builtin/cat-file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  */
   6 #define USE_THE_INDEX_VARIABLE
   7 #include "cache.h"
   8 #include "config.h"
   9 #include "builtin.h"
  10 #include "diff.h"
  11 #include "parse-options.h"
  12 #include "userdiff.h"
  13 #include "streaming.h"
  14 #include "tree-walk.h"
  15 #include "oid-array.h"
  16 #include "packfile.h"
  17 #include "object-store.h"
  18 #include "promisor-remote.h"
  19 #include "mailmap.h"
  20
  21 enum batch_mode {
  22         BATCH_MODE_CONTENTS,
  23         BATCH_MODE_INFO,
  24         BATCH_MODE_QUEUE_AND_DISPATCH,
  25 };
  26
  27 struct batch_options {
  28         int enabled;
  29         int follow_symlinks;
  30         enum batch_mode batch_mode;
  31         int buffer_output;
  32         int all_objects;
  33         int unordered;
  34         int transform_mode; /* may be 'w' or 'c' for --filters or --textconv */
  35         int nul_terminated;
  36         const char *format;
  37 };
  38
  39 static const char *force_path;
  40
  41 static struct string_list mailmap = STRING_LIST_INIT_NODUP;
  42 static int use_mailmap;
  43
  44 static char *replace_idents_using_mailmap(char *, size_t *);
  45
  46 static char *replace_idents_using_mailmap(char *object_buf, size_t *size)
  47 {
  48         struct strbuf sb = STRBUF_INIT;
  49         const char *headers[] = { "author ", "committer ", "tagger ", NULL };
  50
  51         strbuf_attach(&sb, object_buf, *size, *size + 1);
  52         apply_mailmap_to_header(&sb, headers, &mailmap);
  53         *size = sb.len;
  54         return strbuf_detach(&sb, NULL);
  55 }
  56
  57 static int filter_object(const char *path, unsigned mode,
  58                          const struct object_id *oid,
  59                          char **buf, unsigned long *size)
  60 {
  61         enum object_type type;
  62
  63         *buf = read_object_file(oid, &type, size);
  64         if (!*buf)
  65                 return error(_("cannot read object %s '%s'"),
  66                              oid_to_hex(oid), path);
  67         if ((type == OBJ_BLOB) && S_ISREG(mode)) {
  68                 struct strbuf strbuf = STRBUF_INIT;
  69                 struct checkout_metadata meta;
  70
  71                 init_checkout_metadata(&meta, NULL, NULL, oid);
  72                 if (convert_to_working_tree(&the_index, path, *buf, *size, &strbuf, &meta)) {
  73                         free(*buf);
  74                         *size = strbuf.len;
  75                         *buf = strbuf_detach(&strbuf, NULL);
  76                 }
  77         }
  78
  79         return 0;
  80 }
  81
  82 static int stream_blob(const struct object_id *oid)
  83 {
  84         if (stream_blob_to_fd(1, oid, NULL, 0))
  85                 die("unable to stream %s to stdout", oid_to_hex(oid));
  86         return 0;
  87 }
  88
  89 static int cat_one_file(int opt, const char *exp_type, const char *obj_name,
  90                         int unknown_type)
  91 {
  92         int ret;
  93         struct object_id oid;
  94         enum object_type type;
  95         char *buf;
  96         unsigned long size;
  97         struct object_context obj_context;
  98         struct object_info oi = OBJECT_INFO_INIT;
  99         struct strbuf sb = STRBUF_INIT;
 100         unsigned flags = OBJECT_INFO_LOOKUP_REPLACE;
 101         unsigned get_oid_flags = GET_OID_RECORD_PATH | GET_OID_ONLY_TO_DIE;
 102         const char *path = force_path;
 103         const int opt_cw = (opt == 'c' || opt == 'w');
 104         if (!path && opt_cw)
 105                 get_oid_flags |= GET_OID_REQUIRE_PATH;
 106
 107         if (unknown_type)
 108                 flags |= OBJECT_INFO_ALLOW_UNKNOWN_TYPE;
 109
 110         if (get_oid_with_context(the_repository, obj_name, get_oid_flags, &oid,
 111                                  &obj_context))
 112                 die("Not a valid object name %s", obj_name);
 113
 114         if (!path)
 115                 path = obj_context.path;
 116         if (obj_context.mode == S_IFINVALID)
 117                 obj_context.mode = 0100644;
 118
 119         buf = NULL;
 120         switch (opt) {
 121         case 't':
 122                 oi.type_name = &sb;
 123                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 124                         die("git cat-file: could not get object info");
 125                 if (sb.len) {
 126                         printf("%s\n", sb.buf);
 127                         strbuf_release(&sb);
 128                         ret = 0;
 129                         goto cleanup;
 130                 }
 131                 break;
 132
 133         case 's':
 134                 oi.sizep = &size;
 135
 136                 if (use_mailmap) {
 137                         oi.typep = &type;
 138                         oi.contentp = (void**)&buf;
 139                 }
 140
 141                 if (oid_object_info_extended(the_repository, &oid, &oi, flags) < 0)
 142                         die("git cat-file: could not get object info");
 143
 144                 if (use_mailmap && (type == OBJ_COMMIT || type == OBJ_TAG)) {
 145                         size_t s = size;
 146                         buf = replace_idents_using_mailmap(buf, &s);
 147                         size = cast_size_t_to_ulong(s);
 148                 }
 149
 150                 printf("%"PRIuMAX"\n", (uintmax_t)size);
 151                 ret = 0;
 152                 goto cleanup;
 153
 154         case 'e':
 155                 return !has_object_file(&oid);
 156
 157         case 'w':
 158
 159                 if (filter_object(path, obj_context.mode,
 160                                   &oid, &buf, &size)) {
 161                         ret = -1;
 162                         goto cleanup;
 163                 }
 164                 break;
 165
 166         case 'c':
 167                 if (textconv_object(the_repository, path, obj_context.mode,
 168                                     &oid, 1, &buf, &size))
 169                         break;
 170                 /* else fallthrough */
 171
 172         case 'p':
 173                 type = oid_object_info(the_repository, &oid, NULL);
 174                 if (type < 0)
 175                         die("Not a valid object name %s", obj_name);
 176
 177                 /* custom pretty-print here */
 178                 if (type == OBJ_TREE) {
 179                         const char *ls_args[3] = { NULL };
 180                         ls_args[0] =  "ls-tree";
 181                         ls_args[1] =  obj_name;
 182                         ret = cmd_ls_tree(2, ls_args, NULL);
 183                         goto cleanup;
 184                 }
 185
 186                 if (type == OBJ_BLOB) {
 187                         ret = stream_blob(&oid);
 188                         goto cleanup;
 189                 }
 190                 buf = read_object_file(&oid, &type, &size);
 191                 if (!buf)
 192                         die("Cannot read object %s", obj_name);
 193
 194                 if (use_mailmap) {
 195                         size_t s = size;
 196                         buf = replace_idents_using_mailmap(buf, &s);
 197                         size = cast_size_t_to_ulong(s);
 198                 }
 199
 200                 /* otherwise just spit out the data */
 201                 break;
 202
 203         case 0:
 204         {
 205                 enum object_type exp_type_id = type_from_string(exp_type);
 206
 207                 if (exp_type_id == OBJ_BLOB) {
 208                         struct object_id blob_oid;
 209                         if (oid_object_info(the_repository, &oid, NULL) == OBJ_TAG) {
 210                                 char *buffer = read_object_file(&oid, &type,
 211                                                                 &size);
 212                                 const char *target;
 213                                 if (!skip_prefix(buffer, "object ", &target) ||
 214                                     get_oid_hex(target, &blob_oid))
 215                                         die("%s not a valid tag", oid_to_hex(&oid));
 216                                 free(buffer);
 217                         } else
 218                                 oidcpy(&blob_oid, &oid);
 219
 220                         if (oid_object_info(the_repository, &blob_oid, NULL) == OBJ_BLOB) {
 221                                 ret = stream_blob(&blob_oid);
 222                                 goto cleanup;
 223                         }
 224                         /*
 225                          * we attempted to dereference a tag to a blob
 226                          * and failed; there may be new dereference
 227                          * mechanisms this code is not aware of.
 228                          * fall-back to the usual case.
 229                          */
 230                 }
 231                 buf = read_object_with_reference(the_repository, &oid,
 232                                                  exp_type_id, &size, NULL);
 233
 234                 if (use_mailmap) {
 235                         size_t s = size;
 236                         buf = replace_idents_using_mailmap(buf, &s);
 237                         size = cast_size_t_to_ulong(s);
 238                 }
 239                 break;
 240         }
 241         default:
 242                 die("git cat-file: unknown option: %s", exp_type);
 243         }
 244
 245         if (!buf)
 246                 die("git cat-file %s: bad file", obj_name);
 247
 248         write_or_die(1, buf, size);
 249         ret = 0;
 250 cleanup:
 251         free(buf);
 252         free(obj_context.path);
 253         return ret;
 254 }
 255
 256 struct expand_data {
 257         struct object_id oid;
 258         enum object_type type;
 259         unsigned long size;
 260         off_t disk_size;
 261         const char *rest;
 262         struct object_id delta_base_oid;
 263
 264         /*
 265          * If mark_query is true, we do not expand anything, but rather
 266          * just mark the object_info with items we wish to query.
 267          */
 268         int mark_query;
 269
 270         /*
 271          * Whether to split the input on whitespace before feeding it to
 272          * get_sha1; this is decided during the mark_query phase based on
 273          * whether we have a %(rest) token in our format.
 274          */
 275         int split_on_whitespace;
 276
 277         /*
 278          * After a mark_query run, this object_info is set up to be
 279          * passed to oid_object_info_extended. It will point to the data
 280          * elements above, so you can retrieve the response from there.
 281          */
 282         struct object_info info;
 283
 284         /*
 285          * This flag will be true if the requested batch format and options
 286          * don't require us to call oid_object_info, which can then be
 287          * optimized out.
 288          */
 289         unsigned skip_object_info : 1;
 290 };
 291
 292 static int is_atom(const char *atom, const char *s, int slen)
 293 {
 294         int alen = strlen(atom);
 295         return alen == slen && !memcmp(atom, s, alen);
 296 }
 297
 298 static void expand_atom(struct strbuf *sb, const char *atom, int len,
 299                         void *vdata)
 300 {
 301         struct expand_data *data = vdata;
 302
 303         if (is_atom("objectname", atom, len)) {
 304                 if (!data->mark_query)
 305                         strbuf_addstr(sb, oid_to_hex(&data->oid));
 306         } else if (is_atom("objecttype", atom, len)) {
 307                 if (data->mark_query)
 308                         data->info.typep = &data->type;
 309                 else
 310                         strbuf_addstr(sb, type_name(data->type));
 311         } else if (is_atom("objectsize", atom, len)) {
 312                 if (data->mark_query)
 313                         data->info.sizep = &data->size;
 314                 else
 315                         strbuf_addf(sb, "%"PRIuMAX , (uintmax_t)data->size);
 316         } else if (is_atom("objectsize:disk", atom, len)) {
 317                 if (data->mark_query)
 318                         data->info.disk_sizep = &data->disk_size;
 319                 else
 320                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 321         } else if (is_atom("rest", atom, len)) {
 322                 if (data->mark_query)
 323                         data->split_on_whitespace = 1;
 324                 else if (data->rest)
 325                         strbuf_addstr(sb, data->rest);
 326         } else if (is_atom("deltabase", atom, len)) {
 327                 if (data->mark_query)
 328                         data->info.delta_base_oid = &data->delta_base_oid;
 329                 else
 330                         strbuf_addstr(sb,
 331                                       oid_to_hex(&data->delta_base_oid));
 332         } else
 333                 die("unknown format element: %.*s", len, atom);
 334 }
 335
 336 static size_t expand_format(struct strbuf *sb, const char *start, void *data)
 337 {
 338         const char *end;
 339
 340         if (*start != '(')
 341                 return 0;
 342         end = strchr(start + 1, ')');
 343         if (!end)
 344                 die("format element '%s' does not end in ')'", start);
 345
 346         expand_atom(sb, start + 1, end - start - 1, data);
 347
 348         return end - start + 1;
 349 }
 350
 351 static void batch_write(struct batch_options *opt, const void *data, int len)
 352 {
 353         if (opt->buffer_output) {
 354                 if (fwrite(data, 1, len, stdout) != len)
 355                         die_errno("unable to write to stdout");
 356         } else
 357                 write_or_die(1, data, len);
 358 }
 359
 360 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
 361 {
 362         const struct object_id *oid = &data->oid;
 363
 364         assert(data->info.typep);
 365
 366         if (data->type == OBJ_BLOB) {
 367                 if (opt->buffer_output)
 368                         fflush(stdout);
 369                 if (opt->transform_mode) {
 370                         char *contents;
 371                         unsigned long size;
 372
 373                         if (!data->rest)
 374                                 die("missing path for '%s'", oid_to_hex(oid));
 375
 376                         if (opt->transform_mode == 'w') {
 377                                 if (filter_object(data->rest, 0100644, oid,
 378                                                   &contents, &size))
 379                                         die("could not convert '%s' %s",
 380                                             oid_to_hex(oid), data->rest);
 381                         } else if (opt->transform_mode == 'c') {
 382                                 enum object_type type;
 383                                 if (!textconv_object(the_repository,
 384                                                      data->rest, 0100644, oid,
 385                                                      1, &contents, &size))
 386                                         contents = read_object_file(oid,
 387                                                                     &type,
 388                                                                     &size);
 389                                 if (!contents)
 390                                         die("could not convert '%s' %s",
 391                                             oid_to_hex(oid), data->rest);
 392                         } else
 393                                 BUG("invalid transform_mode: %c", opt->transform_mode);
 394                         batch_write(opt, contents, size);
 395                         free(contents);
 396                 } else {
 397                         stream_blob(oid);
 398                 }
 399         }
 400         else {
 401                 enum object_type type;
 402                 unsigned long size;
 403                 void *contents;
 404
 405                 contents = read_object_file(oid, &type, &size);
 406
 407                 if (use_mailmap) {
 408                         size_t s = size;
 409                         contents = replace_idents_using_mailmap(contents, &s);
 410                         size = cast_size_t_to_ulong(s);
 411                 }
 412
 413                 if (!contents)
 414                         die("object %s disappeared", oid_to_hex(oid));
 415                 if (type != data->type)
 416                         die("object %s changed type!?", oid_to_hex(oid));
 417                 if (data->info.sizep && size != data->size && !use_mailmap)
 418                         die("object %s changed size!?", oid_to_hex(oid));
 419
 420                 batch_write(opt, contents, size);
 421                 free(contents);
 422         }
 423 }
 424
 425 static void print_default_format(struct strbuf *scratch, struct expand_data *data)
 426 {
 427         strbuf_addf(scratch, "%s %s %"PRIuMAX"\n", oid_to_hex(&data->oid),
 428                     type_name(data->type),
 429                     (uintmax_t)data->size);
 430 }
 431
 432 /*
 433  * If "pack" is non-NULL, then "offset" is the byte offset within the pack from
 434  * which the object may be accessed (though note that we may also rely on
 435  * data->oid, too). If "pack" is NULL, then offset is ignored.
 436  */
 437 static void batch_object_write(const char *obj_name,
 438                                struct strbuf *scratch,
 439                                struct batch_options *opt,
 440                                struct expand_data *data,
 441                                struct packed_git *pack,
 442                                off_t offset)
 443 {
 444         if (!data->skip_object_info) {
 445                 int ret;
 446
 447                 if (use_mailmap)
 448                         data->info.typep = &data->type;
 449
 450                 if (pack)
 451                         ret = packed_object_info(the_repository, pack, offset,
 452                                                  &data->info);
 453                 else
 454                         ret = oid_object_info_extended(the_repository,
 455                                                        &data->oid, &data->info,
 456                                                        OBJECT_INFO_LOOKUP_REPLACE);
 457                 if (ret < 0) {
 458                         printf("%s missing\n",
 459                                obj_name ? obj_name : oid_to_hex(&data->oid));
 460                         fflush(stdout);
 461                         return;
 462                 }
 463
 464                 if (use_mailmap && (data->type == OBJ_COMMIT || data->type == OBJ_TAG)) {
 465                         size_t s = data->size;
 466                         char *buf = NULL;
 467
 468                         buf = repo_read_object_file(the_repository, &data->oid, &data->type,
 469                                                     &data->size);
 470                         buf = replace_idents_using_mailmap(buf, &s);
 471                         data->size = cast_size_t_to_ulong(s);
 472
 473                         free(buf);
 474                 }
 475         }
 476
 477         strbuf_reset(scratch);
 478
 479         if (!opt->format) {
 480                 print_default_format(scratch, data);
 481         } else {
 482                 strbuf_expand(scratch, opt->format, expand_format, data);
 483                 strbuf_addch(scratch, '\n');
 484         }
 485
 486         batch_write(opt, scratch->buf, scratch->len);
 487
 488         if (opt->batch_mode == BATCH_MODE_CONTENTS) {
 489                 print_object_or_die(opt, data);
 490                 batch_write(opt, "\n", 1);
 491         }
 492 }
 493
 494 static void batch_one_object(const char *obj_name,
 495                              struct strbuf *scratch,
 496                              struct batch_options *opt,
 497                              struct expand_data *data)
 498 {
 499         struct object_context ctx;
 500         int flags = opt->follow_symlinks ? GET_OID_FOLLOW_SYMLINKS : 0;
 501         enum get_oid_result result;
 502
 503         result = get_oid_with_context(the_repository, obj_name,
 504                                       flags, &data->oid, &ctx);
 505         if (result != FOUND) {
 506                 switch (result) {
 507                 case MISSING_OBJECT:
 508                         printf("%s missing\n", obj_name);
 509                         break;
 510                 case SHORT_NAME_AMBIGUOUS:
 511                         printf("%s ambiguous\n", obj_name);
 512                         break;
 513                 case DANGLING_SYMLINK:
 514                         printf("dangling %"PRIuMAX"\n%s\n",
 515                                (uintmax_t)strlen(obj_name), obj_name);
 516                         break;
 517                 case SYMLINK_LOOP:
 518                         printf("loop %"PRIuMAX"\n%s\n",
 519                                (uintmax_t)strlen(obj_name), obj_name);
 520                         break;
 521                 case NOT_DIR:
 522                         printf("notdir %"PRIuMAX"\n%s\n",
 523                                (uintmax_t)strlen(obj_name), obj_name);
 524                         break;
 525                 default:
 526                         BUG("unknown get_sha1_with_context result %d\n",
 527                                result);
 528                         break;
 529                 }
 530                 fflush(stdout);
 531                 return;
 532         }
 533
 534         if (ctx.mode == 0) {
 535                 printf("symlink %"PRIuMAX"\n%s\n",
 536                        (uintmax_t)ctx.symlink_path.len,
 537                        ctx.symlink_path.buf);
 538                 fflush(stdout);
 539                 return;
 540         }
 541
 542         batch_object_write(obj_name, scratch, opt, data, NULL, 0);
 543 }
 544
 545 struct object_cb_data {
 546         struct batch_options *opt;
 547         struct expand_data *expand;
 548         struct oidset *seen;
 549         struct strbuf *scratch;
 550 };
 551
 552 static int batch_object_cb(const struct object_id *oid, void *vdata)
 553 {
 554         struct object_cb_data *data = vdata;
 555         oidcpy(&data->expand->oid, oid);
 556         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 557                            NULL, 0);
 558         return 0;
 559 }
 560
 561 static int collect_loose_object(const struct object_id *oid,
 562                                 const char *path,
 563                                 void *data)
 564 {
 565         oid_array_append(data, oid);
 566         return 0;
 567 }
 568
 569 static int collect_packed_object(const struct object_id *oid,
 570                                  struct packed_git *pack,
 571                                  uint32_t pos,
 572                                  void *data)
 573 {
 574         oid_array_append(data, oid);
 575         return 0;
 576 }
 577
 578 static int batch_unordered_object(const struct object_id *oid,
 579                                   struct packed_git *pack, off_t offset,
 580                                   void *vdata)
 581 {
 582         struct object_cb_data *data = vdata;
 583
 584         if (oidset_insert(data->seen, oid))
 585                 return 0;
 586
 587         oidcpy(&data->expand->oid, oid);
 588         batch_object_write(NULL, data->scratch, data->opt, data->expand,
 589                            pack, offset);
 590         return 0;
 591 }
 592
 593 static int batch_unordered_loose(const struct object_id *oid,
 594                                  const char *path,
 595                                  void *data)
 596 {
 597         return batch_unordered_object(oid, NULL, 0, data);
 598 }
 599
 600 static int batch_unordered_packed(const struct object_id *oid,
 601                                   struct packed_git *pack,
 602                                   uint32_t pos,
 603                                   void *data)
 604 {
 605         return batch_unordered_object(oid, pack,
 606                                       nth_packed_object_offset(pack, pos),
 607                                       data);
 608 }
 609
 610 typedef void (*parse_cmd_fn_t)(struct batch_options *, const char *,
 611                                struct strbuf *, struct expand_data *);
 612
 613 struct queued_cmd {
 614         parse_cmd_fn_t fn;
 615         char *line;
 616 };
 617
 618 static void parse_cmd_contents(struct batch_options *opt,
 619                              const char *line,
 620                              struct strbuf *output,
 621                              struct expand_data *data)
 622 {
 623         opt->batch_mode = BATCH_MODE_CONTENTS;
 624         batch_one_object(line, output, opt, data);
 625 }
 626
 627 static void parse_cmd_info(struct batch_options *opt,
 628                            const char *line,
 629                            struct strbuf *output,
 630                            struct expand_data *data)
 631 {
 632         opt->batch_mode = BATCH_MODE_INFO;
 633         batch_one_object(line, output, opt, data);
 634 }
 635
 636 static void dispatch_calls(struct batch_options *opt,
 637                 struct strbuf *output,
 638                 struct expand_data *data,
 639                 struct queued_cmd *cmd,
 640                 int nr)
 641 {
 642         int i;
 643
 644         if (!opt->buffer_output)
 645                 die(_("flush is only for --buffer mode"));
 646
 647         for (i = 0; i < nr; i++)
 648                 cmd[i].fn(opt, cmd[i].line, output, data);
 649
 650         fflush(stdout);
 651 }
 652
 653 static void free_cmds(struct queued_cmd *cmd, size_t *nr)
 654 {
 655         size_t i;
 656
 657         for (i = 0; i < *nr; i++)
 658                 FREE_AND_NULL(cmd[i].line);
 659
 660         *nr = 0;
 661 }
 662
 663
 664 static const struct parse_cmd {
 665         const char *name;
 666         parse_cmd_fn_t fn;
 667         unsigned takes_args;
 668 } commands[] = {
 669         { "contents", parse_cmd_contents, 1},
 670         { "info", parse_cmd_info, 1},
 671         { "flush", NULL, 0},
 672 };
 673
 674 static void batch_objects_command(struct batch_options *opt,
 675                                     struct strbuf *output,
 676                                     struct expand_data *data)
 677 {
 678         struct strbuf input = STRBUF_INIT;
 679         struct queued_cmd *queued_cmd = NULL;
 680         size_t alloc = 0, nr = 0;
 681
 682         while (1) {
 683                 int i, ret;
 684                 const struct parse_cmd *cmd = NULL;
 685                 const char *p = NULL, *cmd_end;
 686                 struct queued_cmd call = {0};
 687
 688                 if (opt->nul_terminated)
 689                         ret = strbuf_getline_nul(&input, stdin);
 690                 else
 691                         ret = strbuf_getline(&input, stdin);
 692
 693                 if (ret)
 694                         break;
 695
 696                 if (!input.len)
 697                         die(_("empty command in input"));
 698                 if (isspace(*input.buf))
 699                         die(_("whitespace before command: '%s'"), input.buf);
 700
 701                 for (i = 0; i < ARRAY_SIZE(commands); i++) {
 702                         if (!skip_prefix(input.buf, commands[i].name, &cmd_end))
 703                                 continue;
 704
 705                         cmd = &commands[i];
 706                         if (cmd->takes_args) {
 707                                 if (*cmd_end != ' ')
 708                                         die(_("%s requires arguments"),
 709                                             commands[i].name);
 710
 711                                 p = cmd_end + 1;
 712                         } else if (*cmd_end) {
 713                                 die(_("%s takes no arguments"),
 714                                     commands[i].name);
 715                         }
 716
 717                         break;
 718                 }
 719
 720                 if (!cmd)
 721                         die(_("unknown command: '%s'"), input.buf);
 722
 723                 if (!strcmp(cmd->name, "flush")) {
 724                         dispatch_calls(opt, output, data, queued_cmd, nr);
 725                         free_cmds(queued_cmd, &nr);
 726                 } else if (!opt->buffer_output) {
 727                         cmd->fn(opt, p, output, data);
 728                 } else {
 729                         ALLOC_GROW(queued_cmd, nr + 1, alloc);
 730                         call.fn = cmd->fn;
 731                         call.line = xstrdup_or_null(p);
 732                         queued_cmd[nr++] = call;
 733                 }
 734         }
 735
 736         if (opt->buffer_output &&
 737             nr &&
 738             !git_env_bool("GIT_TEST_CAT_FILE_NO_FLUSH_ON_EXIT", 0)) {
 739                 dispatch_calls(opt, output, data, queued_cmd, nr);
 740                 free_cmds(queued_cmd, &nr);
 741         }
 742
 743         free_cmds(queued_cmd, &nr);
 744         free(queued_cmd);
 745         strbuf_release(&input);
 746 }
 747
 748 #define DEFAULT_FORMAT "%(objectname) %(objecttype) %(objectsize)"
 749
 750 static int batch_objects(struct batch_options *opt)
 751 {
 752         struct strbuf input = STRBUF_INIT;
 753         struct strbuf output = STRBUF_INIT;
 754         struct expand_data data;
 755         int save_warning;
 756         int retval = 0;
 757
 758         /*
 759          * Expand once with our special mark_query flag, which will prime the
 760          * object_info to be handed to oid_object_info_extended for each
 761          * object.
 762          */
 763         memset(&data, 0, sizeof(data));
 764         data.mark_query = 1;
 765         strbuf_expand(&output,
 766                       opt->format ? opt->format : DEFAULT_FORMAT,
 767                       expand_format,
 768                       &data);
 769         data.mark_query = 0;
 770         strbuf_release(&output);
 771         if (opt->transform_mode)
 772                 data.split_on_whitespace = 1;
 773
 774         if (opt->format && !strcmp(opt->format, DEFAULT_FORMAT))
 775                 opt->format = NULL;
 776         /*
 777          * If we are printing out the object, then always fill in the type,
 778          * since we will want to decide whether or not to stream.
 779          */
 780         if (opt->batch_mode == BATCH_MODE_CONTENTS)
 781                 data.info.typep = &data.type;
 782
 783         if (opt->all_objects) {
 784                 struct object_cb_data cb;
 785                 struct object_info empty = OBJECT_INFO_INIT;
 786
 787                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 788                         data.skip_object_info = 1;
 789
 790                 if (has_promisor_remote())
 791                         warning("This repository uses promisor remotes. Some objects may not be loaded.");
 792
 793                 read_replace_refs = 0;
 794
 795                 cb.opt = opt;
 796                 cb.expand = &data;
 797                 cb.scratch = &output;
 798
 799                 if (opt->unordered) {
 800                         struct oidset seen = OIDSET_INIT;
 801
 802                         cb.seen = &seen;
 803
 804                         for_each_loose_object(batch_unordered_loose, &cb, 0);
 805                         for_each_packed_object(batch_unordered_packed, &cb,
 806                                                FOR_EACH_OBJECT_PACK_ORDER);
 807
 808                         oidset_clear(&seen);
 809                 } else {
 810                         struct oid_array sa = OID_ARRAY_INIT;
 811
 812                         for_each_loose_object(collect_loose_object, &sa, 0);
 813                         for_each_packed_object(collect_packed_object, &sa, 0);
 814
 815                         oid_array_for_each_unique(&sa, batch_object_cb, &cb);
 816
 817                         oid_array_clear(&sa);
 818                 }
 819
 820                 strbuf_release(&output);
 821                 return 0;
 822         }
 823
 824         /*
 825          * We are going to call get_sha1 on a potentially very large number of
 826          * objects. In most large cases, these will be actual object sha1s. The
 827          * cost to double-check that each one is not also a ref (just so we can
 828          * warn) ends up dwarfing the actual cost of the object lookups
 829          * themselves. We can work around it by just turning off the warning.
 830          */
 831         save_warning = warn_on_object_refname_ambiguity;
 832         warn_on_object_refname_ambiguity = 0;
 833
 834         if (opt->batch_mode == BATCH_MODE_QUEUE_AND_DISPATCH) {
 835                 batch_objects_command(opt, &output, &data);
 836                 goto cleanup;
 837         }
 838
 839         while (1) {
 840                 int ret;
 841                 if (opt->nul_terminated)
 842                         ret = strbuf_getline_nul(&input, stdin);
 843                 else
 844                         ret = strbuf_getline(&input, stdin);
 845
 846                 if (ret == EOF)
 847                         break;
 848
 849                 if (data.split_on_whitespace) {
 850                         /*
 851                          * Split at first whitespace, tying off the beginning
 852                          * of the string and saving the remainder (or NULL) in
 853                          * data.rest.
 854                          */
 855                         char *p = strpbrk(input.buf, " \t");
 856                         if (p) {
 857                                 while (*p && strchr(" \t", *p))
 858                                         *p++ = '\0';
 859                         }
 860                         data.rest = p;
 861                 }
 862
 863                 batch_one_object(input.buf, &output, opt, &data);
 864         }
 865
 866  cleanup:
 867         strbuf_release(&input);
 868         strbuf_release(&output);
 869         warn_on_object_refname_ambiguity = save_warning;
 870         return retval;
 871 }
 872
 873 static int git_cat_file_config(const char *var, const char *value, void *cb)
 874 {
 875         if (userdiff_config(var, value) < 0)
 876                 return -1;
 877
 878         return git_default_config(var, value, 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
 918         const char * const usage[] = {
 919                 N_("git cat-file <type> <object>"),
 920                 N_("git cat-file (-e | -p) <object>"),
 921                 N_("git cat-file (-t | -s) [--allow-unknown-type] <object>"),
 922                 N_("git cat-file (--batch | --batch-check | --batch-command) [--batch-all-objects]\n"
 923                    "             [--buffer] [--follow-symlinks] [--unordered]\n"
 924                    "             [--textconv | --filters] [-z]"),
 925                 N_("git cat-file (--textconv | --filters)\n"
 926                    "             [<rev>:<path|tree-ish> | --path=<path|tree-ish> <rev>]"),
 927                 NULL
 928         };
 929         const struct option options[] = {
 930                 /* Simple queries */
 931                 OPT_GROUP(N_("Check object existence or emit object contents")),
 932                 OPT_CMDMODE('e', NULL, &opt,
 933                             N_("check if <object> exists"), 'e'),
 934                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print <object> content"), 'p'),
 935
 936                 OPT_GROUP(N_("Emit [broken] object attributes")),
 937                 OPT_CMDMODE('t', NULL, &opt, N_("show object type (one of 'blob', 'tree', 'commit', 'tag', ...)"), 't'),
 938                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 939                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 940                           N_("allow -s and -t to work with broken/corrupt objects")),
 941                 OPT_BOOL(0, "use-mailmap", &use_mailmap, N_("use mail map file")),
 942                 OPT_ALIAS(0, "mailmap", "use-mailmap"),
 943                 /* Batch mode */
 944                 OPT_GROUP(N_("Batch objects requested on stdin (or --batch-all-objects)")),
 945                 OPT_CALLBACK_F(0, "batch", &batch, N_("format"),
 946                         N_("show full <object> or <rev> contents"),
 947                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 948                         batch_option_callback),
 949                 OPT_CALLBACK_F(0, "batch-check", &batch, N_("format"),
 950                         N_("like --batch, but don't emit <contents>"),
 951                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 952                         batch_option_callback),
 953                 OPT_BOOL('z', NULL, &batch.nul_terminated, N_("stdin is NUL-terminated")),
 954                 OPT_CALLBACK_F(0, "batch-command", &batch, N_("format"),
 955                         N_("read commands from stdin"),
 956                         PARSE_OPT_OPTARG | PARSE_OPT_NONEG,
 957                         batch_option_callback),
 958                 OPT_CMDMODE(0, "batch-all-objects", &opt,
 959                             N_("with --batch[-check]: ignores stdin, batches all known objects"), 'b'),
 960                 /* Batch-specific options */
 961                 OPT_GROUP(N_("Change or optimize batch output")),
 962                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 963                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 964                          N_("follow in-tree symlinks")),
 965                 OPT_BOOL(0, "unordered", &batch.unordered,
 966                          N_("do not order objects before emitting them")),
 967                 /* Textconv options, stand-ole*/
 968                 OPT_GROUP(N_("Emit object (blob or tree) with conversion or filter (stand-alone, or with batch)")),
 969                 OPT_CMDMODE(0, "textconv", &opt,
 970                             N_("run textconv on object's content"), 'c'),
 971                 OPT_CMDMODE(0, "filters", &opt,
 972                             N_("run filters on object's content"), 'w'),
 973                 OPT_STRING(0, "path", &force_path, N_("blob|tree"),
 974                            N_("use a <path> for (--textconv | --filters); Not with 'batch'")),
 975                 OPT_END()
 976         };
 977
 978         git_config(git_cat_file_config, NULL);
 979
 980         batch.buffer_output = -1;
 981
 982         argc = parse_options(argc, argv, prefix, options, usage, 0);
 983         opt_cw = (opt == 'c' || opt == 'w');
 984         opt_epts = (opt == 'e' || opt == 'p' || opt == 't' || opt == 's');
 985
 986         if (use_mailmap)
 987                 read_mailmap(&mailmap);
 988
 989         /* --batch-all-objects? */
 990         if (opt == 'b')
 991                 batch.all_objects = 1;
 992
 993         /* Option compatibility */
 994         if (force_path && !opt_cw)
 995                 usage_msg_optf(_("'%s=<%s>' needs '%s' or '%s'"),
 996                                usage, options,
 997                                "--path", _("path|tree-ish"), "--filters",
 998                                "--textconv");
 999
1000         /* Option compatibility with batch mode */
1001         if (batch.enabled)
1002                 ;
1003         else if (batch.follow_symlinks)
1004                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1005                                "--follow-symlinks");
1006         else if (batch.buffer_output >= 0)
1007                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1008                                "--buffer");
1009         else if (batch.all_objects)
1010                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1011                                "--batch-all-objects");
1012         else if (batch.nul_terminated)
1013                 usage_msg_optf(_("'%s' requires a batch mode"), usage, options,
1014                                "-z");
1015
1016         /* Batch defaults */
1017         if (batch.buffer_output < 0)
1018                 batch.buffer_output = batch.all_objects;
1019
1020         /* Return early if we're in batch mode? */
1021         if (batch.enabled) {
1022                 if (opt_cw)
1023                         batch.transform_mode = opt;
1024                 else if (opt && opt != 'b')
1025                         usage_msg_optf(_("'-%c' is incompatible with batch mode"),
1026                                        usage, options, opt);
1027                 else if (argc)
1028                         usage_msg_opt(_("batch modes take no arguments"), usage,
1029                                       options);
1030
1031                 return batch_objects(&batch);
1032         }
1033
1034         if (opt) {
1035                 if (!argc && opt == 'c')
1036                         usage_msg_optf(_("<rev> required with '%s'"),
1037                                        usage, options, "--textconv");
1038                 else if (!argc && opt == 'w')
1039                         usage_msg_optf(_("<rev> required with '%s'"),
1040                                        usage, options, "--filters");
1041                 else if (!argc && opt_epts)
1042                         usage_msg_optf(_("<object> required with '-%c'"),
1043                                        usage, options, opt);
1044                 else if (argc == 1)
1045                         obj_name = argv[0];
1046                 else
1047                         usage_msg_opt(_("too many arguments"), usage, options);
1048         } else if (!argc) {
1049                 usage_with_options(usage, options);
1050         } else if (argc != 2) {
1051                 usage_msg_optf(_("only two arguments allowed in <type> <object> mode, not %d"),
1052                               usage, options, argc);
1053         } else if (argc) {
1054                 exp_type = argv[0];
1055                 obj_name = argv[1];
1056         }
1057
1058         if (unknown_type && opt != 't' && opt != 's')
1059                 die("git cat-file --allow-unknown-type: use with -s or -t");
1060         return cat_one_file(opt, exp_type, obj_name, unknown_type);
1061 }