git.oblomov.eu Git - git/blob - 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, GET_SHA1_RECORD_PATH,
  65                                   oid.hash, &obj_context))
  66                 die("Not a valid object name %s", obj_name);
  67
  68         if (!path)
  69                 path = obj_context.path;
  70         if (obj_context.mode == S_IFINVALID)
  71                 obj_context.mode = 0100644;
  72
  73         buf = NULL;
  74         switch (opt) {
  75         case 't':
  76                 oi.typename = &sb;
  77                 if (sha1_object_info_extended(oid.hash, &oi, flags) < 0)
  78                         die("git cat-file: could not get object info");
  79                 if (sb.len) {
  80                         printf("%s\n", sb.buf);
  81                         strbuf_release(&sb);
  82                         return 0;
  83                 }
  84                 break;
  85
  86         case 's':
  87                 oi.sizep = &size;
  88                 if (sha1_object_info_extended(oid.hash, &oi, flags) < 0)
  89                         die("git cat-file: could not get object info");
  90                 printf("%lu\n", size);
  91                 return 0;
  92
  93         case 'e':
  94                 return !has_object_file(&oid);
  95
  96         case 'w':
  97                 if (!path)
  98                         die("git cat-file --filters %s: <object> must be "
  99                             "<sha1:path>", obj_name);
 100
 101                 if (filter_object(path, obj_context.mode,
 102                                   &oid, &buf, &size))
 103                         return -1;
 104                 break;
 105
 106         case 'c':
 107                 if (!path)
 108                         die("git cat-file --textconv %s: <object> must be <sha1:path>",
 109                             obj_name);
 110
 111                 if (textconv_object(path, obj_context.mode, &oid, 1, &buf, &size))
 112                         break;
 113
 114         case 'p':
 115                 type = sha1_object_info(oid.hash, NULL);
 116                 if (type < 0)
 117                         die("Not a valid object name %s", obj_name);
 118
 119                 /* custom pretty-print here */
 120                 if (type == OBJ_TREE) {
 121                         const char *ls_args[3] = { NULL };
 122                         ls_args[0] =  "ls-tree";
 123                         ls_args[1] =  obj_name;
 124                         return cmd_ls_tree(2, ls_args, NULL);
 125                 }
 126
 127                 if (type == OBJ_BLOB)
 128                         return stream_blob_to_fd(1, &oid, NULL, 0);
 129                 buf = read_sha1_file(oid.hash, &type, &size);
 130                 if (!buf)
 131                         die("Cannot read object %s", obj_name);
 132
 133                 /* otherwise just spit out the data */
 134                 break;
 135
 136         case 0:
 137                 if (type_from_string(exp_type) == OBJ_BLOB) {
 138                         struct object_id blob_oid;
 139                         if (sha1_object_info(oid.hash, NULL) == OBJ_TAG) {
 140                                 char *buffer = read_sha1_file(oid.hash, &type, &size);
 141                                 const char *target;
 142                                 if (!skip_prefix(buffer, "object ", &target) ||
 143                                     get_oid_hex(target, &blob_oid))
 144                                         die("%s not a valid tag", oid_to_hex(&oid));
 145                                 free(buffer);
 146                         } else
 147                                 oidcpy(&blob_oid, &oid);
 148
 149                         if (sha1_object_info(blob_oid.hash, NULL) == OBJ_BLOB)
 150                                 return stream_blob_to_fd(1, &blob_oid, NULL, 0);
 151                         /*
 152                          * we attempted to dereference a tag to a blob
 153                          * and failed; there may be new dereference
 154                          * mechanisms this code is not aware of.
 155                          * fall-back to the usual case.
 156                          */
 157                 }
 158                 buf = read_object_with_reference(oid.hash, exp_type, &size, NULL);
 159                 break;
 160
 161         default:
 162                 die("git cat-file: unknown option: %s", exp_type);
 163         }
 164
 165         if (!buf)
 166                 die("git cat-file %s: bad file", obj_name);
 167
 168         write_or_die(1, buf, size);
 169         free(obj_context.path);
 170         return 0;
 171 }
 172
 173 struct expand_data {
 174         struct object_id oid;
 175         enum object_type type;
 176         unsigned long size;
 177         off_t disk_size;
 178         const char *rest;
 179         struct object_id delta_base_oid;
 180
 181         /*
 182          * If mark_query is true, we do not expand anything, but rather
 183          * just mark the object_info with items we wish to query.
 184          */
 185         int mark_query;
 186
 187         /*
 188          * Whether to split the input on whitespace before feeding it to
 189          * get_sha1; this is decided during the mark_query phase based on
 190          * whether we have a %(rest) token in our format.
 191          */
 192         int split_on_whitespace;
 193
 194         /*
 195          * After a mark_query run, this object_info is set up to be
 196          * passed to sha1_object_info_extended. It will point to the data
 197          * elements above, so you can retrieve the response from there.
 198          */
 199         struct object_info info;
 200
 201         /*
 202          * This flag will be true if the requested batch format and options
 203          * don't require us to call sha1_object_info, which can then be
 204          * optimized out.
 205          */
 206         unsigned skip_object_info : 1;
 207 };
 208
 209 static int is_atom(const char *atom, const char *s, int slen)
 210 {
 211         int alen = strlen(atom);
 212         return alen == slen && !memcmp(atom, s, alen);
 213 }
 214
 215 static void expand_atom(struct strbuf *sb, const char *atom, int len,
 216                         void *vdata)
 217 {
 218         struct expand_data *data = vdata;
 219
 220         if (is_atom("objectname", atom, len)) {
 221                 if (!data->mark_query)
 222                         strbuf_addstr(sb, oid_to_hex(&data->oid));
 223         } else if (is_atom("objecttype", atom, len)) {
 224                 if (data->mark_query)
 225                         data->info.typep = &data->type;
 226                 else
 227                         strbuf_addstr(sb, typename(data->type));
 228         } else if (is_atom("objectsize", atom, len)) {
 229                 if (data->mark_query)
 230                         data->info.sizep = &data->size;
 231                 else
 232                         strbuf_addf(sb, "%lu", data->size);
 233         } else if (is_atom("objectsize:disk", atom, len)) {
 234                 if (data->mark_query)
 235                         data->info.disk_sizep = &data->disk_size;
 236                 else
 237                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 238         } else if (is_atom("rest", atom, len)) {
 239                 if (data->mark_query)
 240                         data->split_on_whitespace = 1;
 241                 else if (data->rest)
 242                         strbuf_addstr(sb, data->rest);
 243         } else if (is_atom("deltabase", atom, len)) {
 244                 if (data->mark_query)
 245                         data->info.delta_base_sha1 = data->delta_base_oid.hash;
 246                 else
 247                         strbuf_addstr(sb,
 248                                       oid_to_hex(&data->delta_base_oid));
 249         } else
 250                 die("unknown format element: %.*s", len, atom);
 251 }
 252
 253 static size_t expand_format(struct strbuf *sb, const char *start, void *data)
 254 {
 255         const char *end;
 256
 257         if (*start != '(')
 258                 return 0;
 259         end = strchr(start + 1, ')');
 260         if (!end)
 261                 die("format element '%s' does not end in ')'", start);
 262
 263         expand_atom(sb, start + 1, end - start - 1, data);
 264
 265         return end - start + 1;
 266 }
 267
 268 static void batch_write(struct batch_options *opt, const void *data, int len)
 269 {
 270         if (opt->buffer_output) {
 271                 if (fwrite(data, 1, len, stdout) != len)
 272                         die_errno("unable to write to stdout");
 273         } else
 274                 write_or_die(1, data, len);
 275 }
 276
 277 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
 278 {
 279         const struct object_id *oid = &data->oid;
 280
 281         assert(data->info.typep);
 282
 283         if (data->type == OBJ_BLOB) {
 284                 if (opt->buffer_output)
 285                         fflush(stdout);
 286                 if (opt->cmdmode) {
 287                         char *contents;
 288                         unsigned long size;
 289
 290                         if (!data->rest)
 291                                 die("missing path for '%s'", oid_to_hex(oid));
 292
 293                         if (opt->cmdmode == 'w') {
 294                                 if (filter_object(data->rest, 0100644, oid,
 295                                                   &contents, &size))
 296                                         die("could not convert '%s' %s",
 297                                             oid_to_hex(oid), data->rest);
 298                         } else if (opt->cmdmode == 'c') {
 299                                 enum object_type type;
 300                                 if (!textconv_object(data->rest, 0100644, oid,
 301                                                      1, &contents, &size))
 302                                         contents = read_sha1_file(oid->hash, &type,
 303                                                                   &size);
 304                                 if (!contents)
 305                                         die("could not convert '%s' %s",
 306                                             oid_to_hex(oid), data->rest);
 307                         } else
 308                                 die("BUG: invalid cmdmode: %c", opt->cmdmode);
 309                         batch_write(opt, contents, size);
 310                         free(contents);
 311                 } else if (stream_blob_to_fd(1, oid, NULL, 0) < 0)
 312                         die("unable to stream %s to stdout", oid_to_hex(oid));
 313         }
 314         else {
 315                 enum object_type type;
 316                 unsigned long size;
 317                 void *contents;
 318
 319                 contents = read_sha1_file(oid->hash, &type, &size);
 320                 if (!contents)
 321                         die("object %s disappeared", oid_to_hex(oid));
 322                 if (type != data->type)
 323                         die("object %s changed type!?", oid_to_hex(oid));
 324                 if (data->info.sizep && size != data->size)
 325                         die("object %s changed size!?", oid_to_hex(oid));
 326
 327                 batch_write(opt, contents, size);
 328                 free(contents);
 329         }
 330 }
 331
 332 static void batch_object_write(const char *obj_name, struct batch_options *opt,
 333                                struct expand_data *data)
 334 {
 335         struct strbuf buf = STRBUF_INIT;
 336
 337         if (!data->skip_object_info &&
 338             sha1_object_info_extended(data->oid.hash, &data->info, LOOKUP_REPLACE_OBJECT) < 0) {
 339                 printf("%s missing\n",
 340                        obj_name ? obj_name : oid_to_hex(&data->oid));
 341                 fflush(stdout);
 342                 return;
 343         }
 344
 345         strbuf_expand(&buf, opt->format, expand_format, data);
 346         strbuf_addch(&buf, '\n');
 347         batch_write(opt, buf.buf, buf.len);
 348         strbuf_release(&buf);
 349
 350         if (opt->print_contents) {
 351                 print_object_or_die(opt, data);
 352                 batch_write(opt, "\n", 1);
 353         }
 354 }
 355
 356 static void batch_one_object(const char *obj_name, struct batch_options *opt,
 357                              struct expand_data *data)
 358 {
 359         struct object_context ctx;
 360         int flags = opt->follow_symlinks ? GET_SHA1_FOLLOW_SYMLINKS : 0;
 361         enum follow_symlinks_result result;
 362
 363         result = get_sha1_with_context(obj_name, flags, data->oid.hash, &ctx);
 364         if (result != FOUND) {
 365                 switch (result) {
 366                 case MISSING_OBJECT:
 367                         printf("%s missing\n", obj_name);
 368                         break;
 369                 case DANGLING_SYMLINK:
 370                         printf("dangling %"PRIuMAX"\n%s\n",
 371                                (uintmax_t)strlen(obj_name), obj_name);
 372                         break;
 373                 case SYMLINK_LOOP:
 374                         printf("loop %"PRIuMAX"\n%s\n",
 375                                (uintmax_t)strlen(obj_name), obj_name);
 376                         break;
 377                 case NOT_DIR:
 378                         printf("notdir %"PRIuMAX"\n%s\n",
 379                                (uintmax_t)strlen(obj_name), obj_name);
 380                         break;
 381                 default:
 382                         die("BUG: unknown get_sha1_with_context result %d\n",
 383                                result);
 384                         break;
 385                 }
 386                 fflush(stdout);
 387                 return;
 388         }
 389
 390         if (ctx.mode == 0) {
 391                 printf("symlink %"PRIuMAX"\n%s\n",
 392                        (uintmax_t)ctx.symlink_path.len,
 393                        ctx.symlink_path.buf);
 394                 fflush(stdout);
 395                 return;
 396         }
 397
 398         batch_object_write(obj_name, opt, data);
 399 }
 400
 401 struct object_cb_data {
 402         struct batch_options *opt;
 403         struct expand_data *expand;
 404 };
 405
 406 static int batch_object_cb(const struct object_id *oid, void *vdata)
 407 {
 408         struct object_cb_data *data = vdata;
 409         oidcpy(&data->expand->oid, oid);
 410         batch_object_write(NULL, data->opt, data->expand);
 411         return 0;
 412 }
 413
 414 static int batch_loose_object(const struct object_id *oid,
 415                               const char *path,
 416                               void *data)
 417 {
 418         oid_array_append(data, oid);
 419         return 0;
 420 }
 421
 422 static int batch_packed_object(const struct object_id *oid,
 423                                struct packed_git *pack,
 424                                uint32_t pos,
 425                                void *data)
 426 {
 427         oid_array_append(data, oid);
 428         return 0;
 429 }
 430
 431 static int batch_objects(struct batch_options *opt)
 432 {
 433         struct strbuf buf = STRBUF_INIT;
 434         struct expand_data data;
 435         int save_warning;
 436         int retval = 0;
 437
 438         if (!opt->format)
 439                 opt->format = "%(objectname) %(objecttype) %(objectsize)";
 440
 441         /*
 442          * Expand once with our special mark_query flag, which will prime the
 443          * object_info to be handed to sha1_object_info_extended for each
 444          * object.
 445          */
 446         memset(&data, 0, sizeof(data));
 447         data.mark_query = 1;
 448         strbuf_expand(&buf, opt->format, expand_format, &data);
 449         data.mark_query = 0;
 450         if (opt->cmdmode)
 451                 data.split_on_whitespace = 1;
 452
 453         if (opt->all_objects) {
 454                 struct object_info empty = OBJECT_INFO_INIT;
 455                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 456                         data.skip_object_info = 1;
 457         }
 458
 459         /*
 460          * If we are printing out the object, then always fill in the type,
 461          * since we will want to decide whether or not to stream.
 462          */
 463         if (opt->print_contents)
 464                 data.info.typep = &data.type;
 465
 466         if (opt->all_objects) {
 467                 struct oid_array sa = OID_ARRAY_INIT;
 468                 struct object_cb_data cb;
 469
 470                 for_each_loose_object(batch_loose_object, &sa, 0);
 471                 for_each_packed_object(batch_packed_object, &sa, 0);
 472
 473                 cb.opt = opt;
 474                 cb.expand = &data;
 475                 oid_array_for_each_unique(&sa, batch_object_cb, &cb);
 476
 477                 oid_array_clear(&sa);
 478                 return 0;
 479         }
 480
 481         /*
 482          * We are going to call get_sha1 on a potentially very large number of
 483          * objects. In most large cases, these will be actual object sha1s. The
 484          * cost to double-check that each one is not also a ref (just so we can
 485          * warn) ends up dwarfing the actual cost of the object lookups
 486          * themselves. We can work around it by just turning off the warning.
 487          */
 488         save_warning = warn_on_object_refname_ambiguity;
 489         warn_on_object_refname_ambiguity = 0;
 490
 491         while (strbuf_getline(&buf, stdin) != EOF) {
 492                 if (data.split_on_whitespace) {
 493                         /*
 494                          * Split at first whitespace, tying off the beginning
 495                          * of the string and saving the remainder (or NULL) in
 496                          * data.rest.
 497                          */
 498                         char *p = strpbrk(buf.buf, " \t");
 499                         if (p) {
 500                                 while (*p && strchr(" \t", *p))
 501                                         *p++ = '\0';
 502                         }
 503                         data.rest = p;
 504                 }
 505
 506                 batch_one_object(buf.buf, opt, &data);
 507         }
 508
 509         strbuf_release(&buf);
 510         warn_on_object_refname_ambiguity = save_warning;
 511         return retval;
 512 }
 513
 514 static const char * const cat_file_usage[] = {
 515         N_("git cat-file (-t [--allow-unknown-type] | -s [--allow-unknown-type] | -e | -p | <type> | --textconv | --filters) [--path=<path>] <object>"),
 516         N_("git cat-file (--batch | --batch-check) [--follow-symlinks] [--textconv | --filters]"),
 517         NULL
 518 };
 519
 520 static int git_cat_file_config(const char *var, const char *value, void *cb)
 521 {
 522         if (userdiff_config(var, value) < 0)
 523                 return -1;
 524
 525         return git_default_config(var, value, cb);
 526 }
 527
 528 static int batch_option_callback(const struct option *opt,
 529                                  const char *arg,
 530                                  int unset)
 531 {
 532         struct batch_options *bo = opt->value;
 533
 534         if (bo->enabled) {
 535                 return 1;
 536         }
 537
 538         bo->enabled = 1;
 539         bo->print_contents = !strcmp(opt->long_name, "batch");
 540         bo->format = arg;
 541
 542         return 0;
 543 }
 544
 545 int cmd_cat_file(int argc, const char **argv, const char *prefix)
 546 {
 547         int opt = 0;
 548         const char *exp_type = NULL, *obj_name = NULL;
 549         struct batch_options batch = {0};
 550         int unknown_type = 0;
 551
 552         const struct option options[] = {
 553                 OPT_GROUP(N_("<type> can be one of: blob, tree, commit, tag")),
 554                 OPT_CMDMODE('t', NULL, &opt, N_("show object type"), 't'),
 555                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 556                 OPT_CMDMODE('e', NULL, &opt,
 557                             N_("exit with zero when there's no error"), 'e'),
 558                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print object's content"), 'p'),
 559                 OPT_CMDMODE(0, "textconv", &opt,
 560                             N_("for blob objects, run textconv on object's content"), 'c'),
 561                 OPT_CMDMODE(0, "filters", &opt,
 562                             N_("for blob objects, run filters on object's content"), 'w'),
 563                 OPT_STRING(0, "path", &force_path, N_("blob"),
 564                            N_("use a specific path for --textconv/--filters")),
 565                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 566                           N_("allow -s and -t to work with broken/corrupt objects")),
 567                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 568                 { OPTION_CALLBACK, 0, "batch", &batch, "format",
 569                         N_("show info and content of objects fed from the standard input"),
 570                         PARSE_OPT_OPTARG, batch_option_callback },
 571                 { OPTION_CALLBACK, 0, "batch-check", &batch, "format",
 572                         N_("show info about objects fed from the standard input"),
 573                         PARSE_OPT_OPTARG, batch_option_callback },
 574                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 575                          N_("follow in-tree symlinks (used with --batch or --batch-check)")),
 576                 OPT_BOOL(0, "batch-all-objects", &batch.all_objects,
 577                          N_("show all objects with --batch or --batch-check")),
 578                 OPT_END()
 579         };
 580
 581         git_config(git_cat_file_config, NULL);
 582
 583         batch.buffer_output = -1;
 584         argc = parse_options(argc, argv, prefix, options, cat_file_usage, 0);
 585
 586         if (opt) {
 587                 if (batch.enabled && (opt == 'c' || opt == 'w'))
 588                         batch.cmdmode = opt;
 589                 else if (argc == 1)
 590                         obj_name = argv[0];
 591                 else
 592                         usage_with_options(cat_file_usage, options);
 593         }
 594         if (!opt && !batch.enabled) {
 595                 if (argc == 2) {
 596                         exp_type = argv[0];
 597                         obj_name = argv[1];
 598                 } else
 599                         usage_with_options(cat_file_usage, options);
 600         }
 601         if (batch.enabled) {
 602                 if (batch.cmdmode != opt || argc)
 603                         usage_with_options(cat_file_usage, options);
 604                 if (batch.cmdmode && batch.all_objects)
 605                         die("--batch-all-objects cannot be combined with "
 606                             "--textconv nor with --filters");
 607         }
 608
 609         if ((batch.follow_symlinks || batch.all_objects) && !batch.enabled) {
 610                 usage_with_options(cat_file_usage, options);
 611         }
 612
 613         if (force_path && opt != 'c' && opt != 'w') {
 614                 error("--path=<path> needs --textconv or --filters");
 615                 usage_with_options(cat_file_usage, options);
 616         }
 617
 618         if (force_path && batch.enabled) {
 619                 error("--path=<path> incompatible with --batch");
 620                 usage_with_options(cat_file_usage, options);
 621         }
 622
 623         if (batch.buffer_output < 0)
 624                 batch.buffer_output = batch.all_objects;
 625
 626         if (batch.enabled)
 627                 return batch_objects(&batch);
 628
 629         if (unknown_type && opt != 't' && opt != 's')
 630                 die("git cat-file --allow-unknown-type: use with -s or -t");
 631         return cat_one_file(opt, exp_type, obj_name, unknown_type);
 632 }