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 #include "attr.h"
  22
  23 static char const * const grep_usage[] = {
  24         "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
  25         NULL
  26 };
  27
  28 static int use_threads = 1;
  29
  30 #ifndef NO_PTHREADS
  31 #define THREADS 8
  32 static pthread_t threads[THREADS];
  33
  34 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  35                        const char *name);
  36 static void *load_file(const char *filename, size_t *sz);
  37
  38 enum work_type {WORK_SHA1, WORK_FILE};
  39
  40 /* We use one producer thread and THREADS consumer
  41  * threads. The producer adds struct work_items to 'todo' and the
  42  * consumers pick work items from the same array.
  43  */
  44 struct work_item {
  45         enum work_type type;
  46         char *name;
  47
  48         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  49          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  50          * terminated filename.
  51          */
  52         void *identifier;
  53         char done;
  54         struct strbuf out;
  55 };
  56
  57 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  58  * that have been or are processed by a consumer thread. We haven't
  59  * written the result for these to stdout yet.
  60  *
  61  * The work_items in [todo_start, todo_end) are waiting to be picked
  62  * up by a consumer thread.
  63  *
  64  * The ranges are modulo TODO_SIZE.
  65  */
  66 #define TODO_SIZE 128
  67 static struct work_item todo[TODO_SIZE];
  68 static int todo_start;
  69 static int todo_end;
  70 static int todo_done;
  71
  72 /* Has all work items been added? */
  73 static int all_work_added;
  74
  75 /* This lock protects all the variables above. */
  76 static pthread_mutex_t grep_mutex;
  77
  78 /* Used to serialize calls to read_sha1_file. */
  79 static pthread_mutex_t read_sha1_mutex;
  80
  81 #define grep_lock() pthread_mutex_lock(&grep_mutex)
  82 #define grep_unlock() pthread_mutex_unlock(&grep_mutex)
  83 #define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
  84 #define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
  85
  86 /* Signalled when a new work_item is added to todo. */
  87 static pthread_cond_t cond_add;
  88
  89 /* Signalled when the result from one work_item is written to
  90  * stdout.
  91  */
  92 static pthread_cond_t cond_write;
  93
  94 /* Signalled when we are finished with everything. */
  95 static pthread_cond_t cond_result;
  96
  97 static int skip_first_line;
  98
  99 static void add_work(enum work_type type, char *name, void *id)
 100 {
 101         grep_lock();
 102
 103         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 104                 pthread_cond_wait(&cond_write, &grep_mutex);
 105         }
 106
 107         todo[todo_end].type = type;
 108         todo[todo_end].name = name;
 109         todo[todo_end].identifier = id;
 110         todo[todo_end].done = 0;
 111         strbuf_reset(&todo[todo_end].out);
 112         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 113
 114         pthread_cond_signal(&cond_add);
 115         grep_unlock();
 116 }
 117
 118 static struct work_item *get_work(void)
 119 {
 120         struct work_item *ret;
 121
 122         grep_lock();
 123         while (todo_start == todo_end && !all_work_added) {
 124                 pthread_cond_wait(&cond_add, &grep_mutex);
 125         }
 126
 127         if (todo_start == todo_end && all_work_added) {
 128                 ret = NULL;
 129         } else {
 130                 ret = &todo[todo_start];
 131                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 132         }
 133         grep_unlock();
 134         return ret;
 135 }
 136
 137 static void grep_sha1_async(struct grep_opt *opt, char *name,
 138                             const unsigned char *sha1)
 139 {
 140         unsigned char *s;
 141         s = xmalloc(20);
 142         memcpy(s, sha1, 20);
 143         add_work(WORK_SHA1, name, s);
 144 }
 145
 146 static void grep_file_async(struct grep_opt *opt, char *name,
 147                             const char *filename)
 148 {
 149         add_work(WORK_FILE, name, xstrdup(filename));
 150 }
 151
 152 static void work_done(struct work_item *w)
 153 {
 154         int old_done;
 155
 156         grep_lock();
 157         w->done = 1;
 158         old_done = todo_done;
 159         for(; todo[todo_done].done && todo_done != todo_start;
 160             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 161                 w = &todo[todo_done];
 162                 if (w->out.len) {
 163                         const char *p = w->out.buf;
 164                         size_t len = w->out.len;
 165
 166                         /* Skip the leading hunk mark of the first file. */
 167                         if (skip_first_line) {
 168                                 while (len) {
 169                                         len--;
 170                                         if (*p++ == '\n')
 171                                                 break;
 172                                 }
 173                                 skip_first_line = 0;
 174                         }
 175
 176                         write_or_die(1, p, len);
 177                 }
 178                 free(w->name);
 179                 free(w->identifier);
 180         }
 181
 182         if (old_done != todo_done)
 183                 pthread_cond_signal(&cond_write);
 184
 185         if (all_work_added && todo_done == todo_end)
 186                 pthread_cond_signal(&cond_result);
 187
 188         grep_unlock();
 189 }
 190
 191 static int skip_binary(struct grep_opt *opt, const char *filename)
 192 {
 193         if ((opt->binary & GREP_BINARY_NOMATCH)) {
 194                 static struct git_attr *attr_text;
 195                 struct git_attr_check check;
 196
 197                 if (!attr_text)
 198                         attr_text = git_attr("text");
 199                 memset(&check, 0, sizeof(check));
 200                 check.attr = attr_text;
 201                 return !git_checkattr(filename, 1, &check) &&
 202                                 ATTR_FALSE(check.value);
 203         }
 204         return 0;
 205 }
 206
 207 static void *run(void *arg)
 208 {
 209         int hit = 0;
 210         struct grep_opt *opt = arg;
 211
 212         while (1) {
 213                 struct work_item *w = get_work();
 214                 if (!w)
 215                         break;
 216
 217                 if (skip_binary(opt, (const char *)w->identifier))
 218                         continue;
 219
 220                 opt->output_priv = w;
 221                 if (w->type == WORK_SHA1) {
 222                         unsigned long sz;
 223                         void* data = load_sha1(w->identifier, &sz, w->name);
 224
 225                         if (data) {
 226                                 hit |= grep_buffer(opt, w->name, data, sz);
 227                                 free(data);
 228                         }
 229                 } else if (w->type == WORK_FILE) {
 230                         size_t sz;
 231                         void* data = load_file(w->identifier, &sz);
 232                         if (data) {
 233                                 hit |= grep_buffer(opt, w->name, data, sz);
 234                                 free(data);
 235                         }
 236                 } else {
 237                         assert(0);
 238                 }
 239
 240                 work_done(w);
 241         }
 242         free_grep_patterns(arg);
 243         free(arg);
 244
 245         return (void*) (intptr_t) hit;
 246 }
 247
 248 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 249 {
 250         struct work_item *w = opt->output_priv;
 251         strbuf_add(&w->out, buf, size);
 252 }
 253
 254 static void start_threads(struct grep_opt *opt)
 255 {
 256         int i;
 257
 258         pthread_mutex_init(&grep_mutex, NULL);
 259         pthread_mutex_init(&read_sha1_mutex, NULL);
 260         pthread_cond_init(&cond_add, NULL);
 261         pthread_cond_init(&cond_write, NULL);
 262         pthread_cond_init(&cond_result, NULL);
 263
 264         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 265                 strbuf_init(&todo[i].out, 0);
 266         }
 267
 268         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 269                 int err;
 270                 struct grep_opt *o = grep_opt_dup(opt);
 271                 o->output = strbuf_out;
 272                 compile_grep_patterns(o);
 273                 err = pthread_create(&threads[i], NULL, run, o);
 274
 275                 if (err)
 276                         die(_("grep: failed to create thread: %s"),
 277                             strerror(err));
 278         }
 279 }
 280
 281 static int wait_all(void)
 282 {
 283         int hit = 0;
 284         int i;
 285
 286         grep_lock();
 287         all_work_added = 1;
 288
 289         /* Wait until all work is done. */
 290         while (todo_done != todo_end)
 291                 pthread_cond_wait(&cond_result, &grep_mutex);
 292
 293         /* Wake up all the consumer threads so they can see that there
 294          * is no more work to do.
 295          */
 296         pthread_cond_broadcast(&cond_add);
 297         grep_unlock();
 298
 299         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 300                 void *h;
 301                 pthread_join(threads[i], &h);
 302                 hit |= (int) (intptr_t) h;
 303         }
 304
 305         pthread_mutex_destroy(&grep_mutex);
 306         pthread_mutex_destroy(&read_sha1_mutex);
 307         pthread_cond_destroy(&cond_add);
 308         pthread_cond_destroy(&cond_write);
 309         pthread_cond_destroy(&cond_result);
 310
 311         return hit;
 312 }
 313 #else /* !NO_PTHREADS */
 314 #define read_sha1_lock()
 315 #define read_sha1_unlock()
 316
 317 static int wait_all(void)
 318 {
 319         return 0;
 320 }
 321 #endif
 322
 323 static int grep_config(const char *var, const char *value, void *cb)
 324 {
 325         struct grep_opt *opt = cb;
 326         char *color = NULL;
 327
 328         switch (userdiff_config(var, value)) {
 329         case 0: break;
 330         case -1: return -1;
 331         default: return 0;
 332         }
 333
 334         if (!strcmp(var, "grep.extendedregexp")) {
 335                 if (git_config_bool(var, value))
 336                         opt->regflags |= REG_EXTENDED;
 337                 else
 338                         opt->regflags &= ~REG_EXTENDED;
 339                 return 0;
 340         }
 341
 342         if (!strcmp(var, "grep.linenumber")) {
 343                 opt->linenum = git_config_bool(var, value);
 344                 return 0;
 345         }
 346
 347         if (!strcmp(var, "color.grep"))
 348                 opt->color = git_config_colorbool(var, value);
 349         else if (!strcmp(var, "color.grep.context"))
 350                 color = opt->color_context;
 351         else if (!strcmp(var, "color.grep.filename"))
 352                 color = opt->color_filename;
 353         else if (!strcmp(var, "color.grep.function"))
 354                 color = opt->color_function;
 355         else if (!strcmp(var, "color.grep.linenumber"))
 356                 color = opt->color_lineno;
 357         else if (!strcmp(var, "color.grep.match"))
 358                 color = opt->color_match;
 359         else if (!strcmp(var, "color.grep.selected"))
 360                 color = opt->color_selected;
 361         else if (!strcmp(var, "color.grep.separator"))
 362                 color = opt->color_sep;
 363         else
 364                 return git_color_default_config(var, value, cb);
 365         if (color) {
 366                 if (!value)
 367                         return config_error_nonbool(var);
 368                 color_parse(value, var, color);
 369         }
 370         return 0;
 371 }
 372
 373 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 374 {
 375         void *data;
 376
 377         if (use_threads) {
 378                 read_sha1_lock();
 379                 data = read_sha1_file(sha1, type, size);
 380                 read_sha1_unlock();
 381         } else {
 382                 data = read_sha1_file(sha1, type, size);
 383         }
 384         return data;
 385 }
 386
 387 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 388                        const char *name)
 389 {
 390         enum object_type type;
 391         void *data = lock_and_read_sha1_file(sha1, &type, size);
 392
 393         if (!data)
 394                 error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
 395
 396         return data;
 397 }
 398
 399 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 400                      const char *filename, int tree_name_len)
 401 {
 402         struct strbuf pathbuf = STRBUF_INIT;
 403         char *name;
 404
 405         if (opt->relative && opt->prefix_length) {
 406                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 407                                     opt->prefix);
 408                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 409         } else {
 410                 strbuf_addstr(&pathbuf, filename);
 411         }
 412
 413         name = strbuf_detach(&pathbuf, NULL);
 414
 415 #ifndef NO_PTHREADS
 416         if (use_threads) {
 417                 grep_sha1_async(opt, name, sha1);
 418                 return 0;
 419         } else
 420 #endif
 421         {
 422                 int hit;
 423                 unsigned long sz;
 424                 void *data = load_sha1(sha1, &sz, name);
 425                 if (!data)
 426                         hit = 0;
 427                 else
 428                         hit = grep_buffer(opt, name, data, sz);
 429
 430                 free(data);
 431                 free(name);
 432                 return hit;
 433         }
 434 }
 435
 436 static void *load_file(const char *filename, size_t *sz)
 437 {
 438         struct stat st;
 439         char *data;
 440         int i;
 441
 442         if (lstat(filename, &st) < 0) {
 443         err_ret:
 444                 if (errno != ENOENT)
 445                         error(_("'%s': %s"), filename, strerror(errno));
 446                 return NULL;
 447         }
 448         if (!S_ISREG(st.st_mode))
 449                 return NULL;
 450         *sz = xsize_t(st.st_size);
 451         i = open(filename, O_RDONLY);
 452         if (i < 0)
 453                 goto err_ret;
 454         data = xmalloc(*sz + 1);
 455         if (st.st_size != read_in_full(i, data, *sz)) {
 456                 error(_("'%s': short read %s"), filename, strerror(errno));
 457                 close(i);
 458                 free(data);
 459                 return NULL;
 460         }
 461         close(i);
 462         data[*sz] = 0;
 463         return data;
 464 }
 465
 466 static int grep_file(struct grep_opt *opt, const char *filename)
 467 {
 468         struct strbuf buf = STRBUF_INIT;
 469         char *name;
 470
 471         if (opt->relative && opt->prefix_length)
 472                 quote_path_relative(filename, -1, &buf, opt->prefix);
 473         else
 474                 strbuf_addstr(&buf, filename);
 475         name = strbuf_detach(&buf, NULL);
 476
 477 #ifndef NO_PTHREADS
 478         if (use_threads) {
 479                 grep_file_async(opt, name, filename);
 480                 return 0;
 481         } else
 482 #endif
 483         {
 484                 int hit;
 485                 size_t sz;
 486                 void *data = load_file(filename, &sz);
 487                 if (!data)
 488                         hit = 0;
 489                 else
 490                         hit = grep_buffer(opt, name, data, sz);
 491
 492                 free(data);
 493                 free(name);
 494                 return hit;
 495         }
 496 }
 497
 498 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 499 {
 500         struct string_list *path_list = opt->output_priv;
 501
 502         if (len == 1 && *(const char *)data == '\0')
 503                 return;
 504         string_list_append(path_list, xstrndup(data, len));
 505 }
 506
 507 static void run_pager(struct grep_opt *opt, const char *prefix)
 508 {
 509         struct string_list *path_list = opt->output_priv;
 510         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 511         int i, status;
 512
 513         for (i = 0; i < path_list->nr; i++)
 514                 argv[i] = path_list->items[i].string;
 515         argv[path_list->nr] = NULL;
 516
 517         if (prefix && chdir(prefix))
 518                 die(_("Failed to chdir: %s"), prefix);
 519         status = run_command_v_opt(argv, RUN_USING_SHELL);
 520         if (status)
 521                 exit(status);
 522         free(argv);
 523 }
 524
 525 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 526 {
 527         int hit = 0;
 528         int nr;
 529         read_cache();
 530
 531         for (nr = 0; nr < active_nr; nr++) {
 532                 struct cache_entry *ce = active_cache[nr];
 533                 if (!S_ISREG(ce->ce_mode))
 534                         continue;
 535                 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 536                         continue;
 537                 if (skip_binary(opt, ce->name))
 538                         continue;
 539
 540                 /*
 541                  * If CE_VALID is on, we assume worktree file and its cache entry
 542                  * are identical, even if worktree file has been modified, so use
 543                  * cache version instead
 544                  */
 545                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 546                         if (ce_stage(ce))
 547                                 continue;
 548                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 549                 }
 550                 else
 551                         hit |= grep_file(opt, ce->name);
 552                 if (ce_stage(ce)) {
 553                         do {
 554                                 nr++;
 555                         } while (nr < active_nr &&
 556                                  !strcmp(ce->name, active_cache[nr]->name));
 557                         nr--; /* compensate for loop control */
 558                 }
 559                 if (hit && opt->status_only)
 560                         break;
 561         }
 562         return hit;
 563 }
 564
 565 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 566                      struct tree_desc *tree, struct strbuf *base, int tn_len)
 567 {
 568         int hit = 0, match = 0;
 569         struct name_entry entry;
 570         int old_baselen = base->len;
 571
 572         while (tree_entry(tree, &entry)) {
 573                 int te_len = tree_entry_len(entry.path, entry.sha1);
 574
 575                 if (match != 2) {
 576                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 577                         if (match < 0)
 578                                 break;
 579                         if (match == 0)
 580                                 continue;
 581                 }
 582
 583                 strbuf_add(base, entry.path, te_len);
 584
 585                 if (S_ISREG(entry.mode)) {
 586                         hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 587                 }
 588                 else if (S_ISDIR(entry.mode)) {
 589                         enum object_type type;
 590                         struct tree_desc sub;
 591                         void *data;
 592                         unsigned long size;
 593
 594                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 595                         if (!data)
 596                                 die(_("unable to read tree (%s)"),
 597                                     sha1_to_hex(entry.sha1));
 598
 599                         strbuf_addch(base, '/');
 600                         init_tree_desc(&sub, data, size);
 601                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 602                         free(data);
 603                 }
 604                 strbuf_setlen(base, old_baselen);
 605
 606                 if (hit && opt->status_only)
 607                         break;
 608         }
 609         return hit;
 610 }
 611
 612 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 613                        struct object *obj, const char *name)
 614 {
 615         if (obj->type == OBJ_BLOB)
 616                 return grep_sha1(opt, obj->sha1, name, 0);
 617         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 618                 struct tree_desc tree;
 619                 void *data;
 620                 unsigned long size;
 621                 struct strbuf base;
 622                 int hit, len;
 623
 624                 read_sha1_lock();
 625                 data = read_object_with_reference(obj->sha1, tree_type,
 626                                                   &size, NULL);
 627                 read_sha1_unlock();
 628
 629                 if (!data)
 630                         die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 631
 632                 len = name ? strlen(name) : 0;
 633                 strbuf_init(&base, PATH_MAX + len + 1);
 634                 if (len) {
 635                         strbuf_add(&base, name, len);
 636                         strbuf_addch(&base, ':');
 637                 }
 638                 init_tree_desc(&tree, data, size);
 639                 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 640                 strbuf_release(&base);
 641                 free(data);
 642                 return hit;
 643         }
 644         die(_("unable to grep from object of type %s"), typename(obj->type));
 645 }
 646
 647 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 648                         const struct object_array *list)
 649 {
 650         unsigned int i;
 651         int hit = 0;
 652         const unsigned int nr = list->nr;
 653
 654         for (i = 0; i < nr; i++) {
 655                 struct object *real_obj;
 656                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 657                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 658                         hit = 1;
 659                         if (opt->status_only)
 660                                 break;
 661                 }
 662         }
 663         return hit;
 664 }
 665
 666 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 667                           int exc_std)
 668 {
 669         struct dir_struct dir;
 670         int i, hit = 0;
 671
 672         memset(&dir, 0, sizeof(dir));
 673         if (exc_std)
 674                 setup_standard_excludes(&dir);
 675
 676         fill_directory(&dir, pathspec->raw);
 677         for (i = 0; i < dir.nr; i++) {
 678                 const char *name = dir.entries[i]->name;
 679                 int namelen = strlen(name);
 680                 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 681                         continue;
 682                 hit |= grep_file(opt, dir.entries[i]->name);
 683                 if (hit && opt->status_only)
 684                         break;
 685         }
 686         return hit;
 687 }
 688
 689 static int context_callback(const struct option *opt, const char *arg,
 690                             int unset)
 691 {
 692         struct grep_opt *grep_opt = opt->value;
 693         int value;
 694         const char *endp;
 695
 696         if (unset) {
 697                 grep_opt->pre_context = grep_opt->post_context = 0;
 698                 return 0;
 699         }
 700         value = strtol(arg, (char **)&endp, 10);
 701         if (*endp) {
 702                 return error(_("switch `%c' expects a numerical value"),
 703                              opt->short_name);
 704         }
 705         grep_opt->pre_context = grep_opt->post_context = value;
 706         return 0;
 707 }
 708
 709 static int file_callback(const struct option *opt, const char *arg, int unset)
 710 {
 711         struct grep_opt *grep_opt = opt->value;
 712         int from_stdin = !strcmp(arg, "-");
 713         FILE *patterns;
 714         int lno = 0;
 715         struct strbuf sb = STRBUF_INIT;
 716
 717         patterns = from_stdin ? stdin : fopen(arg, "r");
 718         if (!patterns)
 719                 die_errno(_("cannot open '%s'"), arg);
 720         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 721                 char *s;
 722                 size_t len;
 723
 724                 /* ignore empty line like grep does */
 725                 if (sb.len == 0)
 726                         continue;
 727
 728                 s = strbuf_detach(&sb, &len);
 729                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 730         }
 731         if (!from_stdin)
 732                 fclose(patterns);
 733         strbuf_release(&sb);
 734         return 0;
 735 }
 736
 737 static int not_callback(const struct option *opt, const char *arg, int unset)
 738 {
 739         struct grep_opt *grep_opt = opt->value;
 740         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 741         return 0;
 742 }
 743
 744 static int and_callback(const struct option *opt, const char *arg, int unset)
 745 {
 746         struct grep_opt *grep_opt = opt->value;
 747         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 748         return 0;
 749 }
 750
 751 static int open_callback(const struct option *opt, const char *arg, int unset)
 752 {
 753         struct grep_opt *grep_opt = opt->value;
 754         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 755         return 0;
 756 }
 757
 758 static int close_callback(const struct option *opt, const char *arg, int unset)
 759 {
 760         struct grep_opt *grep_opt = opt->value;
 761         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 762         return 0;
 763 }
 764
 765 static int pattern_callback(const struct option *opt, const char *arg,
 766                             int unset)
 767 {
 768         struct grep_opt *grep_opt = opt->value;
 769         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 770         return 0;
 771 }
 772
 773 static int help_callback(const struct option *opt, const char *arg, int unset)
 774 {
 775         return -1;
 776 }
 777
 778 int cmd_grep(int argc, const char **argv, const char *prefix)
 779 {
 780         int hit = 0;
 781         int cached = 0, untracked = 0, opt_exclude = -1;
 782         int seen_dashdash = 0;
 783         int external_grep_allowed__ignored;
 784         const char *show_in_pager = NULL, *default_pager = "dummy";
 785         struct grep_opt opt;
 786         struct object_array list = OBJECT_ARRAY_INIT;
 787         const char **paths = NULL;
 788         struct pathspec pathspec;
 789         struct string_list path_list = STRING_LIST_INIT_NODUP;
 790         int i;
 791         int dummy;
 792         int use_index = 1;
 793         enum {
 794                 pattern_type_unspecified = 0,
 795                 pattern_type_bre,
 796                 pattern_type_ere,
 797                 pattern_type_fixed,
 798                 pattern_type_pcre,
 799         };
 800         int pattern_type = pattern_type_unspecified;
 801
 802         struct option options[] = {
 803                 OPT_BOOLEAN(0, "cached", &cached,
 804                         "search in index instead of in the work tree"),
 805                 { OPTION_BOOLEAN, 0, "index", &use_index, NULL,
 806                         "finds in contents not managed by git",
 807                         PARSE_OPT_NOARG | PARSE_OPT_NEGHELP },
 808                 OPT_BOOLEAN(0, "untracked", &untracked,
 809                         "search in both tracked and untracked files"),
 810                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 811                             "search also in ignored files", 1),
 812                 OPT_GROUP(""),
 813                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 814                         "show non-matching lines"),
 815                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 816                         "case insensitive matching"),
 817                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 818                         "match patterns only at word boundaries"),
 819                 OPT_SET_INT('a', "text", &opt.binary,
 820                         "process binary files as text", GREP_BINARY_TEXT),
 821                 OPT_SET_INT('I', NULL, &opt.binary,
 822                         "don't match patterns in binary files",
 823                         GREP_BINARY_NOMATCH),
 824                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 825                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 826                         NULL, 1 },
 827                 OPT_GROUP(""),
 828                 OPT_SET_INT('E', "extended-regexp", &pattern_type,
 829                             "use extended POSIX regular expressions",
 830                             pattern_type_ere),
 831                 OPT_SET_INT('G', "basic-regexp", &pattern_type,
 832                             "use basic POSIX regular expressions (default)",
 833                             pattern_type_bre),
 834                 OPT_SET_INT('F', "fixed-strings", &pattern_type,
 835                             "interpret patterns as fixed strings",
 836                             pattern_type_fixed),
 837                 OPT_SET_INT('P', "perl-regexp", &pattern_type,
 838                             "use Perl-compatible regular expressions",
 839                             pattern_type_pcre),
 840                 OPT_GROUP(""),
 841                 OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
 842                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 843                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 844                 OPT_NEGBIT(0, "full-name", &opt.relative,
 845                         "show filenames relative to top directory", 1),
 846                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 847                         "show only filenames instead of matching lines"),
 848                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 849                         "synonym for --files-with-matches"),
 850                 OPT_BOOLEAN('L', "files-without-match",
 851                         &opt.unmatch_name_only,
 852                         "show only the names of files without match"),
 853                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 854                         "print NUL after filenames"),
 855                 OPT_BOOLEAN('c', "count", &opt.count,
 856                         "show the number of matches instead of matching lines"),
 857                 OPT__COLOR(&opt.color, "highlight matches"),
 858                 OPT_BOOLEAN(0, "break", &opt.file_break,
 859                         "print empty line between matches from different files"),
 860                 OPT_BOOLEAN(0, "heading", &opt.heading,
 861                         "show filename only once above matches from same file"),
 862                 OPT_GROUP(""),
 863                 OPT_CALLBACK('C', "context", &opt, "n",
 864                         "show <n> context lines before and after matches",
 865                         context_callback),
 866                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 867                         "show <n> context lines before matches"),
 868                 OPT_INTEGER('A', "after-context", &opt.post_context,
 869                         "show <n> context lines after matches"),
 870                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 871                         context_callback),
 872                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 873                         "show a line with the function name before matches"),
 874                 OPT_BOOLEAN('W', "function-context", &opt.funcbody,
 875                         "show the surrounding function"),
 876                 OPT_GROUP(""),
 877                 OPT_CALLBACK('f', NULL, &opt, "file",
 878                         "read patterns from file", file_callback),
 879                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 880                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 881                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 882                   "combine patterns specified with -e",
 883                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 884                 OPT_BOOLEAN(0, "or", &dummy, ""),
 885                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 886                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 887                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 888                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 889                   open_callback },
 890                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 891                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 892                   close_callback },
 893                 OPT__QUIET(&opt.status_only,
 894                            "indicate hit with exit status without output"),
 895                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 896                         "show only matches from files that match all patterns"),
 897                 OPT_GROUP(""),
 898                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 899                         "pager", "show matching files in the pager",
 900                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 901                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 902                             "allow calling of grep(1) (ignored by this build)"),
 903                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 904                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 905                 OPT_END()
 906         };
 907
 908         /*
 909          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 910          * to show usage information and exit.
 911          */
 912         if (argc == 2 && !strcmp(argv[1], "-h"))
 913                 usage_with_options(grep_usage, options);
 914
 915         memset(&opt, 0, sizeof(opt));
 916         opt.prefix = prefix;
 917         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 918         opt.relative = 1;
 919         opt.pathname = 1;
 920         opt.pattern_tail = &opt.pattern_list;
 921         opt.header_tail = &opt.header_list;
 922         opt.regflags = REG_NEWLINE;
 923         opt.max_depth = -1;
 924
 925         strcpy(opt.color_context, "");
 926         strcpy(opt.color_filename, "");
 927         strcpy(opt.color_function, "");
 928         strcpy(opt.color_lineno, "");
 929         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 930         strcpy(opt.color_selected, "");
 931         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 932         opt.color = -1;
 933         git_config(grep_config, &opt);
 934
 935         /*
 936          * If there is no -- then the paths must exist in the working
 937          * tree.  If there is no explicit pattern specified with -e or
 938          * -f, we take the first unrecognized non option to be the
 939          * pattern, but then what follows it must be zero or more
 940          * valid refs up to the -- (if exists), and then existing
 941          * paths.  If there is an explicit pattern, then the first
 942          * unrecognized non option is the beginning of the refs list
 943          * that continues up to the -- (if exists), and then paths.
 944          */
 945         argc = parse_options(argc, argv, prefix, options, grep_usage,
 946                              PARSE_OPT_KEEP_DASHDASH |
 947                              PARSE_OPT_STOP_AT_NON_OPTION |
 948                              PARSE_OPT_NO_INTERNAL_HELP);
 949         switch (pattern_type) {
 950         case pattern_type_fixed:
 951                 opt.fixed = 1;
 952                 opt.pcre = 0;
 953                 break;
 954         case pattern_type_bre:
 955                 opt.fixed = 0;
 956                 opt.pcre = 0;
 957                 opt.regflags &= ~REG_EXTENDED;
 958                 break;
 959         case pattern_type_ere:
 960                 opt.fixed = 0;
 961                 opt.pcre = 0;
 962                 opt.regflags |= REG_EXTENDED;
 963                 break;
 964         case pattern_type_pcre:
 965                 opt.fixed = 0;
 966                 opt.pcre = 1;
 967                 break;
 968         default:
 969                 break; /* nothing */
 970         }
 971
 972         if (use_index && !startup_info->have_repository)
 973                 /* die the same way as if we did it at the beginning */
 974                 setup_git_directory();
 975
 976         /*
 977          * skip a -- separator; we know it cannot be
 978          * separating revisions from pathnames if
 979          * we haven't even had any patterns yet
 980          */
 981         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 982                 argv++;
 983                 argc--;
 984         }
 985
 986         /* First unrecognized non-option token */
 987         if (argc > 0 && !opt.pattern_list) {
 988                 append_grep_pattern(&opt, argv[0], "command line", 0,
 989                                     GREP_PATTERN);
 990                 argv++;
 991                 argc--;
 992         }
 993
 994         if (show_in_pager == default_pager)
 995                 show_in_pager = git_pager(1);
 996         if (show_in_pager) {
 997                 opt.color = 0;
 998                 opt.name_only = 1;
 999                 opt.null_following_name = 1;
1000                 opt.output_priv = &path_list;
1001                 opt.output = append_path;
1002                 string_list_append(&path_list, show_in_pager);
1003                 use_threads = 0;
1004         }
1005         if ((opt.binary & GREP_BINARY_NOMATCH))
1006                 use_threads = 0;
1007
1008         if (!opt.pattern_list)
1009                 die(_("no pattern given."));
1010         if (!opt.fixed && opt.ignore_case)
1011                 opt.regflags |= REG_ICASE;
1012
1013 #ifndef NO_PTHREADS
1014         if (online_cpus() == 1 || !grep_threads_ok(&opt))
1015                 use_threads = 0;
1016
1017         if (use_threads) {
1018                 if (opt.pre_context || opt.post_context || opt.file_break ||
1019                     opt.funcbody)
1020                         skip_first_line = 1;
1021                 start_threads(&opt);
1022         }
1023 #else
1024         use_threads = 0;
1025 #endif
1026
1027         compile_grep_patterns(&opt);
1028
1029         /* Check revs and then paths */
1030         for (i = 0; i < argc; i++) {
1031                 const char *arg = argv[i];
1032                 unsigned char sha1[20];
1033                 /* Is it a rev? */
1034                 if (!get_sha1(arg, sha1)) {
1035                         struct object *object = parse_object(sha1);
1036                         if (!object)
1037                                 die(_("bad object %s"), arg);
1038                         add_object_array(object, arg, &list);
1039                         continue;
1040                 }
1041                 if (!strcmp(arg, "--")) {
1042                         i++;
1043                         seen_dashdash = 1;
1044                 }
1045                 break;
1046         }
1047
1048         /* The rest are paths */
1049         if (!seen_dashdash) {
1050                 int j;
1051                 for (j = i; j < argc; j++)
1052                         verify_filename(prefix, argv[j]);
1053         }
1054
1055         paths = get_pathspec(prefix, argv + i);
1056         init_pathspec(&pathspec, paths);
1057         pathspec.max_depth = opt.max_depth;
1058         pathspec.recursive = 1;
1059
1060         if (show_in_pager && (cached || list.nr))
1061                 die(_("--open-files-in-pager only works on the worktree"));
1062
1063         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1064                 const char *pager = path_list.items[0].string;
1065                 int len = strlen(pager);
1066
1067                 if (len > 4 && is_dir_sep(pager[len - 5]))
1068                         pager += len - 4;
1069
1070                 if (opt.ignore_case && !strcmp("less", pager))
1071                         string_list_append(&path_list, "-i");
1072
1073                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1074                         struct strbuf buf = STRBUF_INIT;
1075                         strbuf_addf(&buf, "+/%s%s",
1076                                         strcmp("less", pager) ? "" : "*",
1077                                         opt.pattern_list->pattern);
1078                         string_list_append(&path_list, buf.buf);
1079                         strbuf_detach(&buf, NULL);
1080                 }
1081         }
1082
1083         if (!show_in_pager)
1084                 setup_pager();
1085
1086         if (!use_index && (untracked || cached))
1087                 die(_("--cached or --untracked cannot be used with --no-index."));
1088
1089         if (!use_index || untracked) {
1090                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1091                 if (list.nr)
1092                         die(_("--no-index or --untracked cannot be used with revs."));
1093                 hit = grep_directory(&opt, &pathspec, use_exclude);
1094         } else if (0 <= opt_exclude) {
1095                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1096         } else if (!list.nr) {
1097                 if (!cached)
1098                         setup_work_tree();
1099
1100                 hit = grep_cache(&opt, &pathspec, cached);
1101         } else {
1102                 if (cached)
1103                         die(_("both --cached and trees are given."));
1104                 hit = grep_objects(&opt, &pathspec, &list);
1105         }
1106
1107         if (use_threads)
1108                 hit |= wait_all();
1109         if (hit && show_in_pager)
1110                 run_pager(&opt, prefix);
1111         free_grep_patterns(&opt);
1112         return !hit;
1113 }