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 "thread-utils.h"
  21
  22 static char const * const grep_usage[] = {
  23         "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 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  34                        const char *name);
  35 static void *load_file(const char *filename, size_t *sz);
  36
  37 enum work_type {WORK_SHA1, WORK_FILE};
  38
  39 /* We use one producer thread and THREADS consumer
  40  * threads. The producer adds struct work_items to 'todo' and the
  41  * consumers pick work items from the same array.
  42  */
  43 struct work_item {
  44         enum work_type type;
  45         char *name;
  46
  47         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  48          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  49          * terminated filename.
  50          */
  51         void *identifier;
  52         char done;
  53         struct strbuf out;
  54 };
  55
  56 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  57  * that have been or are processed by a consumer thread. We haven't
  58  * written the result for these to stdout yet.
  59  *
  60  * The work_items in [todo_start, todo_end) are waiting to be picked
  61  * up by a consumer thread.
  62  *
  63  * The ranges are modulo TODO_SIZE.
  64  */
  65 #define TODO_SIZE 128
  66 static struct work_item todo[TODO_SIZE];
  67 static int todo_start;
  68 static int todo_end;
  69 static int todo_done;
  70
  71 /* Has all work items been added? */
  72 static int all_work_added;
  73
  74 /* This lock protects all the variables above. */
  75 static pthread_mutex_t grep_mutex;
  76
  77 /* Used to serialize calls to read_sha1_file. */
  78 static pthread_mutex_t read_sha1_mutex;
  79
  80 #define grep_lock() pthread_mutex_lock(&grep_mutex)
  81 #define grep_unlock() pthread_mutex_unlock(&grep_mutex)
  82 #define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
  83 #define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
  84
  85 /* Signalled when a new work_item is added to todo. */
  86 static pthread_cond_t cond_add;
  87
  88 /* Signalled when the result from one work_item is written to
  89  * stdout.
  90  */
  91 static pthread_cond_t cond_write;
  92
  93 /* Signalled when we are finished with everything. */
  94 static pthread_cond_t cond_result;
  95
  96 static int skip_first_line;
  97
  98 static void add_work(enum work_type type, char *name, void *id)
  99 {
 100         grep_lock();
 101
 102         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 103                 pthread_cond_wait(&cond_write, &grep_mutex);
 104         }
 105
 106         todo[todo_end].type = type;
 107         todo[todo_end].name = name;
 108         todo[todo_end].identifier = id;
 109         todo[todo_end].done = 0;
 110         strbuf_reset(&todo[todo_end].out);
 111         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 112
 113         pthread_cond_signal(&cond_add);
 114         grep_unlock();
 115 }
 116
 117 static struct work_item *get_work(void)
 118 {
 119         struct work_item *ret;
 120
 121         grep_lock();
 122         while (todo_start == todo_end && !all_work_added) {
 123                 pthread_cond_wait(&cond_add, &grep_mutex);
 124         }
 125
 126         if (todo_start == todo_end && all_work_added) {
 127                 ret = NULL;
 128         } else {
 129                 ret = &todo[todo_start];
 130                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 131         }
 132         grep_unlock();
 133         return ret;
 134 }
 135
 136 static void grep_sha1_async(struct grep_opt *opt, char *name,
 137                             const unsigned char *sha1)
 138 {
 139         unsigned char *s;
 140         s = xmalloc(20);
 141         memcpy(s, sha1, 20);
 142         add_work(WORK_SHA1, name, s);
 143 }
 144
 145 static void grep_file_async(struct grep_opt *opt, char *name,
 146                             const char *filename)
 147 {
 148         add_work(WORK_FILE, name, xstrdup(filename));
 149 }
 150
 151 static void work_done(struct work_item *w)
 152 {
 153         int old_done;
 154
 155         grep_lock();
 156         w->done = 1;
 157         old_done = todo_done;
 158         for(; todo[todo_done].done && todo_done != todo_start;
 159             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 160                 w = &todo[todo_done];
 161                 if (w->out.len) {
 162                         const char *p = w->out.buf;
 163                         size_t len = w->out.len;
 164
 165                         /* Skip the leading hunk mark of the first file. */
 166                         if (skip_first_line) {
 167                                 while (len) {
 168                                         len--;
 169                                         if (*p++ == '\n')
 170                                                 break;
 171                                 }
 172                                 skip_first_line = 0;
 173                         }
 174
 175                         write_or_die(1, p, len);
 176                 }
 177                 free(w->name);
 178                 free(w->identifier);
 179         }
 180
 181         if (old_done != todo_done)
 182                 pthread_cond_signal(&cond_write);
 183
 184         if (all_work_added && todo_done == todo_end)
 185                 pthread_cond_signal(&cond_result);
 186
 187         grep_unlock();
 188 }
 189
 190 static void *run(void *arg)
 191 {
 192         int hit = 0;
 193         struct grep_opt *opt = arg;
 194
 195         while (1) {
 196                 struct work_item *w = get_work();
 197                 if (!w)
 198                         break;
 199
 200                 opt->output_priv = w;
 201                 if (w->type == WORK_SHA1) {
 202                         unsigned long sz;
 203                         void* data = load_sha1(w->identifier, &sz, w->name);
 204
 205                         if (data) {
 206                                 hit |= grep_buffer(opt, w->name, data, sz);
 207                                 free(data);
 208                         }
 209                 } else if (w->type == WORK_FILE) {
 210                         size_t sz;
 211                         void* data = load_file(w->identifier, &sz);
 212                         if (data) {
 213                                 hit |= grep_buffer(opt, w->name, data, sz);
 214                                 free(data);
 215                         }
 216                 } else {
 217                         assert(0);
 218                 }
 219
 220                 work_done(w);
 221         }
 222         free_grep_patterns(arg);
 223         free(arg);
 224
 225         return (void*) (intptr_t) hit;
 226 }
 227
 228 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 229 {
 230         struct work_item *w = opt->output_priv;
 231         strbuf_add(&w->out, buf, size);
 232 }
 233
 234 static void start_threads(struct grep_opt *opt)
 235 {
 236         int i;
 237
 238         pthread_mutex_init(&grep_mutex, NULL);
 239         pthread_mutex_init(&read_sha1_mutex, NULL);
 240         pthread_cond_init(&cond_add, NULL);
 241         pthread_cond_init(&cond_write, NULL);
 242         pthread_cond_init(&cond_result, NULL);
 243
 244         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 245                 strbuf_init(&todo[i].out, 0);
 246         }
 247
 248         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 249                 int err;
 250                 struct grep_opt *o = grep_opt_dup(opt);
 251                 o->output = strbuf_out;
 252                 compile_grep_patterns(o);
 253                 err = pthread_create(&threads[i], NULL, run, o);
 254
 255                 if (err)
 256                         die(_("grep: failed to create thread: %s"),
 257                             strerror(err));
 258         }
 259 }
 260
 261 static int wait_all(void)
 262 {
 263         int hit = 0;
 264         int i;
 265
 266         grep_lock();
 267         all_work_added = 1;
 268
 269         /* Wait until all work is done. */
 270         while (todo_done != todo_end)
 271                 pthread_cond_wait(&cond_result, &grep_mutex);
 272
 273         /* Wake up all the consumer threads so they can see that there
 274          * is no more work to do.
 275          */
 276         pthread_cond_broadcast(&cond_add);
 277         grep_unlock();
 278
 279         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 280                 void *h;
 281                 pthread_join(threads[i], &h);
 282                 hit |= (int) (intptr_t) h;
 283         }
 284
 285         pthread_mutex_destroy(&grep_mutex);
 286         pthread_mutex_destroy(&read_sha1_mutex);
 287         pthread_cond_destroy(&cond_add);
 288         pthread_cond_destroy(&cond_write);
 289         pthread_cond_destroy(&cond_result);
 290
 291         return hit;
 292 }
 293 #else /* !NO_PTHREADS */
 294 #define read_sha1_lock()
 295 #define read_sha1_unlock()
 296
 297 static int wait_all(void)
 298 {
 299         return 0;
 300 }
 301 #endif
 302
 303 static int grep_config(const char *var, const char *value, void *cb)
 304 {
 305         struct grep_opt *opt = cb;
 306         char *color = NULL;
 307
 308         switch (userdiff_config(var, value)) {
 309         case 0: break;
 310         case -1: return -1;
 311         default: return 0;
 312         }
 313
 314         if (!strcmp(var, "grep.extendedregexp")) {
 315                 if (git_config_bool(var, value))
 316                         opt->regflags |= REG_EXTENDED;
 317                 else
 318                         opt->regflags &= ~REG_EXTENDED;
 319                 return 0;
 320         }
 321
 322         if (!strcmp(var, "grep.linenumber")) {
 323                 opt->linenum = git_config_bool(var, value);
 324                 return 0;
 325         }
 326
 327         if (!strcmp(var, "color.grep"))
 328                 opt->color = git_config_colorbool(var, value);
 329         else if (!strcmp(var, "color.grep.context"))
 330                 color = opt->color_context;
 331         else if (!strcmp(var, "color.grep.filename"))
 332                 color = opt->color_filename;
 333         else if (!strcmp(var, "color.grep.function"))
 334                 color = opt->color_function;
 335         else if (!strcmp(var, "color.grep.linenumber"))
 336                 color = opt->color_lineno;
 337         else if (!strcmp(var, "color.grep.match"))
 338                 color = opt->color_match;
 339         else if (!strcmp(var, "color.grep.selected"))
 340                 color = opt->color_selected;
 341         else if (!strcmp(var, "color.grep.separator"))
 342                 color = opt->color_sep;
 343         else
 344                 return git_color_default_config(var, value, cb);
 345         if (color) {
 346                 if (!value)
 347                         return config_error_nonbool(var);
 348                 color_parse(value, var, color);
 349         }
 350         return 0;
 351 }
 352
 353 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 354 {
 355         void *data;
 356
 357         if (use_threads) {
 358                 read_sha1_lock();
 359                 data = read_sha1_file(sha1, type, size);
 360                 read_sha1_unlock();
 361         } else {
 362                 data = read_sha1_file(sha1, type, size);
 363         }
 364         return data;
 365 }
 366
 367 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 368                        const char *name)
 369 {
 370         enum object_type type;
 371         void *data = lock_and_read_sha1_file(sha1, &type, size);
 372
 373         if (!data)
 374                 error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
 375
 376         return data;
 377 }
 378
 379 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 380                      const char *filename, int tree_name_len)
 381 {
 382         struct strbuf pathbuf = STRBUF_INIT;
 383         char *name;
 384
 385         if (opt->relative && opt->prefix_length) {
 386                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 387                                     opt->prefix);
 388                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 389         } else {
 390                 strbuf_addstr(&pathbuf, filename);
 391         }
 392
 393         name = strbuf_detach(&pathbuf, NULL);
 394
 395 #ifndef NO_PTHREADS
 396         if (use_threads) {
 397                 grep_sha1_async(opt, name, sha1);
 398                 return 0;
 399         } else
 400 #endif
 401         {
 402                 int hit;
 403                 unsigned long sz;
 404                 void *data = load_sha1(sha1, &sz, name);
 405                 if (!data)
 406                         hit = 0;
 407                 else
 408                         hit = grep_buffer(opt, name, data, sz);
 409
 410                 free(data);
 411                 free(name);
 412                 return hit;
 413         }
 414 }
 415
 416 static void *load_file(const char *filename, size_t *sz)
 417 {
 418         struct stat st;
 419         char *data;
 420         int i;
 421
 422         if (lstat(filename, &st) < 0) {
 423         err_ret:
 424                 if (errno != ENOENT)
 425                         error(_("'%s': %s"), filename, strerror(errno));
 426                 return NULL;
 427         }
 428         if (!S_ISREG(st.st_mode))
 429                 return NULL;
 430         *sz = xsize_t(st.st_size);
 431         i = open(filename, O_RDONLY);
 432         if (i < 0)
 433                 goto err_ret;
 434         data = xmalloc(*sz + 1);
 435         if (st.st_size != read_in_full(i, data, *sz)) {
 436                 error(_("'%s': short read %s"), filename, strerror(errno));
 437                 close(i);
 438                 free(data);
 439                 return NULL;
 440         }
 441         close(i);
 442         data[*sz] = 0;
 443         return data;
 444 }
 445
 446 static int grep_file(struct grep_opt *opt, const char *filename)
 447 {
 448         struct strbuf buf = STRBUF_INIT;
 449         char *name;
 450
 451         if (opt->relative && opt->prefix_length)
 452                 quote_path_relative(filename, -1, &buf, opt->prefix);
 453         else
 454                 strbuf_addstr(&buf, filename);
 455         name = strbuf_detach(&buf, NULL);
 456
 457 #ifndef NO_PTHREADS
 458         if (use_threads) {
 459                 grep_file_async(opt, name, filename);
 460                 return 0;
 461         } else
 462 #endif
 463         {
 464                 int hit;
 465                 size_t sz;
 466                 void *data = load_file(filename, &sz);
 467                 if (!data)
 468                         hit = 0;
 469                 else
 470                         hit = grep_buffer(opt, name, data, sz);
 471
 472                 free(data);
 473                 free(name);
 474                 return hit;
 475         }
 476 }
 477
 478 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 479 {
 480         struct string_list *path_list = opt->output_priv;
 481
 482         if (len == 1 && *(const char *)data == '\0')
 483                 return;
 484         string_list_append(path_list, xstrndup(data, len));
 485 }
 486
 487 static void run_pager(struct grep_opt *opt, const char *prefix)
 488 {
 489         struct string_list *path_list = opt->output_priv;
 490         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 491         int i, status;
 492
 493         for (i = 0; i < path_list->nr; i++)
 494                 argv[i] = path_list->items[i].string;
 495         argv[path_list->nr] = NULL;
 496
 497         if (prefix && chdir(prefix))
 498                 die(_("Failed to chdir: %s"), prefix);
 499         status = run_command_v_opt(argv, RUN_USING_SHELL);
 500         if (status)
 501                 exit(status);
 502         free(argv);
 503 }
 504
 505 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 506 {
 507         int hit = 0;
 508         int nr;
 509         read_cache();
 510
 511         for (nr = 0; nr < active_nr; nr++) {
 512                 struct cache_entry *ce = active_cache[nr];
 513                 if (!S_ISREG(ce->ce_mode))
 514                         continue;
 515                 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 516                         continue;
 517                 /*
 518                  * If CE_VALID is on, we assume worktree file and its cache entry
 519                  * are identical, even if worktree file has been modified, so use
 520                  * cache version instead
 521                  */
 522                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 523                         if (ce_stage(ce))
 524                                 continue;
 525                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 526                 }
 527                 else
 528                         hit |= grep_file(opt, ce->name);
 529                 if (ce_stage(ce)) {
 530                         do {
 531                                 nr++;
 532                         } while (nr < active_nr &&
 533                                  !strcmp(ce->name, active_cache[nr]->name));
 534                         nr--; /* compensate for loop control */
 535                 }
 536                 if (hit && opt->status_only)
 537                         break;
 538         }
 539         return hit;
 540 }
 541
 542 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 543                      struct tree_desc *tree, struct strbuf *base, int tn_len)
 544 {
 545         int hit = 0, match = 0;
 546         struct name_entry entry;
 547         int old_baselen = base->len;
 548
 549         while (tree_entry(tree, &entry)) {
 550                 int te_len = tree_entry_len(entry.path, entry.sha1);
 551
 552                 if (match != 2) {
 553                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 554                         if (match < 0)
 555                                 break;
 556                         if (match == 0)
 557                                 continue;
 558                 }
 559
 560                 strbuf_add(base, entry.path, te_len);
 561
 562                 if (S_ISREG(entry.mode)) {
 563                         hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 564                 }
 565                 else if (S_ISDIR(entry.mode)) {
 566                         enum object_type type;
 567                         struct tree_desc sub;
 568                         void *data;
 569                         unsigned long size;
 570
 571                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 572                         if (!data)
 573                                 die(_("unable to read tree (%s)"),
 574                                     sha1_to_hex(entry.sha1));
 575
 576                         strbuf_addch(base, '/');
 577                         init_tree_desc(&sub, data, size);
 578                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 579                         free(data);
 580                 }
 581                 strbuf_setlen(base, old_baselen);
 582
 583                 if (hit && opt->status_only)
 584                         break;
 585         }
 586         return hit;
 587 }
 588
 589 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 590                        struct object *obj, const char *name)
 591 {
 592         if (obj->type == OBJ_BLOB)
 593                 return grep_sha1(opt, obj->sha1, name, 0);
 594         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 595                 struct tree_desc tree;
 596                 void *data;
 597                 unsigned long size;
 598                 struct strbuf base;
 599                 int hit, len;
 600
 601                 read_sha1_lock();
 602                 data = read_object_with_reference(obj->sha1, tree_type,
 603                                                   &size, NULL);
 604                 read_sha1_unlock();
 605
 606                 if (!data)
 607                         die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 608
 609                 len = name ? strlen(name) : 0;
 610                 strbuf_init(&base, PATH_MAX + len + 1);
 611                 if (len) {
 612                         strbuf_add(&base, name, len);
 613                         strbuf_addch(&base, ':');
 614                 }
 615                 init_tree_desc(&tree, data, size);
 616                 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 617                 strbuf_release(&base);
 618                 free(data);
 619                 return hit;
 620         }
 621         die(_("unable to grep from object of type %s"), typename(obj->type));
 622 }
 623
 624 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 625                         const struct object_array *list)
 626 {
 627         unsigned int i;
 628         int hit = 0;
 629         const unsigned int nr = list->nr;
 630
 631         for (i = 0; i < nr; i++) {
 632                 struct object *real_obj;
 633                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 634                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 635                         hit = 1;
 636                         if (opt->status_only)
 637                                 break;
 638                 }
 639         }
 640         return hit;
 641 }
 642
 643 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec)
 644 {
 645         struct dir_struct dir;
 646         int i, hit = 0;
 647
 648         memset(&dir, 0, sizeof(dir));
 649         setup_standard_excludes(&dir);
 650
 651         fill_directory(&dir, pathspec->raw);
 652         for (i = 0; i < dir.nr; i++) {
 653                 const char *name = dir.entries[i]->name;
 654                 int namelen = strlen(name);
 655                 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 656                         continue;
 657                 hit |= grep_file(opt, dir.entries[i]->name);
 658                 if (hit && opt->status_only)
 659                         break;
 660         }
 661         return hit;
 662 }
 663
 664 static int context_callback(const struct option *opt, const char *arg,
 665                             int unset)
 666 {
 667         struct grep_opt *grep_opt = opt->value;
 668         int value;
 669         const char *endp;
 670
 671         if (unset) {
 672                 grep_opt->pre_context = grep_opt->post_context = 0;
 673                 return 0;
 674         }
 675         value = strtol(arg, (char **)&endp, 10);
 676         if (*endp) {
 677                 return error(_("switch `%c' expects a numerical value"),
 678                              opt->short_name);
 679         }
 680         grep_opt->pre_context = grep_opt->post_context = value;
 681         return 0;
 682 }
 683
 684 static int file_callback(const struct option *opt, const char *arg, int unset)
 685 {
 686         struct grep_opt *grep_opt = opt->value;
 687         int from_stdin = !strcmp(arg, "-");
 688         FILE *patterns;
 689         int lno = 0;
 690         struct strbuf sb = STRBUF_INIT;
 691
 692         patterns = from_stdin ? stdin : fopen(arg, "r");
 693         if (!patterns)
 694                 die_errno(_("cannot open '%s'"), arg);
 695         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 696                 char *s;
 697                 size_t len;
 698
 699                 /* ignore empty line like grep does */
 700                 if (sb.len == 0)
 701                         continue;
 702
 703                 s = strbuf_detach(&sb, &len);
 704                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 705         }
 706         if (!from_stdin)
 707                 fclose(patterns);
 708         strbuf_release(&sb);
 709         return 0;
 710 }
 711
 712 static int not_callback(const struct option *opt, const char *arg, int unset)
 713 {
 714         struct grep_opt *grep_opt = opt->value;
 715         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 716         return 0;
 717 }
 718
 719 static int and_callback(const struct option *opt, const char *arg, int unset)
 720 {
 721         struct grep_opt *grep_opt = opt->value;
 722         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 723         return 0;
 724 }
 725
 726 static int open_callback(const struct option *opt, const char *arg, int unset)
 727 {
 728         struct grep_opt *grep_opt = opt->value;
 729         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 730         return 0;
 731 }
 732
 733 static int close_callback(const struct option *opt, const char *arg, int unset)
 734 {
 735         struct grep_opt *grep_opt = opt->value;
 736         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 737         return 0;
 738 }
 739
 740 static int pattern_callback(const struct option *opt, const char *arg,
 741                             int unset)
 742 {
 743         struct grep_opt *grep_opt = opt->value;
 744         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 745         return 0;
 746 }
 747
 748 static int help_callback(const struct option *opt, const char *arg, int unset)
 749 {
 750         return -1;
 751 }
 752
 753 int cmd_grep(int argc, const char **argv, const char *prefix)
 754 {
 755         int hit = 0;
 756         int cached = 0;
 757         int seen_dashdash = 0;
 758         int external_grep_allowed__ignored;
 759         const char *show_in_pager = NULL, *default_pager = "dummy";
 760         struct grep_opt opt;
 761         struct object_array list = OBJECT_ARRAY_INIT;
 762         const char **paths = NULL;
 763         struct pathspec pathspec;
 764         struct string_list path_list = STRING_LIST_INIT_NODUP;
 765         int i;
 766         int dummy;
 767         int use_index = 1;
 768         enum {
 769                 pattern_type_unspecified = 0,
 770                 pattern_type_bre,
 771                 pattern_type_ere,
 772                 pattern_type_fixed,
 773                 pattern_type_pcre,
 774         };
 775         int pattern_type = pattern_type_unspecified;
 776
 777         struct option options[] = {
 778                 OPT_BOOLEAN(0, "cached", &cached,
 779                         "search in index instead of in the work tree"),
 780                 OPT_BOOLEAN(0, "index", &use_index,
 781                         "--no-index finds in contents not managed by git"),
 782                 OPT_GROUP(""),
 783                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 784                         "show non-matching lines"),
 785                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 786                         "case insensitive matching"),
 787                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 788                         "match patterns only at word boundaries"),
 789                 OPT_SET_INT('a', "text", &opt.binary,
 790                         "process binary files as text", GREP_BINARY_TEXT),
 791                 OPT_SET_INT('I', NULL, &opt.binary,
 792                         "don't match patterns in binary files",
 793                         GREP_BINARY_NOMATCH),
 794                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 795                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 796                         NULL, 1 },
 797                 OPT_GROUP(""),
 798                 OPT_SET_INT('E', "extended-regexp", &pattern_type,
 799                             "use extended POSIX regular expressions",
 800                             pattern_type_ere),
 801                 OPT_SET_INT('G', "basic-regexp", &pattern_type,
 802                             "use basic POSIX regular expressions (default)",
 803                             pattern_type_bre),
 804                 OPT_SET_INT('F', "fixed-strings", &pattern_type,
 805                             "interpret patterns as fixed strings",
 806                             pattern_type_fixed),
 807                 OPT_SET_INT('P', "perl-regexp", &pattern_type,
 808                             "use Perl-compatible regular expressions",
 809                             pattern_type_pcre),
 810                 OPT_GROUP(""),
 811                 OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
 812                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 813                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 814                 OPT_NEGBIT(0, "full-name", &opt.relative,
 815                         "show filenames relative to top directory", 1),
 816                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 817                         "show only filenames instead of matching lines"),
 818                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 819                         "synonym for --files-with-matches"),
 820                 OPT_BOOLEAN('L', "files-without-match",
 821                         &opt.unmatch_name_only,
 822                         "show only the names of files without match"),
 823                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 824                         "print NUL after filenames"),
 825                 OPT_BOOLEAN('c', "count", &opt.count,
 826                         "show the number of matches instead of matching lines"),
 827                 OPT__COLOR(&opt.color, "highlight matches"),
 828                 OPT_BOOLEAN(0, "break", &opt.file_break,
 829                         "print empty line between matches from different files"),
 830                 OPT_BOOLEAN(0, "heading", &opt.heading,
 831                         "show filename only once above matches from same file"),
 832                 OPT_GROUP(""),
 833                 OPT_CALLBACK('C', "context", &opt, "n",
 834                         "show <n> context lines before and after matches",
 835                         context_callback),
 836                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 837                         "show <n> context lines before matches"),
 838                 OPT_INTEGER('A', "after-context", &opt.post_context,
 839                         "show <n> context lines after matches"),
 840                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 841                         context_callback),
 842                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 843                         "show a line with the function name before matches"),
 844                 OPT_BOOLEAN('W', "function-context", &opt.funcbody,
 845                         "show the surrounding function"),
 846                 OPT_GROUP(""),
 847                 OPT_CALLBACK('f', NULL, &opt, "file",
 848                         "read patterns from file", file_callback),
 849                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 850                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 851                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 852                   "combine patterns specified with -e",
 853                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 854                 OPT_BOOLEAN(0, "or", &dummy, ""),
 855                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 856                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 857                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 858                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 859                   open_callback },
 860                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 861                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 862                   close_callback },
 863                 OPT__QUIET(&opt.status_only,
 864                            "indicate hit with exit status without output"),
 865                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 866                         "show only matches from files that match all patterns"),
 867                 OPT_GROUP(""),
 868                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 869                         "pager", "show matching files in the pager",
 870                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 871                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 872                             "allow calling of grep(1) (ignored by this build)"),
 873                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 874                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 875                 OPT_END()
 876         };
 877
 878         /*
 879          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 880          * to show usage information and exit.
 881          */
 882         if (argc == 2 && !strcmp(argv[1], "-h"))
 883                 usage_with_options(grep_usage, options);
 884
 885         memset(&opt, 0, sizeof(opt));
 886         opt.prefix = prefix;
 887         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 888         opt.relative = 1;
 889         opt.pathname = 1;
 890         opt.pattern_tail = &opt.pattern_list;
 891         opt.header_tail = &opt.header_list;
 892         opt.regflags = REG_NEWLINE;
 893         opt.max_depth = -1;
 894
 895         strcpy(opt.color_context, "");
 896         strcpy(opt.color_filename, "");
 897         strcpy(opt.color_function, "");
 898         strcpy(opt.color_lineno, "");
 899         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 900         strcpy(opt.color_selected, "");
 901         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 902         opt.color = -1;
 903         git_config(grep_config, &opt);
 904
 905         /*
 906          * If there is no -- then the paths must exist in the working
 907          * tree.  If there is no explicit pattern specified with -e or
 908          * -f, we take the first unrecognized non option to be the
 909          * pattern, but then what follows it must be zero or more
 910          * valid refs up to the -- (if exists), and then existing
 911          * paths.  If there is an explicit pattern, then the first
 912          * unrecognized non option is the beginning of the refs list
 913          * that continues up to the -- (if exists), and then paths.
 914          */
 915         argc = parse_options(argc, argv, prefix, options, grep_usage,
 916                              PARSE_OPT_KEEP_DASHDASH |
 917                              PARSE_OPT_STOP_AT_NON_OPTION |
 918                              PARSE_OPT_NO_INTERNAL_HELP);
 919         switch (pattern_type) {
 920         case pattern_type_fixed:
 921                 opt.fixed = 1;
 922                 opt.pcre = 0;
 923                 break;
 924         case pattern_type_bre:
 925                 opt.fixed = 0;
 926                 opt.pcre = 0;
 927                 opt.regflags &= ~REG_EXTENDED;
 928                 break;
 929         case pattern_type_ere:
 930                 opt.fixed = 0;
 931                 opt.pcre = 0;
 932                 opt.regflags |= REG_EXTENDED;
 933                 break;
 934         case pattern_type_pcre:
 935                 opt.fixed = 0;
 936                 opt.pcre = 1;
 937                 break;
 938         default:
 939                 break; /* nothing */
 940         }
 941
 942         if (use_index && !startup_info->have_repository)
 943                 /* die the same way as if we did it at the beginning */
 944                 setup_git_directory();
 945
 946         /*
 947          * skip a -- separator; we know it cannot be
 948          * separating revisions from pathnames if
 949          * we haven't even had any patterns yet
 950          */
 951         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 952                 argv++;
 953                 argc--;
 954         }
 955
 956         /* First unrecognized non-option token */
 957         if (argc > 0 && !opt.pattern_list) {
 958                 append_grep_pattern(&opt, argv[0], "command line", 0,
 959                                     GREP_PATTERN);
 960                 argv++;
 961                 argc--;
 962         }
 963
 964         if (show_in_pager == default_pager)
 965                 show_in_pager = git_pager(1);
 966         if (show_in_pager) {
 967                 opt.color = 0;
 968                 opt.name_only = 1;
 969                 opt.null_following_name = 1;
 970                 opt.output_priv = &path_list;
 971                 opt.output = append_path;
 972                 string_list_append(&path_list, show_in_pager);
 973                 use_threads = 0;
 974         }
 975
 976         if (!opt.pattern_list)
 977                 die(_("no pattern given."));
 978         if (!opt.fixed && opt.ignore_case)
 979                 opt.regflags |= REG_ICASE;
 980
 981 #ifndef NO_PTHREADS
 982         if (online_cpus() == 1 || !grep_threads_ok(&opt))
 983                 use_threads = 0;
 984
 985         if (use_threads) {
 986                 if (opt.pre_context || opt.post_context || opt.file_break ||
 987                     opt.funcbody)
 988                         skip_first_line = 1;
 989                 start_threads(&opt);
 990         }
 991 #else
 992         use_threads = 0;
 993 #endif
 994
 995         compile_grep_patterns(&opt);
 996
 997         /* Check revs and then paths */
 998         for (i = 0; i < argc; i++) {
 999                 const char *arg = argv[i];
1000                 unsigned char sha1[20];
1001                 /* Is it a rev? */
1002                 if (!get_sha1(arg, sha1)) {
1003                         struct object *object = parse_object(sha1);
1004                         if (!object)
1005                                 die(_("bad object %s"), arg);
1006                         add_object_array(object, arg, &list);
1007                         continue;
1008                 }
1009                 if (!strcmp(arg, "--")) {
1010                         i++;
1011                         seen_dashdash = 1;
1012                 }
1013                 break;
1014         }
1015
1016         /* The rest are paths */
1017         if (!seen_dashdash) {
1018                 int j;
1019                 for (j = i; j < argc; j++)
1020                         verify_filename(prefix, argv[j]);
1021         }
1022
1023         paths = get_pathspec(prefix, argv + i);
1024         init_pathspec(&pathspec, paths);
1025         pathspec.max_depth = opt.max_depth;
1026         pathspec.recursive = 1;
1027
1028         if (show_in_pager && (cached || list.nr))
1029                 die(_("--open-files-in-pager only works on the worktree"));
1030
1031         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1032                 const char *pager = path_list.items[0].string;
1033                 int len = strlen(pager);
1034
1035                 if (len > 4 && is_dir_sep(pager[len - 5]))
1036                         pager += len - 4;
1037
1038                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1039                         struct strbuf buf = STRBUF_INIT;
1040                         strbuf_addf(&buf, "+/%s%s",
1041                                         strcmp("less", pager) ? "" : "*",
1042                                         opt.pattern_list->pattern);
1043                         string_list_append(&path_list, buf.buf);
1044                         strbuf_detach(&buf, NULL);
1045                 }
1046         }
1047
1048         if (!show_in_pager)
1049                 setup_pager();
1050
1051
1052         if (!use_index) {
1053                 if (cached)
1054                         die(_("--cached cannot be used with --no-index."));
1055                 if (list.nr)
1056                         die(_("--no-index cannot be used with revs."));
1057                 hit = grep_directory(&opt, &pathspec);
1058         } else if (!list.nr) {
1059                 if (!cached)
1060                         setup_work_tree();
1061
1062                 hit = grep_cache(&opt, &pathspec, cached);
1063         } else {
1064                 if (cached)
1065                         die(_("both --cached and trees are given."));
1066                 hit = grep_objects(&opt, &pathspec, &list);
1067         }
1068
1069         if (use_threads)
1070                 hit |= wait_all();
1071         if (hit && show_in_pager)
1072                 run_pager(&opt, prefix);
1073         free_grep_patterns(&opt);
1074         return !hit;
1075 }