builtin/cat-file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  */
   6 #include "cache.h"
   7 #include "builtin.h"
   8 #include "parse-options.h"
   9 #include "userdiff.h"
  10 #include "streaming.h"
  11 #include "tree-walk.h"
  12 #include "sha1-array.h"
  13
  14 struct batch_options {
  15         int enabled;
  16         int follow_symlinks;
  17         int print_contents;
  18         int buffer_output;
  19         int all_objects;
  20         int cmdmode; /* may be 'w' or 'c' for --filters or --textconv */
  21         const char *format;
  22 };
  23
  24 static const char *force_path;
  25
  26 static int filter_object(const char *path, unsigned mode,
  27                          const struct object_id *oid,
  28                          char **buf, unsigned long *size)
  29 {
  30         enum object_type type;
  31
  32         *buf = read_sha1_file(oid->hash, &type, size);
  33         if (!*buf)
  34                 return error(_("cannot read object %s '%s'"),
  35                              oid_to_hex(oid), path);
  36         if ((type == OBJ_BLOB) && S_ISREG(mode)) {
  37                 struct strbuf strbuf = STRBUF_INIT;
  38                 if (convert_to_working_tree(path, *buf, *size, &strbuf)) {
  39                         free(*buf);
  40                         *size = strbuf.len;
  41                         *buf = strbuf_detach(&strbuf, NULL);
  42                 }
  43         }
  44
  45         return 0;
  46 }
  47
  48 static int cat_one_file(int opt, const char *exp_type, const char *obj_name,
  49                         int unknown_type)
  50 {
  51         struct object_id oid;
  52         enum object_type type;
  53         char *buf;
  54         unsigned long size;
  55         struct object_context obj_context;
  56         struct object_info oi = OBJECT_INFO_INIT;
  57         struct strbuf sb = STRBUF_INIT;
  58         unsigned flags = LOOKUP_REPLACE_OBJECT;
  59         const char *path = force_path;
  60
  61         if (unknown_type)
  62                 flags |= LOOKUP_UNKNOWN_OBJECT;
  63
  64         if (get_sha1_with_context(obj_name, 0, oid.hash, &obj_context))
  65                 die("Not a valid object name %s", obj_name);
  66
  67         if (!path)
  68                 path = obj_context.path;
  69         if (obj_context.mode == S_IFINVALID)
  70                 obj_context.mode = 0100644;
  71
  72         buf = NULL;
  73         switch (opt) {
  74         case 't':
  75                 oi.typename = &sb;
  76                 if (sha1_object_info_extended(oid.hash, &oi, flags) < 0)
  77                         die("git cat-file: could not get object info");
  78                 if (sb.len) {
  79                         printf("%s\n", sb.buf);
  80                         strbuf_release(&sb);
  81                         return 0;
  82                 }
  83                 break;
  84
  85         case 's':
  86                 oi.sizep = &size;
  87                 if (sha1_object_info_extended(oid.hash, &oi, flags) < 0)
  88                         die("git cat-file: could not get object info");
  89                 printf("%lu\n", size);
  90                 return 0;
  91
  92         case 'e':
  93                 return !has_object_file(&oid);
  94
  95         case 'w':
  96                 if (!path[0])
  97                         die("git cat-file --filters %s: <object> must be "
  98                             "<sha1:path>", obj_name);
  99
 100                 if (filter_object(path, obj_context.mode,
 101                                   &oid, &buf, &size))
 102                         return -1;
 103                 break;
 104
 105         case 'c':
 106                 if (!path[0])
 107                         die("git cat-file --textconv %s: <object> must be <sha1:path>",
 108                             obj_name);
 109
 110                 if (textconv_object(path, obj_context.mode, &oid, 1, &buf, &size))
 111                         break;
 112
 113         case 'p':
 114                 type = sha1_object_info(oid.hash, NULL);
 115                 if (type < 0)
 116                         die("Not a valid object name %s", obj_name);
 117
 118                 /* custom pretty-print here */
 119                 if (type == OBJ_TREE) {
 120                         const char *ls_args[3] = { NULL };
 121                         ls_args[0] =  "ls-tree";
 122                         ls_args[1] =  obj_name;
 123                         return cmd_ls_tree(2, ls_args, NULL);
 124                 }
 125
 126                 if (type == OBJ_BLOB)
 127                         return stream_blob_to_fd(1, &oid, NULL, 0);
 128                 buf = read_sha1_file(oid.hash, &type, &size);
 129                 if (!buf)
 130                         die("Cannot read object %s", obj_name);
 131
 132                 /* otherwise just spit out the data */
 133                 break;
 134
 135         case 0:
 136                 if (type_from_string(exp_type) == OBJ_BLOB) {
 137                         struct object_id blob_oid;
 138                         if (sha1_object_info(oid.hash, NULL) == OBJ_TAG) {
 139                                 char *buffer = read_sha1_file(oid.hash, &type, &size);
 140                                 const char *target;
 141                                 if (!skip_prefix(buffer, "object ", &target) ||
 142                                     get_oid_hex(target, &blob_oid))
 143                                         die("%s not a valid tag", oid_to_hex(&oid));
 144                                 free(buffer);
 145                         } else
 146                                 oidcpy(&blob_oid, &oid);
 147
 148                         if (sha1_object_info(blob_oid.hash, NULL) == OBJ_BLOB)
 149                                 return stream_blob_to_fd(1, &blob_oid, NULL, 0);
 150                         /*
 151                          * we attempted to dereference a tag to a blob
 152                          * and failed; there may be new dereference
 153                          * mechanisms this code is not aware of.
 154                          * fall-back to the usual case.
 155                          */
 156                 }
 157                 buf = read_object_with_reference(oid.hash, exp_type, &size, NULL);
 158                 break;
 159
 160         default:
 161                 die("git cat-file: unknown option: %s", exp_type);
 162         }
 163
 164         if (!buf)
 165                 die("git cat-file %s: bad file", obj_name);
 166
 167         write_or_die(1, buf, size);
 168         return 0;
 169 }
 170
 171 struct expand_data {
 172         struct object_id oid;
 173         enum object_type type;
 174         unsigned long size;
 175         off_t disk_size;
 176         const char *rest;
 177         struct object_id delta_base_oid;
 178
 179         /*
 180          * If mark_query is true, we do not expand anything, but rather
 181          * just mark the object_info with items we wish to query.
 182          */
 183         int mark_query;
 184
 185         /*
 186          * Whether to split the input on whitespace before feeding it to
 187          * get_sha1; this is decided during the mark_query phase based on
 188          * whether we have a %(rest) token in our format.
 189          */
 190         int split_on_whitespace;
 191
 192         /*
 193          * After a mark_query run, this object_info is set up to be
 194          * passed to sha1_object_info_extended. It will point to the data
 195          * elements above, so you can retrieve the response from there.
 196          */
 197         struct object_info info;
 198
 199         /*
 200          * This flag will be true if the requested batch format and options
 201          * don't require us to call sha1_object_info, which can then be
 202          * optimized out.
 203          */
 204         unsigned skip_object_info : 1;
 205 };
 206
 207 static int is_atom(const char *atom, const char *s, int slen)
 208 {
 209         int alen = strlen(atom);
 210         return alen == slen && !memcmp(atom, s, alen);
 211 }
 212
 213 static void expand_atom(struct strbuf *sb, const char *atom, int len,
 214                         void *vdata)
 215 {
 216         struct expand_data *data = vdata;
 217
 218         if (is_atom("objectname", atom, len)) {
 219                 if (!data->mark_query)
 220                         strbuf_addstr(sb, oid_to_hex(&data->oid));
 221         } else if (is_atom("objecttype", atom, len)) {
 222                 if (data->mark_query)
 223                         data->info.typep = &data->type;
 224                 else
 225                         strbuf_addstr(sb, typename(data->type));
 226         } else if (is_atom("objectsize", atom, len)) {
 227                 if (data->mark_query)
 228                         data->info.sizep = &data->size;
 229                 else
 230                         strbuf_addf(sb, "%lu", data->size);
 231         } else if (is_atom("objectsize:disk", atom, len)) {
 232                 if (data->mark_query)
 233                         data->info.disk_sizep = &data->disk_size;
 234                 else
 235                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 236         } else if (is_atom("rest", atom, len)) {
 237                 if (data->mark_query)
 238                         data->split_on_whitespace = 1;
 239                 else if (data->rest)
 240                         strbuf_addstr(sb, data->rest);
 241         } else if (is_atom("deltabase", atom, len)) {
 242                 if (data->mark_query)
 243                         data->info.delta_base_sha1 = data->delta_base_oid.hash;
 244                 else
 245                         strbuf_addstr(sb,
 246                                       oid_to_hex(&data->delta_base_oid));
 247         } else
 248                 die("unknown format element: %.*s", len, atom);
 249 }
 250
 251 static size_t expand_format(struct strbuf *sb, const char *start, void *data)
 252 {
 253         const char *end;
 254
 255         if (*start != '(')
 256                 return 0;
 257         end = strchr(start + 1, ')');
 258         if (!end)
 259                 die("format element '%s' does not end in ')'", start);
 260
 261         expand_atom(sb, start + 1, end - start - 1, data);
 262
 263         return end - start + 1;
 264 }
 265
 266 static void batch_write(struct batch_options *opt, const void *data, int len)
 267 {
 268         if (opt->buffer_output) {
 269                 if (fwrite(data, 1, len, stdout) != len)
 270                         die_errno("unable to write to stdout");
 271         } else
 272                 write_or_die(1, data, len);
 273 }
 274
 275 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
 276 {
 277         const struct object_id *oid = &data->oid;
 278
 279         assert(data->info.typep);
 280
 281         if (data->type == OBJ_BLOB) {
 282                 if (opt->buffer_output)
 283                         fflush(stdout);
 284                 if (opt->cmdmode) {
 285                         char *contents;
 286                         unsigned long size;
 287
 288                         if (!data->rest)
 289                                 die("missing path for '%s'", oid_to_hex(oid));
 290
 291                         if (opt->cmdmode == 'w') {
 292                                 if (filter_object(data->rest, 0100644, oid,
 293                                                   &contents, &size))
 294                                         die("could not convert '%s' %s",
 295                                             oid_to_hex(oid), data->rest);
 296                         } else if (opt->cmdmode == 'c') {
 297                                 enum object_type type;
 298                                 if (!textconv_object(data->rest, 0100644, oid,
 299                                                      1, &contents, &size))
 300                                         contents = read_sha1_file(oid->hash, &type,
 301                                                                   &size);
 302                                 if (!contents)
 303                                         die("could not convert '%s' %s",
 304                                             oid_to_hex(oid), data->rest);
 305                         } else
 306                                 die("BUG: invalid cmdmode: %c", opt->cmdmode);
 307                         batch_write(opt, contents, size);
 308                         free(contents);
 309                 } else if (stream_blob_to_fd(1, oid, NULL, 0) < 0)
 310                         die("unable to stream %s to stdout", oid_to_hex(oid));
 311         }
 312         else {
 313                 enum object_type type;
 314                 unsigned long size;
 315                 void *contents;
 316
 317                 contents = read_sha1_file(oid->hash, &type, &size);
 318                 if (!contents)
 319                         die("object %s disappeared", oid_to_hex(oid));
 320                 if (type != data->type)
 321                         die("object %s changed type!?", oid_to_hex(oid));
 322                 if (data->info.sizep && size != data->size)
 323                         die("object %s changed size!?", oid_to_hex(oid));
 324
 325                 batch_write(opt, contents, size);
 326                 free(contents);
 327         }
 328 }
 329
 330 static void batch_object_write(const char *obj_name, struct batch_options *opt,
 331                                struct expand_data *data)
 332 {
 333         struct strbuf buf = STRBUF_INIT;
 334
 335         if (!data->skip_object_info &&
 336             sha1_object_info_extended(data->oid.hash, &data->info, LOOKUP_REPLACE_OBJECT) < 0) {
 337                 printf("%s missing\n",
 338                        obj_name ? obj_name : oid_to_hex(&data->oid));
 339                 fflush(stdout);
 340                 return;
 341         }
 342
 343         strbuf_expand(&buf, opt->format, expand_format, data);
 344         strbuf_addch(&buf, '\n');
 345         batch_write(opt, buf.buf, buf.len);
 346         strbuf_release(&buf);
 347
 348         if (opt->print_contents) {
 349                 print_object_or_die(opt, data);
 350                 batch_write(opt, "\n", 1);
 351         }
 352 }
 353
 354 static void batch_one_object(const char *obj_name, struct batch_options *opt,
 355                              struct expand_data *data)
 356 {
 357         struct object_context ctx;
 358         int flags = opt->follow_symlinks ? GET_SHA1_FOLLOW_SYMLINKS : 0;
 359         enum follow_symlinks_result result;
 360
 361         result = get_sha1_with_context(obj_name, flags, data->oid.hash, &ctx);
 362         if (result != FOUND) {
 363                 switch (result) {
 364                 case MISSING_OBJECT:
 365                         printf("%s missing\n", obj_name);
 366                         break;
 367                 case DANGLING_SYMLINK:
 368                         printf("dangling %"PRIuMAX"\n%s\n",
 369                                (uintmax_t)strlen(obj_name), obj_name);
 370                         break;
 371                 case SYMLINK_LOOP:
 372                         printf("loop %"PRIuMAX"\n%s\n",
 373                                (uintmax_t)strlen(obj_name), obj_name);
 374                         break;
 375                 case NOT_DIR:
 376                         printf("notdir %"PRIuMAX"\n%s\n",
 377                                (uintmax_t)strlen(obj_name), obj_name);
 378                         break;
 379                 default:
 380                         die("BUG: unknown get_sha1_with_context result %d\n",
 381                                result);
 382                         break;
 383                 }
 384                 fflush(stdout);
 385                 return;
 386         }
 387
 388         if (ctx.mode == 0) {
 389                 printf("symlink %"PRIuMAX"\n%s\n",
 390                        (uintmax_t)ctx.symlink_path.len,
 391                        ctx.symlink_path.buf);
 392                 fflush(stdout);
 393                 return;
 394         }
 395
 396         batch_object_write(obj_name, opt, data);
 397 }
 398
 399 struct object_cb_data {
 400         struct batch_options *opt;
 401         struct expand_data *expand;
 402 };
 403
 404 static int batch_object_cb(const unsigned char sha1[20], void *vdata)
 405 {
 406         struct object_cb_data *data = vdata;
 407         hashcpy(data->expand->oid.hash, sha1);
 408         batch_object_write(NULL, data->opt, data->expand);
 409         return 0;
 410 }
 411
 412 static int batch_loose_object(const unsigned char *sha1,
 413                               const char *path,
 414                               void *data)
 415 {
 416         sha1_array_append(data, sha1);
 417         return 0;
 418 }
 419
 420 static int batch_packed_object(const unsigned char *sha1,
 421                                struct packed_git *pack,
 422                                uint32_t pos,
 423                                void *data)
 424 {
 425         sha1_array_append(data, sha1);
 426         return 0;
 427 }
 428
 429 static int batch_objects(struct batch_options *opt)
 430 {
 431         struct strbuf buf = STRBUF_INIT;
 432         struct expand_data data;
 433         int save_warning;
 434         int retval = 0;
 435
 436         if (!opt->format)
 437                 opt->format = "%(objectname) %(objecttype) %(objectsize)";
 438
 439         /*
 440          * Expand once with our special mark_query flag, which will prime the
 441          * object_info to be handed to sha1_object_info_extended for each
 442          * object.
 443          */
 444         memset(&data, 0, sizeof(data));
 445         data.mark_query = 1;
 446         strbuf_expand(&buf, opt->format, expand_format, &data);
 447         data.mark_query = 0;
 448         if (opt->cmdmode)
 449                 data.split_on_whitespace = 1;
 450
 451         if (opt->all_objects) {
 452                 struct object_info empty = OBJECT_INFO_INIT;
 453                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 454                         data.skip_object_info = 1;
 455         }
 456
 457         /*
 458          * If we are printing out the object, then always fill in the type,
 459          * since we will want to decide whether or not to stream.
 460          */
 461         if (opt->print_contents)
 462                 data.info.typep = &data.type;
 463
 464         if (opt->all_objects) {
 465                 struct sha1_array sa = SHA1_ARRAY_INIT;
 466                 struct object_cb_data cb;
 467
 468                 for_each_loose_object(batch_loose_object, &sa, 0);
 469                 for_each_packed_object(batch_packed_object, &sa, 0);
 470
 471                 cb.opt = opt;
 472                 cb.expand = &data;
 473                 sha1_array_for_each_unique(&sa, batch_object_cb, &cb);
 474
 475                 sha1_array_clear(&sa);
 476                 return 0;
 477         }
 478
 479         /*
 480          * We are going to call get_sha1 on a potentially very large number of
 481          * objects. In most large cases, these will be actual object sha1s. The
 482          * cost to double-check that each one is not also a ref (just so we can
 483          * warn) ends up dwarfing the actual cost of the object lookups
 484          * themselves. We can work around it by just turning off the warning.
 485          */
 486         save_warning = warn_on_object_refname_ambiguity;
 487         warn_on_object_refname_ambiguity = 0;
 488
 489         while (strbuf_getline(&buf, stdin) != EOF) {
 490                 if (data.split_on_whitespace) {
 491                         /*
 492                          * Split at first whitespace, tying off the beginning
 493                          * of the string and saving the remainder (or NULL) in
 494                          * data.rest.
 495                          */
 496                         char *p = strpbrk(buf.buf, " \t");
 497                         if (p) {
 498                                 while (*p && strchr(" \t", *p))
 499                                         *p++ = '\0';
 500                         }
 501                         data.rest = p;
 502                 }
 503
 504                 batch_one_object(buf.buf, opt, &data);
 505         }
 506
 507         strbuf_release(&buf);
 508         warn_on_object_refname_ambiguity = save_warning;
 509         return retval;
 510 }
 511
 512 static const char * const cat_file_usage[] = {
 513         N_("git cat-file (-t [--allow-unknown-type] | -s [--allow-unknown-type] | -e | -p | <type> | --textconv | --filters) [--path=<path>] <object>"),
 514         N_("git cat-file (--batch | --batch-check) [--follow-symlinks] [--textconv | --filters]"),
 515         NULL
 516 };
 517
 518 static int git_cat_file_config(const char *var, const char *value, void *cb)
 519 {
 520         if (userdiff_config(var, value) < 0)
 521                 return -1;
 522
 523         return git_default_config(var, value, cb);
 524 }
 525
 526 static int batch_option_callback(const struct option *opt,
 527                                  const char *arg,
 528                                  int unset)
 529 {
 530         struct batch_options *bo = opt->value;
 531
 532         if (bo->enabled) {
 533                 return 1;
 534         }
 535
 536         bo->enabled = 1;
 537         bo->print_contents = !strcmp(opt->long_name, "batch");
 538         bo->format = arg;
 539
 540         return 0;
 541 }
 542
 543 int cmd_cat_file(int argc, const char **argv, const char *prefix)
 544 {
 545         int opt = 0;
 546         const char *exp_type = NULL, *obj_name = NULL;
 547         struct batch_options batch = {0};
 548         int unknown_type = 0;
 549
 550         const struct option options[] = {
 551                 OPT_GROUP(N_("<type> can be one of: blob, tree, commit, tag")),
 552                 OPT_CMDMODE('t', NULL, &opt, N_("show object type"), 't'),
 553                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 554                 OPT_CMDMODE('e', NULL, &opt,
 555                             N_("exit with zero when there's no error"), 'e'),
 556                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print object's content"), 'p'),
 557                 OPT_CMDMODE(0, "textconv", &opt,
 558                             N_("for blob objects, run textconv on object's content"), 'c'),
 559                 OPT_CMDMODE(0, "filters", &opt,
 560                             N_("for blob objects, run filters on object's content"), 'w'),
 561                 OPT_STRING(0, "path", &force_path, N_("blob"),
 562                            N_("use a specific path for --textconv/--filters")),
 563                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 564                           N_("allow -s and -t to work with broken/corrupt objects")),
 565                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 566                 { OPTION_CALLBACK, 0, "batch", &batch, "format",
 567                         N_("show info and content of objects fed from the standard input"),
 568                         PARSE_OPT_OPTARG, batch_option_callback },
 569                 { OPTION_CALLBACK, 0, "batch-check", &batch, "format",
 570                         N_("show info about objects fed from the standard input"),
 571                         PARSE_OPT_OPTARG, batch_option_callback },
 572                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 573                          N_("follow in-tree symlinks (used with --batch or --batch-check)")),
 574                 OPT_BOOL(0, "batch-all-objects", &batch.all_objects,
 575                          N_("show all objects with --batch or --batch-check")),
 576                 OPT_END()
 577         };
 578
 579         git_config(git_cat_file_config, NULL);
 580
 581         batch.buffer_output = -1;
 582         argc = parse_options(argc, argv, prefix, options, cat_file_usage, 0);
 583
 584         if (opt) {
 585                 if (batch.enabled && (opt == 'c' || opt == 'w'))
 586                         batch.cmdmode = opt;
 587                 else if (argc == 1)
 588                         obj_name = argv[0];
 589                 else
 590                         usage_with_options(cat_file_usage, options);
 591         }
 592         if (!opt && !batch.enabled) {
 593                 if (argc == 2) {
 594                         exp_type = argv[0];
 595                         obj_name = argv[1];
 596                 } else
 597                         usage_with_options(cat_file_usage, options);
 598         }
 599         if (batch.enabled) {
 600                 if (batch.cmdmode != opt || argc)
 601                         usage_with_options(cat_file_usage, options);
 602                 if (batch.cmdmode && batch.all_objects)
 603                         die("--batch-all-objects cannot be combined with "
 604                             "--textconv nor with --filters");
 605         }
 606
 607         if ((batch.follow_symlinks || batch.all_objects) && !batch.enabled) {
 608                 usage_with_options(cat_file_usage, options);
 609         }
 610
 611         if (force_path && opt != 'c' && opt != 'w') {
 612                 error("--path=<path> needs --textconv or --filters");
 613                 usage_with_options(cat_file_usage, options);
 614         }
 615
 616         if (force_path && batch.enabled) {
 617                 error("--path=<path> incompatible with --batch");
 618                 usage_with_options(cat_file_usage, options);
 619         }
 620
 621         if (batch.buffer_output < 0)
 622                 batch.buffer_output = batch.all_objects;
 623
 624         if (batch.enabled)
 625                 return batch_objects(&batch);
 626
 627         if (unknown_type && opt != 't' && opt != 's')
 628                 die("git cat-file --allow-unknown-type: use with -s or -t");
 629         return cat_one_file(opt, exp_type, obj_name, unknown_type);
 630 }