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
  17 struct batch_options {
  18         int enabled;
  19         int follow_symlinks;
  20         int print_contents;
  21         int buffer_output;
  22         int all_objects;
  23         int cmdmode; /* may be 'w' or 'c' for --filters or --textconv */
  24         const char *format;
  25 };
  26
  27 static const char *force_path;
  28
  29 static int filter_object(const char *path, unsigned mode,
  30                          const struct object_id *oid,
  31                          char **buf, unsigned long *size)
  32 {
  33         enum object_type type;
  34
  35         *buf = read_sha1_file(oid->hash, &type, size);
  36         if (!*buf)
  37                 return error(_("cannot read object %s '%s'"),
  38                              oid_to_hex(oid), path);
  39         if ((type == OBJ_BLOB) && S_ISREG(mode)) {
  40                 struct strbuf strbuf = STRBUF_INIT;
  41                 if (convert_to_working_tree(path, *buf, *size, &strbuf)) {
  42                         free(*buf);
  43                         *size = strbuf.len;
  44                         *buf = strbuf_detach(&strbuf, NULL);
  45                 }
  46         }
  47
  48         return 0;
  49 }
  50
  51 static int cat_one_file(int opt, const char *exp_type, const char *obj_name,
  52                         int unknown_type)
  53 {
  54         struct object_id oid;
  55         enum object_type type;
  56         char *buf;
  57         unsigned long size;
  58         struct object_context obj_context;
  59         struct object_info oi = OBJECT_INFO_INIT;
  60         struct strbuf sb = STRBUF_INIT;
  61         unsigned flags = OBJECT_INFO_LOOKUP_REPLACE;
  62         const char *path = force_path;
  63
  64         if (unknown_type)
  65                 flags |= OBJECT_INFO_ALLOW_UNKNOWN_TYPE;
  66
  67         if (get_oid_with_context(obj_name, GET_OID_RECORD_PATH,
  68                                  &oid, &obj_context))
  69                 die("Not a valid object name %s", obj_name);
  70
  71         if (!path)
  72                 path = obj_context.path;
  73         if (obj_context.mode == S_IFINVALID)
  74                 obj_context.mode = 0100644;
  75
  76         buf = NULL;
  77         switch (opt) {
  78         case 't':
  79                 oi.typename = &sb;
  80                 if (sha1_object_info_extended(oid.hash, &oi, flags) < 0)
  81                         die("git cat-file: could not get object info");
  82                 if (sb.len) {
  83                         printf("%s\n", sb.buf);
  84                         strbuf_release(&sb);
  85                         return 0;
  86                 }
  87                 break;
  88
  89         case 's':
  90                 oi.sizep = &size;
  91                 if (sha1_object_info_extended(oid.hash, &oi, flags) < 0)
  92                         die("git cat-file: could not get object info");
  93                 printf("%lu\n", size);
  94                 return 0;
  95
  96         case 'e':
  97                 return !has_object_file(&oid);
  98
  99         case 'w':
 100                 if (!path[0])
 101                         die("git cat-file --filters %s: <object> must be "
 102                             "<sha1:path>", obj_name);
 103
 104                 if (filter_object(path, obj_context.mode,
 105                                   &oid, &buf, &size))
 106                         return -1;
 107                 break;
 108
 109         case 'c':
 110                 if (!path[0])
 111                         die("git cat-file --textconv %s: <object> must be <sha1:path>",
 112                             obj_name);
 113
 114                 if (textconv_object(path, obj_context.mode, &oid, 1, &buf, &size))
 115                         break;
 116
 117         case 'p':
 118                 type = sha1_object_info(oid.hash, NULL);
 119                 if (type < 0)
 120                         die("Not a valid object name %s", obj_name);
 121
 122                 /* custom pretty-print here */
 123                 if (type == OBJ_TREE) {
 124                         const char *ls_args[3] = { NULL };
 125                         ls_args[0] =  "ls-tree";
 126                         ls_args[1] =  obj_name;
 127                         return cmd_ls_tree(2, ls_args, NULL);
 128                 }
 129
 130                 if (type == OBJ_BLOB)
 131                         return stream_blob_to_fd(1, &oid, NULL, 0);
 132                 buf = read_sha1_file(oid.hash, &type, &size);
 133                 if (!buf)
 134                         die("Cannot read object %s", obj_name);
 135
 136                 /* otherwise just spit out the data */
 137                 break;
 138
 139         case 0:
 140                 if (type_from_string(exp_type) == OBJ_BLOB) {
 141                         struct object_id blob_oid;
 142                         if (sha1_object_info(oid.hash, NULL) == OBJ_TAG) {
 143                                 char *buffer = read_sha1_file(oid.hash, &type, &size);
 144                                 const char *target;
 145                                 if (!skip_prefix(buffer, "object ", &target) ||
 146                                     get_oid_hex(target, &blob_oid))
 147                                         die("%s not a valid tag", oid_to_hex(&oid));
 148                                 free(buffer);
 149                         } else
 150                                 oidcpy(&blob_oid, &oid);
 151
 152                         if (sha1_object_info(blob_oid.hash, NULL) == OBJ_BLOB)
 153                                 return stream_blob_to_fd(1, &blob_oid, NULL, 0);
 154                         /*
 155                          * we attempted to dereference a tag to a blob
 156                          * and failed; there may be new dereference
 157                          * mechanisms this code is not aware of.
 158                          * fall-back to the usual case.
 159                          */
 160                 }
 161                 buf = read_object_with_reference(oid.hash, exp_type, &size, NULL);
 162                 break;
 163
 164         default:
 165                 die("git cat-file: unknown option: %s", exp_type);
 166         }
 167
 168         if (!buf)
 169                 die("git cat-file %s: bad file", obj_name);
 170
 171         write_or_die(1, buf, size);
 172         free(buf);
 173         free(obj_context.path);
 174         return 0;
 175 }
 176
 177 struct expand_data {
 178         struct object_id oid;
 179         enum object_type type;
 180         unsigned long size;
 181         off_t disk_size;
 182         const char *rest;
 183         struct object_id delta_base_oid;
 184
 185         /*
 186          * If mark_query is true, we do not expand anything, but rather
 187          * just mark the object_info with items we wish to query.
 188          */
 189         int mark_query;
 190
 191         /*
 192          * Whether to split the input on whitespace before feeding it to
 193          * get_sha1; this is decided during the mark_query phase based on
 194          * whether we have a %(rest) token in our format.
 195          */
 196         int split_on_whitespace;
 197
 198         /*
 199          * After a mark_query run, this object_info is set up to be
 200          * passed to sha1_object_info_extended. It will point to the data
 201          * elements above, so you can retrieve the response from there.
 202          */
 203         struct object_info info;
 204
 205         /*
 206          * This flag will be true if the requested batch format and options
 207          * don't require us to call sha1_object_info, which can then be
 208          * optimized out.
 209          */
 210         unsigned skip_object_info : 1;
 211 };
 212
 213 static int is_atom(const char *atom, const char *s, int slen)
 214 {
 215         int alen = strlen(atom);
 216         return alen == slen && !memcmp(atom, s, alen);
 217 }
 218
 219 static void expand_atom(struct strbuf *sb, const char *atom, int len,
 220                         void *vdata)
 221 {
 222         struct expand_data *data = vdata;
 223
 224         if (is_atom("objectname", atom, len)) {
 225                 if (!data->mark_query)
 226                         strbuf_addstr(sb, oid_to_hex(&data->oid));
 227         } else if (is_atom("objecttype", atom, len)) {
 228                 if (data->mark_query)
 229                         data->info.typep = &data->type;
 230                 else
 231                         strbuf_addstr(sb, typename(data->type));
 232         } else if (is_atom("objectsize", atom, len)) {
 233                 if (data->mark_query)
 234                         data->info.sizep = &data->size;
 235                 else
 236                         strbuf_addf(sb, "%lu", data->size);
 237         } else if (is_atom("objectsize:disk", atom, len)) {
 238                 if (data->mark_query)
 239                         data->info.disk_sizep = &data->disk_size;
 240                 else
 241                         strbuf_addf(sb, "%"PRIuMAX, (uintmax_t)data->disk_size);
 242         } else if (is_atom("rest", atom, len)) {
 243                 if (data->mark_query)
 244                         data->split_on_whitespace = 1;
 245                 else if (data->rest)
 246                         strbuf_addstr(sb, data->rest);
 247         } else if (is_atom("deltabase", atom, len)) {
 248                 if (data->mark_query)
 249                         data->info.delta_base_sha1 = data->delta_base_oid.hash;
 250                 else
 251                         strbuf_addstr(sb,
 252                                       oid_to_hex(&data->delta_base_oid));
 253         } else
 254                 die("unknown format element: %.*s", len, atom);
 255 }
 256
 257 static size_t expand_format(struct strbuf *sb, const char *start, void *data)
 258 {
 259         const char *end;
 260
 261         if (*start != '(')
 262                 return 0;
 263         end = strchr(start + 1, ')');
 264         if (!end)
 265                 die("format element '%s' does not end in ')'", start);
 266
 267         expand_atom(sb, start + 1, end - start - 1, data);
 268
 269         return end - start + 1;
 270 }
 271
 272 static void batch_write(struct batch_options *opt, const void *data, int len)
 273 {
 274         if (opt->buffer_output) {
 275                 if (fwrite(data, 1, len, stdout) != len)
 276                         die_errno("unable to write to stdout");
 277         } else
 278                 write_or_die(1, data, len);
 279 }
 280
 281 static void print_object_or_die(struct batch_options *opt, struct expand_data *data)
 282 {
 283         const struct object_id *oid = &data->oid;
 284
 285         assert(data->info.typep);
 286
 287         if (data->type == OBJ_BLOB) {
 288                 if (opt->buffer_output)
 289                         fflush(stdout);
 290                 if (opt->cmdmode) {
 291                         char *contents;
 292                         unsigned long size;
 293
 294                         if (!data->rest)
 295                                 die("missing path for '%s'", oid_to_hex(oid));
 296
 297                         if (opt->cmdmode == 'w') {
 298                                 if (filter_object(data->rest, 0100644, oid,
 299                                                   &contents, &size))
 300                                         die("could not convert '%s' %s",
 301                                             oid_to_hex(oid), data->rest);
 302                         } else if (opt->cmdmode == 'c') {
 303                                 enum object_type type;
 304                                 if (!textconv_object(data->rest, 0100644, oid,
 305                                                      1, &contents, &size))
 306                                         contents = read_sha1_file(oid->hash, &type,
 307                                                                   &size);
 308                                 if (!contents)
 309                                         die("could not convert '%s' %s",
 310                                             oid_to_hex(oid), data->rest);
 311                         } else
 312                                 die("BUG: invalid cmdmode: %c", opt->cmdmode);
 313                         batch_write(opt, contents, size);
 314                         free(contents);
 315                 } else if (stream_blob_to_fd(1, oid, NULL, 0) < 0)
 316                         die("unable to stream %s to stdout", oid_to_hex(oid));
 317         }
 318         else {
 319                 enum object_type type;
 320                 unsigned long size;
 321                 void *contents;
 322
 323                 contents = read_sha1_file(oid->hash, &type, &size);
 324                 if (!contents)
 325                         die("object %s disappeared", oid_to_hex(oid));
 326                 if (type != data->type)
 327                         die("object %s changed type!?", oid_to_hex(oid));
 328                 if (data->info.sizep && size != data->size)
 329                         die("object %s changed size!?", oid_to_hex(oid));
 330
 331                 batch_write(opt, contents, size);
 332                 free(contents);
 333         }
 334 }
 335
 336 static void batch_object_write(const char *obj_name, struct batch_options *opt,
 337                                struct expand_data *data)
 338 {
 339         struct strbuf buf = STRBUF_INIT;
 340
 341         if (!data->skip_object_info &&
 342             sha1_object_info_extended(data->oid.hash, &data->info,
 343                                       OBJECT_INFO_LOOKUP_REPLACE) < 0) {
 344                 printf("%s missing\n",
 345                        obj_name ? obj_name : oid_to_hex(&data->oid));
 346                 fflush(stdout);
 347                 return;
 348         }
 349
 350         strbuf_expand(&buf, opt->format, expand_format, data);
 351         strbuf_addch(&buf, '\n');
 352         batch_write(opt, buf.buf, buf.len);
 353         strbuf_release(&buf);
 354
 355         if (opt->print_contents) {
 356                 print_object_or_die(opt, data);
 357                 batch_write(opt, "\n", 1);
 358         }
 359 }
 360
 361 static void batch_one_object(const char *obj_name, struct batch_options *opt,
 362                              struct expand_data *data)
 363 {
 364         struct object_context ctx;
 365         int flags = opt->follow_symlinks ? GET_OID_FOLLOW_SYMLINKS : 0;
 366         enum follow_symlinks_result result;
 367
 368         result = get_oid_with_context(obj_name, flags, &data->oid, &ctx);
 369         if (result != FOUND) {
 370                 switch (result) {
 371                 case MISSING_OBJECT:
 372                         printf("%s missing\n", obj_name);
 373                         break;
 374                 case DANGLING_SYMLINK:
 375                         printf("dangling %"PRIuMAX"\n%s\n",
 376                                (uintmax_t)strlen(obj_name), obj_name);
 377                         break;
 378                 case SYMLINK_LOOP:
 379                         printf("loop %"PRIuMAX"\n%s\n",
 380                                (uintmax_t)strlen(obj_name), obj_name);
 381                         break;
 382                 case NOT_DIR:
 383                         printf("notdir %"PRIuMAX"\n%s\n",
 384                                (uintmax_t)strlen(obj_name), obj_name);
 385                         break;
 386                 default:
 387                         die("BUG: unknown get_sha1_with_context result %d\n",
 388                                result);
 389                         break;
 390                 }
 391                 fflush(stdout);
 392                 return;
 393         }
 394
 395         if (ctx.mode == 0) {
 396                 printf("symlink %"PRIuMAX"\n%s\n",
 397                        (uintmax_t)ctx.symlink_path.len,
 398                        ctx.symlink_path.buf);
 399                 fflush(stdout);
 400                 return;
 401         }
 402
 403         batch_object_write(obj_name, opt, data);
 404 }
 405
 406 struct object_cb_data {
 407         struct batch_options *opt;
 408         struct expand_data *expand;
 409 };
 410
 411 static int batch_object_cb(const struct object_id *oid, void *vdata)
 412 {
 413         struct object_cb_data *data = vdata;
 414         oidcpy(&data->expand->oid, oid);
 415         batch_object_write(NULL, data->opt, data->expand);
 416         return 0;
 417 }
 418
 419 static int batch_loose_object(const struct object_id *oid,
 420                               const char *path,
 421                               void *data)
 422 {
 423         oid_array_append(data, oid);
 424         return 0;
 425 }
 426
 427 static int batch_packed_object(const struct object_id *oid,
 428                                struct packed_git *pack,
 429                                uint32_t pos,
 430                                void *data)
 431 {
 432         oid_array_append(data, oid);
 433         return 0;
 434 }
 435
 436 static int batch_objects(struct batch_options *opt)
 437 {
 438         struct strbuf buf = STRBUF_INIT;
 439         struct expand_data data;
 440         int save_warning;
 441         int retval = 0;
 442
 443         if (!opt->format)
 444                 opt->format = "%(objectname) %(objecttype) %(objectsize)";
 445
 446         /*
 447          * Expand once with our special mark_query flag, which will prime the
 448          * object_info to be handed to sha1_object_info_extended for each
 449          * object.
 450          */
 451         memset(&data, 0, sizeof(data));
 452         data.mark_query = 1;
 453         strbuf_expand(&buf, opt->format, expand_format, &data);
 454         data.mark_query = 0;
 455         if (opt->cmdmode)
 456                 data.split_on_whitespace = 1;
 457
 458         if (opt->all_objects) {
 459                 struct object_info empty = OBJECT_INFO_INIT;
 460                 if (!memcmp(&data.info, &empty, sizeof(empty)))
 461                         data.skip_object_info = 1;
 462         }
 463
 464         /*
 465          * If we are printing out the object, then always fill in the type,
 466          * since we will want to decide whether or not to stream.
 467          */
 468         if (opt->print_contents)
 469                 data.info.typep = &data.type;
 470
 471         if (opt->all_objects) {
 472                 struct oid_array sa = OID_ARRAY_INIT;
 473                 struct object_cb_data cb;
 474
 475                 for_each_loose_object(batch_loose_object, &sa, 0);
 476                 for_each_packed_object(batch_packed_object, &sa, 0);
 477
 478                 cb.opt = opt;
 479                 cb.expand = &data;
 480                 oid_array_for_each_unique(&sa, batch_object_cb, &cb);
 481
 482                 oid_array_clear(&sa);
 483                 return 0;
 484         }
 485
 486         /*
 487          * We are going to call get_sha1 on a potentially very large number of
 488          * objects. In most large cases, these will be actual object sha1s. The
 489          * cost to double-check that each one is not also a ref (just so we can
 490          * warn) ends up dwarfing the actual cost of the object lookups
 491          * themselves. We can work around it by just turning off the warning.
 492          */
 493         save_warning = warn_on_object_refname_ambiguity;
 494         warn_on_object_refname_ambiguity = 0;
 495
 496         while (strbuf_getline(&buf, stdin) != EOF) {
 497                 if (data.split_on_whitespace) {
 498                         /*
 499                          * Split at first whitespace, tying off the beginning
 500                          * of the string and saving the remainder (or NULL) in
 501                          * data.rest.
 502                          */
 503                         char *p = strpbrk(buf.buf, " \t");
 504                         if (p) {
 505                                 while (*p && strchr(" \t", *p))
 506                                         *p++ = '\0';
 507                         }
 508                         data.rest = p;
 509                 }
 510
 511                 batch_one_object(buf.buf, opt, &data);
 512         }
 513
 514         strbuf_release(&buf);
 515         warn_on_object_refname_ambiguity = save_warning;
 516         return retval;
 517 }
 518
 519 static const char * const cat_file_usage[] = {
 520         N_("git cat-file (-t [--allow-unknown-type] | -s [--allow-unknown-type] | -e | -p | <type> | --textconv | --filters) [--path=<path>] <object>"),
 521         N_("git cat-file (--batch | --batch-check) [--follow-symlinks] [--textconv | --filters]"),
 522         NULL
 523 };
 524
 525 static int git_cat_file_config(const char *var, const char *value, void *cb)
 526 {
 527         if (userdiff_config(var, value) < 0)
 528                 return -1;
 529
 530         return git_default_config(var, value, cb);
 531 }
 532
 533 static int batch_option_callback(const struct option *opt,
 534                                  const char *arg,
 535                                  int unset)
 536 {
 537         struct batch_options *bo = opt->value;
 538
 539         if (bo->enabled) {
 540                 return 1;
 541         }
 542
 543         bo->enabled = 1;
 544         bo->print_contents = !strcmp(opt->long_name, "batch");
 545         bo->format = arg;
 546
 547         return 0;
 548 }
 549
 550 int cmd_cat_file(int argc, const char **argv, const char *prefix)
 551 {
 552         int opt = 0;
 553         const char *exp_type = NULL, *obj_name = NULL;
 554         struct batch_options batch = {0};
 555         int unknown_type = 0;
 556
 557         const struct option options[] = {
 558                 OPT_GROUP(N_("<type> can be one of: blob, tree, commit, tag")),
 559                 OPT_CMDMODE('t', NULL, &opt, N_("show object type"), 't'),
 560                 OPT_CMDMODE('s', NULL, &opt, N_("show object size"), 's'),
 561                 OPT_CMDMODE('e', NULL, &opt,
 562                             N_("exit with zero when there's no error"), 'e'),
 563                 OPT_CMDMODE('p', NULL, &opt, N_("pretty-print object's content"), 'p'),
 564                 OPT_CMDMODE(0, "textconv", &opt,
 565                             N_("for blob objects, run textconv on object's content"), 'c'),
 566                 OPT_CMDMODE(0, "filters", &opt,
 567                             N_("for blob objects, run filters on object's content"), 'w'),
 568                 OPT_STRING(0, "path", &force_path, N_("blob"),
 569                            N_("use a specific path for --textconv/--filters")),
 570                 OPT_BOOL(0, "allow-unknown-type", &unknown_type,
 571                           N_("allow -s and -t to work with broken/corrupt objects")),
 572                 OPT_BOOL(0, "buffer", &batch.buffer_output, N_("buffer --batch output")),
 573                 { OPTION_CALLBACK, 0, "batch", &batch, "format",
 574                         N_("show info and content of objects fed from the standard input"),
 575                         PARSE_OPT_OPTARG, batch_option_callback },
 576                 { OPTION_CALLBACK, 0, "batch-check", &batch, "format",
 577                         N_("show info about objects fed from the standard input"),
 578                         PARSE_OPT_OPTARG, batch_option_callback },
 579                 OPT_BOOL(0, "follow-symlinks", &batch.follow_symlinks,
 580                          N_("follow in-tree symlinks (used with --batch or --batch-check)")),
 581                 OPT_BOOL(0, "batch-all-objects", &batch.all_objects,
 582                          N_("show all objects with --batch or --batch-check")),
 583                 OPT_END()
 584         };
 585
 586         git_config(git_cat_file_config, NULL);
 587
 588         batch.buffer_output = -1;
 589         argc = parse_options(argc, argv, prefix, options, cat_file_usage, 0);
 590
 591         if (opt) {
 592                 if (batch.enabled && (opt == 'c' || opt == 'w'))
 593                         batch.cmdmode = opt;
 594                 else if (argc == 1)
 595                         obj_name = argv[0];
 596                 else
 597                         usage_with_options(cat_file_usage, options);
 598         }
 599         if (!opt && !batch.enabled) {
 600                 if (argc == 2) {
 601                         exp_type = argv[0];
 602                         obj_name = argv[1];
 603                 } else
 604                         usage_with_options(cat_file_usage, options);
 605         }
 606         if (batch.enabled) {
 607                 if (batch.cmdmode != opt || argc)
 608                         usage_with_options(cat_file_usage, options);
 609                 if (batch.cmdmode && batch.all_objects)
 610                         die("--batch-all-objects cannot be combined with "
 611                             "--textconv nor with --filters");
 612         }
 613
 614         if ((batch.follow_symlinks || batch.all_objects) && !batch.enabled) {
 615                 usage_with_options(cat_file_usage, options);
 616         }
 617
 618         if (force_path && opt != 'c' && opt != 'w') {
 619                 error("--path=<path> needs --textconv or --filters");
 620                 usage_with_options(cat_file_usage, options);
 621         }
 622
 623         if (force_path && batch.enabled) {
 624                 error("--path=<path> incompatible with --batch");
 625                 usage_with_options(cat_file_usage, options);
 626         }
 627
 628         if (batch.buffer_output < 0)
 629                 batch.buffer_output = batch.all_objects;
 630
 631         if (batch.enabled)
 632                 return batch_objects(&batch);
 633
 634         if (unknown_type && opt != 't' && opt != 's')
 635                 die("git cat-file --allow-unknown-type: use with -s or -t");
 636         return cat_one_file(opt, exp_type, obj_name, unknown_type);
 637 }