builtin/cat-file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  */
   6 #define USE_THE_REPOSITORY_VARIABLE
   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 = {0};
 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                 ret = !repo_has_object_file(the_repository, &oid);
 167                 goto cleanup;
 168
 169         case 'w':
 170
 171                 if (filter_object(path, obj_context.mode,
 172                                   &oid, &buf, &size)) {
 173                         ret = -1;
 174                         goto cleanup;
 175                 }
 176                 break;
 177
 178         case 'c':
 179                 if (textconv_object(the_repository, path, obj_context.mode,
 180                                     &oid, 1, &buf, &size))
 181                         break;
 182                 /* else fallthrough */
 183
 184         case 'p':
 185                 type = oid_object_info(the_repository, &oid, NULL);
 186                 if (type < 0)
 187                         die("Not a valid object name %s", obj_name);
 188
 189                 /* custom pretty-print here */
 190                 if (type == OBJ_TREE) {
 191                         const char *ls_args[3] = { NULL };
 192                         ls_args[0] =  "ls-tree";
 193                         ls_args[1] =  obj_name;
 194                         ret = cmd_ls_tree(2, ls_args, NULL, the_repository);
 195                         goto cleanup;
 196                 }
 197
 198                 if (type == OBJ_BLOB) {
 199                         ret = stream_blob(&oid);
 200                         goto cleanup;
 201                 }
 202                 buf = repo_read_object_file(the_repository, &oid, &type,
 203                                             &size);
 204                 if (!buf)
 205                         die("Cannot read object %s", obj_name);
 206
 207                 if (use_mailmap) {
 208                         size_t s = size;
 209                         buf = replace_idents_using_mailmap(buf, &s);
 210                         size = cast_size_t_to_ulong(s);
 211                 }
 212
 213                 /* otherwise just spit out the data */
 214                 break;
 215
 216         case 0:
 217         {
 218                 enum object_type exp_type_id = type_from_string(exp_type);
 219
 220                 if (exp_type_id == OBJ_BLOB) {
 221                         struct object_id blob_oid;
 222                         if (oid_object_info(the_repository, &oid, NULL) == OBJ_TAG) {
 223                                 char *buffer = repo_read_object_file(the_repository,
 224                                                                      &oid,
 225                                                                      &type,
 226                                                                      &size);
 227                                 const char *target;
 228
 229                                 if (!buffer)
 230                                         die(_("unable to read %s"), oid_to_hex(&oid));
 231
 232                                 if (!skip_prefix(buffer, "object ", &target) ||
 233                                     get_oid_hex_algop(target, &blob_oid,
 234                                                       &hash_algos[oid.algo]))
 235                                         die("%s not a valid tag", oid_to_hex(&oid));
 236                                 free(buffer);
 237                         } else
 238                                 oidcpy(&blob_oid, &oid);
 239
 240                         if (oid_object_info(the_repository, &blob_oid, NULL) == OBJ_BLOB) {
 241                                 ret = stream_blob(&blob_oid);
 242                                 goto cleanup;
 243                         }
 244                         /*
 245                          * we attempted to dereference a tag to a blob
 246                          * and failed; there may be new dereference
 247                          * mechanisms this code is not aware of.
 248                          * fall-back to the usual case.
 249                          */
 250                 }
 251                 buf = read_object_with_reference(the_repository, &oid,
 252                                                  exp_type_id, &size, NULL);
 253
 254                 if (use_mailmap) {
 255                         size_t s = size;
 256                         buf = replace_idents_using_mailmap(buf, &s);
 257                         size = cast_size_t_to_ulong(s);
 258                 }
 259                 break;
 260         }
 261         default:
 262                 die("git cat-file: unknown option: %s", exp_type);
 263         }
 264
 265         if (!buf)
 266                 die("git cat-file %s: bad file", obj_name);
 267
 268         write_or_die(1, buf, size);
 269         ret = 0;
 270 cleanup:
 271         free(buf);
 272         object_context_release(&obj_context);
 273         return ret;
 274 }
 275
 276 struct expand_data {
 277         struct object_id oid;
 278         enum object_type type;
 279         unsigned long size;
 280         off_t disk_size;
 281         const char *rest;
 282         struct object_id delta_base_oid;
 283
 284         /*
 285          * If mark_query is true, we do not expand anything, but rather
 286          * just mark the object_info with items we wish to query.
 287          */
 288         int mark_query;
 289
 290         /*
 291          * Whether to split the input on whitespace before feeding it to
 292          * get_sha1; this is decided during the mark_query phase based on
 293          * whether we have a %(rest) token in our format.
 294          */
 295         int split_on_whitespace;
 296
 297         /*
 298          * After a mark_query run, this object_info is set up to be
 299          * passed to oid_object_info_extended. It will point to the data
 300          * elements above, so you can retrieve the response from there.
 301          */
 302         struct object_info info;
 303
 304         /*
 305          * This flag will be true if the requested batch format and options
 306          * don't require us to call oid_object_info, which can then be
 307          * optimized out.
 308          */
 309         unsigned skip_object_info : 1;
 310 };
 311
 312 static int is_atom(const char *atom, const char *s, int slen)
 313 {
 314         int alen = strlen(atom);
 315         return alen == slen && !memcmp(atom, s, alen);
 316 }
 317
 318 static int expand_atom(struct strbuf *sb, const char *atom, int len,
 319                        struct expand_data *data)
 320 {
 321         if (is_atom("objectname", atom, len)) {
 322                 if (!data->mark_query)
 323                         strbuf_addstr(sb, oid_to_hex(&data->oid));
 324         } else if (is_atom("objecttype", atom, len)) {
 325                 if (data->mark_query)
 326                         data->info.typep = &data->type;
 327                 else
 328                         strbuf_addstr(sb, type_name(data->type));
 329         } else if (is_atom("objectsize", atom, len)) {
 330                 if (data->mark_query)
 331                         data->info.sizep = &data->size;
 332                 else
 333                         strbuf_addf(sb, "%"PRIuMAX , (uintmax_t)data->size);
 334         } else if (is_atom("objectsize:disk", atom, len)) {
 335                 if (data->mark_query)
 336                         data->info.disk_sizep = &data->disk_size;
 337                 else
 338                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 339         } else if (is_atom("rest", atom, len)) {
 340                 if (data->mark_query)
 341                         data->split_on_whitespace = 1;
 342                 else if (data->rest)
 343                         strbuf_addstr(sb, data->rest);
 344         } else if (is_atom("deltabase", atom, len)) {
 345                 if (data->mark_query)
 346                         data->info.delta_base_oid = &data->delta_base_oid;
 347                 else
 348                         strbuf_addstr(sb,
 349                                       oid_to_hex(&data->delta_base_oid));
 350         } else
 351                 return 0;
 352         return 1;
 353 }
 354
 355 static void expand_format(struct strbuf *sb, const char *start,
 356                           struct expand_data *data)
 357 {
 358         while (strbuf_expand_step(sb, &start)) {
 359                 const char *end;
 360
 361                 if (skip_prefix(start, "%", &start) || *start != '(')
 362                         strbuf_addch(sb, '%');
 363                 else if ((end = strchr(start + 1, ')')) &&
 364                          expand_atom(sb, start + 1, end - start - 1, data))
 365                         start = end + 1;
 366                 else
 367                         strbuf_expand_bad_format(start, "cat-file");
 368         }
 369 }
 370
 371 static void batch_write(struct batch_options *opt, const void *data, int len)
 372 {
 373         if (opt->buffer_output) {
 374                 if (fwrite(data, 1, len, stdout) != len)
 375                         die_errno("unable to write to stdout");
 376         } else
 377                 write_or_die(1, data, len);
 378 }
 379
 380 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
 381 {
 382         const struct object_id *oid = &data->oid;
 383
 384         assert(data->info.typep);
 385
 386         if (data->type == OBJ_BLOB) {
 387                 if (opt->buffer_output)
 388                         fflush(stdout);
 389                 if (opt->transform_mode) {
 390                         char *contents;
 391                         unsigned long size;
 392
 393                         if (!data->rest)
 394                                 die("missing path for '%s'", oid_to_hex(oid));
 395
 396                         if (opt->transform_mode == 'w') {
 397                                 if (filter_object(data->rest, 0100644, oid,
 398                                                   &contents, &size))
 399                                         die("could not convert '%s' %s",
 400                                             oid_to_hex(oid), data->rest);
 401                         } else if (opt->transform_mode == 'c') {
 402                                 enum object_type type;
 403                                 if (!textconv_object(the_repository,
 404                                                      data->rest, 0100644, oid,
 405                                                      1, &contents, &size))
 406                                         contents = repo_read_object_file(the_repository,
 407                                                                          oid,
 408                                                                          &type,
 409                                                                          &size);
 410                                 if (!contents)
 411                                         die("could not convert '%s' %s",
 412                                             oid_to_hex(oid), data->rest);
 413                         } else
 414                                 BUG("invalid transform_mode: %c", opt->transform_mode);
 415                         batch_write(opt, contents, size);
 416                         free(contents);
 417                 } else {
 418                         stream_blob(oid);
 419                 }
 420         }
 421         else {
 422                 enum object_type type;
 423                 unsigned long size;
 424                 void *contents;
 425
 426                 contents = repo_read_object_file(the_repository, oid, &type,
 427                                                  &size);
 428                 if (!contents)
 429                         die("object %s disappeared", oid_to_hex(oid));
 430
 431                 if (use_mailmap) {
 432                         size_t s = size;
 433                         contents = replace_idents_using_mailmap(contents, &s);
 434                         size = cast_size_t_to_ulong(s);
 435                 }
 436
 437                 if (type != data->type)
 438                         die("object %s changed type!?", oid_to_hex(oid));
 439                 if (data->info.sizep && size != data->size && !use_mailmap)
 440                         die("object %s changed size!?", oid_to_hex(oid));
 441
 442                 batch_write(opt, contents, size);
 443                 free(contents);
 444         }
 445 }
 446
 447 static void print_default_format(struct strbuf *scratch, struct expand_data *data,
 448                                  struct batch_options *opt)
 449 {
 450         strbuf_addf(scratch, "%s %s %"PRIuMAX"%c", oid_to_hex(&data->oid),
 451                     type_name(data->type),
 452                     (uintmax_t)data->size, opt->output_delim);
 453 }
 454
 455 /*
 456  * If "pack" is non-NULL, then "offset" is the byte offset within the pack from
 457  * which the object may be accessed (though note that we may also rely on
 458  * data->oid, too). If "pack" is NULL, then offset is ignored.
 459  */
 460 static void batch_object_write(const char *obj_name,
 461                                struct strbuf *scratch,
 462                                struct batch_options *opt,
 463                                struct expand_data *data,
 464                                struct packed_git *pack,
 465                                off_t offset)
 466 {
 467         if (!data->skip_object_info) {
 468                 int ret;
 469
 470                 if (use_mailmap)
 471                         data->info.typep = &data->type;
 472
 473                 if (pack)
 474                         ret = packed_object_info(the_repository, pack, offset,
 475                                                  &data->info);
 476                 else
 477                         ret = oid_object_info_extended(the_repository,
 478                                                        &data->oid, &data->info,
 479                                                        OBJECT_INFO_LOOKUP_REPLACE);
 480                 if (ret < 0) {
 481                         printf("%s missing%c",
 482                                obj_name ? obj_name : oid_to_hex(&data->oid), opt->output_delim);
 483                         fflush(stdout);
 484                         return;
 485                 }
 486
 487                 if (use_mailmap && (data->type == OBJ_COMMIT || data->type == OBJ_TAG)) {
 488                         size_t s = data->size;
 489                         char *buf = NULL;
 490
 491                         buf = repo_read_object_file(the_repository, &data->oid, &data->type,
 492                                                     &data->size);
 493                         if (!buf)
 494                                 die(_("unable to read %s"), oid_to_hex(&data->oid));
 495                         buf = replace_idents_using_mailmap(buf, &s);
 496                         data->size = cast_size_t_to_ulong(s);
 497
 498                         free(buf);
 499                 }
 500         }
 501
 502         strbuf_reset(scratch);
 503
 504         if (!opt->format) {
 505                 print_default_format(scratch, data, opt);
 506         } else {
 507                 expand_format(scratch, opt->format, data);
 508                 strbuf_addch(scratch, opt->output_delim);
 509         }
 510
 511         batch_write(opt, scratch->buf, scratch->len);
 512
 513         if (opt->batch_mode == BATCH_MODE_CONTENTS) {
 514                 print_object_or_die(opt, data);
 515                 batch_write(opt, &opt->output_delim, 1);
 516         }
 517 }
 518
 519 static void batch_one_object(const char *obj_name,
 520                              struct strbuf *scratch,
 521                              struct batch_options *opt,
 522                              struct expand_data *data)
 523 {
 524         struct object_context ctx = {0};
 525         int flags =
 526                 GET_OID_HASH_ANY |
 527                 (opt->follow_symlinks ? GET_OID_FOLLOW_SYMLINKS : 0);
 528         enum get_oid_result result;
 529
 530         result = get_oid_with_context(the_repository, obj_name,
 531                                       flags, &data->oid, &ctx);
 532         if (result != FOUND) {
 533                 switch (result) {
 534                 case MISSING_OBJECT:
 535                         printf("%s missing%c", obj_name, opt->output_delim);
 536                         break;
 537                 case SHORT_NAME_AMBIGUOUS:
 538                         printf("%s ambiguous%c", obj_name, opt->output_delim);
 539                         break;
 540                 case DANGLING_SYMLINK:
 541                         printf("dangling %"PRIuMAX"%c%s%c",
 542                                (uintmax_t)strlen(obj_name),
 543                                opt->output_delim, obj_name, opt->output_delim);
 544                         break;
 545                 case SYMLINK_LOOP:
 546                         printf("loop %"PRIuMAX"%c%s%c",
 547                                (uintmax_t)strlen(obj_name),
 548                                opt->output_delim, obj_name, opt->output_delim);
 549                         break;
 550                 case NOT_DIR:
 551                         printf("notdir %"PRIuMAX"%c%s%c",
 552                                (uintmax_t)strlen(obj_name),
 553                                opt->output_delim, obj_name, opt->output_delim);
 554                         break;
 555                 default:
 556                         BUG("unknown get_sha1_with_context result %d\n",
 557                                result);
 558                         break;
 559                 }
 560                 fflush(stdout);
 561
 562                 goto out;
 563         }
 564
 565         if (ctx.mode == 0) {
 566                 printf("symlink %"PRIuMAX"%c%s%c",
 567                        (uintmax_t)ctx.symlink_path.len,
 568                        opt->output_delim, ctx.symlink_path.buf, opt->output_delim);
 569                 fflush(stdout);
 570                 goto out;
 571         }
 572
 573         batch_object_write(obj_name, scratch, opt, data, NULL, 0);
 574
 575 out:
 576         object_context_release(&ctx);
 577 }
 578
 579 struct object_cb_data {
 580         struct batch_options *opt;
 581         struct expand_data *expand;
 582         struct oidset *seen;
 583         struct strbuf *scratch;
 584 };
 585
 586 static int batch_object_cb(const struct object_id *oid, void *vdata)
 587 {
 588         struct object_cb_data *data = vdata;
 589         oidcpy(&data->expand->oid, oid);
 590         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 591                            NULL, 0);
 592         return 0;
 593 }
 594
 595 static int collect_loose_object(const struct object_id *oid,
 596                                 const char *path UNUSED,
 597                                 void *data)
 598 {
 599         oid_array_append(data, oid);
 600         return 0;
 601 }
 602
 603 static int collect_packed_object(const struct object_id *oid,
 604                                  struct packed_git *pack UNUSED,
 605                                  uint32_t pos UNUSED,
 606                                  void *data)
 607 {
 608         oid_array_append(data, oid);
 609         return 0;
 610 }
 611
 612 static int batch_unordered_object(const struct object_id *oid,
 613                                   struct packed_git *pack, off_t offset,
 614                                   void *vdata)
 615 {
 616         struct object_cb_data *data = vdata;
 617
 618         if (oidset_insert(data->seen, oid))
 619                 return 0;
 620
 621         oidcpy(&data->expand->oid, oid);
 622         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 623                            pack, offset);
 624         return 0;
 625 }
 626
 627 static int batch_unordered_loose(const struct object_id *oid,
 628                                  const char *path UNUSED,
 629                                  void *data)
 630 {
 631         return batch_unordered_object(oid, NULL, 0, data);
 632 }
 633
 634 static int batch_unordered_packed(const struct object_id *oid,
 635                                   struct packed_git *pack,
 636                                   uint32_t pos,
 637                                   void *data)
 638 {
 639         return batch_unordered_object(oid, pack,
 640                                       nth_packed_object_offset(pack, pos),
 641                                       data);
 642 }
 643
 644 typedef void (*parse_cmd_fn_t)(struct batch_options *, const char *,
 645                                struct strbuf *, struct expand_data *);
 646
 647 struct queued_cmd {
 648         parse_cmd_fn_t fn;
 649         char *line;
 650 };
 651
 652 static void parse_cmd_contents(struct batch_options *opt,
 653                              const char *line,
 654                              struct strbuf *output,
 655                              struct expand_data *data)
 656 {
 657         opt->batch_mode = BATCH_MODE_CONTENTS;
 658         batch_one_object(line, output, opt, data);
 659 }
 660
 661 static void parse_cmd_info(struct batch_options *opt,
 662                            const char *line,
 663                            struct strbuf *output,
 664                            struct expand_data *data)
 665 {
 666         opt->batch_mode = BATCH_MODE_INFO;
 667         batch_one_object(line, output, opt, data);
 668 }
 669
 670 static void dispatch_calls(struct batch_options *opt,
 671                 struct strbuf *output,
 672                 struct expand_data *data,
 673                 struct queued_cmd *cmd,
 674                 int nr)
 675 {
 676         int i;
 677
 678         if (!opt->buffer_output)
 679                 die(_("flush is only for --buffer mode"));
 680
 681         for (i = 0; i < nr; i++)
 682                 cmd[i].fn(opt, cmd[i].line, output, data);
 683
 684         fflush(stdout);
 685 }
 686
 687 static void free_cmds(struct queued_cmd *cmd, size_t *nr)
 688 {
 689         size_t i;
 690
 691         for (i = 0; i < *nr; i++)
 692                 FREE_AND_NULL(cmd[i].line);
 693
 694         *nr = 0;
 695 }
 696
 697
 698 static const struct parse_cmd {
 699         const char *name;
 700         parse_cmd_fn_t fn;
 701         unsigned takes_args;
 702 } commands[] = {
 703         { "contents", parse_cmd_contents, 1},
 704         { "info", parse_cmd_info, 1},
 705         { "flush", NULL, 0},
 706 };
 707
 708 static void batch_objects_command(struct batch_options *opt,
 709                                     struct strbuf *output,
 710                                     struct expand_data *data)
 711 {
 712         struct strbuf input = STRBUF_INIT;
 713         struct queued_cmd *queued_cmd = NULL;
 714         size_t alloc = 0, nr = 0;
 715
 716         while (strbuf_getdelim_strip_crlf(&input, stdin, opt->input_delim) != EOF) {
 717                 int i;
 718                 const struct parse_cmd *cmd = NULL;
 719                 const char *p = NULL, *cmd_end;
 720                 struct queued_cmd call = {0};
 721
 722                 if (!input.len)
 723                         die(_("empty command in input"));
 724                 if (isspace(*input.buf))
 725                         die(_("whitespace before command: '%s'"), input.buf);
 726
 727                 for (i = 0; i < ARRAY_SIZE(commands); i++) {
 728                         if (!skip_prefix(input.buf, commands[i].name, &cmd_end))
 729                                 continue;
 730
 731                         cmd = &commands[i];
 732                         if (cmd->takes_args) {
 733                                 if (*cmd_end != ' ')
 734                                         die(_("%s requires arguments"),
 735                                             commands[i].name);
 736
 737                                 p = cmd_end + 1;
 738                         } else if (*cmd_end) {
 739                                 die(_("%s takes no arguments"),
 740                                     commands[i].name);
 741                         }
 742
 743                         break;
 744                 }
 745
 746                 if (!cmd)
 747                         die(_("unknown command: '%s'"), input.buf);
 748
 749                 if (!strcmp(cmd->name, "flush")) {
 750                         dispatch_calls(opt, output, data, queued_cmd, nr);
 751                         free_cmds(queued_cmd, &nr);
 752                 } else if (!opt->buffer_output) {
 753                         cmd->fn(opt, p, output, data);
 754                 } else {
 755                         ALLOC_GROW(queued_cmd, nr + 1, alloc);
 756                         call.fn = cmd->fn;
 757                         call.line = xstrdup_or_null(p);
 758                         queued_cmd[nr++] = call;
 759                 }
 760         }
 761
 762         if (opt->buffer_output &&
 763             nr &&
 764             !git_env_bool("GIT_TEST_CAT_FILE_NO_FLUSH_ON_EXIT", 0)) {
 765                 dispatch_calls(opt, output, data, queued_cmd, nr);
 766                 free_cmds(queued_cmd, &nr);
 767         }
 768
 769         free_cmds(queued_cmd, &nr);
 770         free(queued_cmd);
 771         strbuf_release(&input);
 772 }
 773
 774 #define DEFAULT_FORMAT "%(objectname) %(objecttype) %(objectsize)"
 775
 776 static int batch_objects(struct batch_options *opt)
 777 {
 778         struct strbuf input = STRBUF_INIT;
 779         struct strbuf output = STRBUF_INIT;
 780         struct expand_data data;
 781         int save_warning;
 782         int retval = 0;
 783
 784         /*
 785          * Expand once with our special mark_query flag, which will prime the
 786          * object_info to be handed to oid_object_info_extended for each
 787          * object.
 788          */
 789         memset(&data, 0, sizeof(data));
 790         data.mark_query = 1;
 791         expand_format(&output,
 792                       opt->format ? opt->format : DEFAULT_FORMAT,
 793                       &data);
 794         data.mark_query = 0;
 795         strbuf_release(&output);
 796         if (opt->transform_mode)
 797                 data.split_on_whitespace = 1;
 798
 799         if (opt->format && !strcmp(opt->format, DEFAULT_FORMAT))
 800                 opt->format = NULL;
 801         /*
 802          * If we are printing out the object, then always fill in the type,
 803          * since we will want to decide whether or not to stream.
 804          */
 805         if (opt->batch_mode == BATCH_MODE_CONTENTS)
 806                 data.info.typep = &data.type;
 807
 808         if (opt->all_objects) {
 809                 struct object_cb_data cb;
 810                 struct object_info empty = OBJECT_INFO_INIT;
 811
 812                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 813                         data.skip_object_info = 1;
 814
 815                 if (repo_has_promisor_remote(the_repository))
 816                         warning("This repository uses promisor remotes. Some objects may not be loaded.");
 817
 818                 disable_replace_refs();
 819
 820                 cb.opt = opt;
 821                 cb.expand = &data;
 822                 cb.scratch = &output;
 823
 824                 if (opt->unordered) {
 825                         struct oidset seen = OIDSET_INIT;
 826
 827                         cb.seen = &seen;
 828
 829                         for_each_loose_object(batch_unordered_loose, &cb, 0);
 830                         for_each_packed_object(batch_unordered_packed, &cb,
 831                                                FOR_EACH_OBJECT_PACK_ORDER);
 832
 833                         oidset_clear(&seen);
 834                 } else {
 835                         struct oid_array sa = OID_ARRAY_INIT;
 836
 837                         for_each_loose_object(collect_loose_object, &sa, 0);
 838                         for_each_packed_object(collect_packed_object, &sa, 0);
 839
 840                         oid_array_for_each_unique(&sa, batch_object_cb, &cb);
 841
 842                         oid_array_clear(&sa);
 843                 }
 844
 845                 strbuf_release(&output);
 846                 return 0;
 847         }
 848
 849         /*
 850          * We are going to call get_sha1 on a potentially very large number of
 851          * objects. In most large cases, these will be actual object sha1s. The
 852          * cost to double-check that each one is not also a ref (just so we can
 853          * warn) ends up dwarfing the actual cost of the object lookups
 854          * themselves. We can work around it by just turning off the warning.
 855          */
 856         save_warning = warn_on_object_refname_ambiguity;
 857         warn_on_object_refname_ambiguity = 0;
 858
 859         if (opt->batch_mode == BATCH_MODE_QUEUE_AND_DISPATCH) {
 860                 batch_objects_command(opt, &output, &data);
 861                 goto cleanup;
 862         }
 863
 864         while (strbuf_getdelim_strip_crlf(&input, stdin, opt->input_delim) != EOF) {
 865                 if (data.split_on_whitespace) {
 866                         /*
 867                          * Split at first whitespace, tying off the beginning
 868                          * of the string and saving the remainder (or NULL) in
 869                          * data.rest.
 870                          */
 871                         char *p = strpbrk(input.buf, " \t");
 872                         if (p) {
 873                                 while (*p && strchr(" \t", *p))
 874                                         *p++ = '\0';
 875                         }
 876                         data.rest = p;
 877                 }
 878
 879                 batch_one_object(input.buf, &output, opt, &data);
 880         }
 881
 882  cleanup:
 883         strbuf_release(&input);
 884         strbuf_release(&output);
 885         warn_on_object_refname_ambiguity = save_warning;
 886         return retval;
 887 }
 888
 889 static int git_cat_file_config(const char *var, const char *value,
 890                                const struct config_context *ctx, void *cb)
 891 {
 892         if (userdiff_config(var, value) < 0)
 893                 return -1;
 894
 895         return git_default_config(var, value, ctx, cb);
 896 }
 897
 898 static int batch_option_callback(const struct option *opt,
 899                                  const char *arg,
 900                                  int unset)
 901 {
 902         struct batch_options *bo = opt->value;
 903
 904         BUG_ON_OPT_NEG(unset);
 905
 906         if (bo->enabled) {
 907                 return error(_("only one batch option may be specified"));
 908         }
 909
 910         bo->enabled = 1;
 911
 912         if (!strcmp(opt->long_name, "batch"))
 913                 bo->batch_mode = BATCH_MODE_CONTENTS;
 914         else if (!strcmp(opt->long_name, "batch-check"))
 915                 bo->batch_mode = BATCH_MODE_INFO;
 916         else if (!strcmp(opt->long_name, "batch-command"))
 917                 bo->batch_mode = BATCH_MODE_QUEUE_AND_DISPATCH;
 918         else
 919                 BUG("%s given to batch-option-callback", opt->long_name);
 920
 921         bo->format = arg;
 922
 923         return 0;
 924 }
 925
 926 int cmd_cat_file(int argc,
 927                  const char **argv,
 928                  const char *prefix,
 929                  struct repository *repo UNUSED)
 930 {
 931         int opt = 0;
 932         int opt_cw = 0;
 933         int opt_epts = 0;
 934         const char *exp_type = NULL, *obj_name = NULL;
 935         struct batch_options batch = {0};
 936         int unknown_type = 0;
 937         int input_nul_terminated = 0;
 938         int nul_terminated = 0;
 939
 940         const char * const usage[] = {
 941                 N_("git cat-file <type> <object>"),
 942                 N_("git cat-file (-e | -p) <object>"),
 943                 N_("git cat-file (-t | -s) [--allow-unknown-type] <object>"),
 944                 N_("git cat-file (--textconv | --filters)\n"
 945                    "             [<rev>:<path|tree-ish> | --path=<path|tree-ish> <rev>]"),
 946                 N_("git cat-file (--batch | --batch-check | --batch-command) [--batch-all-objects]\n"
 947                    "             [--buffer] [--follow-symlinks] [--unordered]\n"
 948                    "             [--textconv | --filters] [-Z]"),
 949                 NULL
 950         };
 951         const struct option options[] = {
 952                 /* Simple queries */
 953                 OPT_GROUP(N_("Check object existence or emit object contents")),
 954                 OPT_CMDMODE('e', NULL, &opt,
 955                             N_("check if <object> exists"), 'e'),
 956                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print <object> content"), 'p'),
 957
 958                 OPT_GROUP(N_("Emit [broken] object attributes")),
 959                 OPT_CMDMODE('t', NULL, &opt, N_("show object type (one of 'blob', 'tree', 'commit', 'tag', ...)"), 't'),
 960                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 961                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 962                           N_("allow -s and -t to work with broken/corrupt objects")),
 963                 OPT_BOOL(0, "use-mailmap", &use_mailmap, N_("use mail map file")),
 964                 OPT_ALIAS(0, "mailmap", "use-mailmap"),
 965                 /* Batch mode */
 966                 OPT_GROUP(N_("Batch objects requested on stdin (or --batch-all-objects)")),
 967                 OPT_CALLBACK_F(0, "batch", &batch, N_("format"),
 968                         N_("show full <object> or <rev> contents"),
 969                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 970                         batch_option_callback),
 971                 OPT_CALLBACK_F(0, "batch-check", &batch, N_("format"),
 972                         N_("like --batch, but don't emit <contents>"),
 973                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 974                         batch_option_callback),
 975                 OPT_BOOL_F('z', NULL, &input_nul_terminated, N_("stdin is NUL-terminated"),
 976                         PARSE_OPT_HIDDEN),
 977                 OPT_BOOL('Z', NULL, &nul_terminated, N_("stdin and stdout is NUL-terminated")),
 978                 OPT_CALLBACK_F(0, "batch-command", &batch, N_("format"),
 979                         N_("read commands from stdin"),
 980                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 981                         batch_option_callback),
 982                 OPT_CMDMODE(0, "batch-all-objects", &opt,
 983                             N_("with --batch[-check]: ignores stdin, batches all known objects"), 'b'),
 984                 /* Batch-specific options */
 985                 OPT_GROUP(N_("Change or optimize batch output")),
 986                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 987                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 988                          N_("follow in-tree symlinks")),
 989                 OPT_BOOL(0, "unordered", &batch.unordered,
 990                          N_("do not order objects before emitting them")),
 991                 /* Textconv options, stand-ole*/
 992                 OPT_GROUP(N_("Emit object (blob or tree) with conversion or filter (stand-alone, or with batch)")),
 993                 OPT_CMDMODE(0, "textconv", &opt,
 994                             N_("run textconv on object's content"), 'c'),
 995                 OPT_CMDMODE(0, "filters", &opt,
 996                             N_("run filters on object's content"), 'w'),
 997                 OPT_STRING(0, "path", &force_path, N_("blob|tree"),
 998                            N_("use a <path> for (--textconv | --filters); Not with 'batch'")),
 999                 OPT_END()
1000         };
1001
1002         git_config(git_cat_file_config, NULL);
1003
1004         batch.buffer_output = -1;
1005
1006         argc = parse_options(argc, argv, prefix, options, usage, 0);
1007         opt_cw = (opt == 'c' || opt == 'w');
1008         opt_epts = (opt == 'e' || opt == 'p' || opt == 't' || opt == 's');
1009
1010         if (use_mailmap)
1011                 read_mailmap(&mailmap);
1012
1013         /* --batch-all-objects? */
1014         if (opt == 'b')
1015                 batch.all_objects = 1;
1016
1017         /* Option compatibility */
1018         if (force_path && !opt_cw)
1019                 usage_msg_optf(_("'%s=<%s>' needs '%s' or '%s'"),
1020                                usage, options,
1021                                "--path", _("path|tree-ish"), "--filters",
1022                                "--textconv");
1023
1024         /* Option compatibility with batch mode */
1025         if (batch.enabled)
1026                 ;
1027         else if (batch.follow_symlinks)
1028                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1029                                "--follow-symlinks");
1030         else if (batch.buffer_output >= 0)
1031                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1032                                "--buffer");
1033         else if (batch.all_objects)
1034                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1035                                "--batch-all-objects");
1036         else if (input_nul_terminated)
1037                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1038                                "-z");
1039         else if (nul_terminated)
1040                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1041                                "-Z");
1042
1043         batch.input_delim = batch.output_delim = '\n';
1044         if (input_nul_terminated)
1045                 batch.input_delim = '\0';
1046         if (nul_terminated)
1047                 batch.input_delim = batch.output_delim = '\0';
1048
1049         /* Batch defaults */
1050         if (batch.buffer_output < 0)
1051                 batch.buffer_output = batch.all_objects;
1052
1053         prepare_repo_settings(the_repository);
1054         the_repository->settings.command_requires_full_index = 0;
1055
1056         /* Return early if we're in batch mode? */
1057         if (batch.enabled) {
1058                 if (opt_cw)
1059                         batch.transform_mode = opt;
1060                 else if (opt && opt != 'b')
1061                         usage_msg_optf(_("'-%c' is incompatible with batch mode"),
1062                                        usage, options, opt);
1063                 else if (argc)
1064                         usage_msg_opt(_("batch modes take no arguments"), usage,
1065                                       options);
1066
1067                 return batch_objects(&batch);
1068         }
1069
1070         if (opt) {
1071                 if (!argc && opt == 'c')
1072                         usage_msg_optf(_("<rev> required with '%s'"),
1073                                        usage, options, "--textconv");
1074                 else if (!argc && opt == 'w')
1075                         usage_msg_optf(_("<rev> required with '%s'"),
1076                                        usage, options, "--filters");
1077                 else if (!argc && opt_epts)
1078                         usage_msg_optf(_("<object> required with '-%c'"),
1079                                        usage, options, opt);
1080                 else if (argc == 1)
1081                         obj_name = argv[0];
1082                 else
1083                         usage_msg_opt(_("too many arguments"), usage, options);
1084         } else if (!argc) {
1085                 usage_with_options(usage, options);
1086         } else if (argc != 2) {
1087                 usage_msg_optf(_("only two arguments allowed in <type> <object> mode, not %d"),
1088                               usage, options, argc);
1089         } else if (argc) {
1090                 exp_type = argv[0];
1091                 obj_name = argv[1];
1092         }
1093
1094         if (unknown_type && opt != 't' && opt != 's')
1095                 die("git cat-file --allow-unknown-type: use with -s or -t");
1096         return cat_one_file(opt, exp_type, obj_name, unknown_type);
1097 }