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