builtin/grep.c

   1 /*
   2  * Builtin "git grep"
   3  *
   4  * Copyright (c) 2006 Junio C Hamano
   5  */
   6 #include "cache.h"
   7 #include "blob.h"
   8 #include "tree.h"
   9 #include "commit.h"
  10 #include "tag.h"
  11 #include "tree-walk.h"
  12 #include "builtin.h"
  13 #include "parse-options.h"
  14 #include "string-list.h"
  15 #include "run-command.h"
  16 #include "userdiff.h"
  17 #include "grep.h"
  18 #include "quote.h"
  19 #include "dir.h"
  20 #include "attr.h"
  21
  22 #ifndef NO_PTHREADS
  23 #include <pthread.h>
  24 #include "thread-utils.h"
  25 #endif
  26
  27 static char const * const grep_usage[] = {
  28         "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
  29         NULL
  30 };
  31
  32 static int use_threads = 1;
  33
  34 #ifndef NO_PTHREADS
  35 #define THREADS 8
  36 static pthread_t threads[THREADS];
  37
  38 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  39                        const char *name);
  40 static void *load_file(const char *filename, size_t *sz);
  41
  42 enum work_type {WORK_SHA1, WORK_FILE};
  43
  44 /* We use one producer thread and THREADS consumer
  45  * threads. The producer adds struct work_items to 'todo' and the
  46  * consumers pick work items from the same array.
  47  */
  48 struct work_item
  49 {
  50         enum work_type type;
  51         char *name;
  52
  53         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  54          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  55          * terminated filename.
  56          */
  57         void *identifier;
  58         char done;
  59         struct strbuf out;
  60 };
  61
  62 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  63  * that have been or are processed by a consumer thread. We haven't
  64  * written the result for these to stdout yet.
  65  *
  66  * The work_items in [todo_start, todo_end) are waiting to be picked
  67  * up by a consumer thread.
  68  *
  69  * The ranges are modulo TODO_SIZE.
  70  */
  71 #define TODO_SIZE 128
  72 static struct work_item todo[TODO_SIZE];
  73 static int todo_start;
  74 static int todo_end;
  75 static int todo_done;
  76
  77 /* Has all work items been added? */
  78 static int all_work_added;
  79
  80 /* This lock protects all the variables above. */
  81 static pthread_mutex_t grep_mutex;
  82
  83 /* Used to serialize calls to read_sha1_file. */
  84 static pthread_mutex_t read_sha1_mutex;
  85
  86 #define grep_lock() pthread_mutex_lock(&grep_mutex)
  87 #define grep_unlock() pthread_mutex_unlock(&grep_mutex)
  88 #define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
  89 #define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
  90
  91 /* Signalled when a new work_item is added to todo. */
  92 static pthread_cond_t cond_add;
  93
  94 /* Signalled when the result from one work_item is written to
  95  * stdout.
  96  */
  97 static pthread_cond_t cond_write;
  98
  99 /* Signalled when we are finished with everything. */
 100 static pthread_cond_t cond_result;
 101
 102 static int print_hunk_marks_between_files;
 103 static int printed_something;
 104
 105 static void add_work(enum work_type type, char *name, void *id)
 106 {
 107         grep_lock();
 108
 109         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 110                 pthread_cond_wait(&cond_write, &grep_mutex);
 111         }
 112
 113         todo[todo_end].type = type;
 114         todo[todo_end].name = name;
 115         todo[todo_end].identifier = id;
 116         todo[todo_end].done = 0;
 117         strbuf_reset(&todo[todo_end].out);
 118         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 119
 120         pthread_cond_signal(&cond_add);
 121         grep_unlock();
 122 }
 123
 124 static struct work_item *get_work(void)
 125 {
 126         struct work_item *ret;
 127
 128         grep_lock();
 129         while (todo_start == todo_end && !all_work_added) {
 130                 pthread_cond_wait(&cond_add, &grep_mutex);
 131         }
 132
 133         if (todo_start == todo_end && all_work_added) {
 134                 ret = NULL;
 135         } else {
 136                 ret = &todo[todo_start];
 137                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 138         }
 139         grep_unlock();
 140         return ret;
 141 }
 142
 143 static void grep_sha1_async(struct grep_opt *opt, char *name,
 144                             const unsigned char *sha1)
 145 {
 146         unsigned char *s;
 147         s = xmalloc(20);
 148         memcpy(s, sha1, 20);
 149         add_work(WORK_SHA1, name, s);
 150 }
 151
 152 static void grep_file_async(struct grep_opt *opt, char *name,
 153                             const char *filename)
 154 {
 155         add_work(WORK_FILE, name, xstrdup(filename));
 156 }
 157
 158 static void work_done(struct work_item *w)
 159 {
 160         int old_done;
 161
 162         grep_lock();
 163         w->done = 1;
 164         old_done = todo_done;
 165         for(; todo[todo_done].done && todo_done != todo_start;
 166             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 167                 w = &todo[todo_done];
 168                 if (w->out.len) {
 169                         if (print_hunk_marks_between_files && printed_something)
 170                                 write_or_die(1, "--\n", 3);
 171                         write_or_die(1, w->out.buf, w->out.len);
 172                         printed_something = 1;
 173                 }
 174                 free(w->name);
 175                 free(w->identifier);
 176         }
 177
 178         if (old_done != todo_done)
 179                 pthread_cond_signal(&cond_write);
 180
 181         if (all_work_added && todo_done == todo_end)
 182                 pthread_cond_signal(&cond_result);
 183
 184         grep_unlock();
 185 }
 186
 187 static int skip_binary(struct grep_opt *opt, const char *filename)
 188 {
 189         if ((opt->binary & GREP_BINARY_NOMATCH)) {
 190                 static struct git_attr *attr_text;
 191                 struct git_attr_check check;
 192
 193                 if (!attr_text)
 194                         attr_text = git_attr("text");
 195                 memset(&check, 0, sizeof(check));
 196                 check.attr = attr_text;
 197                 return !git_checkattr(filename, 1, &check) &&
 198                                 ATTR_FALSE(check.value);
 199         }
 200         return 0;
 201 }
 202
 203 static void *run(void *arg)
 204 {
 205         int hit = 0;
 206         struct grep_opt *opt = arg;
 207
 208         while (1) {
 209                 struct work_item *w = get_work();
 210                 if (!w)
 211                         break;
 212
 213                 if (skip_binary(opt, (const char *)w->identifier))
 214                         continue;
 215
 216                 opt->output_priv = w;
 217                 if (w->type == WORK_SHA1) {
 218                         unsigned long sz;
 219                         void* data = load_sha1(w->identifier, &sz, w->name);
 220
 221                         if (data) {
 222                                 hit |= grep_buffer(opt, w->name, data, sz);
 223                                 free(data);
 224                         }
 225                 } else if (w->type == WORK_FILE) {
 226                         size_t sz;
 227                         void* data = load_file(w->identifier, &sz);
 228                         if (data) {
 229                                 hit |= grep_buffer(opt, w->name, data, sz);
 230                                 free(data);
 231                         }
 232                 } else {
 233                         assert(0);
 234                 }
 235
 236                 work_done(w);
 237         }
 238         free_grep_patterns(arg);
 239         free(arg);
 240
 241         return (void*) (intptr_t) hit;
 242 }
 243
 244 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 245 {
 246         struct work_item *w = opt->output_priv;
 247         strbuf_add(&w->out, buf, size);
 248 }
 249
 250 static void start_threads(struct grep_opt *opt)
 251 {
 252         int i;
 253
 254         pthread_mutex_init(&grep_mutex, NULL);
 255         pthread_mutex_init(&read_sha1_mutex, NULL);
 256         pthread_cond_init(&cond_add, NULL);
 257         pthread_cond_init(&cond_write, NULL);
 258         pthread_cond_init(&cond_result, NULL);
 259
 260         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 261                 strbuf_init(&todo[i].out, 0);
 262         }
 263
 264         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 265                 int err;
 266                 struct grep_opt *o = grep_opt_dup(opt);
 267                 o->output = strbuf_out;
 268                 compile_grep_patterns(o);
 269                 err = pthread_create(&threads[i], NULL, run, o);
 270
 271                 if (err)
 272                         die("grep: failed to create thread: %s",
 273                             strerror(err));
 274         }
 275 }
 276
 277 static int wait_all(void)
 278 {
 279         int hit = 0;
 280         int i;
 281
 282         grep_lock();
 283         all_work_added = 1;
 284
 285         /* Wait until all work is done. */
 286         while (todo_done != todo_end)
 287                 pthread_cond_wait(&cond_result, &grep_mutex);
 288
 289         /* Wake up all the consumer threads so they can see that there
 290          * is no more work to do.
 291          */
 292         pthread_cond_broadcast(&cond_add);
 293         grep_unlock();
 294
 295         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 296                 void *h;
 297                 pthread_join(threads[i], &h);
 298                 hit |= (int) (intptr_t) h;
 299         }
 300
 301         pthread_mutex_destroy(&grep_mutex);
 302         pthread_mutex_destroy(&read_sha1_mutex);
 303         pthread_cond_destroy(&cond_add);
 304         pthread_cond_destroy(&cond_write);
 305         pthread_cond_destroy(&cond_result);
 306
 307         return hit;
 308 }
 309 #else /* !NO_PTHREADS */
 310 #define read_sha1_lock()
 311 #define read_sha1_unlock()
 312
 313 static int wait_all(void)
 314 {
 315         return 0;
 316 }
 317 #endif
 318
 319 static int grep_config(const char *var, const char *value, void *cb)
 320 {
 321         struct grep_opt *opt = cb;
 322         char *color = NULL;
 323
 324         switch (userdiff_config(var, value)) {
 325         case 0: break;
 326         case -1: return -1;
 327         default: return 0;
 328         }
 329
 330         if (!strcmp(var, "color.grep"))
 331                 opt->color = git_config_colorbool(var, value, -1);
 332         else if (!strcmp(var, "color.grep.context"))
 333                 color = opt->color_context;
 334         else if (!strcmp(var, "color.grep.filename"))
 335                 color = opt->color_filename;
 336         else if (!strcmp(var, "color.grep.function"))
 337                 color = opt->color_function;
 338         else if (!strcmp(var, "color.grep.linenumber"))
 339                 color = opt->color_lineno;
 340         else if (!strcmp(var, "color.grep.match"))
 341                 color = opt->color_match;
 342         else if (!strcmp(var, "color.grep.selected"))
 343                 color = opt->color_selected;
 344         else if (!strcmp(var, "color.grep.separator"))
 345                 color = opt->color_sep;
 346         else
 347                 return git_color_default_config(var, value, cb);
 348         if (color) {
 349                 if (!value)
 350                         return config_error_nonbool(var);
 351                 color_parse(value, var, color);
 352         }
 353         return 0;
 354 }
 355
 356 /*
 357  * Return non-zero if max_depth is negative or path has no more then max_depth
 358  * slashes.
 359  */
 360 static int accept_subdir(const char *path, int max_depth)
 361 {
 362         if (max_depth < 0)
 363                 return 1;
 364
 365         while ((path = strchr(path, '/')) != NULL) {
 366                 max_depth--;
 367                 if (max_depth < 0)
 368                         return 0;
 369                 path++;
 370         }
 371         return 1;
 372 }
 373
 374 /*
 375  * Return non-zero if name is a subdirectory of match and is not too deep.
 376  */
 377 static int is_subdir(const char *name, int namelen,
 378                 const char *match, int matchlen, int max_depth)
 379 {
 380         if (matchlen > namelen || strncmp(name, match, matchlen))
 381                 return 0;
 382
 383         if (name[matchlen] == '\0') /* exact match */
 384                 return 1;
 385
 386         if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
 387                 return accept_subdir(name + matchlen + 1, max_depth);
 388
 389         return 0;
 390 }
 391
 392 /*
 393  * git grep pathspecs are somewhat different from diff-tree pathspecs;
 394  * pathname wildcards are allowed.
 395  */
 396 static int pathspec_matches(const char **paths, const char *name, int max_depth)
 397 {
 398         int namelen, i;
 399         if (!paths || !*paths)
 400                 return accept_subdir(name, max_depth);
 401         namelen = strlen(name);
 402         for (i = 0; paths[i]; i++) {
 403                 const char *match = paths[i];
 404                 int matchlen = strlen(match);
 405                 const char *cp, *meta;
 406
 407                 if (is_subdir(name, namelen, match, matchlen, max_depth))
 408                         return 1;
 409                 if (!fnmatch(match, name, 0))
 410                         return 1;
 411                 if (name[namelen-1] != '/')
 412                         continue;
 413
 414                 /* We are being asked if the directory ("name") is worth
 415                  * descending into.
 416                  *
 417                  * Find the longest leading directory name that does
 418                  * not have metacharacter in the pathspec; the name
 419                  * we are looking at must overlap with that directory.
 420                  */
 421                 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
 422                         char ch = *cp;
 423                         if (ch == '*' || ch == '[' || ch == '?') {
 424                                 meta = cp;
 425                                 break;
 426                         }
 427                 }
 428                 if (!meta)
 429                         meta = cp; /* fully literal */
 430
 431                 if (namelen <= meta - match) {
 432                         /* Looking at "Documentation/" and
 433                          * the pattern says "Documentation/howto/", or
 434                          * "Documentation/diff*.txt".  The name we
 435                          * have should match prefix.
 436                          */
 437                         if (!memcmp(match, name, namelen))
 438                                 return 1;
 439                         continue;
 440                 }
 441
 442                 if (meta - match < namelen) {
 443                         /* Looking at "Documentation/howto/" and
 444                          * the pattern says "Documentation/h*";
 445                          * match up to "Do.../h"; this avoids descending
 446                          * into "Documentation/technical/".
 447                          */
 448                         if (!memcmp(match, name, meta - match))
 449                                 return 1;
 450                         continue;
 451                 }
 452         }
 453         return 0;
 454 }
 455
 456 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 457 {
 458         void *data;
 459
 460         if (use_threads) {
 461                 read_sha1_lock();
 462                 data = read_sha1_file(sha1, type, size);
 463                 read_sha1_unlock();
 464         } else {
 465                 data = read_sha1_file(sha1, type, size);
 466         }
 467         return data;
 468 }
 469
 470 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 471                        const char *name)
 472 {
 473         enum object_type type;
 474         void *data = lock_and_read_sha1_file(sha1, &type, size);
 475
 476         if (!data)
 477                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 478
 479         return data;
 480 }
 481
 482 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 483                      const char *filename, int tree_name_len)
 484 {
 485         struct strbuf pathbuf = STRBUF_INIT;
 486         char *name;
 487
 488         if (opt->relative && opt->prefix_length) {
 489                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 490                                     opt->prefix);
 491                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 492         } else {
 493                 strbuf_addstr(&pathbuf, filename);
 494         }
 495
 496         name = strbuf_detach(&pathbuf, NULL);
 497
 498 #ifndef NO_PTHREADS
 499         if (use_threads) {
 500                 grep_sha1_async(opt, name, sha1);
 501                 return 0;
 502         } else
 503 #endif
 504         {
 505                 int hit;
 506                 unsigned long sz;
 507                 void *data = load_sha1(sha1, &sz, name);
 508                 if (!data)
 509                         hit = 0;
 510                 else
 511                         hit = grep_buffer(opt, name, data, sz);
 512
 513                 free(data);
 514                 free(name);
 515                 return hit;
 516         }
 517 }
 518
 519 static void *load_file(const char *filename, size_t *sz)
 520 {
 521         struct stat st;
 522         char *data;
 523         int i;
 524
 525         if (lstat(filename, &st) < 0) {
 526         err_ret:
 527                 if (errno != ENOENT)
 528                         error("'%s': %s", filename, strerror(errno));
 529                 return 0;
 530         }
 531         if (!S_ISREG(st.st_mode))
 532                 return 0;
 533         *sz = xsize_t(st.st_size);
 534         i = open(filename, O_RDONLY);
 535         if (i < 0)
 536                 goto err_ret;
 537         data = xmalloc(*sz + 1);
 538         if (st.st_size != read_in_full(i, data, *sz)) {
 539                 error("'%s': short read %s", filename, strerror(errno));
 540                 close(i);
 541                 free(data);
 542                 return 0;
 543         }
 544         close(i);
 545         data[*sz] = 0;
 546         return data;
 547 }
 548
 549 static int grep_file(struct grep_opt *opt, const char *filename)
 550 {
 551         struct strbuf buf = STRBUF_INIT;
 552         char *name;
 553
 554         if (opt->relative && opt->prefix_length)
 555                 quote_path_relative(filename, -1, &buf, opt->prefix);
 556         else
 557                 strbuf_addstr(&buf, filename);
 558         name = strbuf_detach(&buf, NULL);
 559
 560 #ifndef NO_PTHREADS
 561         if (use_threads) {
 562                 grep_file_async(opt, name, filename);
 563                 return 0;
 564         } else
 565 #endif
 566         {
 567                 int hit;
 568                 size_t sz;
 569                 void *data = load_file(filename, &sz);
 570                 if (!data)
 571                         hit = 0;
 572                 else
 573                         hit = grep_buffer(opt, name, data, sz);
 574
 575                 free(data);
 576                 free(name);
 577                 return hit;
 578         }
 579 }
 580
 581 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 582 {
 583         struct string_list *path_list = opt->output_priv;
 584
 585         if (len == 1 && *(const char *)data == '\0')
 586                 return;
 587         string_list_append(path_list, xstrndup(data, len));
 588 }
 589
 590 static void run_pager(struct grep_opt *opt, const char *prefix)
 591 {
 592         struct string_list *path_list = opt->output_priv;
 593         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 594         int i, status;
 595
 596         for (i = 0; i < path_list->nr; i++)
 597                 argv[i] = path_list->items[i].string;
 598         argv[path_list->nr] = NULL;
 599
 600         if (prefix && chdir(prefix))
 601                 die("Failed to chdir: %s", prefix);
 602         status = run_command_v_opt(argv, RUN_USING_SHELL);
 603         if (status)
 604                 exit(status);
 605         free(argv);
 606 }
 607
 608 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 609 {
 610         int hit = 0;
 611         int nr;
 612         read_cache();
 613
 614         for (nr = 0; nr < active_nr; nr++) {
 615                 struct cache_entry *ce = active_cache[nr];
 616                 if (!S_ISREG(ce->ce_mode))
 617                         continue;
 618                 if (!pathspec_matches(paths, ce->name, opt->max_depth))
 619                         continue;
 620                 if (skip_binary(opt, ce->name))
 621                         continue;
 622
 623                 /*
 624                  * If CE_VALID is on, we assume worktree file and its cache entry
 625                  * are identical, even if worktree file has been modified, so use
 626                  * cache version instead
 627                  */
 628                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 629                         if (ce_stage(ce))
 630                                 continue;
 631                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 632                 }
 633                 else
 634                         hit |= grep_file(opt, ce->name);
 635                 if (ce_stage(ce)) {
 636                         do {
 637                                 nr++;
 638                         } while (nr < active_nr &&
 639                                  !strcmp(ce->name, active_cache[nr]->name));
 640                         nr--; /* compensate for loop control */
 641                 }
 642                 if (hit && opt->status_only)
 643                         break;
 644         }
 645         return hit;
 646 }
 647
 648 static int grep_tree(struct grep_opt *opt, const char **paths,
 649                      struct tree_desc *tree,
 650                      const char *tree_name, const char *base)
 651 {
 652         int len;
 653         int hit = 0;
 654         struct name_entry entry;
 655         char *down;
 656         int tn_len = strlen(tree_name);
 657         struct strbuf pathbuf;
 658
 659         strbuf_init(&pathbuf, PATH_MAX + tn_len);
 660
 661         if (tn_len) {
 662                 strbuf_add(&pathbuf, tree_name, tn_len);
 663                 strbuf_addch(&pathbuf, ':');
 664                 tn_len = pathbuf.len;
 665         }
 666         strbuf_addstr(&pathbuf, base);
 667         len = pathbuf.len;
 668
 669         while (tree_entry(tree, &entry)) {
 670                 int te_len = tree_entry_len(entry.path, entry.sha1);
 671                 pathbuf.len = len;
 672                 strbuf_add(&pathbuf, entry.path, te_len);
 673
 674                 if (S_ISDIR(entry.mode))
 675                         /* Match "abc/" against pathspec to
 676                          * decide if we want to descend into "abc"
 677                          * directory.
 678                          */
 679                         strbuf_addch(&pathbuf, '/');
 680
 681                 down = pathbuf.buf + tn_len;
 682                 if (!pathspec_matches(paths, down, opt->max_depth))
 683                         ;
 684                 else if (S_ISREG(entry.mode))
 685                         hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
 686                 else if (S_ISDIR(entry.mode)) {
 687                         enum object_type type;
 688                         struct tree_desc sub;
 689                         void *data;
 690                         unsigned long size;
 691
 692                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 693                         if (!data)
 694                                 die("unable to read tree (%s)",
 695                                     sha1_to_hex(entry.sha1));
 696                         init_tree_desc(&sub, data, size);
 697                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 698                         free(data);
 699                 }
 700                 if (hit && opt->status_only)
 701                         break;
 702         }
 703         strbuf_release(&pathbuf);
 704         return hit;
 705 }
 706
 707 static int grep_object(struct grep_opt *opt, const char **paths,
 708                        struct object *obj, const char *name)
 709 {
 710         if (obj->type == OBJ_BLOB)
 711                 return grep_sha1(opt, obj->sha1, name, 0);
 712         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 713                 struct tree_desc tree;
 714                 void *data;
 715                 unsigned long size;
 716                 int hit;
 717                 data = read_object_with_reference(obj->sha1, tree_type,
 718                                                   &size, NULL);
 719                 if (!data)
 720                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 721                 init_tree_desc(&tree, data, size);
 722                 hit = grep_tree(opt, paths, &tree, name, "");
 723                 free(data);
 724                 return hit;
 725         }
 726         die("unable to grep from object of type %s", typename(obj->type));
 727 }
 728
 729 static int grep_objects(struct grep_opt *opt, const char **paths,
 730                         const struct object_array *list)
 731 {
 732         unsigned int i;
 733         int hit = 0;
 734         const unsigned int nr = list->nr;
 735
 736         for (i = 0; i < nr; i++) {
 737                 struct object *real_obj;
 738                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 739                 if (grep_object(opt, paths, real_obj, list->objects[i].name)) {
 740                         hit = 1;
 741                         if (opt->status_only)
 742                                 break;
 743                 }
 744         }
 745         return hit;
 746 }
 747
 748 static int grep_directory(struct grep_opt *opt, const char **paths)
 749 {
 750         struct dir_struct dir;
 751         int i, hit = 0;
 752
 753         memset(&dir, 0, sizeof(dir));
 754         setup_standard_excludes(&dir);
 755
 756         fill_directory(&dir, paths);
 757         for (i = 0; i < dir.nr; i++) {
 758                 hit |= grep_file(opt, dir.entries[i]->name);
 759                 if (hit && opt->status_only)
 760                         break;
 761         }
 762         return hit;
 763 }
 764
 765 static int context_callback(const struct option *opt, const char *arg,
 766                             int unset)
 767 {
 768         struct grep_opt *grep_opt = opt->value;
 769         int value;
 770         const char *endp;
 771
 772         if (unset) {
 773                 grep_opt->pre_context = grep_opt->post_context = 0;
 774                 return 0;
 775         }
 776         value = strtol(arg, (char **)&endp, 10);
 777         if (*endp) {
 778                 return error("switch `%c' expects a numerical value",
 779                              opt->short_name);
 780         }
 781         grep_opt->pre_context = grep_opt->post_context = value;
 782         return 0;
 783 }
 784
 785 static int file_callback(const struct option *opt, const char *arg, int unset)
 786 {
 787         struct grep_opt *grep_opt = opt->value;
 788         FILE *patterns;
 789         int lno = 0;
 790         struct strbuf sb = STRBUF_INIT;
 791
 792         patterns = fopen(arg, "r");
 793         if (!patterns)
 794                 die_errno("cannot open '%s'", arg);
 795         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 796                 char *s;
 797                 size_t len;
 798
 799                 /* ignore empty line like grep does */
 800                 if (sb.len == 0)
 801                         continue;
 802
 803                 s = strbuf_detach(&sb, &len);
 804                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 805         }
 806         fclose(patterns);
 807         strbuf_release(&sb);
 808         return 0;
 809 }
 810
 811 static int not_callback(const struct option *opt, const char *arg, int unset)
 812 {
 813         struct grep_opt *grep_opt = opt->value;
 814         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 815         return 0;
 816 }
 817
 818 static int and_callback(const struct option *opt, const char *arg, int unset)
 819 {
 820         struct grep_opt *grep_opt = opt->value;
 821         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 822         return 0;
 823 }
 824
 825 static int open_callback(const struct option *opt, const char *arg, int unset)
 826 {
 827         struct grep_opt *grep_opt = opt->value;
 828         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 829         return 0;
 830 }
 831
 832 static int close_callback(const struct option *opt, const char *arg, int unset)
 833 {
 834         struct grep_opt *grep_opt = opt->value;
 835         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 836         return 0;
 837 }
 838
 839 static int pattern_callback(const struct option *opt, const char *arg,
 840                             int unset)
 841 {
 842         struct grep_opt *grep_opt = opt->value;
 843         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 844         return 0;
 845 }
 846
 847 static int help_callback(const struct option *opt, const char *arg, int unset)
 848 {
 849         return -1;
 850 }
 851
 852 int cmd_grep(int argc, const char **argv, const char *prefix)
 853 {
 854         int hit = 0;
 855         int cached = 0;
 856         int seen_dashdash = 0;
 857         int external_grep_allowed__ignored;
 858         const char *show_in_pager = NULL, *default_pager = "dummy";
 859         struct grep_opt opt;
 860         struct object_array list = OBJECT_ARRAY_INIT;
 861         const char **paths = NULL;
 862         struct string_list path_list = STRING_LIST_INIT_NODUP;
 863         int i;
 864         int dummy;
 865         int use_index = 1;
 866         struct option options[] = {
 867                 OPT_BOOLEAN(0, "cached", &cached,
 868                         "search in index instead of in the work tree"),
 869                 OPT_BOOLEAN(0, "index", &use_index,
 870                         "--no-index finds in contents not managed by git"),
 871                 OPT_GROUP(""),
 872                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 873                         "show non-matching lines"),
 874                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 875                         "case insensitive matching"),
 876                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 877                         "match patterns only at word boundaries"),
 878                 OPT_SET_INT('a', "text", &opt.binary,
 879                         "process binary files as text", GREP_BINARY_TEXT),
 880                 OPT_SET_INT('I', NULL, &opt.binary,
 881                         "don't match patterns in binary files",
 882                         GREP_BINARY_NOMATCH),
 883                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 884                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 885                         NULL, 1 },
 886                 OPT_GROUP(""),
 887                 OPT_BIT('E', "extended-regexp", &opt.regflags,
 888                         "use extended POSIX regular expressions", REG_EXTENDED),
 889                 OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
 890                         "use basic POSIX regular expressions (default)",
 891                         REG_EXTENDED),
 892                 OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
 893                         "interpret patterns as fixed strings"),
 894                 OPT_GROUP(""),
 895                 OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
 896                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 897                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 898                 OPT_NEGBIT(0, "full-name", &opt.relative,
 899                         "show filenames relative to top directory", 1),
 900                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 901                         "show only filenames instead of matching lines"),
 902                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 903                         "synonym for --files-with-matches"),
 904                 OPT_BOOLEAN('L', "files-without-match",
 905                         &opt.unmatch_name_only,
 906                         "show only the names of files without match"),
 907                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 908                         "print NUL after filenames"),
 909                 OPT_BOOLEAN('c', "count", &opt.count,
 910                         "show the number of matches instead of matching lines"),
 911                 OPT__COLOR(&opt.color, "highlight matches"),
 912                 OPT_GROUP(""),
 913                 OPT_CALLBACK('C', NULL, &opt, "n",
 914                         "show <n> context lines before and after matches",
 915                         context_callback),
 916                 OPT_INTEGER('B', NULL, &opt.pre_context,
 917                         "show <n> context lines before matches"),
 918                 OPT_INTEGER('A', NULL, &opt.post_context,
 919                         "show <n> context lines after matches"),
 920                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 921                         context_callback),
 922                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 923                         "show a line with the function name before matches"),
 924                 OPT_GROUP(""),
 925                 OPT_CALLBACK('f', NULL, &opt, "file",
 926                         "read patterns from file", file_callback),
 927                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 928                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 929                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 930                   "combine patterns specified with -e",
 931                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 932                 OPT_BOOLEAN(0, "or", &dummy, ""),
 933                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 934                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 935                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 936                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 937                   open_callback },
 938                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 939                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 940                   close_callback },
 941                 OPT_BOOLEAN('q', "quiet", &opt.status_only,
 942                             "indicate hit with exit status without output"),
 943                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 944                         "show only matches from files that match all patterns"),
 945                 OPT_GROUP(""),
 946                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 947                         "pager", "show matching files in the pager",
 948                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 949                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 950                             "allow calling of grep(1) (ignored by this build)"),
 951                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 952                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 953                 OPT_END()
 954         };
 955
 956         /*
 957          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 958          * to show usage information and exit.
 959          */
 960         if (argc == 2 && !strcmp(argv[1], "-h"))
 961                 usage_with_options(grep_usage, options);
 962
 963         memset(&opt, 0, sizeof(opt));
 964         opt.prefix = prefix;
 965         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 966         opt.relative = 1;
 967         opt.pathname = 1;
 968         opt.pattern_tail = &opt.pattern_list;
 969         opt.header_tail = &opt.header_list;
 970         opt.regflags = REG_NEWLINE;
 971         opt.max_depth = -1;
 972
 973         strcpy(opt.color_context, "");
 974         strcpy(opt.color_filename, "");
 975         strcpy(opt.color_function, "");
 976         strcpy(opt.color_lineno, "");
 977         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 978         strcpy(opt.color_selected, "");
 979         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 980         opt.color = -1;
 981         git_config(grep_config, &opt);
 982         if (opt.color == -1)
 983                 opt.color = git_use_color_default;
 984
 985         /*
 986          * If there is no -- then the paths must exist in the working
 987          * tree.  If there is no explicit pattern specified with -e or
 988          * -f, we take the first unrecognized non option to be the
 989          * pattern, but then what follows it must be zero or more
 990          * valid refs up to the -- (if exists), and then existing
 991          * paths.  If there is an explicit pattern, then the first
 992          * unrecognized non option is the beginning of the refs list
 993          * that continues up to the -- (if exists), and then paths.
 994          */
 995         argc = parse_options(argc, argv, prefix, options, grep_usage,
 996                              PARSE_OPT_KEEP_DASHDASH |
 997                              PARSE_OPT_STOP_AT_NON_OPTION |
 998                              PARSE_OPT_NO_INTERNAL_HELP);
 999
1000         if (use_index && !startup_info->have_repository)
1001                 /* die the same way as if we did it at the beginning */
1002                 setup_git_directory();
1003
1004         /*
1005          * skip a -- separator; we know it cannot be
1006          * separating revisions from pathnames if
1007          * we haven't even had any patterns yet
1008          */
1009         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1010                 argv++;
1011                 argc--;
1012         }
1013
1014         /* First unrecognized non-option token */
1015         if (argc > 0 && !opt.pattern_list) {
1016                 append_grep_pattern(&opt, argv[0], "command line", 0,
1017                                     GREP_PATTERN);
1018                 argv++;
1019                 argc--;
1020         }
1021
1022         if (show_in_pager == default_pager)
1023                 show_in_pager = git_pager(1);
1024         if (show_in_pager) {
1025                 opt.color = 0;
1026                 opt.name_only = 1;
1027                 opt.null_following_name = 1;
1028                 opt.output_priv = &path_list;
1029                 opt.output = append_path;
1030                 string_list_append(&path_list, show_in_pager);
1031                 use_threads = 0;
1032         }
1033         if ((opt.binary & GREP_BINARY_NOMATCH))
1034                 use_threads = 0;
1035
1036         if (!opt.pattern_list)
1037                 die("no pattern given.");
1038         if (!opt.fixed && opt.ignore_case)
1039                 opt.regflags |= REG_ICASE;
1040         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
1041                 die("cannot mix --fixed-strings and regexp");
1042
1043 #ifndef NO_PTHREADS
1044         if (online_cpus() == 1 || !grep_threads_ok(&opt))
1045                 use_threads = 0;
1046
1047         if (use_threads) {
1048                 if (opt.pre_context || opt.post_context)
1049                         print_hunk_marks_between_files = 1;
1050                 start_threads(&opt);
1051         }
1052 #else
1053         use_threads = 0;
1054 #endif
1055
1056         compile_grep_patterns(&opt);
1057
1058         /* Check revs and then paths */
1059         for (i = 0; i < argc; i++) {
1060                 const char *arg = argv[i];
1061                 unsigned char sha1[20];
1062                 /* Is it a rev? */
1063                 if (!get_sha1(arg, sha1)) {
1064                         struct object *object = parse_object(sha1);
1065                         if (!object)
1066                                 die("bad object %s", arg);
1067                         add_object_array(object, arg, &list);
1068                         continue;
1069                 }
1070                 if (!strcmp(arg, "--")) {
1071                         i++;
1072                         seen_dashdash = 1;
1073                 }
1074                 break;
1075         }
1076
1077         /* The rest are paths */
1078         if (!seen_dashdash) {
1079                 int j;
1080                 for (j = i; j < argc; j++)
1081                         verify_filename(prefix, argv[j]);
1082         }
1083
1084         if (i < argc)
1085                 paths = get_pathspec(prefix, argv + i);
1086         else if (prefix) {
1087                 paths = xcalloc(2, sizeof(const char *));
1088                 paths[0] = prefix;
1089                 paths[1] = NULL;
1090         }
1091
1092         if (show_in_pager && (cached || list.nr))
1093                 die("--open-files-in-pager only works on the worktree");
1094
1095         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1096                 const char *pager = path_list.items[0].string;
1097                 int len = strlen(pager);
1098
1099                 if (len > 4 && is_dir_sep(pager[len - 5]))
1100                         pager += len - 4;
1101
1102                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1103                         struct strbuf buf = STRBUF_INIT;
1104                         strbuf_addf(&buf, "+/%s%s",
1105                                         strcmp("less", pager) ? "" : "*",
1106                                         opt.pattern_list->pattern);
1107                         string_list_append(&path_list, buf.buf);
1108                         strbuf_detach(&buf, NULL);
1109                 }
1110         }
1111
1112         if (!show_in_pager)
1113                 setup_pager();
1114
1115
1116         if (!use_index) {
1117                 if (cached)
1118                         die("--cached cannot be used with --no-index.");
1119                 if (list.nr)
1120                         die("--no-index cannot be used with revs.");
1121                 hit = grep_directory(&opt, paths);
1122         } else if (!list.nr) {
1123                 if (!cached)
1124                         setup_work_tree();
1125
1126                 hit = grep_cache(&opt, paths, cached);
1127         } else {
1128                 if (cached)
1129                         die("both --cached and trees are given.");
1130                 hit = grep_objects(&opt, paths, &list);
1131         }
1132
1133         if (use_threads)
1134                 hit |= wait_all();
1135         if (hit && show_in_pager)
1136                 run_pager(&opt, prefix);
1137         free_grep_patterns(&opt);
1138         return !hit;
1139 }