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 "userdiff.h"
  15 #include "grep.h"
  16 #include "quote.h"
  17 #include "dir.h"
  18 #include "string-list.h"
  19
  20 #ifndef NO_PTHREADS
  21 #include <pthread.h>
  22 #include "thread-utils.h"
  23 #endif
  24
  25 static char const * const grep_usage[] = {
  26         "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
  27         NULL
  28 };
  29
  30 static int use_threads = 1;
  31
  32 #ifndef NO_PTHREADS
  33 #define THREADS 8
  34 static pthread_t threads[THREADS];
  35
  36 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  37                        const char *name);
  38 static void *load_file(const char *filename, size_t *sz);
  39
  40 enum work_type {WORK_SHA1, WORK_FILE};
  41
  42 /* We use one producer thread and THREADS consumer
  43  * threads. The producer adds struct work_items to 'todo' and the
  44  * consumers pick work items from the same array.
  45  */
  46 struct work_item
  47 {
  48         enum work_type type;
  49         char *name;
  50
  51         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  52          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  53          * terminated filename.
  54          */
  55         void *identifier;
  56         char done;
  57         struct strbuf out;
  58 };
  59
  60 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  61  * that have been or are processed by a consumer thread. We haven't
  62  * written the result for these to stdout yet.
  63  *
  64  * The work_items in [todo_start, todo_end) are waiting to be picked
  65  * up by a consumer thread.
  66  *
  67  * The ranges are modulo TODO_SIZE.
  68  */
  69 #define TODO_SIZE 128
  70 static struct work_item todo[TODO_SIZE];
  71 static int todo_start;
  72 static int todo_end;
  73 static int todo_done;
  74
  75 /* Has all work items been added? */
  76 static int all_work_added;
  77
  78 /* This lock protects all the variables above. */
  79 static pthread_mutex_t grep_mutex;
  80
  81 /* Used to serialize calls to read_sha1_file. */
  82 static pthread_mutex_t read_sha1_mutex;
  83
  84 #define grep_lock() pthread_mutex_lock(&grep_mutex)
  85 #define grep_unlock() pthread_mutex_unlock(&grep_mutex)
  86 #define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
  87 #define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
  88
  89 /* Signalled when a new work_item is added to todo. */
  90 static pthread_cond_t cond_add;
  91
  92 /* Signalled when the result from one work_item is written to
  93  * stdout.
  94  */
  95 static pthread_cond_t cond_write;
  96
  97 /* Signalled when we are finished with everything. */
  98 static pthread_cond_t cond_result;
  99
 100 static int print_hunk_marks_between_files;
 101 static int printed_something;
 102
 103 static void add_work(enum work_type type, char *name, void *id)
 104 {
 105         grep_lock();
 106
 107         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 108                 pthread_cond_wait(&cond_write, &grep_mutex);
 109         }
 110
 111         todo[todo_end].type = type;
 112         todo[todo_end].name = name;
 113         todo[todo_end].identifier = id;
 114         todo[todo_end].done = 0;
 115         strbuf_reset(&todo[todo_end].out);
 116         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 117
 118         pthread_cond_signal(&cond_add);
 119         grep_unlock();
 120 }
 121
 122 static struct work_item *get_work(void)
 123 {
 124         struct work_item *ret;
 125
 126         grep_lock();
 127         while (todo_start == todo_end && !all_work_added) {
 128                 pthread_cond_wait(&cond_add, &grep_mutex);
 129         }
 130
 131         if (todo_start == todo_end && all_work_added) {
 132                 ret = NULL;
 133         } else {
 134                 ret = &todo[todo_start];
 135                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 136         }
 137         grep_unlock();
 138         return ret;
 139 }
 140
 141 static void grep_sha1_async(struct grep_opt *opt, char *name,
 142                             const unsigned char *sha1)
 143 {
 144         unsigned char *s;
 145         s = xmalloc(20);
 146         memcpy(s, sha1, 20);
 147         add_work(WORK_SHA1, name, s);
 148 }
 149
 150 static void grep_file_async(struct grep_opt *opt, char *name,
 151                             const char *filename)
 152 {
 153         add_work(WORK_FILE, name, xstrdup(filename));
 154 }
 155
 156 static void work_done(struct work_item *w)
 157 {
 158         int old_done;
 159
 160         grep_lock();
 161         w->done = 1;
 162         old_done = todo_done;
 163         for(; todo[todo_done].done && todo_done != todo_start;
 164             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 165                 w = &todo[todo_done];
 166                 if (w->out.len) {
 167                         if (print_hunk_marks_between_files && printed_something)
 168                                 write_or_die(1, "--\n", 3);
 169                         write_or_die(1, w->out.buf, w->out.len);
 170                         printed_something = 1;
 171                 }
 172                 free(w->name);
 173                 free(w->identifier);
 174         }
 175
 176         if (old_done != todo_done)
 177                 pthread_cond_signal(&cond_write);
 178
 179         if (all_work_added && todo_done == todo_end)
 180                 pthread_cond_signal(&cond_result);
 181
 182         grep_unlock();
 183 }
 184
 185 static void *run(void *arg)
 186 {
 187         int hit = 0;
 188         struct grep_opt *opt = arg;
 189
 190         while (1) {
 191                 struct work_item *w = get_work();
 192                 if (!w)
 193                         break;
 194
 195                 opt->output_priv = w;
 196                 if (w->type == WORK_SHA1) {
 197                         unsigned long sz;
 198                         void* data = load_sha1(w->identifier, &sz, w->name);
 199
 200                         if (data) {
 201                                 hit |= grep_buffer(opt, w->name, data, sz);
 202                                 free(data);
 203                         }
 204                 } else if (w->type == WORK_FILE) {
 205                         size_t sz;
 206                         void* data = load_file(w->identifier, &sz);
 207                         if (data) {
 208                                 hit |= grep_buffer(opt, w->name, data, sz);
 209                                 free(data);
 210                         }
 211                 } else {
 212                         assert(0);
 213                 }
 214
 215                 work_done(w);
 216         }
 217         free_grep_patterns(arg);
 218         free(arg);
 219
 220         return (void*) (intptr_t) hit;
 221 }
 222
 223 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 224 {
 225         struct work_item *w = opt->output_priv;
 226         strbuf_add(&w->out, buf, size);
 227 }
 228
 229 static void start_threads(struct grep_opt *opt)
 230 {
 231         int i;
 232
 233         pthread_mutex_init(&grep_mutex, NULL);
 234         pthread_mutex_init(&read_sha1_mutex, NULL);
 235         pthread_cond_init(&cond_add, NULL);
 236         pthread_cond_init(&cond_write, NULL);
 237         pthread_cond_init(&cond_result, NULL);
 238
 239         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 240                 strbuf_init(&todo[i].out, 0);
 241         }
 242
 243         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 244                 int err;
 245                 struct grep_opt *o = grep_opt_dup(opt);
 246                 o->output = strbuf_out;
 247                 compile_grep_patterns(o);
 248                 err = pthread_create(&threads[i], NULL, run, o);
 249
 250                 if (err)
 251                         die("grep: failed to create thread: %s",
 252                             strerror(err));
 253         }
 254 }
 255
 256 static int wait_all(void)
 257 {
 258         int hit = 0;
 259         int i;
 260
 261         grep_lock();
 262         all_work_added = 1;
 263
 264         /* Wait until all work is done. */
 265         while (todo_done != todo_end)
 266                 pthread_cond_wait(&cond_result, &grep_mutex);
 267
 268         /* Wake up all the consumer threads so they can see that there
 269          * is no more work to do.
 270          */
 271         pthread_cond_broadcast(&cond_add);
 272         grep_unlock();
 273
 274         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 275                 void *h;
 276                 pthread_join(threads[i], &h);
 277                 hit |= (int) (intptr_t) h;
 278         }
 279
 280         pthread_mutex_destroy(&grep_mutex);
 281         pthread_mutex_destroy(&read_sha1_mutex);
 282         pthread_cond_destroy(&cond_add);
 283         pthread_cond_destroy(&cond_write);
 284         pthread_cond_destroy(&cond_result);
 285
 286         return hit;
 287 }
 288 #else /* !NO_PTHREADS */
 289 #define read_sha1_lock()
 290 #define read_sha1_unlock()
 291
 292 static int wait_all(void)
 293 {
 294         return 0;
 295 }
 296 #endif
 297
 298 static int grep_config(const char *var, const char *value, void *cb)
 299 {
 300         struct grep_opt *opt = cb;
 301         char *color = NULL;
 302
 303         switch (userdiff_config(var, value)) {
 304         case 0: break;
 305         case -1: return -1;
 306         default: return 0;
 307         }
 308
 309         if (!strcmp(var, "color.grep"))
 310                 opt->color = git_config_colorbool(var, value, -1);
 311         else if (!strcmp(var, "color.grep.context"))
 312                 color = opt->color_context;
 313         else if (!strcmp(var, "color.grep.filename"))
 314                 color = opt->color_filename;
 315         else if (!strcmp(var, "color.grep.function"))
 316                 color = opt->color_function;
 317         else if (!strcmp(var, "color.grep.linenumber"))
 318                 color = opt->color_lineno;
 319         else if (!strcmp(var, "color.grep.match"))
 320                 color = opt->color_match;
 321         else if (!strcmp(var, "color.grep.selected"))
 322                 color = opt->color_selected;
 323         else if (!strcmp(var, "color.grep.separator"))
 324                 color = opt->color_sep;
 325         else
 326                 return git_color_default_config(var, value, cb);
 327         if (color) {
 328                 if (!value)
 329                         return config_error_nonbool(var);
 330                 color_parse(value, var, color);
 331         }
 332         return 0;
 333 }
 334
 335 /*
 336  * Return non-zero if max_depth is negative or path has no more then max_depth
 337  * slashes.
 338  */
 339 static int accept_subdir(const char *path, int max_depth)
 340 {
 341         if (max_depth < 0)
 342                 return 1;
 343
 344         while ((path = strchr(path, '/')) != NULL) {
 345                 max_depth--;
 346                 if (max_depth < 0)
 347                         return 0;
 348                 path++;
 349         }
 350         return 1;
 351 }
 352
 353 /*
 354  * Return non-zero if name is a subdirectory of match and is not too deep.
 355  */
 356 static int is_subdir(const char *name, int namelen,
 357                 const char *match, int matchlen, int max_depth)
 358 {
 359         if (matchlen > namelen || strncmp(name, match, matchlen))
 360                 return 0;
 361
 362         if (name[matchlen] == '\0') /* exact match */
 363                 return 1;
 364
 365         if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
 366                 return accept_subdir(name + matchlen + 1, max_depth);
 367
 368         return 0;
 369 }
 370
 371 /*
 372  * git grep pathspecs are somewhat different from diff-tree pathspecs;
 373  * pathname wildcards are allowed.
 374  */
 375 static int pathspec_matches(const char **paths, const char *name, int max_depth)
 376 {
 377         int namelen, i;
 378         if (!paths || !*paths)
 379                 return accept_subdir(name, max_depth);
 380         namelen = strlen(name);
 381         for (i = 0; paths[i]; i++) {
 382                 const char *match = paths[i];
 383                 int matchlen = strlen(match);
 384                 const char *cp, *meta;
 385
 386                 if (is_subdir(name, namelen, match, matchlen, max_depth))
 387                         return 1;
 388                 if (!fnmatch(match, name, 0))
 389                         return 1;
 390                 if (name[namelen-1] != '/')
 391                         continue;
 392
 393                 /* We are being asked if the directory ("name") is worth
 394                  * descending into.
 395                  *
 396                  * Find the longest leading directory name that does
 397                  * not have metacharacter in the pathspec; the name
 398                  * we are looking at must overlap with that directory.
 399                  */
 400                 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
 401                         char ch = *cp;
 402                         if (ch == '*' || ch == '[' || ch == '?') {
 403                                 meta = cp;
 404                                 break;
 405                         }
 406                 }
 407                 if (!meta)
 408                         meta = cp; /* fully literal */
 409
 410                 if (namelen <= meta - match) {
 411                         /* Looking at "Documentation/" and
 412                          * the pattern says "Documentation/howto/", or
 413                          * "Documentation/diff*.txt".  The name we
 414                          * have should match prefix.
 415                          */
 416                         if (!memcmp(match, name, namelen))
 417                                 return 1;
 418                         continue;
 419                 }
 420
 421                 if (meta - match < namelen) {
 422                         /* Looking at "Documentation/howto/" and
 423                          * the pattern says "Documentation/h*";
 424                          * match up to "Do.../h"; this avoids descending
 425                          * into "Documentation/technical/".
 426                          */
 427                         if (!memcmp(match, name, meta - match))
 428                                 return 1;
 429                         continue;
 430                 }
 431         }
 432         return 0;
 433 }
 434
 435 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 436 {
 437         void *data;
 438
 439         if (use_threads) {
 440                 read_sha1_lock();
 441                 data = read_sha1_file(sha1, type, size);
 442                 read_sha1_unlock();
 443         } else {
 444                 data = read_sha1_file(sha1, type, size);
 445         }
 446         return data;
 447 }
 448
 449 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 450                        const char *name)
 451 {
 452         enum object_type type;
 453         void *data = lock_and_read_sha1_file(sha1, &type, size);
 454
 455         if (!data)
 456                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 457
 458         return data;
 459 }
 460
 461 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 462                      const char *filename, int tree_name_len)
 463 {
 464         struct strbuf pathbuf = STRBUF_INIT;
 465         char *name;
 466
 467         if (opt->relative && opt->prefix_length) {
 468                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 469                                     opt->prefix);
 470                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 471         } else {
 472                 strbuf_addstr(&pathbuf, filename);
 473         }
 474
 475         name = strbuf_detach(&pathbuf, NULL);
 476
 477 #ifndef NO_PTHREADS
 478         if (use_threads) {
 479                 grep_sha1_async(opt, name, sha1);
 480                 return 0;
 481         } else
 482 #endif
 483         {
 484                 int hit;
 485                 unsigned long sz;
 486                 void *data = load_sha1(sha1, &sz, name);
 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 *load_file(const char *filename, size_t *sz)
 499 {
 500         struct stat st;
 501         char *data;
 502         int i;
 503
 504         if (lstat(filename, &st) < 0) {
 505         err_ret:
 506                 if (errno != ENOENT)
 507                         error("'%s': %s", filename, strerror(errno));
 508                 return 0;
 509         }
 510         if (!S_ISREG(st.st_mode))
 511                 return 0;
 512         *sz = xsize_t(st.st_size);
 513         i = open(filename, O_RDONLY);
 514         if (i < 0)
 515                 goto err_ret;
 516         data = xmalloc(*sz + 1);
 517         if (st.st_size != read_in_full(i, data, *sz)) {
 518                 error("'%s': short read %s", filename, strerror(errno));
 519                 close(i);
 520                 free(data);
 521                 return 0;
 522         }
 523         close(i);
 524         data[*sz] = 0;
 525         return data;
 526 }
 527
 528 static int grep_file(struct grep_opt *opt, const char *filename)
 529 {
 530         struct strbuf buf = STRBUF_INIT;
 531         char *name;
 532
 533         if (opt->relative && opt->prefix_length)
 534                 quote_path_relative(filename, -1, &buf, opt->prefix);
 535         else
 536                 strbuf_addstr(&buf, filename);
 537         name = strbuf_detach(&buf, NULL);
 538
 539 #ifndef NO_PTHREADS
 540         if (use_threads) {
 541                 grep_file_async(opt, name, filename);
 542                 return 0;
 543         } else
 544 #endif
 545         {
 546                 int hit;
 547                 size_t sz;
 548                 void *data = load_file(filename, &sz);
 549                 if (!data)
 550                         hit = 0;
 551                 else
 552                         hit = grep_buffer(opt, name, data, sz);
 553
 554                 free(data);
 555                 free(name);
 556                 return hit;
 557         }
 558 }
 559
 560 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 561 {
 562         struct string_list *path_list = opt->output_priv;
 563
 564         if (len == 1 && *(char *)data == '\0')
 565                 return;
 566         string_list_append(xstrndup(data, len), path_list);
 567 }
 568
 569 static void run_pager(struct grep_opt *opt, const char *prefix)
 570 {
 571         struct string_list *path_list = opt->output_priv;
 572         char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 573         int i;
 574
 575         for (i = 0; i < path_list->nr; i++)
 576                 argv[i] = path_list->items[i].string;
 577         argv[path_list->nr] = NULL;
 578
 579         if (prefix)
 580                 chdir(prefix);
 581         execvp(argv[0], argv);
 582         error("Could not run pager %s: %s", argv[0], strerror(errno));
 583 }
 584
 585 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 586 {
 587         int hit = 0;
 588         int nr;
 589         read_cache();
 590
 591         for (nr = 0; nr < active_nr; nr++) {
 592                 struct cache_entry *ce = active_cache[nr];
 593                 if (!S_ISREG(ce->ce_mode))
 594                         continue;
 595                 if (!pathspec_matches(paths, ce->name, opt->max_depth))
 596                         continue;
 597                 /*
 598                  * If CE_VALID is on, we assume worktree file and its cache entry
 599                  * are identical, even if worktree file has been modified, so use
 600                  * cache version instead
 601                  */
 602                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 603                         if (ce_stage(ce))
 604                                 continue;
 605                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 606                 }
 607                 else
 608                         hit |= grep_file(opt, ce->name);
 609                 if (ce_stage(ce)) {
 610                         do {
 611                                 nr++;
 612                         } while (nr < active_nr &&
 613                                  !strcmp(ce->name, active_cache[nr]->name));
 614                         nr--; /* compensate for loop control */
 615                 }
 616                 if (hit && opt->status_only)
 617                         break;
 618         }
 619         return hit;
 620 }
 621
 622 static int grep_tree(struct grep_opt *opt, const char **paths,
 623                      struct tree_desc *tree,
 624                      const char *tree_name, const char *base)
 625 {
 626         int len;
 627         int hit = 0;
 628         struct name_entry entry;
 629         char *down;
 630         int tn_len = strlen(tree_name);
 631         struct strbuf pathbuf;
 632
 633         strbuf_init(&pathbuf, PATH_MAX + tn_len);
 634
 635         if (tn_len) {
 636                 strbuf_add(&pathbuf, tree_name, tn_len);
 637                 strbuf_addch(&pathbuf, ':');
 638                 tn_len = pathbuf.len;
 639         }
 640         strbuf_addstr(&pathbuf, base);
 641         len = pathbuf.len;
 642
 643         while (tree_entry(tree, &entry)) {
 644                 int te_len = tree_entry_len(entry.path, entry.sha1);
 645                 pathbuf.len = len;
 646                 strbuf_add(&pathbuf, entry.path, te_len);
 647
 648                 if (S_ISDIR(entry.mode))
 649                         /* Match "abc/" against pathspec to
 650                          * decide if we want to descend into "abc"
 651                          * directory.
 652                          */
 653                         strbuf_addch(&pathbuf, '/');
 654
 655                 down = pathbuf.buf + tn_len;
 656                 if (!pathspec_matches(paths, down, opt->max_depth))
 657                         ;
 658                 else if (S_ISREG(entry.mode))
 659                         hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
 660                 else if (S_ISDIR(entry.mode)) {
 661                         enum object_type type;
 662                         struct tree_desc sub;
 663                         void *data;
 664                         unsigned long size;
 665
 666                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 667                         if (!data)
 668                                 die("unable to read tree (%s)",
 669                                     sha1_to_hex(entry.sha1));
 670                         init_tree_desc(&sub, data, size);
 671                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 672                         free(data);
 673                 }
 674                 if (hit && opt->status_only)
 675                         break;
 676         }
 677         strbuf_release(&pathbuf);
 678         return hit;
 679 }
 680
 681 static int grep_object(struct grep_opt *opt, const char **paths,
 682                        struct object *obj, const char *name)
 683 {
 684         if (obj->type == OBJ_BLOB)
 685                 return grep_sha1(opt, obj->sha1, name, 0);
 686         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 687                 struct tree_desc tree;
 688                 void *data;
 689                 unsigned long size;
 690                 int hit;
 691                 data = read_object_with_reference(obj->sha1, tree_type,
 692                                                   &size, NULL);
 693                 if (!data)
 694                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 695                 init_tree_desc(&tree, data, size);
 696                 hit = grep_tree(opt, paths, &tree, name, "");
 697                 free(data);
 698                 return hit;
 699         }
 700         die("unable to grep from object of type %s", typename(obj->type));
 701 }
 702
 703 static int grep_directory(struct grep_opt *opt, const char **paths)
 704 {
 705         struct dir_struct dir;
 706         int i, hit = 0;
 707
 708         memset(&dir, 0, sizeof(dir));
 709         setup_standard_excludes(&dir);
 710
 711         fill_directory(&dir, paths);
 712         for (i = 0; i < dir.nr; i++) {
 713                 hit |= grep_file(opt, dir.entries[i]->name);
 714                 if (hit && opt->status_only)
 715                         break;
 716         }
 717         free_grep_patterns(opt);
 718         return hit;
 719 }
 720
 721 static int context_callback(const struct option *opt, const char *arg,
 722                             int unset)
 723 {
 724         struct grep_opt *grep_opt = opt->value;
 725         int value;
 726         const char *endp;
 727
 728         if (unset) {
 729                 grep_opt->pre_context = grep_opt->post_context = 0;
 730                 return 0;
 731         }
 732         value = strtol(arg, (char **)&endp, 10);
 733         if (*endp) {
 734                 return error("switch `%c' expects a numerical value",
 735                              opt->short_name);
 736         }
 737         grep_opt->pre_context = grep_opt->post_context = value;
 738         return 0;
 739 }
 740
 741 static int file_callback(const struct option *opt, const char *arg, int unset)
 742 {
 743         struct grep_opt *grep_opt = opt->value;
 744         FILE *patterns;
 745         int lno = 0;
 746         struct strbuf sb = STRBUF_INIT;
 747
 748         patterns = fopen(arg, "r");
 749         if (!patterns)
 750                 die_errno("cannot open '%s'", arg);
 751         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 752                 char *s;
 753                 size_t len;
 754
 755                 /* ignore empty line like grep does */
 756                 if (sb.len == 0)
 757                         continue;
 758
 759                 s = strbuf_detach(&sb, &len);
 760                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 761         }
 762         fclose(patterns);
 763         strbuf_release(&sb);
 764         return 0;
 765 }
 766
 767 static int not_callback(const struct option *opt, const char *arg, int unset)
 768 {
 769         struct grep_opt *grep_opt = opt->value;
 770         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 771         return 0;
 772 }
 773
 774 static int and_callback(const struct option *opt, const char *arg, int unset)
 775 {
 776         struct grep_opt *grep_opt = opt->value;
 777         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 778         return 0;
 779 }
 780
 781 static int open_callback(const struct option *opt, const char *arg, int unset)
 782 {
 783         struct grep_opt *grep_opt = opt->value;
 784         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 785         return 0;
 786 }
 787
 788 static int close_callback(const struct option *opt, const char *arg, int unset)
 789 {
 790         struct grep_opt *grep_opt = opt->value;
 791         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 792         return 0;
 793 }
 794
 795 static int pattern_callback(const struct option *opt, const char *arg,
 796                             int unset)
 797 {
 798         struct grep_opt *grep_opt = opt->value;
 799         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 800         return 0;
 801 }
 802
 803 static int help_callback(const struct option *opt, const char *arg, int unset)
 804 {
 805         return -1;
 806 }
 807
 808 int cmd_grep(int argc, const char **argv, const char *prefix)
 809 {
 810         int hit = 0;
 811         int cached = 0;
 812         int seen_dashdash = 0;
 813         int external_grep_allowed__ignored;
 814         const char *show_in_pager = NULL, *default_pager = "dummy";
 815         struct grep_opt opt;
 816         struct object_array list = { 0, 0, NULL };
 817         const char **paths = NULL;
 818         struct string_list path_list = { NULL, 0, 0, 0 };
 819         int i;
 820         int dummy;
 821         int nongit = 0, use_index = 1;
 822         struct option options[] = {
 823                 OPT_BOOLEAN(0, "cached", &cached,
 824                         "search in index instead of in the work tree"),
 825                 OPT_BOOLEAN(0, "index", &use_index,
 826                         "--no-index finds in contents not managed by git"),
 827                 OPT_GROUP(""),
 828                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 829                         "show non-matching lines"),
 830                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 831                         "case insensitive matching"),
 832                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 833                         "match patterns only at word boundaries"),
 834                 OPT_SET_INT('a', "text", &opt.binary,
 835                         "process binary files as text", GREP_BINARY_TEXT),
 836                 OPT_SET_INT('I', NULL, &opt.binary,
 837                         "don't match patterns in binary files",
 838                         GREP_BINARY_NOMATCH),
 839                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 840                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 841                         NULL, 1 },
 842                 OPT_GROUP(""),
 843                 OPT_BIT('E', "extended-regexp", &opt.regflags,
 844                         "use extended POSIX regular expressions", REG_EXTENDED),
 845                 OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
 846                         "use basic POSIX regular expressions (default)",
 847                         REG_EXTENDED),
 848                 OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
 849                         "interpret patterns as fixed strings"),
 850                 OPT_GROUP(""),
 851                 OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
 852                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 853                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 854                 OPT_NEGBIT(0, "full-name", &opt.relative,
 855                         "show filenames relative to top directory", 1),
 856                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 857                         "show only filenames instead of matching lines"),
 858                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 859                         "synonym for --files-with-matches"),
 860                 OPT_BOOLEAN('L', "files-without-match",
 861                         &opt.unmatch_name_only,
 862                         "show only the names of files without match"),
 863                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 864                         "print NUL after filenames"),
 865                 OPT_BOOLEAN('c', "count", &opt.count,
 866                         "show the number of matches instead of matching lines"),
 867                 OPT__COLOR(&opt.color, "highlight matches"),
 868                 OPT_GROUP(""),
 869                 OPT_CALLBACK('C', NULL, &opt, "n",
 870                         "show <n> context lines before and after matches",
 871                         context_callback),
 872                 OPT_INTEGER('B', NULL, &opt.pre_context,
 873                         "show <n> context lines before matches"),
 874                 OPT_INTEGER('A', NULL, &opt.post_context,
 875                         "show <n> context lines after matches"),
 876                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 877                         context_callback),
 878                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 879                         "show a line with the function name before matches"),
 880                 OPT_GROUP(""),
 881                 OPT_CALLBACK('f', NULL, &opt, "file",
 882                         "read patterns from file", file_callback),
 883                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 884                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 885                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 886                   "combine patterns specified with -e",
 887                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 888                 OPT_BOOLEAN(0, "or", &dummy, ""),
 889                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 890                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 891                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 892                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 893                   open_callback },
 894                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 895                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 896                   close_callback },
 897                 OPT_BOOLEAN('q', "quiet", &opt.status_only,
 898                             "indicate hit with exit status without output"),
 899                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 900                         "show only matches from files that match all patterns"),
 901                 OPT_GROUP(""),
 902                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 903                         "pager", "show matching files in the pager",
 904                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 905                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 906                             "allow calling of grep(1) (ignored by this build)"),
 907                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 908                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 909                 OPT_END()
 910         };
 911
 912         prefix = setup_git_directory_gently(&nongit);
 913
 914         /*
 915          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 916          * to show usage information and exit.
 917          */
 918         if (argc == 2 && !strcmp(argv[1], "-h"))
 919                 usage_with_options(grep_usage, options);
 920
 921         memset(&opt, 0, sizeof(opt));
 922         opt.prefix = prefix;
 923         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 924         opt.relative = 1;
 925         opt.pathname = 1;
 926         opt.pattern_tail = &opt.pattern_list;
 927         opt.header_tail = &opt.header_list;
 928         opt.regflags = REG_NEWLINE;
 929         opt.max_depth = -1;
 930
 931         strcpy(opt.color_context, "");
 932         strcpy(opt.color_filename, "");
 933         strcpy(opt.color_function, "");
 934         strcpy(opt.color_lineno, "");
 935         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 936         strcpy(opt.color_selected, "");
 937         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 938         opt.color = -1;
 939         git_config(grep_config, &opt);
 940         if (opt.color == -1)
 941                 opt.color = git_use_color_default;
 942
 943         /*
 944          * If there is no -- then the paths must exist in the working
 945          * tree.  If there is no explicit pattern specified with -e or
 946          * -f, we take the first unrecognized non option to be the
 947          * pattern, but then what follows it must be zero or more
 948          * valid refs up to the -- (if exists), and then existing
 949          * paths.  If there is an explicit pattern, then the first
 950          * unrecognized non option is the beginning of the refs list
 951          * that continues up to the -- (if exists), and then paths.
 952          */
 953         argc = parse_options(argc, argv, prefix, options, grep_usage,
 954                              PARSE_OPT_KEEP_DASHDASH |
 955                              PARSE_OPT_STOP_AT_NON_OPTION |
 956                              PARSE_OPT_NO_INTERNAL_HELP);
 957
 958         if (use_index && nongit)
 959                 /* die the same way as if we did it at the beginning */
 960                 setup_git_directory();
 961
 962         /*
 963          * skip a -- separator; we know it cannot be
 964          * separating revisions from pathnames if
 965          * we haven't even had any patterns yet
 966          */
 967         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 968                 argv++;
 969                 argc--;
 970         }
 971
 972         /* First unrecognized non-option token */
 973         if (argc > 0 && !opt.pattern_list) {
 974                 append_grep_pattern(&opt, argv[0], "command line", 0,
 975                                     GREP_PATTERN);
 976                 argv++;
 977                 argc--;
 978         }
 979
 980         if (show_in_pager) {
 981                 if (show_in_pager == default_pager) {
 982                         show_in_pager = getenv("GIT_PAGER");
 983                         if (!show_in_pager)
 984                                 show_in_pager = getenv("PAGER");
 985                         if (!show_in_pager)
 986                                 show_in_pager = "less";
 987                 }
 988                 opt.name_only = 1;
 989                 opt.null_following_name = 1;
 990                 opt.output_priv = &path_list;
 991                 opt.output = append_path;
 992                 string_list_append(show_in_pager, &path_list);
 993                 use_threads = 0;
 994         }
 995
 996         if (!opt.pattern_list)
 997                 die("no pattern given.");
 998         if (!opt.fixed && opt.ignore_case)
 999                 opt.regflags |= REG_ICASE;
1000         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
1001                 die("cannot mix --fixed-strings and regexp");
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)
1009                         print_hunk_marks_between_files = 1;
1010                 start_threads(&opt);
1011         }
1012 #else
1013         use_threads = 0;
1014 #endif
1015
1016         compile_grep_patterns(&opt);
1017
1018         /* Check revs and then paths */
1019         for (i = 0; i < argc; i++) {
1020                 const char *arg = argv[i];
1021                 unsigned char sha1[20];
1022                 /* Is it a rev? */
1023                 if (!get_sha1(arg, sha1)) {
1024                         struct object *object = parse_object(sha1);
1025                         if (!object)
1026                                 die("bad object %s", arg);
1027                         add_object_array(object, arg, &list);
1028                         continue;
1029                 }
1030                 if (!strcmp(arg, "--")) {
1031                         i++;
1032                         seen_dashdash = 1;
1033                 }
1034                 break;
1035         }
1036
1037         /* The rest are paths */
1038         if (!seen_dashdash) {
1039                 int j;
1040                 for (j = i; j < argc; j++)
1041                         verify_filename(prefix, argv[j]);
1042         }
1043
1044         if (i < argc)
1045                 paths = get_pathspec(prefix, argv + i);
1046         else if (prefix) {
1047                 paths = xcalloc(2, sizeof(const char *));
1048                 paths[0] = prefix;
1049                 paths[1] = NULL;
1050         }
1051
1052         if (!use_index) {
1053                 int hit;
1054                 if (cached)
1055                         die("--cached cannot be used with --no-index.");
1056                 if (list.nr)
1057                         die("--no-index cannot be used with revs.");
1058                 hit = grep_directory(&opt, paths);
1059                 if (use_threads)
1060                         hit |= wait_all();
1061                 return !hit;
1062         }
1063
1064         if (show_in_pager && (cached || list.nr))
1065                 die ("--open-files-in-pager only works on the worktree");
1066
1067         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1068                 const char *pager = path_list.items[0].string;
1069                 int len = strlen(pager);
1070
1071                 if (len > 4 && is_dir_sep(pager[len - 5]))
1072                         pager += len - 4;
1073
1074                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1075                         struct strbuf buf = STRBUF_INIT;
1076                         strbuf_addf(&buf, "+/%s%s",
1077                                         strcmp("less", pager) ? "" : "*",
1078                                         opt.pattern_list->pattern);
1079                         string_list_append(buf.buf, &path_list);
1080                         strbuf_detach(&buf, NULL);
1081                 }
1082         }
1083
1084         if (!show_in_pager)
1085                 setup_pager();
1086
1087         if (!list.nr) {
1088                 if (!cached)
1089                         setup_work_tree();
1090
1091                 hit = grep_cache(&opt, paths, cached);
1092         }
1093         else if (cached)
1094                 die("both --cached and trees are given.");
1095         else
1096                 for (i = 0; i < list.nr; i++) {
1097                         struct object *real_obj;
1098                         real_obj = deref_tag(list.objects[i].item, NULL, 0);
1099                         if (grep_object(&opt, paths, real_obj,
1100                                                 list.objects[i].name)) {
1101                                 hit = 1;
1102                                 if (opt.status_only)
1103                                         break;
1104                         }
1105                 }
1106
1107         if (use_threads)
1108                 hit |= wait_all();
1109
1110         if (hit && show_in_pager)
1111                 run_pager(&opt, prefix);
1112
1113         free_grep_patterns(&opt);
1114         return !hit;
1115 }