builtin/cat-file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  */
   6 #define USE_THE_INDEX_VARIABLE
   7 #include "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_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 = GET_OID_RECORD_PATH | GET_OID_ONLY_TO_DIE;
 110         const char *path = force_path;
 111         const int opt_cw = (opt == 'c' || opt == 'w');
 112         if (!path && opt_cw)
 113                 get_oid_flags |= GET_OID_REQUIRE_PATH;
 114
 115         if (unknown_type)
 116                 flags |= OBJECT_INFO_ALLOW_UNKNOWN_TYPE;
 117
 118         if (get_oid_with_context(the_repository, obj_name, get_oid_flags, &oid,
 119                                  &obj_context))
 120                 die("Not a valid object name %s", obj_name);
 121
 122         if (!path)
 123                 path = obj_context.path;
 124         if (obj_context.mode == S_IFINVALID)
 125                 obj_context.mode = 0100644;
 126
 127         buf = NULL;
 128         switch (opt) {
 129         case 't':
 130                 oi.type_name = &sb;
 131                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 132                         die("git cat-file: could not get object info");
 133                 if (sb.len) {
 134                         printf("%s\n", sb.buf);
 135                         strbuf_release(&sb);
 136                         ret = 0;
 137                         goto cleanup;
 138                 }
 139                 break;
 140
 141         case 's':
 142                 oi.sizep = &size;
 143
 144                 if (use_mailmap) {
 145                         oi.typep = &type;
 146                         oi.contentp = (void**)&buf;
 147                 }
 148
 149                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 150                         die("git cat-file: could not get object info");
 151
 152                 if (use_mailmap && (type == OBJ_COMMIT || type == OBJ_TAG)) {
 153                         size_t s = size;
 154                         buf = replace_idents_using_mailmap(buf, &s);
 155                         size = cast_size_t_to_ulong(s);
 156                 }
 157
 158                 printf("%"PRIuMAX"\n", (uintmax_t)size);
 159                 ret = 0;
 160                 goto cleanup;
 161
 162         case 'e':
 163                 return !repo_has_object_file(the_repository, &oid);
 164
 165         case 'w':
 166
 167                 if (filter_object(path, obj_context.mode,
 168                                   &oid, &buf, &size)) {
 169                         ret = -1;
 170                         goto cleanup;
 171                 }
 172                 break;
 173
 174         case 'c':
 175                 if (textconv_object(the_repository, path, obj_context.mode,
 176                                     &oid, 1, &buf, &size))
 177                         break;
 178                 /* else fallthrough */
 179
 180         case 'p':
 181                 type = oid_object_info(the_repository, &oid, NULL);
 182                 if (type < 0)
 183                         die("Not a valid object name %s", obj_name);
 184
 185                 /* custom pretty-print here */
 186                 if (type == OBJ_TREE) {
 187                         const char *ls_args[3] = { NULL };
 188                         ls_args[0] =  "ls-tree";
 189                         ls_args[1] =  obj_name;
 190                         ret = cmd_ls_tree(2, ls_args, NULL);
 191                         goto cleanup;
 192                 }
 193
 194                 if (type == OBJ_BLOB) {
 195                         ret = stream_blob(&oid);
 196                         goto cleanup;
 197                 }
 198                 buf = repo_read_object_file(the_repository, &oid, &type,
 199                                             &size);
 200                 if (!buf)
 201                         die("Cannot read object %s", obj_name);
 202
 203                 if (use_mailmap) {
 204                         size_t s = size;
 205                         buf = replace_idents_using_mailmap(buf, &s);
 206                         size = cast_size_t_to_ulong(s);
 207                 }
 208
 209                 /* otherwise just spit out the data */
 210                 break;
 211
 212         case 0:
 213         {
 214                 enum object_type exp_type_id = type_from_string(exp_type);
 215
 216                 if (exp_type_id == OBJ_BLOB) {
 217                         struct object_id blob_oid;
 218                         if (oid_object_info(the_repository, &oid, NULL) == OBJ_TAG) {
 219                                 char *buffer = repo_read_object_file(the_repository,
 220                                                                      &oid,
 221                                                                      &type,
 222                                                                      &size);
 223                                 const char *target;
 224                                 if (!skip_prefix(buffer, "object ", &target) ||
 225                                     get_oid_hex(target, &blob_oid))
 226                                         die("%s not a valid tag", oid_to_hex(&oid));
 227                                 free(buffer);
 228                         } else
 229                                 oidcpy(&blob_oid, &oid);
 230
 231                         if (oid_object_info(the_repository, &blob_oid, NULL) == OBJ_BLOB) {
 232                                 ret = stream_blob(&blob_oid);
 233                                 goto cleanup;
 234                         }
 235                         /*
 236                          * we attempted to dereference a tag to a blob
 237                          * and failed; there may be new dereference
 238                          * mechanisms this code is not aware of.
 239                          * fall-back to the usual case.
 240                          */
 241                 }
 242                 buf = read_object_with_reference(the_repository, &oid,
 243                                                  exp_type_id, &size, NULL);
 244
 245                 if (use_mailmap) {
 246                         size_t s = size;
 247                         buf = replace_idents_using_mailmap(buf, &s);
 248                         size = cast_size_t_to_ulong(s);
 249                 }
 250                 break;
 251         }
 252         default:
 253                 die("git cat-file: unknown option: %s", exp_type);
 254         }
 255
 256         if (!buf)
 257                 die("git cat-file %s: bad file", obj_name);
 258
 259         write_or_die(1, buf, size);
 260         ret = 0;
 261 cleanup:
 262         free(buf);
 263         free(obj_context.path);
 264         return ret;
 265 }
 266
 267 struct expand_data {
 268         struct object_id oid;
 269         enum object_type type;
 270         unsigned long size;
 271         off_t disk_size;
 272         const char *rest;
 273         struct object_id delta_base_oid;
 274
 275         /*
 276          * If mark_query is true, we do not expand anything, but rather
 277          * just mark the object_info with items we wish to query.
 278          */
 279         int mark_query;
 280
 281         /*
 282          * Whether to split the input on whitespace before feeding it to
 283          * get_sha1; this is decided during the mark_query phase based on
 284          * whether we have a %(rest) token in our format.
 285          */
 286         int split_on_whitespace;
 287
 288         /*
 289          * After a mark_query run, this object_info is set up to be
 290          * passed to oid_object_info_extended. It will point to the data
 291          * elements above, so you can retrieve the response from there.
 292          */
 293         struct object_info info;
 294
 295         /*
 296          * This flag will be true if the requested batch format and options
 297          * don't require us to call oid_object_info, which can then be
 298          * optimized out.
 299          */
 300         unsigned skip_object_info : 1;
 301 };
 302
 303 static int is_atom(const char *atom, const char *s, int slen)
 304 {
 305         int alen = strlen(atom);
 306         return alen == slen && !memcmp(atom, s, alen);
 307 }
 308
 309 static void expand_atom(struct strbuf *sb, const char *atom, int len,
 310                         struct expand_data *data)
 311 {
 312         if (is_atom("objectname", atom, len)) {
 313                 if (!data->mark_query)
 314                         strbuf_addstr(sb, oid_to_hex(&data->oid));
 315         } else if (is_atom("objecttype", atom, len)) {
 316                 if (data->mark_query)
 317                         data->info.typep = &data->type;
 318                 else
 319                         strbuf_addstr(sb, type_name(data->type));
 320         } else if (is_atom("objectsize", atom, len)) {
 321                 if (data->mark_query)
 322                         data->info.sizep = &data->size;
 323                 else
 324                         strbuf_addf(sb, "%"PRIuMAX , (uintmax_t)data->size);
 325         } else if (is_atom("objectsize:disk", atom, len)) {
 326                 if (data->mark_query)
 327                         data->info.disk_sizep = &data->disk_size;
 328                 else
 329                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 330         } else if (is_atom("rest", atom, len)) {
 331                 if (data->mark_query)
 332                         data->split_on_whitespace = 1;
 333                 else if (data->rest)
 334                         strbuf_addstr(sb, data->rest);
 335         } else if (is_atom("deltabase", atom, len)) {
 336                 if (data->mark_query)
 337                         data->info.delta_base_oid = &data->delta_base_oid;
 338                 else
 339                         strbuf_addstr(sb,
 340                                       oid_to_hex(&data->delta_base_oid));
 341         } else
 342                 die("unknown format element: %.*s", len, atom);
 343 }
 344
 345 static void expand_format(struct strbuf *sb, const char *start,
 346                           struct expand_data *data)
 347 {
 348         while (strbuf_expand_step(sb, &start)) {
 349                 const char *end;
 350
 351                 if (skip_prefix(start, "%", &start) || *start != '(')
 352                         strbuf_addch(sb, '%');
 353                 else if (!(end = strchr(start + 1, ')')))
 354                         die("format element '%s' does not end in ')'", start);
 355                 else {
 356                         expand_atom(sb, start + 1, end - start - 1, data);
 357                         start = end + 1;
 358                 }
 359         }
 360 }
 361
 362 static void batch_write(struct batch_options *opt, const void *data, int len)
 363 {
 364         if (opt->buffer_output) {
 365                 if (fwrite(data, 1, len, stdout) != len)
 366                         die_errno("unable to write to stdout");
 367         } else
 368                 write_or_die(1, data, len);
 369 }
 370
 371 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
 372 {
 373         const struct object_id *oid = &data->oid;
 374
 375         assert(data->info.typep);
 376
 377         if (data->type == OBJ_BLOB) {
 378                 if (opt->buffer_output)
 379                         fflush(stdout);
 380                 if (opt->transform_mode) {
 381                         char *contents;
 382                         unsigned long size;
 383
 384                         if (!data->rest)
 385                                 die("missing path for '%s'", oid_to_hex(oid));
 386
 387                         if (opt->transform_mode == 'w') {
 388                                 if (filter_object(data->rest, 0100644, oid,
 389                                                   &contents, &size))
 390                                         die("could not convert '%s' %s",
 391                                             oid_to_hex(oid), data->rest);
 392                         } else if (opt->transform_mode == 'c') {
 393                                 enum object_type type;
 394                                 if (!textconv_object(the_repository,
 395                                                      data->rest, 0100644, oid,
 396                                                      1, &contents, &size))
 397                                         contents = repo_read_object_file(the_repository,
 398                                                                          oid,
 399                                                                          &type,
 400                                                                          &size);
 401                                 if (!contents)
 402                                         die("could not convert '%s' %s",
 403                                             oid_to_hex(oid), data->rest);
 404                         } else
 405                                 BUG("invalid transform_mode: %c", opt->transform_mode);
 406                         batch_write(opt, contents, size);
 407                         free(contents);
 408                 } else {
 409                         stream_blob(oid);
 410                 }
 411         }
 412         else {
 413                 enum object_type type;
 414                 unsigned long size;
 415                 void *contents;
 416
 417                 contents = repo_read_object_file(the_repository, oid, &type,
 418                                                  &size);
 419
 420                 if (use_mailmap) {
 421                         size_t s = size;
 422                         contents = replace_idents_using_mailmap(contents, &s);
 423                         size = cast_size_t_to_ulong(s);
 424                 }
 425
 426                 if (!contents)
 427                         die("object %s disappeared", oid_to_hex(oid));
 428                 if (type != data->type)
 429                         die("object %s changed type!?", oid_to_hex(oid));
 430                 if (data->info.sizep && size != data->size && !use_mailmap)
 431                         die("object %s changed size!?", oid_to_hex(oid));
 432
 433                 batch_write(opt, contents, size);
 434                 free(contents);
 435         }
 436 }
 437
 438 static void print_default_format(struct strbuf *scratch, struct expand_data *data,
 439                                  struct batch_options *opt)
 440 {
 441         strbuf_addf(scratch, "%s %s %"PRIuMAX"%c", oid_to_hex(&data->oid),
 442                     type_name(data->type),
 443                     (uintmax_t)data->size, opt->output_delim);
 444 }
 445
 446 /*
 447  * If "pack" is non-NULL, then "offset" is the byte offset within the pack from
 448  * which the object may be accessed (though note that we may also rely on
 449  * data->oid, too). If "pack" is NULL, then offset is ignored.
 450  */
 451 static void batch_object_write(const char *obj_name,
 452                                struct strbuf *scratch,
 453                                struct batch_options *opt,
 454                                struct expand_data *data,
 455                                struct packed_git *pack,
 456                                off_t offset)
 457 {
 458         if (!data->skip_object_info) {
 459                 int ret;
 460
 461                 if (use_mailmap)
 462                         data->info.typep = &data->type;
 463
 464                 if (pack)
 465                         ret = packed_object_info(the_repository, pack, offset,
 466                                                  &data->info);
 467                 else
 468                         ret = oid_object_info_extended(the_repository,
 469                                                        &data->oid, &data->info,
 470                                                        OBJECT_INFO_LOOKUP_REPLACE);
 471                 if (ret < 0) {
 472                         printf("%s missing%c",
 473                                obj_name ? obj_name : oid_to_hex(&data->oid), opt->output_delim);
 474                         fflush(stdout);
 475                         return;
 476                 }
 477
 478                 if (use_mailmap && (data->type == OBJ_COMMIT || data->type == OBJ_TAG)) {
 479                         size_t s = data->size;
 480                         char *buf = NULL;
 481
 482                         buf = repo_read_object_file(the_repository, &data->oid, &data->type,
 483                                                     &data->size);
 484                         buf = replace_idents_using_mailmap(buf, &s);
 485                         data->size = cast_size_t_to_ulong(s);
 486
 487                         free(buf);
 488                 }
 489         }
 490
 491         strbuf_reset(scratch);
 492
 493         if (!opt->format) {
 494                 print_default_format(scratch, data, opt);
 495         } else {
 496                 expand_format(scratch, opt->format, data);
 497                 strbuf_addch(scratch, opt->output_delim);
 498         }
 499
 500         batch_write(opt, scratch->buf, scratch->len);
 501
 502         if (opt->batch_mode == BATCH_MODE_CONTENTS) {
 503                 print_object_or_die(opt, data);
 504                 batch_write(opt, &opt->output_delim, 1);
 505         }
 506 }
 507
 508 static void batch_one_object(const char *obj_name,
 509                              struct strbuf *scratch,
 510                              struct batch_options *opt,
 511                              struct expand_data *data)
 512 {
 513         struct object_context ctx;
 514         int flags = opt->follow_symlinks ? GET_OID_FOLLOW_SYMLINKS : 0;
 515         enum get_oid_result result;
 516
 517         result = get_oid_with_context(the_repository, obj_name,
 518                                       flags, &data->oid, &ctx);
 519         if (result != FOUND) {
 520                 switch (result) {
 521                 case MISSING_OBJECT:
 522                         printf("%s missing%c", obj_name, opt->output_delim);
 523                         break;
 524                 case SHORT_NAME_AMBIGUOUS:
 525                         printf("%s ambiguous%c", obj_name, opt->output_delim);
 526                         break;
 527                 case DANGLING_SYMLINK:
 528                         printf("dangling %"PRIuMAX"%c%s%c",
 529                                (uintmax_t)strlen(obj_name),
 530                                opt->output_delim, obj_name, opt->output_delim);
 531                         break;
 532                 case SYMLINK_LOOP:
 533                         printf("loop %"PRIuMAX"%c%s%c",
 534                                (uintmax_t)strlen(obj_name),
 535                                opt->output_delim, obj_name, opt->output_delim);
 536                         break;
 537                 case NOT_DIR:
 538                         printf("notdir %"PRIuMAX"%c%s%c",
 539                                (uintmax_t)strlen(obj_name),
 540                                opt->output_delim, obj_name, opt->output_delim);
 541                         break;
 542                 default:
 543                         BUG("unknown get_sha1_with_context result %d\n",
 544                                result);
 545                         break;
 546                 }
 547                 fflush(stdout);
 548                 return;
 549         }
 550
 551         if (ctx.mode == 0) {
 552                 printf("symlink %"PRIuMAX"%c%s%c",
 553                        (uintmax_t)ctx.symlink_path.len,
 554                        opt->output_delim, ctx.symlink_path.buf, opt->output_delim);
 555                 fflush(stdout);
 556                 return;
 557         }
 558
 559         batch_object_write(obj_name, scratch, opt, data, NULL, 0);
 560 }
 561
 562 struct object_cb_data {
 563         struct batch_options *opt;
 564         struct expand_data *expand;
 565         struct oidset *seen;
 566         struct strbuf *scratch;
 567 };
 568
 569 static int batch_object_cb(const struct object_id *oid, void *vdata)
 570 {
 571         struct object_cb_data *data = vdata;
 572         oidcpy(&data->expand->oid, oid);
 573         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 574                            NULL, 0);
 575         return 0;
 576 }
 577
 578 static int collect_loose_object(const struct object_id *oid,
 579                                 const char *path UNUSED,
 580                                 void *data)
 581 {
 582         oid_array_append(data, oid);
 583         return 0;
 584 }
 585
 586 static int collect_packed_object(const struct object_id *oid,
 587                                  struct packed_git *pack UNUSED,
 588                                  uint32_t pos UNUSED,
 589                                  void *data)
 590 {
 591         oid_array_append(data, oid);
 592         return 0;
 593 }
 594
 595 static int batch_unordered_object(const struct object_id *oid,
 596                                   struct packed_git *pack, off_t offset,
 597                                   void *vdata)
 598 {
 599         struct object_cb_data *data = vdata;
 600
 601         if (oidset_insert(data->seen, oid))
 602                 return 0;
 603
 604         oidcpy(&data->expand->oid, oid);
 605         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 606                            pack, offset);
 607         return 0;
 608 }
 609
 610 static int batch_unordered_loose(const struct object_id *oid,
 611                                  const char *path UNUSED,
 612                                  void *data)
 613 {
 614         return batch_unordered_object(oid, NULL, 0, data);
 615 }
 616
 617 static int batch_unordered_packed(const struct object_id *oid,
 618                                   struct packed_git *pack,
 619                                   uint32_t pos,
 620                                   void *data)
 621 {
 622         return batch_unordered_object(oid, pack,
 623                                       nth_packed_object_offset(pack, pos),
 624                                       data);
 625 }
 626
 627 typedef void (*parse_cmd_fn_t)(struct batch_options *, const char *,
 628                                struct strbuf *, struct expand_data *);
 629
 630 struct queued_cmd {
 631         parse_cmd_fn_t fn;
 632         char *line;
 633 };
 634
 635 static void parse_cmd_contents(struct batch_options *opt,
 636                              const char *line,
 637                              struct strbuf *output,
 638                              struct expand_data *data)
 639 {
 640         opt->batch_mode = BATCH_MODE_CONTENTS;
 641         batch_one_object(line, output, opt, data);
 642 }
 643
 644 static void parse_cmd_info(struct batch_options *opt,
 645                            const char *line,
 646                            struct strbuf *output,
 647                            struct expand_data *data)
 648 {
 649         opt->batch_mode = BATCH_MODE_INFO;
 650         batch_one_object(line, output, opt, data);
 651 }
 652
 653 static void dispatch_calls(struct batch_options *opt,
 654                 struct strbuf *output,
 655                 struct expand_data *data,
 656                 struct queued_cmd *cmd,
 657                 int nr)
 658 {
 659         int i;
 660
 661         if (!opt->buffer_output)
 662                 die(_("flush is only for --buffer mode"));
 663
 664         for (i = 0; i < nr; i++)
 665                 cmd[i].fn(opt, cmd[i].line, output, data);
 666
 667         fflush(stdout);
 668 }
 669
 670 static void free_cmds(struct queued_cmd *cmd, size_t *nr)
 671 {
 672         size_t i;
 673
 674         for (i = 0; i < *nr; i++)
 675                 FREE_AND_NULL(cmd[i].line);
 676
 677         *nr = 0;
 678 }
 679
 680
 681 static const struct parse_cmd {
 682         const char *name;
 683         parse_cmd_fn_t fn;
 684         unsigned takes_args;
 685 } commands[] = {
 686         { "contents", parse_cmd_contents, 1},
 687         { "info", parse_cmd_info, 1},
 688         { "flush", NULL, 0},
 689 };
 690
 691 static void batch_objects_command(struct batch_options *opt,
 692                                     struct strbuf *output,
 693                                     struct expand_data *data)
 694 {
 695         struct strbuf input = STRBUF_INIT;
 696         struct queued_cmd *queued_cmd = NULL;
 697         size_t alloc = 0, nr = 0;
 698
 699         while (strbuf_getdelim_strip_crlf(&input, stdin, opt->input_delim) != EOF) {
 700                 int i;
 701                 const struct parse_cmd *cmd = NULL;
 702                 const char *p = NULL, *cmd_end;
 703                 struct queued_cmd call = {0};
 704
 705                 if (!input.len)
 706                         die(_("empty command in input"));
 707                 if (isspace(*input.buf))
 708                         die(_("whitespace before command: '%s'"), input.buf);
 709
 710                 for (i = 0; i < ARRAY_SIZE(commands); i++) {
 711                         if (!skip_prefix(input.buf, commands[i].name, &cmd_end))
 712                                 continue;
 713
 714                         cmd = &commands[i];
 715                         if (cmd->takes_args) {
 716                                 if (*cmd_end != ' ')
 717                                         die(_("%s requires arguments"),
 718                                             commands[i].name);
 719
 720                                 p = cmd_end + 1;
 721                         } else if (*cmd_end) {
 722                                 die(_("%s takes no arguments"),
 723                                     commands[i].name);
 724                         }
 725
 726                         break;
 727                 }
 728
 729                 if (!cmd)
 730                         die(_("unknown command: '%s'"), input.buf);
 731
 732                 if (!strcmp(cmd->name, "flush")) {
 733                         dispatch_calls(opt, output, data, queued_cmd, nr);
 734                         free_cmds(queued_cmd, &nr);
 735                 } else if (!opt->buffer_output) {
 736                         cmd->fn(opt, p, output, data);
 737                 } else {
 738                         ALLOC_GROW(queued_cmd, nr + 1, alloc);
 739                         call.fn = cmd->fn;
 740                         call.line = xstrdup_or_null(p);
 741                         queued_cmd[nr++] = call;
 742                 }
 743         }
 744
 745         if (opt->buffer_output &&
 746             nr &&
 747             !git_env_bool("GIT_TEST_CAT_FILE_NO_FLUSH_ON_EXIT", 0)) {
 748                 dispatch_calls(opt, output, data, queued_cmd, nr);
 749                 free_cmds(queued_cmd, &nr);
 750         }
 751
 752         free_cmds(queued_cmd, &nr);
 753         free(queued_cmd);
 754         strbuf_release(&input);
 755 }
 756
 757 #define DEFAULT_FORMAT "%(objectname) %(objecttype) %(objectsize)"
 758
 759 static int batch_objects(struct batch_options *opt)
 760 {
 761         struct strbuf input = STRBUF_INIT;
 762         struct strbuf output = STRBUF_INIT;
 763         struct expand_data data;
 764         int save_warning;
 765         int retval = 0;
 766
 767         /*
 768          * Expand once with our special mark_query flag, which will prime the
 769          * object_info to be handed to oid_object_info_extended for each
 770          * object.
 771          */
 772         memset(&data, 0, sizeof(data));
 773         data.mark_query = 1;
 774         expand_format(&output,
 775                       opt->format ? opt->format : DEFAULT_FORMAT,
 776                       &data);
 777         data.mark_query = 0;
 778         strbuf_release(&output);
 779         if (opt->transform_mode)
 780                 data.split_on_whitespace = 1;
 781
 782         if (opt->format && !strcmp(opt->format, DEFAULT_FORMAT))
 783                 opt->format = NULL;
 784         /*
 785          * If we are printing out the object, then always fill in the type,
 786          * since we will want to decide whether or not to stream.
 787          */
 788         if (opt->batch_mode == BATCH_MODE_CONTENTS)
 789                 data.info.typep = &data.type;
 790
 791         if (opt->all_objects) {
 792                 struct object_cb_data cb;
 793                 struct object_info empty = OBJECT_INFO_INIT;
 794
 795                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 796                         data.skip_object_info = 1;
 797
 798                 if (repo_has_promisor_remote(the_repository))
 799                         warning("This repository uses promisor remotes. Some objects may not be loaded.");
 800
 801                 disable_replace_refs();
 802
 803                 cb.opt = opt;
 804                 cb.expand = &data;
 805                 cb.scratch = &output;
 806
 807                 if (opt->unordered) {
 808                         struct oidset seen = OIDSET_INIT;
 809
 810                         cb.seen = &seen;
 811
 812                         for_each_loose_object(batch_unordered_loose, &cb, 0);
 813                         for_each_packed_object(batch_unordered_packed, &cb,
 814                                                FOR_EACH_OBJECT_PACK_ORDER);
 815
 816                         oidset_clear(&seen);
 817                 } else {
 818                         struct oid_array sa = OID_ARRAY_INIT;
 819
 820                         for_each_loose_object(collect_loose_object, &sa, 0);
 821                         for_each_packed_object(collect_packed_object, &sa, 0);
 822
 823                         oid_array_for_each_unique(&sa, batch_object_cb, &cb);
 824
 825                         oid_array_clear(&sa);
 826                 }
 827
 828                 strbuf_release(&output);
 829                 return 0;
 830         }
 831
 832         /*
 833          * We are going to call get_sha1 on a potentially very large number of
 834          * objects. In most large cases, these will be actual object sha1s. The
 835          * cost to double-check that each one is not also a ref (just so we can
 836          * warn) ends up dwarfing the actual cost of the object lookups
 837          * themselves. We can work around it by just turning off the warning.
 838          */
 839         save_warning = warn_on_object_refname_ambiguity;
 840         warn_on_object_refname_ambiguity = 0;
 841
 842         if (opt->batch_mode == BATCH_MODE_QUEUE_AND_DISPATCH) {
 843                 batch_objects_command(opt, &output, &data);
 844                 goto cleanup;
 845         }
 846
 847         while (strbuf_getdelim_strip_crlf(&input, stdin, opt->input_delim) != EOF) {
 848                 if (data.split_on_whitespace) {
 849                         /*
 850                          * Split at first whitespace, tying off the beginning
 851                          * of the string and saving the remainder (or NULL) in
 852                          * data.rest.
 853                          */
 854                         char *p = strpbrk(input.buf, " \t");
 855                         if (p) {
 856                                 while (*p && strchr(" \t", *p))
 857                                         *p++ = '\0';
 858                         }
 859                         data.rest = p;
 860                 }
 861
 862                 batch_one_object(input.buf, &output, opt, &data);
 863         }
 864
 865  cleanup:
 866         strbuf_release(&input);
 867         strbuf_release(&output);
 868         warn_on_object_refname_ambiguity = save_warning;
 869         return retval;
 870 }
 871
 872 static int git_cat_file_config(const char *var, const char *value,
 873                                const struct config_context *ctx, void *cb)
 874 {
 875         if (userdiff_config(var, value) < 0)
 876                 return -1;
 877
 878         return git_default_config(var, value, ctx, cb);
 879 }
 880
 881 static int batch_option_callback(const struct option *opt,
 882                                  const char *arg,
 883                                  int unset)
 884 {
 885         struct batch_options *bo = opt->value;
 886
 887         BUG_ON_OPT_NEG(unset);
 888
 889         if (bo->enabled) {
 890                 return error(_("only one batch option may be specified"));
 891         }
 892
 893         bo->enabled = 1;
 894
 895         if (!strcmp(opt->long_name, "batch"))
 896                 bo->batch_mode = BATCH_MODE_CONTENTS;
 897         else if (!strcmp(opt->long_name, "batch-check"))
 898                 bo->batch_mode = BATCH_MODE_INFO;
 899         else if (!strcmp(opt->long_name, "batch-command"))
 900                 bo->batch_mode = BATCH_MODE_QUEUE_AND_DISPATCH;
 901         else
 902                 BUG("%s given to batch-option-callback", opt->long_name);
 903
 904         bo->format = arg;
 905
 906         return 0;
 907 }
 908
 909 int cmd_cat_file(int argc, const char **argv, const char *prefix)
 910 {
 911         int opt = 0;
 912         int opt_cw = 0;
 913         int opt_epts = 0;
 914         const char *exp_type = NULL, *obj_name = NULL;
 915         struct batch_options batch = {0};
 916         int unknown_type = 0;
 917         int input_nul_terminated = 0;
 918         int nul_terminated = 0;
 919
 920         const char * const usage[] = {
 921                 N_("git cat-file <type> <object>"),
 922                 N_("git cat-file (-e | -p) <object>"),
 923                 N_("git cat-file (-t | -s) [--allow-unknown-type] <object>"),
 924                 N_("git cat-file (--textconv | --filters)\n"
 925                    "             [<rev>:<path|tree-ish> | --path=<path|tree-ish> <rev>]"),
 926                 N_("git cat-file (--batch | --batch-check | --batch-command) [--batch-all-objects]\n"
 927                    "             [--buffer] [--follow-symlinks] [--unordered]\n"
 928                    "             [--textconv | --filters] [-Z]"),
 929                 NULL
 930         };
 931         const struct option options[] = {
 932                 /* Simple queries */
 933                 OPT_GROUP(N_("Check object existence or emit object contents")),
 934                 OPT_CMDMODE('e', NULL, &opt,
 935                             N_("check if <object> exists"), 'e'),
 936                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print <object> content"), 'p'),
 937
 938                 OPT_GROUP(N_("Emit [broken] object attributes")),
 939                 OPT_CMDMODE('t', NULL, &opt, N_("show object type (one of 'blob', 'tree', 'commit', 'tag', ...)"), 't'),
 940                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 941                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 942                           N_("allow -s and -t to work with broken/corrupt objects")),
 943                 OPT_BOOL(0, "use-mailmap", &use_mailmap, N_("use mail map file")),
 944                 OPT_ALIAS(0, "mailmap", "use-mailmap"),
 945                 /* Batch mode */
 946                 OPT_GROUP(N_("Batch objects requested on stdin (or --batch-all-objects)")),
 947                 OPT_CALLBACK_F(0, "batch", &batch, N_("format"),
 948                         N_("show full <object> or <rev> contents"),
 949                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 950                         batch_option_callback),
 951                 OPT_CALLBACK_F(0, "batch-check", &batch, N_("format"),
 952                         N_("like --batch, but don't emit <contents>"),
 953                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 954                         batch_option_callback),
 955                 OPT_BOOL_F('z', NULL, &input_nul_terminated, N_("stdin is NUL-terminated"),
 956                         PARSE_OPT_HIDDEN),
 957                 OPT_BOOL('Z', NULL, &nul_terminated, N_("stdin and stdout is NUL-terminated")),
 958                 OPT_CALLBACK_F(0, "batch-command", &batch, N_("format"),
 959                         N_("read commands from stdin"),
 960                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 961                         batch_option_callback),
 962                 OPT_CMDMODE(0, "batch-all-objects", &opt,
 963                             N_("with --batch[-check]: ignores stdin, batches all known objects"), 'b'),
 964                 /* Batch-specific options */
 965                 OPT_GROUP(N_("Change or optimize batch output")),
 966                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 967                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 968                          N_("follow in-tree symlinks")),
 969                 OPT_BOOL(0, "unordered", &batch.unordered,
 970                          N_("do not order objects before emitting them")),
 971                 /* Textconv options, stand-ole*/
 972                 OPT_GROUP(N_("Emit object (blob or tree) with conversion or filter (stand-alone, or with batch)")),
 973                 OPT_CMDMODE(0, "textconv", &opt,
 974                             N_("run textconv on object's content"), 'c'),
 975                 OPT_CMDMODE(0, "filters", &opt,
 976                             N_("run filters on object's content"), 'w'),
 977                 OPT_STRING(0, "path", &force_path, N_("blob|tree"),
 978                            N_("use a <path> for (--textconv | --filters); Not with 'batch'")),
 979                 OPT_END()
 980         };
 981
 982         git_config(git_cat_file_config, NULL);
 983
 984         batch.buffer_output = -1;
 985
 986         argc = parse_options(argc, argv, prefix, options, usage, 0);
 987         opt_cw = (opt == 'c' || opt == 'w');
 988         opt_epts = (opt == 'e' || opt == 'p' || opt == 't' || opt == 's');
 989
 990         if (use_mailmap)
 991                 read_mailmap(&mailmap);
 992
 993         /* --batch-all-objects? */
 994         if (opt == 'b')
 995                 batch.all_objects = 1;
 996
 997         /* Option compatibility */
 998         if (force_path && !opt_cw)
 999                 usage_msg_optf(_("'%s=<%s>' needs '%s' or '%s'"),
1000                                usage, options,
1001                                "--path", _("path|tree-ish"), "--filters",
1002                                "--textconv");
1003
1004         /* Option compatibility with batch mode */
1005         if (batch.enabled)
1006                 ;
1007         else if (batch.follow_symlinks)
1008                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1009                                "--follow-symlinks");
1010         else if (batch.buffer_output >= 0)
1011                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1012                                "--buffer");
1013         else if (batch.all_objects)
1014                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1015                                "--batch-all-objects");
1016         else if (input_nul_terminated)
1017                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1018                                "-z");
1019         else if (nul_terminated)
1020                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1021                                "-Z");
1022
1023         batch.input_delim = batch.output_delim = '\n';
1024         if (input_nul_terminated)
1025                 batch.input_delim = '\0';
1026         if (nul_terminated)
1027                 batch.input_delim = batch.output_delim = '\0';
1028
1029         /* Batch defaults */
1030         if (batch.buffer_output < 0)
1031                 batch.buffer_output = batch.all_objects;
1032
1033         /* Return early if we're in batch mode? */
1034         if (batch.enabled) {
1035                 if (opt_cw)
1036                         batch.transform_mode = opt;
1037                 else if (opt && opt != 'b')
1038                         usage_msg_optf(_("'-%c' is incompatible with batch mode"),
1039                                        usage, options, opt);
1040                 else if (argc)
1041                         usage_msg_opt(_("batch modes take no arguments"), usage,
1042                                       options);
1043
1044                 return batch_objects(&batch);
1045         }
1046
1047         if (opt) {
1048                 if (!argc && opt == 'c')
1049                         usage_msg_optf(_("<rev> required with '%s'"),
1050                                        usage, options, "--textconv");
1051                 else if (!argc && opt == 'w')
1052                         usage_msg_optf(_("<rev> required with '%s'"),
1053                                        usage, options, "--filters");
1054                 else if (!argc && opt_epts)
1055                         usage_msg_optf(_("<object> required with '-%c'"),
1056                                        usage, options, opt);
1057                 else if (argc == 1)
1058                         obj_name = argv[0];
1059                 else
1060                         usage_msg_opt(_("too many arguments"), usage, options);
1061         } else if (!argc) {
1062                 usage_with_options(usage, options);
1063         } else if (argc != 2) {
1064                 usage_msg_optf(_("only two arguments allowed in <type> <object> mode, not %d"),
1065                               usage, options, argc);
1066         } else if (argc) {
1067                 exp_type = argv[0];
1068                 obj_name = argv[1];
1069         }
1070
1071         if (unknown_type && opt != 't' && opt != 's')
1072                 die("git cat-file --allow-unknown-type: use with -s or -t");
1073         return cat_one_file(opt, exp_type, obj_name, unknown_type);
1074 }