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