builtin/cat-file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  */
   6
   7 #include "builtin.h"
   8 #include "config.h"
   9 #include "convert.h"
  10 #include "diff.h"
  11 #include "environment.h"
  12 #include "gettext.h"
  13 #include "hex.h"
  14 #include "ident.h"
  15 #include "parse-options.h"
  16 #include "userdiff.h"
  17 #include "streaming.h"
  18 #include "oid-array.h"
  19 #include "packfile.h"
  20 #include "object-file.h"
  21 #include "object-name.h"
  22 #include "object-store-ll.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         char input_delim;
  43         char output_delim;
  44         const char *format;
  45 };
  46
  47 static const char *force_path;
  48
  49 static struct string_list mailmap = STRING_LIST_INIT_NODUP;
  50 static int use_mailmap;
  51
  52 static char *replace_idents_using_mailmap(char *, size_t *);
  53
  54 static char *replace_idents_using_mailmap(char *object_buf, size_t *size)
  55 {
  56         struct strbuf sb = STRBUF_INIT;
  57         const char *headers[] = { "author ", "committer ", "tagger ", NULL };
  58
  59         strbuf_attach(&sb, object_buf, *size, *size + 1);
  60         apply_mailmap_to_header(&sb, headers, &mailmap);
  61         *size = sb.len;
  62         return strbuf_detach(&sb, NULL);
  63 }
  64
  65 static int filter_object(const char *path, unsigned mode,
  66                          const struct object_id *oid,
  67                          char **buf, unsigned long *size)
  68 {
  69         enum object_type type;
  70
  71         *buf = repo_read_object_file(the_repository, oid, &type, size);
  72         if (!*buf)
  73                 return error(_("cannot read object %s '%s'"),
  74                              oid_to_hex(oid), path);
  75         if ((type == OBJ_BLOB) && S_ISREG(mode)) {
  76                 struct strbuf strbuf = STRBUF_INIT;
  77                 struct checkout_metadata meta;
  78
  79                 init_checkout_metadata(&meta, NULL, NULL, oid);
  80                 if (convert_to_working_tree(the_repository->index, path, *buf, *size, &strbuf, &meta)) {
  81                         free(*buf);
  82                         *size = strbuf.len;
  83                         *buf = strbuf_detach(&strbuf, NULL);
  84                 }
  85         }
  86
  87         return 0;
  88 }
  89
  90 static int stream_blob(const struct object_id *oid)
  91 {
  92         if (stream_blob_to_fd(1, oid, NULL, 0))
  93                 die("unable to stream %s to stdout", oid_to_hex(oid));
  94         return 0;
  95 }
  96
  97 static int cat_one_file(int opt, const char *exp_type, const char *obj_name,
  98                         int unknown_type)
  99 {
 100         int ret;
 101         struct object_id oid;
 102         enum object_type type;
 103         char *buf;
 104         unsigned long size;
 105         struct object_context obj_context;
 106         struct object_info oi = OBJECT_INFO_INIT;
 107         struct strbuf sb = STRBUF_INIT;
 108         unsigned flags = OBJECT_INFO_LOOKUP_REPLACE;
 109         unsigned get_oid_flags =
 110                 GET_OID_RECORD_PATH |
 111                 GET_OID_ONLY_TO_DIE |
 112                 GET_OID_HASH_ANY;
 113         const char *path = force_path;
 114         const int opt_cw = (opt == 'c' || opt == 'w');
 115         if (!path && opt_cw)
 116                 get_oid_flags |= GET_OID_REQUIRE_PATH;
 117
 118         if (unknown_type)
 119                 flags |= OBJECT_INFO_ALLOW_UNKNOWN_TYPE;
 120
 121         if (get_oid_with_context(the_repository, obj_name, get_oid_flags, &oid,
 122                                  &obj_context))
 123                 die("Not a valid object name %s", obj_name);
 124
 125         if (!path)
 126                 path = obj_context.path;
 127         if (obj_context.mode == S_IFINVALID)
 128                 obj_context.mode = 0100644;
 129
 130         buf = NULL;
 131         switch (opt) {
 132         case 't':
 133                 oi.type_name = &sb;
 134                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 135                         die("git cat-file: could not get object info");
 136                 if (sb.len) {
 137                         printf("%s\n", sb.buf);
 138                         strbuf_release(&sb);
 139                         ret = 0;
 140                         goto cleanup;
 141                 }
 142                 break;
 143
 144         case 's':
 145                 oi.sizep = &size;
 146
 147                 if (use_mailmap) {
 148                         oi.typep = &type;
 149                         oi.contentp = (void**)&buf;
 150                 }
 151
 152                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 153                         die("git cat-file: could not get object info");
 154
 155                 if (use_mailmap && (type == OBJ_COMMIT || type == OBJ_TAG)) {
 156                         size_t s = size;
 157                         buf = replace_idents_using_mailmap(buf, &s);
 158                         size = cast_size_t_to_ulong(s);
 159                 }
 160
 161                 printf("%"PRIuMAX"\n", (uintmax_t)size);
 162                 ret = 0;
 163                 goto cleanup;
 164
 165         case 'e':
 166                 return !repo_has_object_file(the_repository, &oid);
 167
 168         case 'w':
 169
 170                 if (filter_object(path, obj_context.mode,
 171                                   &oid, &buf, &size)) {
 172                         ret = -1;
 173                         goto cleanup;
 174                 }
 175                 break;
 176
 177         case 'c':
 178                 if (textconv_object(the_repository, path, obj_context.mode,
 179                                     &oid, 1, &buf, &size))
 180                         break;
 181                 /* else fallthrough */
 182
 183         case 'p':
 184                 type = oid_object_info(the_repository, &oid, NULL);
 185                 if (type < 0)
 186                         die("Not a valid object name %s", obj_name);
 187
 188                 /* custom pretty-print here */
 189                 if (type == OBJ_TREE) {
 190                         const char *ls_args[3] = { NULL };
 191                         ls_args[0] =  "ls-tree";
 192                         ls_args[1] =  obj_name;
 193                         ret = cmd_ls_tree(2, ls_args, NULL);
 194                         goto cleanup;
 195                 }
 196
 197                 if (type == OBJ_BLOB) {
 198                         ret = stream_blob(&oid);
 199                         goto cleanup;
 200                 }
 201                 buf = repo_read_object_file(the_repository, &oid, &type,
 202                                             &size);
 203                 if (!buf)
 204                         die("Cannot read object %s", obj_name);
 205
 206                 if (use_mailmap) {
 207                         size_t s = size;
 208                         buf = replace_idents_using_mailmap(buf, &s);
 209                         size = cast_size_t_to_ulong(s);
 210                 }
 211
 212                 /* otherwise just spit out the data */
 213                 break;
 214
 215         case 0:
 216         {
 217                 enum object_type exp_type_id = type_from_string(exp_type);
 218
 219                 if (exp_type_id == OBJ_BLOB) {
 220                         struct object_id blob_oid;
 221                         if (oid_object_info(the_repository, &oid, NULL) == OBJ_TAG) {
 222                                 char *buffer = repo_read_object_file(the_repository,
 223                                                                      &oid,
 224                                                                      &type,
 225                                                                      &size);
 226                                 const char *target;
 227
 228                                 if (!buffer)
 229                                         die(_("unable to read %s"), oid_to_hex(&oid));
 230
 231                                 if (!skip_prefix(buffer, "object ", &target) ||
 232                                     get_oid_hex_algop(target, &blob_oid,
 233                                                       &hash_algos[oid.algo]))
 234                                         die("%s not a valid tag", oid_to_hex(&oid));
 235                                 free(buffer);
 236                         } else
 237                                 oidcpy(&blob_oid, &oid);
 238
 239                         if (oid_object_info(the_repository, &blob_oid, NULL) == OBJ_BLOB) {
 240                                 ret = stream_blob(&blob_oid);
 241                                 goto cleanup;
 242                         }
 243                         /*
 244                          * we attempted to dereference a tag to a blob
 245                          * and failed; there may be new dereference
 246                          * mechanisms this code is not aware of.
 247                          * fall-back to the usual case.
 248                          */
 249                 }
 250                 buf = read_object_with_reference(the_repository, &oid,
 251                                                  exp_type_id, &size, NULL);
 252
 253                 if (use_mailmap) {
 254                         size_t s = size;
 255                         buf = replace_idents_using_mailmap(buf, &s);
 256                         size = cast_size_t_to_ulong(s);
 257                 }
 258                 break;
 259         }
 260         default:
 261                 die("git cat-file: unknown option: %s", exp_type);
 262         }
 263
 264         if (!buf)
 265                 die("git cat-file %s: bad file", obj_name);
 266
 267         write_or_die(1, buf, size);
 268         ret = 0;
 269 cleanup:
 270         free(buf);
 271         free(obj_context.path);
 272         return ret;
 273 }
 274
 275 struct expand_data {
 276         struct object_id oid;
 277         enum object_type type;
 278         unsigned long size;
 279         off_t disk_size;
 280         const char *rest;
 281         struct object_id delta_base_oid;
 282
 283         /*
 284          * If mark_query is true, we do not expand anything, but rather
 285          * just mark the object_info with items we wish to query.
 286          */
 287         int mark_query;
 288
 289         /*
 290          * Whether to split the input on whitespace before feeding it to
 291          * get_sha1; this is decided during the mark_query phase based on
 292          * whether we have a %(rest) token in our format.
 293          */
 294         int split_on_whitespace;
 295
 296         /*
 297          * After a mark_query run, this object_info is set up to be
 298          * passed to oid_object_info_extended. It will point to the data
 299          * elements above, so you can retrieve the response from there.
 300          */
 301         struct object_info info;
 302
 303         /*
 304          * This flag will be true if the requested batch format and options
 305          * don't require us to call oid_object_info, which can then be
 306          * optimized out.
 307          */
 308         unsigned skip_object_info : 1;
 309 };
 310
 311 static int is_atom(const char *atom, const char *s, int slen)
 312 {
 313         int alen = strlen(atom);
 314         return alen == slen && !memcmp(atom, s, alen);
 315 }
 316
 317 static int expand_atom(struct strbuf *sb, const char *atom, int len,
 318                        struct expand_data *data)
 319 {
 320         if (is_atom("objectname", atom, len)) {
 321                 if (!data->mark_query)
 322                         strbuf_addstr(sb, oid_to_hex(&data->oid));
 323         } else if (is_atom("objecttype", atom, len)) {
 324                 if (data->mark_query)
 325                         data->info.typep = &data->type;
 326                 else
 327                         strbuf_addstr(sb, type_name(data->type));
 328         } else if (is_atom("objectsize", atom, len)) {
 329                 if (data->mark_query)
 330                         data->info.sizep = &data->size;
 331                 else
 332                         strbuf_addf(sb, "%"PRIuMAX , (uintmax_t)data->size);
 333         } else if (is_atom("objectsize:disk", atom, len)) {
 334                 if (data->mark_query)
 335                         data->info.disk_sizep = &data->disk_size;
 336                 else
 337                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 338         } else if (is_atom("rest", atom, len)) {
 339                 if (data->mark_query)
 340                         data->split_on_whitespace = 1;
 341                 else if (data->rest)
 342                         strbuf_addstr(sb, data->rest);
 343         } else if (is_atom("deltabase", atom, len)) {
 344                 if (data->mark_query)
 345                         data->info.delta_base_oid = &data->delta_base_oid;
 346                 else
 347                         strbuf_addstr(sb,
 348                                       oid_to_hex(&data->delta_base_oid));
 349         } else
 350                 return 0;
 351         return 1;
 352 }
 353
 354 static void expand_format(struct strbuf *sb, const char *start,
 355                           struct expand_data *data)
 356 {
 357         while (strbuf_expand_step(sb, &start)) {
 358                 const char *end;
 359
 360                 if (skip_prefix(start, "%", &start) || *start != '(')
 361                         strbuf_addch(sb, '%');
 362                 else if ((end = strchr(start + 1, ')')) &&
 363                          expand_atom(sb, start + 1, end - start - 1, data))
 364                         start = end + 1;
 365                 else
 366                         strbuf_expand_bad_format(start, "cat-file");
 367         }
 368 }
 369
 370 static void batch_write(struct batch_options *opt, const void *data, int len)
 371 {
 372         if (opt->buffer_output) {
 373                 if (fwrite(data, 1, len, stdout) != len)
 374                         die_errno("unable to write to stdout");
 375         } else
 376                 write_or_die(1, data, len);
 377 }
 378
 379 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
 380 {
 381         const struct object_id *oid = &data->oid;
 382
 383         assert(data->info.typep);
 384
 385         if (data->type == OBJ_BLOB) {
 386                 if (opt->buffer_output)
 387                         fflush(stdout);
 388                 if (opt->transform_mode) {
 389                         char *contents;
 390                         unsigned long size;
 391
 392                         if (!data->rest)
 393                                 die("missing path for '%s'", oid_to_hex(oid));
 394
 395                         if (opt->transform_mode == 'w') {
 396                                 if (filter_object(data->rest, 0100644, oid,
 397                                                   &contents, &size))
 398                                         die("could not convert '%s' %s",
 399                                             oid_to_hex(oid), data->rest);
 400                         } else if (opt->transform_mode == 'c') {
 401                                 enum object_type type;
 402                                 if (!textconv_object(the_repository,
 403                                                      data->rest, 0100644, oid,
 404                                                      1, &contents, &size))
 405                                         contents = repo_read_object_file(the_repository,
 406                                                                          oid,
 407                                                                          &type,
 408                                                                          &size);
 409                                 if (!contents)
 410                                         die("could not convert '%s' %s",
 411                                             oid_to_hex(oid), data->rest);
 412                         } else
 413                                 BUG("invalid transform_mode: %c", opt->transform_mode);
 414                         batch_write(opt, contents, size);
 415                         free(contents);
 416                 } else {
 417                         stream_blob(oid);
 418                 }
 419         }
 420         else {
 421                 enum object_type type;
 422                 unsigned long size;
 423                 void *contents;
 424
 425                 contents = repo_read_object_file(the_repository, oid, &type,
 426                                                  &size);
 427                 if (!contents)
 428                         die("object %s disappeared", oid_to_hex(oid));
 429
 430                 if (use_mailmap) {
 431                         size_t s = size;
 432                         contents = replace_idents_using_mailmap(contents, &s);
 433                         size = cast_size_t_to_ulong(s);
 434                 }
 435
 436                 if (type != data->type)
 437                         die("object %s changed type!?", oid_to_hex(oid));
 438                 if (data->info.sizep && size != data->size && !use_mailmap)
 439                         die("object %s changed size!?", oid_to_hex(oid));
 440
 441                 batch_write(opt, contents, size);
 442                 free(contents);
 443         }
 444 }
 445
 446 static void print_default_format(struct strbuf *scratch, struct expand_data *data,
 447                                  struct batch_options *opt)
 448 {
 449         strbuf_addf(scratch, "%s %s %"PRIuMAX"%c", oid_to_hex(&data->oid),
 450                     type_name(data->type),
 451                     (uintmax_t)data->size, opt->output_delim);
 452 }
 453
 454 /*
 455  * If "pack" is non-NULL, then "offset" is the byte offset within the pack from
 456  * which the object may be accessed (though note that we may also rely on
 457  * data->oid, too). If "pack" is NULL, then offset is ignored.
 458  */
 459 static void batch_object_write(const char *obj_name,
 460                                struct strbuf *scratch,
 461                                struct batch_options *opt,
 462                                struct expand_data *data,
 463                                struct packed_git *pack,
 464                                off_t offset)
 465 {
 466         if (!data->skip_object_info) {
 467                 int ret;
 468
 469                 if (use_mailmap)
 470                         data->info.typep = &data->type;
 471
 472                 if (pack)
 473                         ret = packed_object_info(the_repository, pack, offset,
 474                                                  &data->info);
 475                 else
 476                         ret = oid_object_info_extended(the_repository,
 477                                                        &data->oid, &data->info,
 478                                                        OBJECT_INFO_LOOKUP_REPLACE);
 479                 if (ret < 0) {
 480                         printf("%s missing%c",
 481                                obj_name ? obj_name : oid_to_hex(&data->oid), opt->output_delim);
 482                         fflush(stdout);
 483                         return;
 484                 }
 485
 486                 if (use_mailmap && (data->type == OBJ_COMMIT || data->type == OBJ_TAG)) {
 487                         size_t s = data->size;
 488                         char *buf = NULL;
 489
 490                         buf = repo_read_object_file(the_repository, &data->oid, &data->type,
 491                                                     &data->size);
 492                         if (!buf)
 493                                 die(_("unable to read %s"), oid_to_hex(&data->oid));
 494                         buf = replace_idents_using_mailmap(buf, &s);
 495                         data->size = cast_size_t_to_ulong(s);
 496
 497                         free(buf);
 498                 }
 499         }
 500
 501         strbuf_reset(scratch);
 502
 503         if (!opt->format) {
 504                 print_default_format(scratch, data, opt);
 505         } else {
 506                 expand_format(scratch, opt->format, data);
 507                 strbuf_addch(scratch, opt->output_delim);
 508         }
 509
 510         batch_write(opt, scratch->buf, scratch->len);
 511
 512         if (opt->batch_mode == BATCH_MODE_CONTENTS) {
 513                 print_object_or_die(opt, data);
 514                 batch_write(opt, &opt->output_delim, 1);
 515         }
 516 }
 517
 518 static void batch_one_object(const char *obj_name,
 519                              struct strbuf *scratch,
 520                              struct batch_options *opt,
 521                              struct expand_data *data)
 522 {
 523         struct object_context ctx;
 524         int flags =
 525                 GET_OID_HASH_ANY |
 526                 (opt->follow_symlinks ? GET_OID_FOLLOW_SYMLINKS : 0);
 527         enum get_oid_result result;
 528
 529         result = get_oid_with_context(the_repository, obj_name,
 530                                       flags, &data->oid, &ctx);
 531         if (result != FOUND) {
 532                 switch (result) {
 533                 case MISSING_OBJECT:
 534                         printf("%s missing%c", obj_name, opt->output_delim);
 535                         break;
 536                 case SHORT_NAME_AMBIGUOUS:
 537                         printf("%s ambiguous%c", obj_name, opt->output_delim);
 538                         break;
 539                 case DANGLING_SYMLINK:
 540                         printf("dangling %"PRIuMAX"%c%s%c",
 541                                (uintmax_t)strlen(obj_name),
 542                                opt->output_delim, obj_name, opt->output_delim);
 543                         break;
 544                 case SYMLINK_LOOP:
 545                         printf("loop %"PRIuMAX"%c%s%c",
 546                                (uintmax_t)strlen(obj_name),
 547                                opt->output_delim, obj_name, opt->output_delim);
 548                         break;
 549                 case NOT_DIR:
 550                         printf("notdir %"PRIuMAX"%c%s%c",
 551                                (uintmax_t)strlen(obj_name),
 552                                opt->output_delim, obj_name, opt->output_delim);
 553                         break;
 554                 default:
 555                         BUG("unknown get_sha1_with_context result %d\n",
 556                                result);
 557                         break;
 558                 }
 559                 fflush(stdout);
 560                 return;
 561         }
 562
 563         if (ctx.mode == 0) {
 564                 printf("symlink %"PRIuMAX"%c%s%c",
 565                        (uintmax_t)ctx.symlink_path.len,
 566                        opt->output_delim, ctx.symlink_path.buf, opt->output_delim);
 567                 fflush(stdout);
 568                 return;
 569         }
 570
 571         batch_object_write(obj_name, scratch, opt, data, NULL, 0);
 572 }
 573
 574 struct object_cb_data {
 575         struct batch_options *opt;
 576         struct expand_data *expand;
 577         struct oidset *seen;
 578         struct strbuf *scratch;
 579 };
 580
 581 static int batch_object_cb(const struct object_id *oid, void *vdata)
 582 {
 583         struct object_cb_data *data = vdata;
 584         oidcpy(&data->expand->oid, oid);
 585         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 586                            NULL, 0);
 587         return 0;
 588 }
 589
 590 static int collect_loose_object(const struct object_id *oid,
 591                                 const char *path UNUSED,
 592                                 void *data)
 593 {
 594         oid_array_append(data, oid);
 595         return 0;
 596 }
 597
 598 static int collect_packed_object(const struct object_id *oid,
 599                                  struct packed_git *pack UNUSED,
 600                                  uint32_t pos UNUSED,
 601                                  void *data)
 602 {
 603         oid_array_append(data, oid);
 604         return 0;
 605 }
 606
 607 static int batch_unordered_object(const struct object_id *oid,
 608                                   struct packed_git *pack, off_t offset,
 609                                   void *vdata)
 610 {
 611         struct object_cb_data *data = vdata;
 612
 613         if (oidset_insert(data->seen, oid))
 614                 return 0;
 615
 616         oidcpy(&data->expand->oid, oid);
 617         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 618                            pack, offset);
 619         return 0;
 620 }
 621
 622 static int batch_unordered_loose(const struct object_id *oid,
 623                                  const char *path UNUSED,
 624                                  void *data)
 625 {
 626         return batch_unordered_object(oid, NULL, 0, data);
 627 }
 628
 629 static int batch_unordered_packed(const struct object_id *oid,
 630                                   struct packed_git *pack,
 631                                   uint32_t pos,
 632                                   void *data)
 633 {
 634         return batch_unordered_object(oid, pack,
 635                                       nth_packed_object_offset(pack, pos),
 636                                       data);
 637 }
 638
 639 typedef void (*parse_cmd_fn_t)(struct batch_options *, const char *,
 640                                struct strbuf *, struct expand_data *);
 641
 642 struct queued_cmd {
 643         parse_cmd_fn_t fn;
 644         char *line;
 645 };
 646
 647 static void parse_cmd_contents(struct batch_options *opt,
 648                              const char *line,
 649                              struct strbuf *output,
 650                              struct expand_data *data)
 651 {
 652         opt->batch_mode = BATCH_MODE_CONTENTS;
 653         batch_one_object(line, output, opt, data);
 654 }
 655
 656 static void parse_cmd_info(struct batch_options *opt,
 657                            const char *line,
 658                            struct strbuf *output,
 659                            struct expand_data *data)
 660 {
 661         opt->batch_mode = BATCH_MODE_INFO;
 662         batch_one_object(line, output, opt, data);
 663 }
 664
 665 static void dispatch_calls(struct batch_options *opt,
 666                 struct strbuf *output,
 667                 struct expand_data *data,
 668                 struct queued_cmd *cmd,
 669                 int nr)
 670 {
 671         int i;
 672
 673         if (!opt->buffer_output)
 674                 die(_("flush is only for --buffer mode"));
 675
 676         for (i = 0; i < nr; i++)
 677                 cmd[i].fn(opt, cmd[i].line, output, data);
 678
 679         fflush(stdout);
 680 }
 681
 682 static void free_cmds(struct queued_cmd *cmd, size_t *nr)
 683 {
 684         size_t i;
 685
 686         for (i = 0; i < *nr; i++)
 687                 FREE_AND_NULL(cmd[i].line);
 688
 689         *nr = 0;
 690 }
 691
 692
 693 static const struct parse_cmd {
 694         const char *name;
 695         parse_cmd_fn_t fn;
 696         unsigned takes_args;
 697 } commands[] = {
 698         { "contents", parse_cmd_contents, 1},
 699         { "info", parse_cmd_info, 1},
 700         { "flush", NULL, 0},
 701 };
 702
 703 static void batch_objects_command(struct batch_options *opt,
 704                                     struct strbuf *output,
 705                                     struct expand_data *data)
 706 {
 707         struct strbuf input = STRBUF_INIT;
 708         struct queued_cmd *queued_cmd = NULL;
 709         size_t alloc = 0, nr = 0;
 710
 711         while (strbuf_getdelim_strip_crlf(&input, stdin, opt->input_delim) != EOF) {
 712                 int i;
 713                 const struct parse_cmd *cmd = NULL;
 714                 const char *p = NULL, *cmd_end;
 715                 struct queued_cmd call = {0};
 716
 717                 if (!input.len)
 718                         die(_("empty command in input"));
 719                 if (isspace(*input.buf))
 720                         die(_("whitespace before command: '%s'"), input.buf);
 721
 722                 for (i = 0; i < ARRAY_SIZE(commands); i++) {
 723                         if (!skip_prefix(input.buf, commands[i].name, &cmd_end))
 724                                 continue;
 725
 726                         cmd = &commands[i];
 727                         if (cmd->takes_args) {
 728                                 if (*cmd_end != ' ')
 729                                         die(_("%s requires arguments"),
 730                                             commands[i].name);
 731
 732                                 p = cmd_end + 1;
 733                         } else if (*cmd_end) {
 734                                 die(_("%s takes no arguments"),
 735                                     commands[i].name);
 736                         }
 737
 738                         break;
 739                 }
 740
 741                 if (!cmd)
 742                         die(_("unknown command: '%s'"), input.buf);
 743
 744                 if (!strcmp(cmd->name, "flush")) {
 745                         dispatch_calls(opt, output, data, queued_cmd, nr);
 746                         free_cmds(queued_cmd, &nr);
 747                 } else if (!opt->buffer_output) {
 748                         cmd->fn(opt, p, output, data);
 749                 } else {
 750                         ALLOC_GROW(queued_cmd, nr + 1, alloc);
 751                         call.fn = cmd->fn;
 752                         call.line = xstrdup_or_null(p);
 753                         queued_cmd[nr++] = call;
 754                 }
 755         }
 756
 757         if (opt->buffer_output &&
 758             nr &&
 759             !git_env_bool("GIT_TEST_CAT_FILE_NO_FLUSH_ON_EXIT", 0)) {
 760                 dispatch_calls(opt, output, data, queued_cmd, nr);
 761                 free_cmds(queued_cmd, &nr);
 762         }
 763
 764         free_cmds(queued_cmd, &nr);
 765         free(queued_cmd);
 766         strbuf_release(&input);
 767 }
 768
 769 #define DEFAULT_FORMAT "%(objectname) %(objecttype) %(objectsize)"
 770
 771 static int batch_objects(struct batch_options *opt)
 772 {
 773         struct strbuf input = STRBUF_INIT;
 774         struct strbuf output = STRBUF_INIT;
 775         struct expand_data data;
 776         int save_warning;
 777         int retval = 0;
 778
 779         /*
 780          * Expand once with our special mark_query flag, which will prime the
 781          * object_info to be handed to oid_object_info_extended for each
 782          * object.
 783          */
 784         memset(&data, 0, sizeof(data));
 785         data.mark_query = 1;
 786         expand_format(&output,
 787                       opt->format ? opt->format : DEFAULT_FORMAT,
 788                       &data);
 789         data.mark_query = 0;
 790         strbuf_release(&output);
 791         if (opt->transform_mode)
 792                 data.split_on_whitespace = 1;
 793
 794         if (opt->format && !strcmp(opt->format, DEFAULT_FORMAT))
 795                 opt->format = NULL;
 796         /*
 797          * If we are printing out the object, then always fill in the type,
 798          * since we will want to decide whether or not to stream.
 799          */
 800         if (opt->batch_mode == BATCH_MODE_CONTENTS)
 801                 data.info.typep = &data.type;
 802
 803         if (opt->all_objects) {
 804                 struct object_cb_data cb;
 805                 struct object_info empty = OBJECT_INFO_INIT;
 806
 807                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 808                         data.skip_object_info = 1;
 809
 810                 if (repo_has_promisor_remote(the_repository))
 811                         warning("This repository uses promisor remotes. Some objects may not be loaded.");
 812
 813                 disable_replace_refs();
 814
 815                 cb.opt = opt;
 816                 cb.expand = &data;
 817                 cb.scratch = &output;
 818
 819                 if (opt->unordered) {
 820                         struct oidset seen = OIDSET_INIT;
 821
 822                         cb.seen = &seen;
 823
 824                         for_each_loose_object(batch_unordered_loose, &cb, 0);
 825                         for_each_packed_object(batch_unordered_packed, &cb,
 826                                                FOR_EACH_OBJECT_PACK_ORDER);
 827
 828                         oidset_clear(&seen);
 829                 } else {
 830                         struct oid_array sa = OID_ARRAY_INIT;
 831
 832                         for_each_loose_object(collect_loose_object, &sa, 0);
 833                         for_each_packed_object(collect_packed_object, &sa, 0);
 834
 835                         oid_array_for_each_unique(&sa, batch_object_cb, &cb);
 836
 837                         oid_array_clear(&sa);
 838                 }
 839
 840                 strbuf_release(&output);
 841                 return 0;
 842         }
 843
 844         /*
 845          * We are going to call get_sha1 on a potentially very large number of
 846          * objects. In most large cases, these will be actual object sha1s. The
 847          * cost to double-check that each one is not also a ref (just so we can
 848          * warn) ends up dwarfing the actual cost of the object lookups
 849          * themselves. We can work around it by just turning off the warning.
 850          */
 851         save_warning = warn_on_object_refname_ambiguity;
 852         warn_on_object_refname_ambiguity = 0;
 853
 854         if (opt->batch_mode == BATCH_MODE_QUEUE_AND_DISPATCH) {
 855                 batch_objects_command(opt, &output, &data);
 856                 goto cleanup;
 857         }
 858
 859         while (strbuf_getdelim_strip_crlf(&input, stdin, opt->input_delim) != EOF) {
 860                 if (data.split_on_whitespace) {
 861                         /*
 862                          * Split at first whitespace, tying off the beginning
 863                          * of the string and saving the remainder (or NULL) in
 864                          * data.rest.
 865                          */
 866                         char *p = strpbrk(input.buf, " \t");
 867                         if (p) {
 868                                 while (*p && strchr(" \t", *p))
 869                                         *p++ = '\0';
 870                         }
 871                         data.rest = p;
 872                 }
 873
 874                 batch_one_object(input.buf, &output, opt, &data);
 875         }
 876
 877  cleanup:
 878         strbuf_release(&input);
 879         strbuf_release(&output);
 880         warn_on_object_refname_ambiguity = save_warning;
 881         return retval;
 882 }
 883
 884 static int git_cat_file_config(const char *var, const char *value,
 885                                const struct config_context *ctx, void *cb)
 886 {
 887         if (userdiff_config(var, value) < 0)
 888                 return -1;
 889
 890         return git_default_config(var, value, ctx, 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         int input_nul_terminated = 0;
 930         int nul_terminated = 0;
 931
 932         const char * const usage[] = {
 933                 N_("git cat-file <type> <object>"),
 934                 N_("git cat-file (-e | -p) <object>"),
 935                 N_("git cat-file (-t | -s) [--allow-unknown-type] <object>"),
 936                 N_("git cat-file (--textconv | --filters)\n"
 937                    "             [<rev>:<path|tree-ish> | --path=<path|tree-ish> <rev>]"),
 938                 N_("git cat-file (--batch | --batch-check | --batch-command) [--batch-all-objects]\n"
 939                    "             [--buffer] [--follow-symlinks] [--unordered]\n"
 940                    "             [--textconv | --filters] [-Z]"),
 941                 NULL
 942         };
 943         const struct option options[] = {
 944                 /* Simple queries */
 945                 OPT_GROUP(N_("Check object existence or emit object contents")),
 946                 OPT_CMDMODE('e', NULL, &opt,
 947                             N_("check if <object> exists"), 'e'),
 948                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print <object> content"), 'p'),
 949
 950                 OPT_GROUP(N_("Emit [broken] object attributes")),
 951                 OPT_CMDMODE('t', NULL, &opt, N_("show object type (one of 'blob', 'tree', 'commit', 'tag', ...)"), 't'),
 952                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 953                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 954                           N_("allow -s and -t to work with broken/corrupt objects")),
 955                 OPT_BOOL(0, "use-mailmap", &use_mailmap, N_("use mail map file")),
 956                 OPT_ALIAS(0, "mailmap", "use-mailmap"),
 957                 /* Batch mode */
 958                 OPT_GROUP(N_("Batch objects requested on stdin (or --batch-all-objects)")),
 959                 OPT_CALLBACK_F(0, "batch", &batch, N_("format"),
 960                         N_("show full <object> or <rev> contents"),
 961                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 962                         batch_option_callback),
 963                 OPT_CALLBACK_F(0, "batch-check", &batch, N_("format"),
 964                         N_("like --batch, but don't emit <contents>"),
 965                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 966                         batch_option_callback),
 967                 OPT_BOOL_F('z', NULL, &input_nul_terminated, N_("stdin is NUL-terminated"),
 968                         PARSE_OPT_HIDDEN),
 969                 OPT_BOOL('Z', NULL, &nul_terminated, N_("stdin and stdout is NUL-terminated")),
 970                 OPT_CALLBACK_F(0, "batch-command", &batch, N_("format"),
 971                         N_("read commands from stdin"),
 972                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 973                         batch_option_callback),
 974                 OPT_CMDMODE(0, "batch-all-objects", &opt,
 975                             N_("with --batch[-check]: ignores stdin, batches all known objects"), 'b'),
 976                 /* Batch-specific options */
 977                 OPT_GROUP(N_("Change or optimize batch output")),
 978                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 979                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 980                          N_("follow in-tree symlinks")),
 981                 OPT_BOOL(0, "unordered", &batch.unordered,
 982                          N_("do not order objects before emitting them")),
 983                 /* Textconv options, stand-ole*/
 984                 OPT_GROUP(N_("Emit object (blob or tree) with conversion or filter (stand-alone, or with batch)")),
 985                 OPT_CMDMODE(0, "textconv", &opt,
 986                             N_("run textconv on object's content"), 'c'),
 987                 OPT_CMDMODE(0, "filters", &opt,
 988                             N_("run filters on object's content"), 'w'),
 989                 OPT_STRING(0, "path", &force_path, N_("blob|tree"),
 990                            N_("use a <path> for (--textconv | --filters); Not with 'batch'")),
 991                 OPT_END()
 992         };
 993
 994         git_config(git_cat_file_config, NULL);
 995
 996         batch.buffer_output = -1;
 997
 998         argc = parse_options(argc, argv, prefix, options, usage, 0);
 999         opt_cw = (opt == 'c' || opt == 'w');
1000         opt_epts = (opt == 'e' || opt == 'p' || opt == 't' || opt == 's');
1001
1002         if (use_mailmap)
1003                 read_mailmap(&mailmap);
1004
1005         /* --batch-all-objects? */
1006         if (opt == 'b')
1007                 batch.all_objects = 1;
1008
1009         /* Option compatibility */
1010         if (force_path && !opt_cw)
1011                 usage_msg_optf(_("'%s=<%s>' needs '%s' or '%s'"),
1012                                usage, options,
1013                                "--path", _("path|tree-ish"), "--filters",
1014                                "--textconv");
1015
1016         /* Option compatibility with batch mode */
1017         if (batch.enabled)
1018                 ;
1019         else if (batch.follow_symlinks)
1020                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1021                                "--follow-symlinks");
1022         else if (batch.buffer_output >= 0)
1023                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1024                                "--buffer");
1025         else if (batch.all_objects)
1026                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1027                                "--batch-all-objects");
1028         else if (input_nul_terminated)
1029                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1030                                "-z");
1031         else if (nul_terminated)
1032                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1033                                "-Z");
1034
1035         batch.input_delim = batch.output_delim = '\n';
1036         if (input_nul_terminated)
1037                 batch.input_delim = '\0';
1038         if (nul_terminated)
1039                 batch.input_delim = batch.output_delim = '\0';
1040
1041         /* Batch defaults */
1042         if (batch.buffer_output < 0)
1043                 batch.buffer_output = batch.all_objects;
1044
1045         /* Return early if we're in batch mode? */
1046         if (batch.enabled) {
1047                 if (opt_cw)
1048                         batch.transform_mode = opt;
1049                 else if (opt && opt != 'b')
1050                         usage_msg_optf(_("'-%c' is incompatible with batch mode"),
1051                                        usage, options, opt);
1052                 else if (argc)
1053                         usage_msg_opt(_("batch modes take no arguments"), usage,
1054                                       options);
1055
1056                 return batch_objects(&batch);
1057         }
1058
1059         if (opt) {
1060                 if (!argc && opt == 'c')
1061                         usage_msg_optf(_("<rev> required with '%s'"),
1062                                        usage, options, "--textconv");
1063                 else if (!argc && opt == 'w')
1064                         usage_msg_optf(_("<rev> required with '%s'"),
1065                                        usage, options, "--filters");
1066                 else if (!argc && opt_epts)
1067                         usage_msg_optf(_("<object> required with '-%c'"),
1068                                        usage, options, opt);
1069                 else if (argc == 1)
1070                         obj_name = argv[0];
1071                 else
1072                         usage_msg_opt(_("too many arguments"), usage, options);
1073         } else if (!argc) {
1074                 usage_with_options(usage, options);
1075         } else if (argc != 2) {
1076                 usage_msg_optf(_("only two arguments allowed in <type> <object> mode, not %d"),
1077                               usage, options, argc);
1078         } else if (argc) {
1079                 exp_type = argv[0];
1080                 obj_name = argv[1];
1081         }
1082
1083         if (unknown_type && opt != 't' && opt != 's')
1084                 die("git cat-file --allow-unknown-type: use with -s or -t");
1085         return cat_one_file(opt, exp_type, obj_name, unknown_type);
1086 }