builtin/grep.c

   1 /*
   2  * Builtin "git grep"
   3  *
   4  * Copyright (c) 2006 Junio C Hamano
   5  */
   6 #include "cache.h"
   7 #include "blob.h"
   8 #include "tree.h"
   9 #include "commit.h"
  10 #include "tag.h"
  11 #include "tree-walk.h"
  12 #include "builtin.h"
  13 #include "parse-options.h"
  14 #include "string-list.h"
  15 #include "run-command.h"
  16 #include "userdiff.h"
  17 #include "grep.h"
  18 #include "quote.h"
  19 #include "dir.h"
  20 #include "attr.h"
  21
  22 static char const * const grep_usage[] = {
  23         N_("git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  24         NULL
  25 };
  26
  27 static int use_threads = 1;
  28
  29 #ifndef NO_PTHREADS
  30 #define THREADS 8
  31 static pthread_t threads[THREADS];
  32
  33 /* We use one producer thread and THREADS consumer
  34  * threads. The producer adds struct work_items to 'todo' and the
  35  * consumers pick work items from the same array.
  36  */
  37 struct work_item {
  38         struct grep_source source;
  39         char done;
  40         struct strbuf out;
  41 };
  42
  43 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  44  * that have been or are processed by a consumer thread. We haven't
  45  * written the result for these to stdout yet.
  46  *
  47  * The work_items in [todo_start, todo_end) are waiting to be picked
  48  * up by a consumer thread.
  49  *
  50  * The ranges are modulo TODO_SIZE.
  51  */
  52 #define TODO_SIZE 128
  53 static struct work_item todo[TODO_SIZE];
  54 static int todo_start;
  55 static int todo_end;
  56 static int todo_done;
  57
  58 /* Has all work items been added? */
  59 static int all_work_added;
  60
  61 /* This lock protects all the variables above. */
  62 static pthread_mutex_t grep_mutex;
  63
  64 static inline void grep_lock(void)
  65 {
  66         if (use_threads)
  67                 pthread_mutex_lock(&grep_mutex);
  68 }
  69
  70 static inline void grep_unlock(void)
  71 {
  72         if (use_threads)
  73                 pthread_mutex_unlock(&grep_mutex);
  74 }
  75
  76 /* Signalled when a new work_item is added to todo. */
  77 static pthread_cond_t cond_add;
  78
  79 /* Signalled when the result from one work_item is written to
  80  * stdout.
  81  */
  82 static pthread_cond_t cond_write;
  83
  84 /* Signalled when we are finished with everything. */
  85 static pthread_cond_t cond_result;
  86
  87 static int skip_first_line;
  88
  89 static void add_work(struct grep_opt *opt, enum grep_source_type type,
  90                      const char *name, const void *id)
  91 {
  92         grep_lock();
  93
  94         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
  95                 pthread_cond_wait(&cond_write, &grep_mutex);
  96         }
  97
  98         grep_source_init(&todo[todo_end].source, type, name, id);
  99         if (opt->binary != GREP_BINARY_TEXT)
 100                 grep_source_load_driver(&todo[todo_end].source);
 101         todo[todo_end].done = 0;
 102         strbuf_reset(&todo[todo_end].out);
 103         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 104
 105         pthread_cond_signal(&cond_add);
 106         grep_unlock();
 107 }
 108
 109 static struct work_item *get_work(void)
 110 {
 111         struct work_item *ret;
 112
 113         grep_lock();
 114         while (todo_start == todo_end && !all_work_added) {
 115                 pthread_cond_wait(&cond_add, &grep_mutex);
 116         }
 117
 118         if (todo_start == todo_end && all_work_added) {
 119                 ret = NULL;
 120         } else {
 121                 ret = &todo[todo_start];
 122                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 123         }
 124         grep_unlock();
 125         return ret;
 126 }
 127
 128 static void work_done(struct work_item *w)
 129 {
 130         int old_done;
 131
 132         grep_lock();
 133         w->done = 1;
 134         old_done = todo_done;
 135         for(; todo[todo_done].done && todo_done != todo_start;
 136             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 137                 w = &todo[todo_done];
 138                 if (w->out.len) {
 139                         const char *p = w->out.buf;
 140                         size_t len = w->out.len;
 141
 142                         /* Skip the leading hunk mark of the first file. */
 143                         if (skip_first_line) {
 144                                 while (len) {
 145                                         len--;
 146                                         if (*p++ == '\n')
 147                                                 break;
 148                                 }
 149                                 skip_first_line = 0;
 150                         }
 151
 152                         write_or_die(1, p, len);
 153                 }
 154                 grep_source_clear(&w->source);
 155         }
 156
 157         if (old_done != todo_done)
 158                 pthread_cond_signal(&cond_write);
 159
 160         if (all_work_added && todo_done == todo_end)
 161                 pthread_cond_signal(&cond_result);
 162
 163         grep_unlock();
 164 }
 165
 166 static int skip_binary(struct grep_opt *opt, const char *filename)
 167 {
 168         if ((opt->binary & GREP_BINARY_NOMATCH)) {
 169                 static struct git_attr *attr_text;
 170                 struct git_attr_check check;
 171
 172                 if (!attr_text)
 173                         attr_text = git_attr("text");
 174                 memset(&check, 0, sizeof(check));
 175                 check.attr = attr_text;
 176                 return !git_check_attr(filename, 1, &check) &&
 177                                 ATTR_FALSE(check.value);
 178         }
 179         return 0;
 180 }
 181
 182 static void *run(void *arg)
 183 {
 184         int hit = 0;
 185         struct grep_opt *opt = arg;
 186
 187         while (1) {
 188                 struct work_item *w = get_work();
 189                 if (!w)
 190                         break;
 191
 192                 if (skip_binary(opt, (const char *)w->source.identifier))
 193                         continue;
 194
 195                 opt->output_priv = w;
 196                 hit |= grep_source(opt, &w->source);
 197                 grep_source_clear_data(&w->source);
 198                 work_done(w);
 199         }
 200         free_grep_patterns(arg);
 201         free(arg);
 202
 203         return (void*) (intptr_t) hit;
 204 }
 205
 206 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 207 {
 208         struct work_item *w = opt->output_priv;
 209         strbuf_add(&w->out, buf, size);
 210 }
 211
 212 static void start_threads(struct grep_opt *opt)
 213 {
 214         int i;
 215
 216         pthread_mutex_init(&grep_mutex, NULL);
 217         pthread_mutex_init(&grep_read_mutex, NULL);
 218         pthread_mutex_init(&grep_attr_mutex, NULL);
 219         pthread_cond_init(&cond_add, NULL);
 220         pthread_cond_init(&cond_write, NULL);
 221         pthread_cond_init(&cond_result, NULL);
 222         grep_use_locks = 1;
 223
 224         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 225                 strbuf_init(&todo[i].out, 0);
 226         }
 227
 228         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 229                 int err;
 230                 struct grep_opt *o = grep_opt_dup(opt);
 231                 o->output = strbuf_out;
 232                 o->debug = 0;
 233                 compile_grep_patterns(o);
 234                 err = pthread_create(&threads[i], NULL, run, o);
 235
 236                 if (err)
 237                         die(_("grep: failed to create thread: %s"),
 238                             strerror(err));
 239         }
 240 }
 241
 242 static int wait_all(void)
 243 {
 244         int hit = 0;
 245         int i;
 246
 247         grep_lock();
 248         all_work_added = 1;
 249
 250         /* Wait until all work is done. */
 251         while (todo_done != todo_end)
 252                 pthread_cond_wait(&cond_result, &grep_mutex);
 253
 254         /* Wake up all the consumer threads so they can see that there
 255          * is no more work to do.
 256          */
 257         pthread_cond_broadcast(&cond_add);
 258         grep_unlock();
 259
 260         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 261                 void *h;
 262                 pthread_join(threads[i], &h);
 263                 hit |= (int) (intptr_t) h;
 264         }
 265
 266         pthread_mutex_destroy(&grep_mutex);
 267         pthread_mutex_destroy(&grep_read_mutex);
 268         pthread_mutex_destroy(&grep_attr_mutex);
 269         pthread_cond_destroy(&cond_add);
 270         pthread_cond_destroy(&cond_write);
 271         pthread_cond_destroy(&cond_result);
 272         grep_use_locks = 0;
 273
 274         return hit;
 275 }
 276 #else /* !NO_PTHREADS */
 277
 278 static int wait_all(void)
 279 {
 280         return 0;
 281 }
 282 #endif
 283
 284 static int parse_pattern_type_arg(const char *opt, const char *arg)
 285 {
 286         if (!strcmp(arg, "default"))
 287                 return GREP_PATTERN_TYPE_UNSPECIFIED;
 288         else if (!strcmp(arg, "basic"))
 289                 return GREP_PATTERN_TYPE_BRE;
 290         else if (!strcmp(arg, "extended"))
 291                 return GREP_PATTERN_TYPE_ERE;
 292         else if (!strcmp(arg, "fixed"))
 293                 return GREP_PATTERN_TYPE_FIXED;
 294         else if (!strcmp(arg, "perl"))
 295                 return GREP_PATTERN_TYPE_PCRE;
 296         die("bad %s argument: %s", opt, arg);
 297 }
 298
 299 static void grep_pattern_type_options(const int pattern_type, struct grep_opt *opt)
 300 {
 301         switch (pattern_type) {
 302         case GREP_PATTERN_TYPE_UNSPECIFIED:
 303                 /* fall through */
 304
 305         case GREP_PATTERN_TYPE_BRE:
 306                 opt->fixed = 0;
 307                 opt->pcre = 0;
 308                 opt->regflags &= ~REG_EXTENDED;
 309                 break;
 310
 311         case GREP_PATTERN_TYPE_ERE:
 312                 opt->fixed = 0;
 313                 opt->pcre = 0;
 314                 opt->regflags |= REG_EXTENDED;
 315                 break;
 316
 317         case GREP_PATTERN_TYPE_FIXED:
 318                 opt->fixed = 1;
 319                 opt->pcre = 0;
 320                 opt->regflags &= ~REG_EXTENDED;
 321                 break;
 322
 323         case GREP_PATTERN_TYPE_PCRE:
 324                 opt->fixed = 0;
 325                 opt->pcre = 1;
 326                 opt->regflags &= ~REG_EXTENDED;
 327                 break;
 328         }
 329 }
 330
 331 static int grep_config(const char *var, const char *value, void *cb)
 332 {
 333         struct grep_opt *opt = cb;
 334         char *color = NULL;
 335
 336         if (userdiff_config(var, value) < 0)
 337                 return -1;
 338
 339         if (!strcmp(var, "grep.extendedregexp")) {
 340                 if (git_config_bool(var, value))
 341                         opt->extended_regexp_option = 1;
 342                 else
 343                         opt->extended_regexp_option = 0;
 344                 return 0;
 345         }
 346
 347         if (!strcmp(var, "grep.patterntype")) {
 348                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
 349                 return 0;
 350   }
 351
 352         if (!strcmp(var, "grep.linenumber")) {
 353                 opt->linenum = git_config_bool(var, value);
 354                 return 0;
 355         }
 356
 357         if (!strcmp(var, "color.grep"))
 358                 opt->color = git_config_colorbool(var, value);
 359         else if (!strcmp(var, "color.grep.context"))
 360                 color = opt->color_context;
 361         else if (!strcmp(var, "color.grep.filename"))
 362                 color = opt->color_filename;
 363         else if (!strcmp(var, "color.grep.function"))
 364                 color = opt->color_function;
 365         else if (!strcmp(var, "color.grep.linenumber"))
 366                 color = opt->color_lineno;
 367         else if (!strcmp(var, "color.grep.match"))
 368                 color = opt->color_match;
 369         else if (!strcmp(var, "color.grep.selected"))
 370                 color = opt->color_selected;
 371         else if (!strcmp(var, "color.grep.separator"))
 372                 color = opt->color_sep;
 373         else
 374                 return git_color_default_config(var, value, cb);
 375         if (color) {
 376                 if (!value)
 377                         return config_error_nonbool(var);
 378                 color_parse(value, var, color);
 379         }
 380         return 0;
 381 }
 382
 383 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 384 {
 385         void *data;
 386
 387         grep_read_lock();
 388         data = read_sha1_file(sha1, type, size);
 389         grep_read_unlock();
 390         return data;
 391 }
 392
 393 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 394                      const char *filename, int tree_name_len)
 395 {
 396         struct strbuf pathbuf = STRBUF_INIT;
 397
 398         if (opt->relative && opt->prefix_length) {
 399                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 400                                     opt->prefix);
 401                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 402         } else {
 403                 strbuf_addstr(&pathbuf, filename);
 404         }
 405
 406 #ifndef NO_PTHREADS
 407         if (use_threads) {
 408                 add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, sha1);
 409                 strbuf_release(&pathbuf);
 410                 return 0;
 411         } else
 412 #endif
 413         {
 414                 struct grep_source gs;
 415                 int hit;
 416
 417                 grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, sha1);
 418                 strbuf_release(&pathbuf);
 419                 hit = grep_source(opt, &gs);
 420
 421                 grep_source_clear(&gs);
 422                 return hit;
 423         }
 424 }
 425
 426 static int grep_file(struct grep_opt *opt, const char *filename)
 427 {
 428         struct strbuf buf = STRBUF_INIT;
 429
 430         if (opt->relative && opt->prefix_length)
 431                 quote_path_relative(filename, -1, &buf, opt->prefix);
 432         else
 433                 strbuf_addstr(&buf, filename);
 434
 435 #ifndef NO_PTHREADS
 436         if (use_threads) {
 437                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename);
 438                 strbuf_release(&buf);
 439                 return 0;
 440         } else
 441 #endif
 442         {
 443                 struct grep_source gs;
 444                 int hit;
 445
 446                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename);
 447                 strbuf_release(&buf);
 448                 hit = grep_source(opt, &gs);
 449
 450                 grep_source_clear(&gs);
 451                 return hit;
 452         }
 453 }
 454
 455 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 456 {
 457         struct string_list *path_list = opt->output_priv;
 458
 459         if (len == 1 && *(const char *)data == '\0')
 460                 return;
 461         string_list_append(path_list, xstrndup(data, len));
 462 }
 463
 464 static void run_pager(struct grep_opt *opt, const char *prefix)
 465 {
 466         struct string_list *path_list = opt->output_priv;
 467         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 468         int i, status;
 469
 470         for (i = 0; i < path_list->nr; i++)
 471                 argv[i] = path_list->items[i].string;
 472         argv[path_list->nr] = NULL;
 473
 474         if (prefix && chdir(prefix))
 475                 die(_("Failed to chdir: %s"), prefix);
 476         status = run_command_v_opt(argv, RUN_USING_SHELL);
 477         if (status)
 478                 exit(status);
 479         free(argv);
 480 }
 481
 482 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 483 {
 484         int hit = 0;
 485         int nr;
 486         read_cache();
 487
 488         for (nr = 0; nr < active_nr; nr++) {
 489                 struct cache_entry *ce = active_cache[nr];
 490                 if (!S_ISREG(ce->ce_mode))
 491                         continue;
 492                 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 493                         continue;
 494                 if (skip_binary(opt, ce->name))
 495                         continue;
 496
 497                 /*
 498                  * If CE_VALID is on, we assume worktree file and its cache entry
 499                  * are identical, even if worktree file has been modified, so use
 500                  * cache version instead
 501                  */
 502                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 503                         if (ce_stage(ce))
 504                                 continue;
 505                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 506                 }
 507                 else
 508                         hit |= grep_file(opt, ce->name);
 509                 if (ce_stage(ce)) {
 510                         do {
 511                                 nr++;
 512                         } while (nr < active_nr &&
 513                                  !strcmp(ce->name, active_cache[nr]->name));
 514                         nr--; /* compensate for loop control */
 515                 }
 516                 if (hit && opt->status_only)
 517                         break;
 518         }
 519         return hit;
 520 }
 521
 522 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 523                      struct tree_desc *tree, struct strbuf *base, int tn_len)
 524 {
 525         int hit = 0;
 526         enum interesting match = entry_not_interesting;
 527         struct name_entry entry;
 528         int old_baselen = base->len;
 529
 530         while (tree_entry(tree, &entry)) {
 531                 int te_len = tree_entry_len(&entry);
 532
 533                 if (match != all_entries_interesting) {
 534                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 535                         if (match == all_entries_not_interesting)
 536                                 break;
 537                         if (match == entry_not_interesting)
 538                                 continue;
 539                 }
 540
 541                 strbuf_add(base, entry.path, te_len);
 542
 543                 if (S_ISREG(entry.mode)) {
 544                         hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 545                 }
 546                 else if (S_ISDIR(entry.mode)) {
 547                         enum object_type type;
 548                         struct tree_desc sub;
 549                         void *data;
 550                         unsigned long size;
 551
 552                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 553                         if (!data)
 554                                 die(_("unable to read tree (%s)"),
 555                                     sha1_to_hex(entry.sha1));
 556
 557                         strbuf_addch(base, '/');
 558                         init_tree_desc(&sub, data, size);
 559                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 560                         free(data);
 561                 }
 562                 strbuf_setlen(base, old_baselen);
 563
 564                 if (hit && opt->status_only)
 565                         break;
 566         }
 567         return hit;
 568 }
 569
 570 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 571                        struct object *obj, const char *name)
 572 {
 573         if (obj->type == OBJ_BLOB)
 574                 return grep_sha1(opt, obj->sha1, name, 0);
 575         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 576                 struct tree_desc tree;
 577                 void *data;
 578                 unsigned long size;
 579                 struct strbuf base;
 580                 int hit, len;
 581
 582                 grep_read_lock();
 583                 data = read_object_with_reference(obj->sha1, tree_type,
 584                                                   &size, NULL);
 585                 grep_read_unlock();
 586
 587                 if (!data)
 588                         die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 589
 590                 len = name ? strlen(name) : 0;
 591                 strbuf_init(&base, PATH_MAX + len + 1);
 592                 if (len) {
 593                         strbuf_add(&base, name, len);
 594                         strbuf_addch(&base, ':');
 595                 }
 596                 init_tree_desc(&tree, data, size);
 597                 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 598                 strbuf_release(&base);
 599                 free(data);
 600                 return hit;
 601         }
 602         die(_("unable to grep from object of type %s"), typename(obj->type));
 603 }
 604
 605 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 606                         const struct object_array *list)
 607 {
 608         unsigned int i;
 609         int hit = 0;
 610         const unsigned int nr = list->nr;
 611
 612         for (i = 0; i < nr; i++) {
 613                 struct object *real_obj;
 614                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 615                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 616                         hit = 1;
 617                         if (opt->status_only)
 618                                 break;
 619                 }
 620         }
 621         return hit;
 622 }
 623
 624 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 625                           int exc_std)
 626 {
 627         struct dir_struct dir;
 628         int i, hit = 0;
 629
 630         memset(&dir, 0, sizeof(dir));
 631         if (exc_std)
 632                 setup_standard_excludes(&dir);
 633
 634         fill_directory(&dir, pathspec->raw);
 635         for (i = 0; i < dir.nr; i++) {
 636                 const char *name = dir.entries[i]->name;
 637                 int namelen = strlen(name);
 638                 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 639                         continue;
 640                 hit |= grep_file(opt, dir.entries[i]->name);
 641                 if (hit && opt->status_only)
 642                         break;
 643         }
 644         return hit;
 645 }
 646
 647 static int context_callback(const struct option *opt, const char *arg,
 648                             int unset)
 649 {
 650         struct grep_opt *grep_opt = opt->value;
 651         int value;
 652         const char *endp;
 653
 654         if (unset) {
 655                 grep_opt->pre_context = grep_opt->post_context = 0;
 656                 return 0;
 657         }
 658         value = strtol(arg, (char **)&endp, 10);
 659         if (*endp) {
 660                 return error(_("switch `%c' expects a numerical value"),
 661                              opt->short_name);
 662         }
 663         grep_opt->pre_context = grep_opt->post_context = value;
 664         return 0;
 665 }
 666
 667 static int file_callback(const struct option *opt, const char *arg, int unset)
 668 {
 669         struct grep_opt *grep_opt = opt->value;
 670         int from_stdin = !strcmp(arg, "-");
 671         FILE *patterns;
 672         int lno = 0;
 673         struct strbuf sb = STRBUF_INIT;
 674
 675         patterns = from_stdin ? stdin : fopen(arg, "r");
 676         if (!patterns)
 677                 die_errno(_("cannot open '%s'"), arg);
 678         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 679                 /* ignore empty line like grep does */
 680                 if (sb.len == 0)
 681                         continue;
 682
 683                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 684                                 GREP_PATTERN);
 685         }
 686         if (!from_stdin)
 687                 fclose(patterns);
 688         strbuf_release(&sb);
 689         return 0;
 690 }
 691
 692 static int not_callback(const struct option *opt, const char *arg, int unset)
 693 {
 694         struct grep_opt *grep_opt = opt->value;
 695         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 696         return 0;
 697 }
 698
 699 static int and_callback(const struct option *opt, const char *arg, int unset)
 700 {
 701         struct grep_opt *grep_opt = opt->value;
 702         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 703         return 0;
 704 }
 705
 706 static int open_callback(const struct option *opt, const char *arg, int unset)
 707 {
 708         struct grep_opt *grep_opt = opt->value;
 709         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 710         return 0;
 711 }
 712
 713 static int close_callback(const struct option *opt, const char *arg, int unset)
 714 {
 715         struct grep_opt *grep_opt = opt->value;
 716         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 717         return 0;
 718 }
 719
 720 static int pattern_callback(const struct option *opt, const char *arg,
 721                             int unset)
 722 {
 723         struct grep_opt *grep_opt = opt->value;
 724         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 725         return 0;
 726 }
 727
 728 static int help_callback(const struct option *opt, const char *arg, int unset)
 729 {
 730         return -1;
 731 }
 732
 733 int cmd_grep(int argc, const char **argv, const char *prefix)
 734 {
 735         int hit = 0;
 736         int cached = 0, untracked = 0, opt_exclude = -1;
 737         int seen_dashdash = 0;
 738         int external_grep_allowed__ignored;
 739         const char *show_in_pager = NULL, *default_pager = "dummy";
 740         struct grep_opt opt;
 741         struct object_array list = OBJECT_ARRAY_INIT;
 742         const char **paths = NULL;
 743         struct pathspec pathspec;
 744         struct string_list path_list = STRING_LIST_INIT_NODUP;
 745         int i;
 746         int dummy;
 747         int use_index = 1;
 748         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 749
 750         struct option options[] = {
 751                 OPT_BOOLEAN(0, "cached", &cached,
 752                         N_("search in index instead of in the work tree")),
 753                 OPT_NEGBIT(0, "no-index", &use_index,
 754                          N_("find in contents not managed by git"), 1),
 755                 OPT_BOOLEAN(0, "untracked", &untracked,
 756                         N_("search in both tracked and untracked files")),
 757                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 758                             N_("search also in ignored files"), 1),
 759                 OPT_GROUP(""),
 760                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 761                         N_("show non-matching lines")),
 762                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 763                         N_("case insensitive matching")),
 764                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 765                         N_("match patterns only at word boundaries")),
 766                 OPT_SET_INT('a', "text", &opt.binary,
 767                         N_("process binary files as text"), GREP_BINARY_TEXT),
 768                 OPT_SET_INT('I', NULL, &opt.binary,
 769                         N_("don't match patterns in binary files"),
 770                         GREP_BINARY_NOMATCH),
 771                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
 772                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
 773                         NULL, 1 },
 774                 OPT_GROUP(""),
 775                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
 776                             N_("use extended POSIX regular expressions"),
 777                             GREP_PATTERN_TYPE_ERE),
 778                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
 779                             N_("use basic POSIX regular expressions (default)"),
 780                             GREP_PATTERN_TYPE_BRE),
 781                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
 782                             N_("interpret patterns as fixed strings"),
 783                             GREP_PATTERN_TYPE_FIXED),
 784                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
 785                             N_("use Perl-compatible regular expressions"),
 786                             GREP_PATTERN_TYPE_PCRE),
 787                 OPT_GROUP(""),
 788                 OPT_BOOLEAN('n', "line-number", &opt.linenum, N_("show line numbers")),
 789                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
 790                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
 791                 OPT_NEGBIT(0, "full-name", &opt.relative,
 792                         N_("show filenames relative to top directory"), 1),
 793                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 794                         N_("show only filenames instead of matching lines")),
 795                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 796                         N_("synonym for --files-with-matches")),
 797                 OPT_BOOLEAN('L', "files-without-match",
 798                         &opt.unmatch_name_only,
 799                         N_("show only the names of files without match")),
 800                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 801                         N_("print NUL after filenames")),
 802                 OPT_BOOLEAN('c', "count", &opt.count,
 803                         N_("show the number of matches instead of matching lines")),
 804                 OPT__COLOR(&opt.color, N_("highlight matches")),
 805                 OPT_BOOLEAN(0, "break", &opt.file_break,
 806                         N_("print empty line between matches from different files")),
 807                 OPT_BOOLEAN(0, "heading", &opt.heading,
 808                         N_("show filename only once above matches from same file")),
 809                 OPT_GROUP(""),
 810                 OPT_CALLBACK('C', "context", &opt, N_("n"),
 811                         N_("show <n> context lines before and after matches"),
 812                         context_callback),
 813                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 814                         N_("show <n> context lines before matches")),
 815                 OPT_INTEGER('A', "after-context", &opt.post_context,
 816                         N_("show <n> context lines after matches")),
 817                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
 818                         context_callback),
 819                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 820                         N_("show a line with the function name before matches")),
 821                 OPT_BOOLEAN('W', "function-context", &opt.funcbody,
 822                         N_("show the surrounding function")),
 823                 OPT_GROUP(""),
 824                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
 825                         N_("read patterns from file"), file_callback),
 826                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
 827                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
 828                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 829                   N_("combine patterns specified with -e"),
 830                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 831                 OPT_BOOLEAN(0, "or", &dummy, ""),
 832                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 833                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 834                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 835                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 836                   open_callback },
 837                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 838                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 839                   close_callback },
 840                 OPT__QUIET(&opt.status_only,
 841                            N_("indicate hit with exit status without output")),
 842                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 843                         N_("show only matches from files that match all patterns")),
 844                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
 845                   N_("show parse tree for grep expression"),
 846                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
 847                 OPT_GROUP(""),
 848                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 849                         N_("pager"), N_("show matching files in the pager"),
 850                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 851                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 852                             N_("allow calling of grep(1) (ignored by this build)")),
 853                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, N_("show usage"),
 854                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 855                 OPT_END()
 856         };
 857
 858         /*
 859          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 860          * to show usage information and exit.
 861          */
 862         if (argc == 2 && !strcmp(argv[1], "-h"))
 863                 usage_with_options(grep_usage, options);
 864
 865         memset(&opt, 0, sizeof(opt));
 866         opt.prefix = prefix;
 867         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 868         opt.relative = 1;
 869         opt.pathname = 1;
 870         opt.pattern_tail = &opt.pattern_list;
 871         opt.header_tail = &opt.header_list;
 872         opt.regflags = REG_NEWLINE;
 873         opt.max_depth = -1;
 874         opt.pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
 875         opt.extended_regexp_option = 0;
 876
 877         strcpy(opt.color_context, "");
 878         strcpy(opt.color_filename, "");
 879         strcpy(opt.color_function, "");
 880         strcpy(opt.color_lineno, "");
 881         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 882         strcpy(opt.color_selected, "");
 883         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 884         opt.color = -1;
 885         git_config(grep_config, &opt);
 886
 887         /*
 888          * If there is no -- then the paths must exist in the working
 889          * tree.  If there is no explicit pattern specified with -e or
 890          * -f, we take the first unrecognized non option to be the
 891          * pattern, but then what follows it must be zero or more
 892          * valid refs up to the -- (if exists), and then existing
 893          * paths.  If there is an explicit pattern, then the first
 894          * unrecognized non option is the beginning of the refs list
 895          * that continues up to the -- (if exists), and then paths.
 896          */
 897         argc = parse_options(argc, argv, prefix, options, grep_usage,
 898                              PARSE_OPT_KEEP_DASHDASH |
 899                              PARSE_OPT_STOP_AT_NON_OPTION |
 900                              PARSE_OPT_NO_INTERNAL_HELP);
 901
 902         if (pattern_type_arg != GREP_PATTERN_TYPE_UNSPECIFIED)
 903                 grep_pattern_type_options(pattern_type_arg, &opt);
 904         else if (opt.pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 905                 grep_pattern_type_options(opt.pattern_type_option, &opt);
 906         else if (opt.extended_regexp_option)
 907                 grep_pattern_type_options(GREP_PATTERN_TYPE_ERE, &opt);
 908
 909         if (use_index && !startup_info->have_repository)
 910                 /* die the same way as if we did it at the beginning */
 911                 setup_git_directory();
 912
 913         /*
 914          * skip a -- separator; we know it cannot be
 915          * separating revisions from pathnames if
 916          * we haven't even had any patterns yet
 917          */
 918         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 919                 argv++;
 920                 argc--;
 921         }
 922
 923         /* First unrecognized non-option token */
 924         if (argc > 0 && !opt.pattern_list) {
 925                 append_grep_pattern(&opt, argv[0], "command line", 0,
 926                                     GREP_PATTERN);
 927                 argv++;
 928                 argc--;
 929         }
 930
 931         if (show_in_pager == default_pager)
 932                 show_in_pager = git_pager(1);
 933         if (show_in_pager) {
 934                 opt.color = 0;
 935                 opt.name_only = 1;
 936                 opt.null_following_name = 1;
 937                 opt.output_priv = &path_list;
 938                 opt.output = append_path;
 939                 string_list_append(&path_list, show_in_pager);
 940                 use_threads = 0;
 941         }
 942         if ((opt.binary & GREP_BINARY_NOMATCH))
 943                 use_threads = 0;
 944
 945         if (!opt.pattern_list)
 946                 die(_("no pattern given."));
 947         if (!opt.fixed && opt.ignore_case)
 948                 opt.regflags |= REG_ICASE;
 949
 950         compile_grep_patterns(&opt);
 951
 952         /* Check revs and then paths */
 953         for (i = 0; i < argc; i++) {
 954                 const char *arg = argv[i];
 955                 unsigned char sha1[20];
 956                 /* Is it a rev? */
 957                 if (!get_sha1(arg, sha1)) {
 958                         struct object *object = parse_object(sha1);
 959                         if (!object)
 960                                 die(_("bad object %s"), arg);
 961                         add_object_array(object, arg, &list);
 962                         continue;
 963                 }
 964                 if (!strcmp(arg, "--")) {
 965                         i++;
 966                         seen_dashdash = 1;
 967                 }
 968                 break;
 969         }
 970
 971 #ifndef NO_PTHREADS
 972         if (list.nr || cached || online_cpus() == 1)
 973                 use_threads = 0;
 974 #else
 975         use_threads = 0;
 976 #endif
 977
 978 #ifndef NO_PTHREADS
 979         if (use_threads) {
 980                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
 981                     && (opt.pre_context || opt.post_context ||
 982                         opt.file_break || opt.funcbody))
 983                         skip_first_line = 1;
 984                 start_threads(&opt);
 985         }
 986 #endif
 987
 988         /* The rest are paths */
 989         if (!seen_dashdash) {
 990                 int j;
 991                 for (j = i; j < argc; j++)
 992                         verify_filename(prefix, argv[j], j == i);
 993         }
 994
 995         paths = get_pathspec(prefix, argv + i);
 996         init_pathspec(&pathspec, paths);
 997         pathspec.max_depth = opt.max_depth;
 998         pathspec.recursive = 1;
 999
1000         if (show_in_pager && (cached || list.nr))
1001                 die(_("--open-files-in-pager only works on the worktree"));
1002
1003         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1004                 const char *pager = path_list.items[0].string;
1005                 int len = strlen(pager);
1006
1007                 if (len > 4 && is_dir_sep(pager[len - 5]))
1008                         pager += len - 4;
1009
1010                 if (opt.ignore_case && !strcmp("less", pager))
1011                         string_list_append(&path_list, "-i");
1012
1013                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1014                         struct strbuf buf = STRBUF_INIT;
1015                         strbuf_addf(&buf, "+/%s%s",
1016                                         strcmp("less", pager) ? "" : "*",
1017                                         opt.pattern_list->pattern);
1018                         string_list_append(&path_list, buf.buf);
1019                         strbuf_detach(&buf, NULL);
1020                 }
1021         }
1022
1023         if (!show_in_pager)
1024                 setup_pager();
1025
1026         if (!use_index && (untracked || cached))
1027                 die(_("--cached or --untracked cannot be used with --no-index."));
1028
1029         if (!use_index || untracked) {
1030                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1031                 if (list.nr)
1032                         die(_("--no-index or --untracked cannot be used with revs."));
1033                 hit = grep_directory(&opt, &pathspec, use_exclude);
1034         } else if (0 <= opt_exclude) {
1035                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1036         } else if (!list.nr) {
1037                 if (!cached)
1038                         setup_work_tree();
1039
1040                 hit = grep_cache(&opt, &pathspec, cached);
1041         } else {
1042                 if (cached)
1043                         die(_("both --cached and trees are given."));
1044                 hit = grep_objects(&opt, &pathspec, &list);
1045         }
1046
1047         if (use_threads)
1048                 hit |= wait_all();
1049         if (hit && show_in_pager)
1050                 run_pager(&opt, prefix);
1051         free_grep_patterns(&opt);
1052         return !hit;
1053 }