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 unsigned char *sha1,
  28                          char **buf, unsigned long *size)
  29 {
  30         enum object_type type;
  31
  32         *buf = read_sha1_file(sha1, &type, size);
  33         if (!*buf)
  34                 return error(_("cannot read object %s '%s'"),
  35                         sha1_to_hex(sha1), 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         unsigned char sha1[20];
  52         enum object_type type;
  53         char *buf;
  54         unsigned long size;
  55         struct object_context obj_context;
  56         struct object_info oi = {NULL};
  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, sha1, &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(sha1, &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(sha1, &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_sha1_file(sha1);
  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                                   sha1, &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,
 111                                     sha1, 1, &buf, &size))
 112                         break;
 113
 114         case 'p':
 115                 type = sha1_object_info(sha1, 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, sha1, NULL, 0);
 129                 buf = read_sha1_file(sha1, &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                         unsigned char blob_sha1[20];
 139                         if (sha1_object_info(sha1, NULL) == OBJ_TAG) {
 140                                 char *buffer = read_sha1_file(sha1, &type, &size);
 141                                 const char *target;
 142                                 if (!skip_prefix(buffer, "object ", &target) ||
 143                                     get_sha1_hex(target, blob_sha1))
 144                                         die("%s not a valid tag", sha1_to_hex(sha1));
 145                                 free(buffer);
 146                         } else
 147                                 hashcpy(blob_sha1, sha1);
 148
 149                         if (sha1_object_info(blob_sha1, NULL) == OBJ_BLOB)
 150                                 return stream_blob_to_fd(1, blob_sha1, 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(sha1, 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         return 0;
 170 }
 171
 172 struct expand_data {
 173         unsigned char sha1[20];
 174         enum object_type type;
 175         unsigned long size;
 176         off_t disk_size;
 177         const char *rest;
 178         unsigned char delta_base_sha1[20];
 179
 180         /*
 181          * If mark_query is true, we do not expand anything, but rather
 182          * just mark the object_info with items we wish to query.
 183          */
 184         int mark_query;
 185
 186         /*
 187          * Whether to split the input on whitespace before feeding it to
 188          * get_sha1; this is decided during the mark_query phase based on
 189          * whether we have a %(rest) token in our format.
 190          */
 191         int split_on_whitespace;
 192
 193         /*
 194          * After a mark_query run, this object_info is set up to be
 195          * passed to sha1_object_info_extended. It will point to the data
 196          * elements above, so you can retrieve the response from there.
 197          */
 198         struct object_info info;
 199
 200         /*
 201          * This flag will be true if the requested batch format and options
 202          * don't require us to call sha1_object_info, which can then be
 203          * optimized out.
 204          */
 205         unsigned skip_object_info : 1;
 206 };
 207
 208 static int is_atom(const char *atom, const char *s, int slen)
 209 {
 210         int alen = strlen(atom);
 211         return alen == slen && !memcmp(atom, s, alen);
 212 }
 213
 214 static void expand_atom(struct strbuf *sb, const char *atom, int len,
 215                         void *vdata)
 216 {
 217         struct expand_data *data = vdata;
 218
 219         if (is_atom("objectname", atom, len)) {
 220                 if (!data->mark_query)
 221                         strbuf_addstr(sb, sha1_to_hex(data->sha1));
 222         } else if (is_atom("objecttype", atom, len)) {
 223                 if (data->mark_query)
 224                         data->info.typep = &data->type;
 225                 else
 226                         strbuf_addstr(sb, typename(data->type));
 227         } else if (is_atom("objectsize", atom, len)) {
 228                 if (data->mark_query)
 229                         data->info.sizep = &data->size;
 230                 else
 231                         strbuf_addf(sb, "%lu", data->size);
 232         } else if (is_atom("objectsize:disk", atom, len)) {
 233                 if (data->mark_query)
 234                         data->info.disk_sizep = &data->disk_size;
 235                 else
 236                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 237         } else if (is_atom("rest", atom, len)) {
 238                 if (data->mark_query)
 239                         data->split_on_whitespace = 1;
 240                 else if (data->rest)
 241                         strbuf_addstr(sb, data->rest);
 242         } else if (is_atom("deltabase", atom, len)) {
 243                 if (data->mark_query)
 244                         data->info.delta_base_sha1 = data->delta_base_sha1;
 245                 else
 246                         strbuf_addstr(sb, sha1_to_hex(data->delta_base_sha1));
 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 unsigned char *sha1 = data->sha1;
 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'", sha1_to_hex(sha1));
 290
 291                         if (opt->cmdmode == 'w') {
 292                                 if (filter_object(data->rest, 0100644, sha1,
 293                                                   &contents, &size))
 294                                         die("could not convert '%s' %s",
 295                                             sha1_to_hex(sha1), data->rest);
 296                         } else if (opt->cmdmode == 'c') {
 297                                 enum object_type type;
 298                                 if (!textconv_object(data->rest, 0100644, sha1,
 299                                                      1, &contents, &size))
 300                                         contents = read_sha1_file(sha1, &type,
 301                                                                   &size);
 302                                 if (!contents)
 303                                         die("could not convert '%s' %s",
 304                                             sha1_to_hex(sha1), 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, sha1, NULL, 0) < 0)
 310                         die("unable to stream %s to stdout", sha1_to_hex(sha1));
 311         }
 312         else {
 313                 enum object_type type;
 314                 unsigned long size;
 315                 void *contents;
 316
 317                 contents = read_sha1_file(sha1, &type, &size);
 318                 if (!contents)
 319                         die("object %s disappeared", sha1_to_hex(sha1));
 320                 if (type != data->type)
 321                         die("object %s changed type!?", sha1_to_hex(sha1));
 322                 if (data->info.sizep && size != data->size)
 323                         die("object %s changed size!?", sha1_to_hex(sha1));
 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->sha1, &data->info, LOOKUP_REPLACE_OBJECT) < 0) {
 337                 printf("%s missing\n", obj_name ? obj_name : sha1_to_hex(data->sha1));
 338                 fflush(stdout);
 339                 return;
 340         }
 341
 342         strbuf_expand(&buf, opt->format, expand_format, data);
 343         strbuf_addch(&buf, '\n');
 344         batch_write(opt, buf.buf, buf.len);
 345         strbuf_release(&buf);
 346
 347         if (opt->print_contents) {
 348                 print_object_or_die(opt, data);
 349                 batch_write(opt, "\n", 1);
 350         }
 351 }
 352
 353 static void batch_one_object(const char *obj_name, struct batch_options *opt,
 354                              struct expand_data *data)
 355 {
 356         struct object_context ctx;
 357         int flags = opt->follow_symlinks ? GET_SHA1_FOLLOW_SYMLINKS : 0;
 358         enum follow_symlinks_result result;
 359
 360         result = get_sha1_with_context(obj_name, flags, data->sha1, &ctx);
 361         if (result != FOUND) {
 362                 switch (result) {
 363                 case MISSING_OBJECT:
 364                         printf("%s missing\n", obj_name);
 365                         break;
 366                 case DANGLING_SYMLINK:
 367                         printf("dangling %"PRIuMAX"\n%s\n",
 368                                (uintmax_t)strlen(obj_name), obj_name);
 369                         break;
 370                 case SYMLINK_LOOP:
 371                         printf("loop %"PRIuMAX"\n%s\n",
 372                                (uintmax_t)strlen(obj_name), obj_name);
 373                         break;
 374                 case NOT_DIR:
 375                         printf("notdir %"PRIuMAX"\n%s\n",
 376                                (uintmax_t)strlen(obj_name), obj_name);
 377                         break;
 378                 default:
 379                         die("BUG: unknown get_sha1_with_context result %d\n",
 380                                result);
 381                         break;
 382                 }
 383                 fflush(stdout);
 384                 return;
 385         }
 386
 387         if (ctx.mode == 0) {
 388                 printf("symlink %"PRIuMAX"\n%s\n",
 389                        (uintmax_t)ctx.symlink_path.len,
 390                        ctx.symlink_path.buf);
 391                 fflush(stdout);
 392                 return;
 393         }
 394
 395         batch_object_write(obj_name, opt, data);
 396 }
 397
 398 struct object_cb_data {
 399         struct batch_options *opt;
 400         struct expand_data *expand;
 401 };
 402
 403 static void batch_object_cb(const unsigned char sha1[20], void *vdata)
 404 {
 405         struct object_cb_data *data = vdata;
 406         hashcpy(data->expand->sha1, sha1);
 407         batch_object_write(NULL, data->opt, data->expand);
 408 }
 409
 410 static int batch_loose_object(const unsigned char *sha1,
 411                               const char *path,
 412                               void *data)
 413 {
 414         sha1_array_append(data, sha1);
 415         return 0;
 416 }
 417
 418 static int batch_packed_object(const unsigned char *sha1,
 419                                struct packed_git *pack,
 420                                uint32_t pos,
 421                                void *data)
 422 {
 423         sha1_array_append(data, sha1);
 424         return 0;
 425 }
 426
 427 static int batch_objects(struct batch_options *opt)
 428 {
 429         struct strbuf buf = STRBUF_INIT;
 430         struct expand_data data;
 431         int save_warning;
 432         int retval = 0;
 433
 434         if (!opt->format)
 435                 opt->format = "%(objectname) %(objecttype) %(objectsize)";
 436
 437         /*
 438          * Expand once with our special mark_query flag, which will prime the
 439          * object_info to be handed to sha1_object_info_extended for each
 440          * object.
 441          */
 442         memset(&data, 0, sizeof(data));
 443         data.mark_query = 1;
 444         strbuf_expand(&buf, opt->format, expand_format, &data);
 445         data.mark_query = 0;
 446         if (opt->cmdmode)
 447                 data.split_on_whitespace = 1;
 448
 449         if (opt->all_objects) {
 450                 struct object_info empty;
 451                 memset(&empty, 0, sizeof(empty));
 452                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 453                         data.skip_object_info = 1;
 454         }
 455
 456         /*
 457          * If we are printing out the object, then always fill in the type,
 458          * since we will want to decide whether or not to stream.
 459          */
 460         if (opt->print_contents)
 461                 data.info.typep = &data.type;
 462
 463         if (opt->all_objects) {
 464                 struct sha1_array sa = SHA1_ARRAY_INIT;
 465                 struct object_cb_data cb;
 466
 467                 for_each_loose_object(batch_loose_object, &sa, 0);
 468                 for_each_packed_object(batch_packed_object, &sa, 0);
 469
 470                 cb.opt = opt;
 471                 cb.expand = &data;
 472                 sha1_array_for_each_unique(&sa, batch_object_cb, &cb);
 473
 474                 sha1_array_clear(&sa);
 475                 return 0;
 476         }
 477
 478         /*
 479          * We are going to call get_sha1 on a potentially very large number of
 480          * objects. In most large cases, these will be actual object sha1s. The
 481          * cost to double-check that each one is not also a ref (just so we can
 482          * warn) ends up dwarfing the actual cost of the object lookups
 483          * themselves. We can work around it by just turning off the warning.
 484          */
 485         save_warning = warn_on_object_refname_ambiguity;
 486         warn_on_object_refname_ambiguity = 0;
 487
 488         while (strbuf_getline(&buf, stdin) != EOF) {
 489                 if (data.split_on_whitespace) {
 490                         /*
 491                          * Split at first whitespace, tying off the beginning
 492                          * of the string and saving the remainder (or NULL) in
 493                          * data.rest.
 494                          */
 495                         char *p = strpbrk(buf.buf, " \t");
 496                         if (p) {
 497                                 while (*p && strchr(" \t", *p))
 498                                         *p++ = '\0';
 499                         }
 500                         data.rest = p;
 501                 }
 502
 503                 batch_one_object(buf.buf, opt, &data);
 504         }
 505
 506         strbuf_release(&buf);
 507         warn_on_object_refname_ambiguity = save_warning;
 508         return retval;
 509 }
 510
 511 static const char * const cat_file_usage[] = {
 512         N_("git cat-file (-t [--allow-unknown-type]|-s [--allow-unknown-type]|-e|-p|<type>|--textconv|--filters) [--path=<path>] <object>"),
 513         N_("git cat-file (--batch | --batch-check) [--follow-symlinks] [--textconv|--filters]"),
 514         NULL
 515 };
 516
 517 static int git_cat_file_config(const char *var, const char *value, void *cb)
 518 {
 519         if (userdiff_config(var, value) < 0)
 520                 return -1;
 521
 522         return git_default_config(var, value, cb);
 523 }
 524
 525 static int batch_option_callback(const struct option *opt,
 526                                  const char *arg,
 527                                  int unset)
 528 {
 529         struct batch_options *bo = opt->value;
 530
 531         if (bo->enabled) {
 532                 return 1;
 533         }
 534
 535         bo->enabled = 1;
 536         bo->print_contents = !strcmp(opt->long_name, "batch");
 537         bo->format = arg;
 538
 539         return 0;
 540 }
 541
 542 int cmd_cat_file(int argc, const char **argv, const char *prefix)
 543 {
 544         int opt = 0;
 545         const char *exp_type = NULL, *obj_name = NULL;
 546         struct batch_options batch = {0};
 547         int unknown_type = 0;
 548
 549         const struct option options[] = {
 550                 OPT_GROUP(N_("<type> can be one of: blob, tree, commit, tag")),
 551                 OPT_CMDMODE('t', NULL, &opt, N_("show object type"), 't'),
 552                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 553                 OPT_CMDMODE('e', NULL, &opt,
 554                             N_("exit with zero when there's no error"), 'e'),
 555                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print object's content"), 'p'),
 556                 OPT_CMDMODE(0, "textconv", &opt,
 557                             N_("for blob objects, run textconv on object's content"), 'c'),
 558                 OPT_CMDMODE(0, "filters", &opt,
 559                             N_("for blob objects, run filters on object's content"), 'w'),
 560                 OPT_STRING(0, "path", &force_path, N_("blob"),
 561                            N_("use a specific path for --textconv/--filters")),
 562                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 563                           N_("allow -s and -t to work with broken/corrupt objects")),
 564                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 565                 { OPTION_CALLBACK, 0, "batch", &batch, "format",
 566                         N_("show info and content of objects fed from the standard input"),
 567                         PARSE_OPT_OPTARG, batch_option_callback },
 568                 { OPTION_CALLBACK, 0, "batch-check", &batch, "format",
 569                         N_("show info about objects fed from the standard input"),
 570                         PARSE_OPT_OPTARG, batch_option_callback },
 571                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 572                          N_("follow in-tree symlinks (used with --batch or --batch-check)")),
 573                 OPT_BOOL(0, "batch-all-objects", &batch.all_objects,
 574                          N_("show all objects with --batch or --batch-check")),
 575                 OPT_END()
 576         };
 577
 578         git_config(git_cat_file_config, NULL);
 579
 580         batch.buffer_output = -1;
 581         argc = parse_options(argc, argv, prefix, options, cat_file_usage, 0);
 582
 583         if (opt) {
 584                 if (batch.enabled && (opt == 'c' || opt == 'w'))
 585                         batch.cmdmode = opt;
 586                 else if (argc == 1)
 587                         obj_name = argv[0];
 588                 else
 589                         usage_with_options(cat_file_usage, options);
 590         }
 591         if (!opt && !batch.enabled) {
 592                 if (argc == 2) {
 593                         exp_type = argv[0];
 594                         obj_name = argv[1];
 595                 } else
 596                         usage_with_options(cat_file_usage, options);
 597         }
 598         if (batch.enabled) {
 599                 if (batch.cmdmode != opt || argc)
 600                         usage_with_options(cat_file_usage, options);
 601                 if (batch.cmdmode && batch.all_objects)
 602                         die("--batch-all-objects cannot be combined with "
 603                             "--textconv nor with --filters");
 604         }
 605
 606         if ((batch.follow_symlinks || batch.all_objects) && !batch.enabled) {
 607                 usage_with_options(cat_file_usage, options);
 608         }
 609
 610         if (force_path && opt != 'c' && opt != 'w') {
 611                 error("--path=<path> needs --textconv or --filters");
 612                 usage_with_options(cat_file_usage, options);
 613         }
 614
 615         if (force_path && batch.enabled) {
 616                 error("--path=<path> incompatible with --batch");
 617                 usage_with_options(cat_file_usage, options);
 618         }
 619
 620         if (batch.buffer_output < 0)
 621                 batch.buffer_output = batch.all_objects;
 622
 623         if (batch.enabled)
 624                 return batch_objects(&batch);
 625
 626         if (unknown_type && opt != 't' && opt != 's')
 627                 die("git cat-file --allow-unknown-type: use with -s or -t");
 628         return cat_one_file(opt, exp_type, obj_name, unknown_type);
 629 }