * This flag will be true if the requested batch format and options
 296          * don't require us to call oid_object_info, which can then be
 297          * optimized out.
 298          */
 299         unsigned skip_object_info : 1;
 300 };
 301
 302 static int is_atom(const char *atom, const char *s, int slen)
 303 {
 304         int alen = strlen(atom);
 305         return alen == slen && !memcmp(atom, s, alen);
 306 }
 307
 308 static void expand_atom(struct strbuf *sb, const char *atom, int len,
 309                         void *vdata)
 310 {
 311         struct expand_data *data = vdata;
 312
 313         if (is_atom("objectname", atom, len)) {
 314                 if (!data->mark_query)
 315                         strbuf_addstr(sb, oid_to_hex(&data->oid));
 316         } else if (is_atom("objecttype", atom, len)) {
 317                 if (data->mark_query)
 318                         data->info.typep = &data->type;
 319                 else
 320                         strbuf_addstr(sb, type_name(data->type));
 321         } else if (is_atom("objectsize", atom, len)) {
 322                 if (data->mark_query)
 323                         data->info.sizep = &data->size;
 324                 else
 325                         strbuf_addf(sb, "%"PRIuMAX , (uintmax_t)data->size);
 326         } else if (is_atom("objectsize:disk", atom, len)) {
 327                 if (data->mark_query)
 328                         data->info.disk_sizep = &data->disk_size;
 329                 else
 330                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 331         } else if (is_atom("rest", atom, len)) {
 332                 if (data->mark_query)
 333                         data->split_on_whitespace = 1;
 334                 else if (data->rest)
 335                         strbuf_addstr(sb, data->rest);
 336         } else if (is_atom("deltabase", atom, len)) {
 337                 if (data->mark_query)
 338                         data->info.delta_base_oid = &data->delta_base_oid;
 339                 else
 340                         strbuf_addstr(sb,
 341                                       oid_to_hex(&data->delta_base_oid));
 342         } else
 343                 die("unknown format element: %.*s", len, atom);
 344 }
 345
 346 static size_t expand_format(struct strbuf *sb, const char *start, void *data)
 347 {
 348         const char *end;
 349
 350         if (*start != '(')
 351                 return 0;
 352         end = strchr(start + 1, ')');
 353         if (!end)
 354                 die("format element '%s' does not end in ')'", start);
 355
 356         expand_atom(sb, start + 1, end - start - 1, data);
 357
 358         return end - start + 1;
 359 }
 360
 361 static void batch_write(struct batch_options *opt, const void *data, int len)
 362 {
 363         if (opt->buffer_output) {
 364                 if (fwrite(data, 1, len, stdout) != len)
 365                         die_errno("unable to write to stdout");
 366         } else
 367                 write_or_die(1, data, len);
 368 }
 369
 370 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
 371 {
 372         const struct object_id *oid = &data->oid;
 373
 374         assert(data->info.typep);
 375
 376         if (data->type == OBJ_BLOB) {
 377                 if (opt->buffer_output)
 378                         fflush(stdout);
 379                 if (opt->transform_mode) {
 380                         char *contents;
 381                         unsigned long size;
 382
 383                         if (!data->rest)
 384                                 die("missing path for '%s'", oid_to_hex(oid));
 385
 386                         if (opt->transform_mode == 'w') {
 387                                 if (filter_object(data->rest, 0100644, oid,
 388                                                   &contents, &size))
 389                                         die("could not convert '%s' %s",
 390                                             oid_to_hex(oid), data->rest);
 391                         } else if (opt->transform_mode == 'c') {
 392                                 enum object_type type;
 393                                 if (!textconv_object(the_repository,
 394                                                      data->rest, 0100644, oid,
 395                                                      1, &contents, &size))
 396                                         contents = repo_read_object_file(the_repository,
 397                                                                          oid,
 398                                                                          &type,
 399                                                                          &size);
 400                                 if (!contents)
 401                                         die("could not convert '%s' %s",
 402                                             oid_to_hex(oid), data->rest);
 403                         } else
 404                                 BUG("invalid transform_mode: %c", opt->transform_mode);
 405                         batch_write(opt, contents, size);
 406                         free(contents);
 407                 } else {
 408                         stream_blob(oid);
 409                 }
 410         }
 411         else {
 412                 enum object_type type;
 413                 unsigned long size;
 414                 void *contents;
 415
 416                 contents = repo_read_object_file(the_repository, oid, &type,
 417                                                  &size);
 418
 419                 if (use_mailmap) {
 420                         size_t s = size;
 421                         contents = replace_idents_using_mailmap(contents, &s);
 422                         size = cast_size_t_to_ulong(s);
 423                 }
 424
 425                 if (!contents)
 426                         die("object %s disappeared", oid_to_hex(oid));
 427                 if (type != data->type)
 428                         die("object %s changed type!?", oid_to_hex(oid));
 429                 if (data->info.sizep && size != data->size && !use_mailmap)
 430                         die("object %s changed size!?", oid_to_hex(oid));
 431
 432                 batch_write(opt, contents, size);
 433                 free(contents);
 434         }
 435 }
 436
 437 static void print_default_format(struct strbuf *scratch, struct expand_data *data)
 438 {
 439         strbuf_addf(scratch, "%s %s %"PRIuMAX"\n", oid_to_hex(&data->oid),
 440                     type_name(data->type),
 441                     (uintmax_t)data->size);
 442 }
 443
 444 /*
 445  * If "pack" is non-NULL, then "offset" is the byte offset within the pack from
 446  * which the object may be accessed (though note that we may also rely on
 447  * data->oid, too). If "pack" is NULL, then offset is ignored.
 448  */
 449 static void batch_object_write(const char *obj_name,
 450                                struct strbuf *scratch,
 451                                struct batch_options *opt,
 452                                struct expand_data *data,
 453                                struct packed_git *pack,
 454                                off_t offset)
 455 {
 456         if (!data->skip_object_info) {
 457                 int ret;
 458
 459                 if (use_mailmap)
 460                         data->info.typep = &data->type;
 461
 462                 if (pack)
 463                         ret = packed_object_info(the_repository, pack, offset,
 464                                                  &data->info);
 465                 else
 466                         ret = oid_object_info_extended(the_repository,
 467                                                        &data->oid, &data->info,
 468                                                        OBJECT_INFO_LOOKUP_REPLACE);
 469                 if (ret < 0) {
 470                         printf("%s missing\n",
 471                                obj_name ? obj_name : oid_to_hex(&data->oid));
 472                         fflush(stdout);
 473                         return;
 474                 }
 475
 476                 if (use_mailmap && (data->type == OBJ_COMMIT || data->type == OBJ_TAG)) {
 477                         size_t s = data->size;
 478                         char *buf = NULL;
 479
 480                         buf = repo_read_object_file(the_repository, &data->oid, &data->type,
 481                                                     &data->size);
 482                         buf = replace_idents_using_mailmap(buf, &s);
 483                         data->size = cast_size_t_to_ulong(s);
 484
 485                         free(buf);
 486                 }
 487         }
 488
 489         strbuf_reset(scratch);
 490
 491         if (!opt->format) {
 492                 print_default_format(scratch, data);
 493         } else {
 494                 strbuf_expand(scratch, opt->format, expand_format, data);
 495                 strbuf_addch(scratch, '\n');
 496         }
 497
 498         batch_write(opt, scratch->buf, scratch->len);
 499
 500         if (opt->batch_mode == BATCH_MODE_CONTENTS) {
 501                 print_object_or_die(opt, data);
 502                 batch_write(opt, "\n", 1);
 503         }
 504 }
 505
 506 static void batch_one_object(const char *obj_name,
 507                              struct strbuf *scratch,
 508                              struct batch_options *opt,
 509                              struct expand_data *data)
 510 {
 511         struct object_context ctx;
 512         int flags = opt->follow_symlinks ? GET_OID_FOLLOW_SYMLINKS : 0;
 513         enum get_oid_result result;
 514
 515         result = get_oid_with_context(the_repository, obj_name,
 516                                       flags, &data->oid, &ctx);
 517         if (result != FOUND) {
 518                 switch (result) {
 519                 case MISSING_OBJECT:
 520                         printf("%s missing\n", obj_name);
 521                         break;
 522                 case SHORT_NAME_AMBIGUOUS:
 523                         printf("%s ambiguous\n", obj_name);
 524                         break;
 525                 case DANGLING_SYMLINK:
 526                         printf("dangling %"PRIuMAX"\n%s\n",
 527                                (uintmax_t)strlen(obj_name), obj_name);
 528                         break;
 529                 case SYMLINK_LOOP:
 530                         printf("loop %"PRIuMAX"\n%s\n",
 531                                (uintmax_t)strlen(obj_name), obj_name);
 532                         break;
 533                 case NOT_DIR:
 534                         printf("notdir %"PRIuMAX"\n%s\n",
 535                                (uintmax_t)strlen(obj_name), obj_name);
 536                         break;
 537                 default:
 538                         BUG("unknown get_sha1_with_context result %d\n",
 539                                result);
 540                         break;
 541                 }
 542                 fflush(stdout);
 543                 return;
 544         }
 545
 546         if (ctx.mode == 0) {
 547                 printf("symlink %"PRIuMAX"\n%s\n",
 548                        (uintmax_t)ctx.symlink_path.len,
 549                        ctx.symlink_path.buf);
 550                 fflush(stdout);
 551                 return;
 552         }
 553
 554         batch_object_write(obj_name, scratch, opt, data, NULL, 0);
 555 }
 556
 557 struct object_cb_data {
 558         struct batch_options *opt;
 559         struct expand_data *expand;
 560         struct oidset *seen;
 561         struct strbuf *scratch;
 562 };
 563
 564 static int batch_object_cb(const struct object_id *oid, void *vdata)
 565 {
 566         struct object_cb_data *data = vdata;
 567         oidcpy(&data->expand->oid, oid);
 568         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 569                            NULL, 0);
 570         return 0;
 571 }
 572
 573 static int collect_loose_object(const struct object_id *oid,
 574                                 const char *path UNUSED,
 575                                 void *data)
 576 {
 577         oid_array_append(data, oid);
 578         return 0;
 579 }
 580
 581 static int collect_packed_object(const struct object_id *oid,
 582                                  struct packed_git *pack UNUSED,
 583                                  uint32_t pos UNUSED,
 584                                  void *data)
 585 {
 586         oid_array_append(data, oid);
 587         return 0;
 588 }
 589
 590 static int batch_unordered_object(const struct object_id *oid,
 591                                   struct packed_git *pack, off_t offset,
 592                                   void *vdata)
 593 {
 594         struct object_cb_data *data = vdata;
 595
 596         if (oidset_insert(data->seen, oid))
 597                 return 0;
 598
 599         oidcpy(&data->expand->oid, oid);
 600         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 601                            pack, offset);
 602         return 0;
 603 }
 604
 605 static int batch_unordered_loose(const struct object_id *oid,
 606                                  const char *path UNUSED,
 607                                  void *data)
 608 {
 609         return batch_unordered_object(oid, NULL, 0, data);
 610 }
 611
 612 static int batch_unordered_packed(const struct object_id *oid,
 613                                   struct packed_git *pack,
 614                                   uint32_t pos,
 615                                   void *data)
 616 {
 617         return batch_unordered_object(oid, pack,
 618                                       nth_packed_object_offset(pack, pos),
 619                                       data);
 620 }
 621
 622 typedef void (*parse_cmd_fn_t)(struct batch_options *, const char *,
 623                                struct strbuf *, struct expand_data *);
 624
 625 struct queued_cmd {
 626         parse_cmd_fn_t fn;
 627         char *line;
 628 };
 629
 630 static void parse_cmd_contents(struct batch_options *opt,
 631                              const char *line,
 632                              struct strbuf *output,
 633                              struct expand_data *data)
 634 {
 635         opt->batch_mode = BATCH_MODE_CONTENTS;
 636         batch_one_object(line, output, opt, data);
 637 }
 638
 639 static void parse_cmd_info(struct batch_options *opt,
 640                            const char *line,
 641                            struct strbuf *output,
 642                            struct expand_data *data)
 643 {
 644         opt->batch_mode = BATCH_MODE_INFO;
 645         batch_one_object(line, output, opt, data);
 646 }
 647
 648 static void dispatch_calls(struct batch_options *opt,
 649                 struct strbuf *output,
 650                 struct expand_data *data,
 651                 struct queued_cmd *cmd,
 652                 int nr)
 653 {
 654         int i;
 655
 656         if (!opt->buffer_output)
 657                 die(_("flush is only for --buffer mode"));
 658
 659         for (i = 0; i < nr; i++)
 660                 cmd[i].fn(opt, cmd[i].line, output, data);
 661
 662         fflush(stdout);
 663 }
 664
 665 static void free_cmds(struct queued_cmd *cmd, size_t *nr)
 666 {
 667         size_t i;
 668
 669         for (i = 0; i < *nr; i++)
 670                 FREE_AND_NULL(cmd[i].line);
 671
 672         *nr = 0;
 673 }
 674
 675
 676 static const struct parse_cmd {
 677         const char *name;
 678         parse_cmd_fn_t fn;
 679         unsigned takes_args;
 680 } commands[] = {
 681         { "contents", parse_cmd_contents, 1},
 682         { "info", parse_cmd_info, 1},
 683         { "flush", NULL, 0},
 684 };
 685
 686 static void batch_objects_command(struct batch_options *opt,
 687                                     struct strbuf *output,
 688                                     struct expand_data *data)
 689 {
 690         struct strbuf input = STRBUF_INIT;
 691         struct queued_cmd *queued_cmd = NULL;
 692         size_t alloc = 0, nr = 0;
 693
 694         while (1) {
 695                 int i, ret;
 696                 const struct parse_cmd *cmd = NULL;
 697                 const char *p = NULL, *cmd_end;
 698                 struct queued_cmd call = {0};
 699
 700                 if (opt->nul_terminated)
 701                         ret = strbuf_getline_nul(&input, stdin);
 702                 else
 703                         ret = strbuf_getline(&input, stdin);
 704
 705                 if (ret)
 706                         break;
 707
 708                 if (!input.len)
 709                         die(_("empty command in input"));
 710                 if (isspace(*input.buf))
 711                         die(_("whitespace before command: '%s'"), input.buf);
 712
 713                 for (i = 0; i < ARRAY_SIZE(commands); i++) {
 714                         if (!skip_prefix(input.buf, commands[i].name, &cmd_end))
 715                                 continue;
 716
 717                         cmd = &commands[i];
 718                         if (cmd->takes_args) {
 719                                 if (*cmd_end != ' ')
 720                                         die(_("%s requires arguments"),
 721                                             commands[i].name);
 722
 723                                 p = cmd_end + 1;
 724                         } else if (*cmd_end) {
 725                                 die(_("%s takes no arguments"),
 726                                     commands[i].name);
 727                         }
 728
 729                         break;
 730                 }
 731
 732                 if (!cmd)
 733                         die(_("unknown command: '%s'"), input.buf);
 734
 735                 if (!strcmp(cmd->name, "flush")) {
 736                         dispatch_calls(opt, output, data, queued_cmd, nr);
 737                         free_cmds(queued_cmd, &nr);
 738                 } else if (!opt->buffer_output) {
 739                         cmd->fn(opt, p, output, data);
 740                 } else {
 741                         ALLOC_GROW(queued_cmd, nr + 1, alloc);
 742                         call.fn = cmd->fn;
 743                         call.line = xstrdup_or_null(p);
 744                         queued_cmd[nr++] = call;
 745                 }
 746         }
 747
 748         if (opt->buffer_output &&
 749             nr &&
 750             !git_env_bool("GIT_TEST_CAT_FILE_NO_FLUSH_ON_EXIT", 0)) {
 751                 dispatch_calls(opt, output, data, queued_cmd, nr);
 752                 free_cmds(queued_cmd, &nr);
 753         }
 754
 755         free_cmds(queued_cmd, &nr);
 756         free(queued_cmd);
 757         strbuf_release(&input);
 758 }
 759
 760 #define DEFAULT_FORMAT "%(objectname) %(objecttype) %(objectsize)"
 761
 762 static int batch_objects(struct batch_options *opt)
 763 {
 764         struct strbuf input = STRBUF_INIT;
 765         struct strbuf output = STRBUF_INIT;
 766         struct expand_data data;
 767         int save_warning;
 768         int retval = 0;
 769
 770         /*
 771          * Expand once with our special mark_query flag, which will prime the
 772          * object_info to be handed to oid_object_info_extended for each
 773          * object.
 774          */
 775         memset(&data, 0, sizeof(data));
 776         data.mark_query = 1;
 777         strbuf_expand(&output,
 778                       opt->format ? opt->format : DEFAULT_FORMAT,
 779                       expand_format,
 780                       &data);
 781         data.mark_query = 0;
 782         strbuf_release(&output);
 783         if (opt->transform_mode)
 784                 data.split_on_whitespace = 1;
 785
 786         if (opt->format && !strcmp(opt->format, DEFAULT_FORMAT))
 787                 opt->format = NULL;
 788         /*
 789          * If we are printing out the object, then always fill in the type,
 790          * since we will want to decide whether or not to stream.
 791          */
 792         if (opt->batch_mode == BATCH_MODE_CONTENTS)
 793                 data.info.typep = &data.type;
 794
 795         if (opt->all_objects) {
 796                 struct object_cb_data cb;
 797                 struct object_info empty = OBJECT_INFO_INIT;
 798
 799                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 800                         data.skip_object_info = 1;
 801
 802                 if (repo_has_promisor_remote(the_repository))
 803                         warning("This repository uses promisor remotes. Some objects may not be loaded.");
 804
 805                 read_replace_refs = 0;
 806
 807                 cb.opt = opt;
 808                 cb.expand = &data;
 809                 cb.scratch = &output;
 810
 811                 if (opt->unordered) {
 812                         struct oidset seen = OIDSET_INIT;
 813
 814                         cb.seen = &seen;
 815
 816                         for_each_loose_object(batch_unordered_loose, &cb, 0);
 817                         for_each_packed_object(batch_unordered_packed, &cb,
 818                                                FOR_EACH_OBJECT_PACK_ORDER);
 819
 820                         oidset_clear(&seen);
 821                 } else {
 822                         struct oid_array sa = OID_ARRAY_INIT;
 823
 824                         for_each_loose_object(collect_loose_object, &sa, 0);
 825                         for_each_packed_object(collect_packed_object, &sa, 0);
 826
 827                         oid_array_for_each_unique(&sa, batch_object_cb, &cb);
 828
 829                         oid_array_clear(&sa);
 830                 }
 831
 832                 strbuf_release(&output);
 833                 return 0;
 834         }
 835
 836         /*
 837          * We are going to call get_sha1 on a potentially very large number of
 838          * objects. In most large cases, these will be actual object sha1s. The
 839          * cost to double-check that each one is not also a ref (just so we can
 840          * warn) ends up dwarfing the actual cost of the object lookups
 841          * themselves. We can work around it by just turning off the warning.
 842          */
 843         save_warning = warn_on_object_refname_ambiguity;
 844         warn_on_object_refname_ambiguity = 0;
 845
 846         if (opt->batch_mode == BATCH_MODE_QUEUE_AND_DISPATCH) {
 847                 batch_objects_command(opt, &output, &data);
 848                 goto cleanup;
 849         }
 850
 851         while (1) {
 852                 int ret;
 853                 if (opt->nul_terminated)
 854                         ret = strbuf_getline_nul(&input, stdin);
 855                 else
 856                         ret = strbuf_getline(&input, stdin);
 857
 858                 if (ret == EOF)
 859                         break;
 860
 861                 if (data.split_on_whitespace) {
 862                         /*
 863                          * Split at first whitespace, tying off the beginning
 864                          * of the string and saving the remainder (or NULL) in
 865                          * data.rest.
 866                          */
 867                         char *p = strpbrk(input.buf, " \t");
 868                         if (p) {
 869                                 while (*p && strchr(" \t", *p))
 870                                         *p++ = '\0';
 871                         }
 872                         data.rest = p;
 873                 }
 874
 875                 batch_one_object(input.buf, &output, opt, &data);
 876         }
 877
 878  cleanup:
 879         strbuf_release(&input);
 880         strbuf_release(&output);
 881         warn_on_object_refname_ambiguity = save_warning;
 882         return retval;
 883 }
 884
 885 static int git_cat_file_config(const char *var, const char *value, void *cb)
 886 {
 887         if (userdiff_config(var, value) < 0)
 888                 return -1;
 889
 890         return git_default_config(var, value, cb);
 891 }
 892
 893 static int batch_option_callback(const struct option *opt,
 894                                  const char *arg,
 895                                  int unset)
 896 {
 897         struct batch_options *bo = opt->value;
 898
 899         BUG_ON_OPT_NEG(unset);
 900
 901         if (bo->enabled) {
 902                 return error(_("only one batch option may be specified"));
 903         }
 904
 905         bo->enabled = 1;
 906
 907         if (!strcmp(opt->long_name, "batch"))
 908                 bo->batch_mode = BATCH_MODE_CONTENTS;
 909         else if (!strcmp(opt->long_name, "batch-check"))
 910                 bo->batch_mode = BATCH_MODE_INFO;
 911         else if (!strcmp(opt->long_name, "batch-command"))
 912                 bo->batch_mode = BATCH_MODE_QUEUE_AND_DISPATCH;
 913         else
 914                 BUG("%s given to batch-option-callback", opt->long_name);
 915
 916         bo->format = arg;
 917
 918         return 0;
 919 }
 920
 921 int cmd_cat_file(int argc, const char **argv, const char *prefix)
 922 {
 923         int opt = 0;
 924         int opt_cw = 0;
 925         int opt_epts = 0;
 926         const char *exp_type = NULL, *obj_name = NULL;
 927         struct batch_options batch = {0};
 928         int unknown_type = 0;
 929
 930         const char * const usage[] = {
 931                 N_("git cat-file <type> <object>"),
 932                 N_("git cat-file (-e | -p) <object>"),
 933                 N_("git cat-file (-t | -s) [--allow-unknown-type] <object>"),
 934                 N_("git cat-file (--batch | --batch-check | --batch-command) [--batch-all-objects]\n"
 935                    "             [--buffer] [--follow-symlinks] [--unordered]\n"
 936                    "             [--textconv | --filters] [-z]"),
 937                 N_("git cat-file (--textconv | --filters)\n"
 938                    "             [<rev>:<path|tree-ish> | --path=<path|tree-ish> <rev>]"),
 939                 NULL
 940         };
 941         const struct option options[] = {
 942                 /* Simple queries */
 943                 OPT_GROUP(N_("Check object existence or emit object contents")),
 944                 OPT_CMDMODE('e', NULL, &opt,
 945                             N_("check if <object> exists"), 'e'),
 946                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print <object> content"), 'p'),
 947
 948                 OPT_GROUP(N_("Emit [broken] object attributes")),
 949                 OPT_CMDMODE('t', NULL, &opt, N_("show object type (one of 'blob', 'tree', 'commit', 'tag', ...)"), 't'),
 950                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 951                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 952                           N_("allow -s and -t to work with broken/corrupt objects")),
 953                 OPT_BOOL(0, "use-mailmap", &use_mailmap, N_("use mail map file")),
 954                 OPT_ALIAS(0, "mailmap", "use-mailmap"),
 955                 /* Batch mode */
 956                 OPT_GROUP(N_("Batch objects requested on stdin (or --batch-all-objects)")),
 957                 OPT_CALLBACK_F(0, "batch", &batch, N_("format"),
 958                         N_("show full <object> or <rev> contents"),
 959                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 960                         batch_option_callback),
 961                 OPT_CALLBACK_F(0, "batch-check", &batch, N_("format"),
 962                         N_("like --batch, but don't emit <contents>"),
 963                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 964                         batch_option_callback),
 965                 OPT_BOOL('z', NULL, &batch.nul_terminated, N_("stdin is NUL-terminated")),
 966                 OPT_CALLBACK_F(0, "batch-command", &batch, N_("format"),
 967                         N_("read commands from stdin"),
 968                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 969                         batch_option_callback),
 970                 OPT_CMDMODE(0, "batch-all-objects", &opt,
 971                             N_("with --batch[-check]: ignores stdin, batches all known objects"), 'b'),
 972                 /* Batch-specific options */
 973                 OPT_GROUP(N_("Change or optimize batch output")),
 974                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 975                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 976                          N_("follow in-tree symlinks")),
 977                 OPT_BOOL(0, "unordered", &batch.unordered,
 978                          N_("do not order objects before emitting them")),
 979                 /* Textconv options, stand-ole*/
 980                 OPT_GROUP(N_("Emit object (blob or tree) with conversion or filter (stand-alone, or with batch)")),
 981                 OPT_CMDMODE(0, "textconv", &opt,
 982                             N_("run textconv on object's content"), 'c'),
 983                 OPT_CMDMODE(0, "filters", &opt,
 984                             N_("run filters on object's content"), 'w'),
 985                 OPT_STRING(0, "path", &force_path, N_("blob|tree"),
 986                            N_("use a <path> for (--textconv | --filters); Not with 'batch'")),
 987                 OPT_END()
 988         };
 989
 990         git_config(git_cat_file_config, NULL);
 991
 992         batch.buffer_output = -1;
 993
 994         argc = parse_options(argc, argv, prefix, options, usage, 0);
 995         opt_cw = (opt == 'c' || opt == 'w');
 996         opt_epts = (opt == 'e' || opt == 'p' || opt == 't' || opt == 's');
 997
 998         if (use_mailmap)
 999                 read_mailmap(&mailmap);
1000
1001         /* --batch-all-objects? */
1002         if (opt == 'b')
1003                 batch.all_objects = 1;
1004
1005         /* Option compatibility */
1006         if (force_path && !opt_cw)
1007                 usage_msg_optf(_("'%s=<%s>' needs '%s' or '%s'"),
1008                                usage, options,
1009                                "--path", _("path|tree-ish"), "--filters",
1010                                "--textconv");
1011
1012         /* Option compatibility with batch mode */
1013         if (batch.enabled)
1014                 ;
1015         else if (batch.follow_symlinks)
1016                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1017                                "--follow-symlinks");
1018         else if (batch.buffer_output >= 0)
1019                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1020                                "--buffer");
1021         else if (batch.all_objects)
1022                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1023                                "--batch-all-objects");
1024         else if (batch.nul_terminated)
1025                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1026                                "-z");
1027
1028         /* Batch defaults */
1029         if (batch.buffer_output < 0)
1030                 batch.buffer_output = batch.all_objects;
1031
1032         /* Return early if we're in batch mode? */
1033         if (batch.enabled) {
1034                 if (opt_cw)
1035                         batch.transform_mode = opt;
1036                 else if (opt && opt != 'b')
1037                         usage_msg_optf(_("'-%c' is incompatible with batch mode"),
1038                                        usage, options, opt);
1039                 else if (argc)
1040                         usage_msg_opt(_("batch modes take no arguments"), usage,
1041                                       options);
1042
1043                 return batch_objects(&batch);
1044         }
1045
1046         if (opt) {
1047                 if (!argc && opt == 'c')
1048                         usage_msg_optf(_("<rev> required with '%s'"),
1049                                        usage, options, "--textconv");
1050                 else if (!argc && opt == 'w')
1051                         usage_msg_optf(_("<rev> required with '%s'"),
1052                                        usage, options, "--filters");
1053                 else if (!argc && opt_epts)
1054                         usage_msg_optf(_("<object> required with '-%c'"),
1055                                        usage, options, opt);
1056                 else if (argc == 1)
1057                         obj_name = argv[0];
1058                 else
1059                         usage_msg_opt(_("too many arguments"), usage, options);
1060         } else if (!argc) {
1061                 usage_with_options(usage, options);
1062         } else if (argc != 2) {
1063                 usage_msg_optf(_("only two arguments allowed in <type> <object> mode, not %d"),
1064                               usage, options, argc);
1065         } else if (argc) {
1066                 exp_type = argv[0];
1067                 obj_name = argv[1];
1068         }
1069
1070         if (unknown_type && opt != 't' && opt != 's')
1071                 die("git cat-file --allow-unknown-type: use with -s or -t");
1072         return cat_one_file(opt, exp_type, obj_name, unknown_type);
1073 }