builtin/grep.c

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