builtin / grep.con commit Unify code paths of threaded greps (685359c)
   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
  19#ifndef NO_PTHREADS
  20#include <pthread.h>
  21#include "thread-utils.h"
  22#endif
  23
  24static char const * const grep_usage[] = {
  25        "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
  26        NULL
  27};
  28
  29static int use_threads = 1;
  30
  31#ifndef NO_PTHREADS
  32#define THREADS 8
  33static pthread_t threads[THREADS];
  34
  35static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  36                       const char *name);
  37static void *load_file(const char *filename, size_t *sz);
  38
  39enum work_type {WORK_SHA1, WORK_FILE};
  40
  41/* We use one producer thread and THREADS consumer
  42 * threads. The producer adds struct work_items to 'todo' and the
  43 * consumers pick work items from the same array.
  44 */
  45struct work_item
  46{
  47        enum work_type type;
  48        char *name;
  49
  50        /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  51         * otherwise type == WORK_FILE, and 'identifier' is a NUL
  52         * terminated filename.
  53         */
  54        void *identifier;
  55        char done;
  56        struct strbuf out;
  57};
  58
  59/* In the range [todo_done, todo_start) in 'todo' we have work_items
  60 * that have been or are processed by a consumer thread. We haven't
  61 * written the result for these to stdout yet.
  62 *
  63 * The work_items in [todo_start, todo_end) are waiting to be picked
  64 * up by a consumer thread.
  65 *
  66 * The ranges are modulo TODO_SIZE.
  67 */
  68#define TODO_SIZE 128
  69static struct work_item todo[TODO_SIZE];
  70static int todo_start;
  71static int todo_end;
  72static int todo_done;
  73
  74/* Has all work items been added? */
  75static int all_work_added;
  76
  77/* This lock protects all the variables above. */
  78static pthread_mutex_t grep_mutex;
  79
  80/* Used to serialize calls to read_sha1_file. */
  81static pthread_mutex_t read_sha1_mutex;
  82
  83#define grep_lock() pthread_mutex_lock(&grep_mutex)
  84#define grep_unlock() pthread_mutex_unlock(&grep_mutex)
  85#define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
  86#define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
  87
  88/* Signalled when a new work_item is added to todo. */
  89static pthread_cond_t cond_add;
  90
  91/* Signalled when the result from one work_item is written to
  92 * stdout.
  93 */
  94static pthread_cond_t cond_write;
  95
  96/* Signalled when we are finished with everything. */
  97static pthread_cond_t cond_result;
  98
  99static int print_hunk_marks_between_files;
 100static int printed_something;
 101
 102static void add_work(enum work_type type, char *name, void *id)
 103{
 104        grep_lock();
 105
 106        while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 107                pthread_cond_wait(&cond_write, &grep_mutex);
 108        }
 109
 110        todo[todo_end].type = type;
 111        todo[todo_end].name = name;
 112        todo[todo_end].identifier = id;
 113        todo[todo_end].done = 0;
 114        strbuf_reset(&todo[todo_end].out);
 115        todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 116
 117        pthread_cond_signal(&cond_add);
 118        grep_unlock();
 119}
 120
 121static struct work_item *get_work(void)
 122{
 123        struct work_item *ret;
 124
 125        grep_lock();
 126        while (todo_start == todo_end && !all_work_added) {
 127                pthread_cond_wait(&cond_add, &grep_mutex);
 128        }
 129
 130        if (todo_start == todo_end && all_work_added) {
 131                ret = NULL;
 132        } else {
 133                ret = &todo[todo_start];
 134                todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 135        }
 136        grep_unlock();
 137        return ret;
 138}
 139
 140static void grep_sha1_async(struct grep_opt *opt, char *name,
 141                            const unsigned char *sha1)
 142{
 143        unsigned char *s;
 144        s = xmalloc(20);
 145        memcpy(s, sha1, 20);
 146        add_work(WORK_SHA1, name, s);
 147}
 148
 149static void grep_file_async(struct grep_opt *opt, char *name,
 150                            const char *filename)
 151{
 152        add_work(WORK_FILE, name, xstrdup(filename));
 153}
 154
 155static void work_done(struct work_item *w)
 156{
 157        int old_done;
 158
 159        grep_lock();
 160        w->done = 1;
 161        old_done = todo_done;
 162        for(; todo[todo_done].done && todo_done != todo_start;
 163            todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 164                w = &todo[todo_done];
 165                if (w->out.len) {
 166                        if (print_hunk_marks_between_files && printed_something)
 167                                write_or_die(1, "--\n", 3);
 168                        write_or_die(1, w->out.buf, w->out.len);
 169                        printed_something = 1;
 170                }
 171                free(w->name);
 172                free(w->identifier);
 173        }
 174
 175        if (old_done != todo_done)
 176                pthread_cond_signal(&cond_write);
 177
 178        if (all_work_added && todo_done == todo_end)
 179                pthread_cond_signal(&cond_result);
 180
 181        grep_unlock();
 182}
 183
 184static void *run(void *arg)
 185{
 186        int hit = 0;
 187        struct grep_opt *opt = arg;
 188
 189        while (1) {
 190                struct work_item *w = get_work();
 191                if (!w)
 192                        break;
 193
 194                opt->output_priv = w;
 195                if (w->type == WORK_SHA1) {
 196                        unsigned long sz;
 197                        void* data = load_sha1(w->identifier, &sz, w->name);
 198
 199                        if (data) {
 200                                hit |= grep_buffer(opt, w->name, data, sz);
 201                                free(data);
 202                        }
 203                } else if (w->type == WORK_FILE) {
 204                        size_t sz;
 205                        void* data = load_file(w->identifier, &sz);
 206                        if (data) {
 207                                hit |= grep_buffer(opt, w->name, data, sz);
 208                                free(data);
 209                        }
 210                } else {
 211                        assert(0);
 212                }
 213
 214                work_done(w);
 215        }
 216        free_grep_patterns(arg);
 217        free(arg);
 218
 219        return (void*) (intptr_t) hit;
 220}
 221
 222static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 223{
 224        struct work_item *w = opt->output_priv;
 225        strbuf_add(&w->out, buf, size);
 226}
 227
 228static void start_threads(struct grep_opt *opt)
 229{
 230        int i;
 231
 232        pthread_mutex_init(&grep_mutex, NULL);
 233        pthread_mutex_init(&read_sha1_mutex, NULL);
 234        pthread_cond_init(&cond_add, NULL);
 235        pthread_cond_init(&cond_write, NULL);
 236        pthread_cond_init(&cond_result, NULL);
 237
 238        for (i = 0; i < ARRAY_SIZE(todo); i++) {
 239                strbuf_init(&todo[i].out, 0);
 240        }
 241
 242        for (i = 0; i < ARRAY_SIZE(threads); i++) {
 243                int err;
 244                struct grep_opt *o = grep_opt_dup(opt);
 245                o->output = strbuf_out;
 246                compile_grep_patterns(o);
 247                err = pthread_create(&threads[i], NULL, run, o);
 248
 249                if (err)
 250                        die("grep: failed to create thread: %s",
 251                            strerror(err));
 252        }
 253}
 254
 255static int wait_all(void)
 256{
 257        int hit = 0;
 258        int i;
 259
 260        grep_lock();
 261        all_work_added = 1;
 262
 263        /* Wait until all work is done. */
 264        while (todo_done != todo_end)
 265                pthread_cond_wait(&cond_result, &grep_mutex);
 266
 267        /* Wake up all the consumer threads so they can see that there
 268         * is no more work to do.
 269         */
 270        pthread_cond_broadcast(&cond_add);
 271        grep_unlock();
 272
 273        for (i = 0; i < ARRAY_SIZE(threads); i++) {
 274                void *h;
 275                pthread_join(threads[i], &h);
 276                hit |= (int) (intptr_t) h;
 277        }
 278
 279        pthread_mutex_destroy(&grep_mutex);
 280        pthread_mutex_destroy(&read_sha1_mutex);
 281        pthread_cond_destroy(&cond_add);
 282        pthread_cond_destroy(&cond_write);
 283        pthread_cond_destroy(&cond_result);
 284
 285        return hit;
 286}
 287#else /* !NO_PTHREADS */
 288#define read_sha1_lock()
 289#define read_sha1_unlock()
 290
 291static int wait_all(void)
 292{
 293        return 0;
 294}
 295#endif
 296
 297static int grep_config(const char *var, const char *value, void *cb)
 298{
 299        struct grep_opt *opt = cb;
 300        char *color = NULL;
 301
 302        switch (userdiff_config(var, value)) {
 303        case 0: break;
 304        case -1: return -1;
 305        default: return 0;
 306        }
 307
 308        if (!strcmp(var, "color.grep"))
 309                opt->color = git_config_colorbool(var, value, -1);
 310        else if (!strcmp(var, "color.grep.context"))
 311                color = opt->color_context;
 312        else if (!strcmp(var, "color.grep.filename"))
 313                color = opt->color_filename;
 314        else if (!strcmp(var, "color.grep.function"))
 315                color = opt->color_function;
 316        else if (!strcmp(var, "color.grep.linenumber"))
 317                color = opt->color_lineno;
 318        else if (!strcmp(var, "color.grep.match"))
 319                color = opt->color_match;
 320        else if (!strcmp(var, "color.grep.selected"))
 321                color = opt->color_selected;
 322        else if (!strcmp(var, "color.grep.separator"))
 323                color = opt->color_sep;
 324        else
 325                return git_color_default_config(var, value, cb);
 326        if (color) {
 327                if (!value)
 328                        return config_error_nonbool(var);
 329                color_parse(value, var, color);
 330        }
 331        return 0;
 332}
 333
 334/*
 335 * Return non-zero if max_depth is negative or path has no more then max_depth
 336 * slashes.
 337 */
 338static int accept_subdir(const char *path, int max_depth)
 339{
 340        if (max_depth < 0)
 341                return 1;
 342
 343        while ((path = strchr(path, '/')) != NULL) {
 344                max_depth--;
 345                if (max_depth < 0)
 346                        return 0;
 347                path++;
 348        }
 349        return 1;
 350}
 351
 352/*
 353 * Return non-zero if name is a subdirectory of match and is not too deep.
 354 */
 355static int is_subdir(const char *name, int namelen,
 356                const char *match, int matchlen, int max_depth)
 357{
 358        if (matchlen > namelen || strncmp(name, match, matchlen))
 359                return 0;
 360
 361        if (name[matchlen] == '\0') /* exact match */
 362                return 1;
 363
 364        if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
 365                return accept_subdir(name + matchlen + 1, max_depth);
 366
 367        return 0;
 368}
 369
 370/*
 371 * git grep pathspecs are somewhat different from diff-tree pathspecs;
 372 * pathname wildcards are allowed.
 373 */
 374static int pathspec_matches(const char **paths, const char *name, int max_depth)
 375{
 376        int namelen, i;
 377        if (!paths || !*paths)
 378                return accept_subdir(name, max_depth);
 379        namelen = strlen(name);
 380        for (i = 0; paths[i]; i++) {
 381                const char *match = paths[i];
 382                int matchlen = strlen(match);
 383                const char *cp, *meta;
 384
 385                if (is_subdir(name, namelen, match, matchlen, max_depth))
 386                        return 1;
 387                if (!fnmatch(match, name, 0))
 388                        return 1;
 389                if (name[namelen-1] != '/')
 390                        continue;
 391
 392                /* We are being asked if the directory ("name") is worth
 393                 * descending into.
 394                 *
 395                 * Find the longest leading directory name that does
 396                 * not have metacharacter in the pathspec; the name
 397                 * we are looking at must overlap with that directory.
 398                 */
 399                for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
 400                        char ch = *cp;
 401                        if (ch == '*' || ch == '[' || ch == '?') {
 402                                meta = cp;
 403                                break;
 404                        }
 405                }
 406                if (!meta)
 407                        meta = cp; /* fully literal */
 408
 409                if (namelen <= meta - match) {
 410                        /* Looking at "Documentation/" and
 411                         * the pattern says "Documentation/howto/", or
 412                         * "Documentation/diff*.txt".  The name we
 413                         * have should match prefix.
 414                         */
 415                        if (!memcmp(match, name, namelen))
 416                                return 1;
 417                        continue;
 418                }
 419
 420                if (meta - match < namelen) {
 421                        /* Looking at "Documentation/howto/" and
 422                         * the pattern says "Documentation/h*";
 423                         * match up to "Do.../h"; this avoids descending
 424                         * into "Documentation/technical/".
 425                         */
 426                        if (!memcmp(match, name, meta - match))
 427                                return 1;
 428                        continue;
 429                }
 430        }
 431        return 0;
 432}
 433
 434static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 435{
 436        void *data;
 437
 438        if (use_threads) {
 439                read_sha1_lock();
 440                data = read_sha1_file(sha1, type, size);
 441                read_sha1_unlock();
 442        } else {
 443                data = read_sha1_file(sha1, type, size);
 444        }
 445        return data;
 446}
 447
 448static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 449                       const char *name)
 450{
 451        enum object_type type;
 452        void *data = lock_and_read_sha1_file(sha1, &type, size);
 453
 454        if (!data)
 455                error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 456
 457        return data;
 458}
 459
 460static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 461                     const char *filename, int tree_name_len)
 462{
 463        struct strbuf pathbuf = STRBUF_INIT;
 464        char *name;
 465
 466        if (opt->relative && opt->prefix_length) {
 467                quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 468                                    opt->prefix);
 469                strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 470        } else {
 471                strbuf_addstr(&pathbuf, filename);
 472        }
 473
 474        name = strbuf_detach(&pathbuf, NULL);
 475
 476#ifndef NO_PTHREADS
 477        if (use_threads) {
 478                grep_sha1_async(opt, name, sha1);
 479                return 0;
 480        } else
 481#endif
 482        {
 483                int hit;
 484                unsigned long sz;
 485                void *data = load_sha1(sha1, &sz, name);
 486                if (!data)
 487                        hit = 0;
 488                else
 489                        hit = grep_buffer(opt, name, data, sz);
 490
 491                free(data);
 492                free(name);
 493                return hit;
 494        }
 495}
 496
 497static void *load_file(const char *filename, size_t *sz)
 498{
 499        struct stat st;
 500        char *data;
 501        int i;
 502
 503        if (lstat(filename, &st) < 0) {
 504        err_ret:
 505                if (errno != ENOENT)
 506                        error("'%s': %s", filename, strerror(errno));
 507                return 0;
 508        }
 509        if (!S_ISREG(st.st_mode))
 510                return 0;
 511        *sz = xsize_t(st.st_size);
 512        i = open(filename, O_RDONLY);
 513        if (i < 0)
 514                goto err_ret;
 515        data = xmalloc(*sz + 1);
 516        if (st.st_size != read_in_full(i, data, *sz)) {
 517                error("'%s': short read %s", filename, strerror(errno));
 518                close(i);
 519                free(data);
 520                return 0;
 521        }
 522        close(i);
 523        data[*sz] = 0;
 524        return data;
 525}
 526
 527static int grep_file(struct grep_opt *opt, const char *filename)
 528{
 529        struct strbuf buf = STRBUF_INIT;
 530        char *name;
 531
 532        if (opt->relative && opt->prefix_length)
 533                quote_path_relative(filename, -1, &buf, opt->prefix);
 534        else
 535                strbuf_addstr(&buf, filename);
 536        name = strbuf_detach(&buf, NULL);
 537
 538#ifndef NO_PTHREADS
 539        if (use_threads) {
 540                grep_file_async(opt, name, filename);
 541                return 0;
 542        } else
 543#endif
 544        {
 545                int hit;
 546                size_t sz;
 547                void *data = load_file(filename, &sz);
 548                if (!data)
 549                        hit = 0;
 550                else
 551                        hit = grep_buffer(opt, name, data, sz);
 552
 553                free(data);
 554                free(name);
 555                return hit;
 556        }
 557}
 558
 559static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 560{
 561        int hit = 0;
 562        int nr;
 563        read_cache();
 564
 565        for (nr = 0; nr < active_nr; nr++) {
 566                struct cache_entry *ce = active_cache[nr];
 567                if (!S_ISREG(ce->ce_mode))
 568                        continue;
 569                if (!pathspec_matches(paths, ce->name, opt->max_depth))
 570                        continue;
 571                /*
 572                 * If CE_VALID is on, we assume worktree file and its cache entry
 573                 * are identical, even if worktree file has been modified, so use
 574                 * cache version instead
 575                 */
 576                if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 577                        if (ce_stage(ce))
 578                                continue;
 579                        hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 580                }
 581                else
 582                        hit |= grep_file(opt, ce->name);
 583                if (ce_stage(ce)) {
 584                        do {
 585                                nr++;
 586                        } while (nr < active_nr &&
 587                                 !strcmp(ce->name, active_cache[nr]->name));
 588                        nr--; /* compensate for loop control */
 589                }
 590                if (hit && opt->status_only)
 591                        break;
 592        }
 593        return hit;
 594}
 595
 596static int grep_tree(struct grep_opt *opt, const char **paths,
 597                     struct tree_desc *tree,
 598                     const char *tree_name, const char *base)
 599{
 600        int len;
 601        int hit = 0;
 602        struct name_entry entry;
 603        char *down;
 604        int tn_len = strlen(tree_name);
 605        struct strbuf pathbuf;
 606
 607        strbuf_init(&pathbuf, PATH_MAX + tn_len);
 608
 609        if (tn_len) {
 610                strbuf_add(&pathbuf, tree_name, tn_len);
 611                strbuf_addch(&pathbuf, ':');
 612                tn_len = pathbuf.len;
 613        }
 614        strbuf_addstr(&pathbuf, base);
 615        len = pathbuf.len;
 616
 617        while (tree_entry(tree, &entry)) {
 618                int te_len = tree_entry_len(entry.path, entry.sha1);
 619                pathbuf.len = len;
 620                strbuf_add(&pathbuf, entry.path, te_len);
 621
 622                if (S_ISDIR(entry.mode))
 623                        /* Match "abc/" against pathspec to
 624                         * decide if we want to descend into "abc"
 625                         * directory.
 626                         */
 627                        strbuf_addch(&pathbuf, '/');
 628
 629                down = pathbuf.buf + tn_len;
 630                if (!pathspec_matches(paths, down, opt->max_depth))
 631                        ;
 632                else if (S_ISREG(entry.mode))
 633                        hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
 634                else if (S_ISDIR(entry.mode)) {
 635                        enum object_type type;
 636                        struct tree_desc sub;
 637                        void *data;
 638                        unsigned long size;
 639
 640                        data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 641                        if (!data)
 642                                die("unable to read tree (%s)",
 643                                    sha1_to_hex(entry.sha1));
 644                        init_tree_desc(&sub, data, size);
 645                        hit |= grep_tree(opt, paths, &sub, tree_name, down);
 646                        free(data);
 647                }
 648                if (hit && opt->status_only)
 649                        break;
 650        }
 651        strbuf_release(&pathbuf);
 652        return hit;
 653}
 654
 655static int grep_object(struct grep_opt *opt, const char **paths,
 656                       struct object *obj, const char *name)
 657{
 658        if (obj->type == OBJ_BLOB)
 659                return grep_sha1(opt, obj->sha1, name, 0);
 660        if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 661                struct tree_desc tree;
 662                void *data;
 663                unsigned long size;
 664                int hit;
 665                data = read_object_with_reference(obj->sha1, tree_type,
 666                                                  &size, NULL);
 667                if (!data)
 668                        die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 669                init_tree_desc(&tree, data, size);
 670                hit = grep_tree(opt, paths, &tree, name, "");
 671                free(data);
 672                return hit;
 673        }
 674        die("unable to grep from object of type %s", typename(obj->type));
 675}
 676
 677static int grep_objects(struct grep_opt *opt, const char **paths,
 678                        const struct object_array *list)
 679{
 680        unsigned int i;
 681        int hit = 0;
 682        const unsigned int nr = list->nr;
 683
 684        for (i = 0; i < nr; i++) {
 685                struct object *real_obj;
 686                real_obj = deref_tag(list->objects[i].item, NULL, 0);
 687                if (grep_object(opt, paths, real_obj, list->objects[i].name)) {
 688                        hit = 1;
 689                        if (opt->status_only)
 690                                break;
 691                }
 692        }
 693        return hit;
 694}
 695
 696static int grep_directory(struct grep_opt *opt, const char **paths)
 697{
 698        struct dir_struct dir;
 699        int i, hit = 0;
 700
 701        memset(&dir, 0, sizeof(dir));
 702        setup_standard_excludes(&dir);
 703
 704        fill_directory(&dir, paths);
 705        for (i = 0; i < dir.nr; i++) {
 706                hit |= grep_file(opt, dir.entries[i]->name);
 707                if (hit && opt->status_only)
 708                        break;
 709        }
 710        return hit;
 711}
 712
 713static int context_callback(const struct option *opt, const char *arg,
 714                            int unset)
 715{
 716        struct grep_opt *grep_opt = opt->value;
 717        int value;
 718        const char *endp;
 719
 720        if (unset) {
 721                grep_opt->pre_context = grep_opt->post_context = 0;
 722                return 0;
 723        }
 724        value = strtol(arg, (char **)&endp, 10);
 725        if (*endp) {
 726                return error("switch `%c' expects a numerical value",
 727                             opt->short_name);
 728        }
 729        grep_opt->pre_context = grep_opt->post_context = value;
 730        return 0;
 731}
 732
 733static int file_callback(const struct option *opt, const char *arg, int unset)
 734{
 735        struct grep_opt *grep_opt = opt->value;
 736        FILE *patterns;
 737        int lno = 0;
 738        struct strbuf sb = STRBUF_INIT;
 739
 740        patterns = fopen(arg, "r");
 741        if (!patterns)
 742                die_errno("cannot open '%s'", arg);
 743        while (strbuf_getline(&sb, patterns, '\n') == 0) {
 744                /* ignore empty line like grep does */
 745                if (sb.len == 0)
 746                        continue;
 747                append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg,
 748                                    ++lno, GREP_PATTERN);
 749        }
 750        fclose(patterns);
 751        strbuf_release(&sb);
 752        return 0;
 753}
 754
 755static int not_callback(const struct option *opt, const char *arg, int unset)
 756{
 757        struct grep_opt *grep_opt = opt->value;
 758        append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 759        return 0;
 760}
 761
 762static int and_callback(const struct option *opt, const char *arg, int unset)
 763{
 764        struct grep_opt *grep_opt = opt->value;
 765        append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 766        return 0;
 767}
 768
 769static int open_callback(const struct option *opt, const char *arg, int unset)
 770{
 771        struct grep_opt *grep_opt = opt->value;
 772        append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 773        return 0;
 774}
 775
 776static int close_callback(const struct option *opt, const char *arg, int unset)
 777{
 778        struct grep_opt *grep_opt = opt->value;
 779        append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 780        return 0;
 781}
 782
 783static int pattern_callback(const struct option *opt, const char *arg,
 784                            int unset)
 785{
 786        struct grep_opt *grep_opt = opt->value;
 787        append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 788        return 0;
 789}
 790
 791static int help_callback(const struct option *opt, const char *arg, int unset)
 792{
 793        return -1;
 794}
 795
 796int cmd_grep(int argc, const char **argv, const char *prefix)
 797{
 798        int hit = 0;
 799        int cached = 0;
 800        int seen_dashdash = 0;
 801        int external_grep_allowed__ignored;
 802        struct grep_opt opt;
 803        struct object_array list = { 0, 0, NULL };
 804        const char **paths = NULL;
 805        int i;
 806        int dummy;
 807        int nongit = 0, use_index = 1;
 808        struct option options[] = {
 809                OPT_BOOLEAN(0, "cached", &cached,
 810                        "search in index instead of in the work tree"),
 811                OPT_BOOLEAN(0, "index", &use_index,
 812                        "--no-index finds in contents not managed by git"),
 813                OPT_GROUP(""),
 814                OPT_BOOLEAN('v', "invert-match", &opt.invert,
 815                        "show non-matching lines"),
 816                OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 817                        "case insensitive matching"),
 818                OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 819                        "match patterns only at word boundaries"),
 820                OPT_SET_INT('a', "text", &opt.binary,
 821                        "process binary files as text", GREP_BINARY_TEXT),
 822                OPT_SET_INT('I', NULL, &opt.binary,
 823                        "don't match patterns in binary files",
 824                        GREP_BINARY_NOMATCH),
 825                { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 826                        "descend at most <depth> levels", PARSE_OPT_NONEG,
 827                        NULL, 1 },
 828                OPT_GROUP(""),
 829                OPT_BIT('E', "extended-regexp", &opt.regflags,
 830                        "use extended POSIX regular expressions", REG_EXTENDED),
 831                OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
 832                        "use basic POSIX regular expressions (default)",
 833                        REG_EXTENDED),
 834                OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
 835                        "interpret patterns as fixed strings"),
 836                OPT_GROUP(""),
 837                OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
 838                OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 839                OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 840                OPT_NEGBIT(0, "full-name", &opt.relative,
 841                        "show filenames relative to top directory", 1),
 842                OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 843                        "show only filenames instead of matching lines"),
 844                OPT_BOOLEAN(0, "name-only", &opt.name_only,
 845                        "synonym for --files-with-matches"),
 846                OPT_BOOLEAN('L', "files-without-match",
 847                        &opt.unmatch_name_only,
 848                        "show only the names of files without match"),
 849                OPT_BOOLEAN('z', "null", &opt.null_following_name,
 850                        "print NUL after filenames"),
 851                OPT_BOOLEAN('c', "count", &opt.count,
 852                        "show the number of matches instead of matching lines"),
 853                OPT__COLOR(&opt.color, "highlight matches"),
 854                OPT_GROUP(""),
 855                OPT_CALLBACK('C', NULL, &opt, "n",
 856                        "show <n> context lines before and after matches",
 857                        context_callback),
 858                OPT_INTEGER('B', NULL, &opt.pre_context,
 859                        "show <n> context lines before matches"),
 860                OPT_INTEGER('A', NULL, &opt.post_context,
 861                        "show <n> context lines after matches"),
 862                OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 863                        context_callback),
 864                OPT_BOOLEAN('p', "show-function", &opt.funcname,
 865                        "show a line with the function name before matches"),
 866                OPT_GROUP(""),
 867                OPT_CALLBACK('f', NULL, &opt, "file",
 868                        "read patterns from file", file_callback),
 869                { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 870                        "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 871                { OPTION_CALLBACK, 0, "and", &opt, NULL,
 872                  "combine patterns specified with -e",
 873                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 874                OPT_BOOLEAN(0, "or", &dummy, ""),
 875                { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 876                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 877                { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 878                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 879                  open_callback },
 880                { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 881                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 882                  close_callback },
 883                OPT_BOOLEAN('q', "quiet", &opt.status_only,
 884                            "indicate hit with exit status without output"),
 885                OPT_BOOLEAN(0, "all-match", &opt.all_match,
 886                        "show only matches from files that match all patterns"),
 887                OPT_GROUP(""),
 888                OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 889                            "allow calling of grep(1) (ignored by this build)"),
 890                { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 891                  PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 892                OPT_END()
 893        };
 894
 895        prefix = setup_git_directory_gently(&nongit);
 896
 897        /*
 898         * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 899         * to show usage information and exit.
 900         */
 901        if (argc == 2 && !strcmp(argv[1], "-h"))
 902                usage_with_options(grep_usage, options);
 903
 904        memset(&opt, 0, sizeof(opt));
 905        opt.prefix = prefix;
 906        opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 907        opt.relative = 1;
 908        opt.pathname = 1;
 909        opt.pattern_tail = &opt.pattern_list;
 910        opt.header_tail = &opt.header_list;
 911        opt.regflags = REG_NEWLINE;
 912        opt.max_depth = -1;
 913
 914        strcpy(opt.color_context, "");
 915        strcpy(opt.color_filename, "");
 916        strcpy(opt.color_function, "");
 917        strcpy(opt.color_lineno, "");
 918        strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 919        strcpy(opt.color_selected, "");
 920        strcpy(opt.color_sep, GIT_COLOR_CYAN);
 921        opt.color = -1;
 922        git_config(grep_config, &opt);
 923        if (opt.color == -1)
 924                opt.color = git_use_color_default;
 925
 926        /*
 927         * If there is no -- then the paths must exist in the working
 928         * tree.  If there is no explicit pattern specified with -e or
 929         * -f, we take the first unrecognized non option to be the
 930         * pattern, but then what follows it must be zero or more
 931         * valid refs up to the -- (if exists), and then existing
 932         * paths.  If there is an explicit pattern, then the first
 933         * unrecognized non option is the beginning of the refs list
 934         * that continues up to the -- (if exists), and then paths.
 935         */
 936        argc = parse_options(argc, argv, prefix, options, grep_usage,
 937                             PARSE_OPT_KEEP_DASHDASH |
 938                             PARSE_OPT_STOP_AT_NON_OPTION |
 939                             PARSE_OPT_NO_INTERNAL_HELP);
 940
 941        if (use_index && nongit)
 942                /* die the same way as if we did it at the beginning */
 943                setup_git_directory();
 944
 945        /*
 946         * skip a -- separator; we know it cannot be
 947         * separating revisions from pathnames if
 948         * we haven't even had any patterns yet
 949         */
 950        if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 951                argv++;
 952                argc--;
 953        }
 954
 955        /* First unrecognized non-option token */
 956        if (argc > 0 && !opt.pattern_list) {
 957                append_grep_pattern(&opt, argv[0], "command line", 0,
 958                                    GREP_PATTERN);
 959                argv++;
 960                argc--;
 961        }
 962
 963        if (!opt.pattern_list)
 964                die("no pattern given.");
 965        if (!opt.fixed && opt.ignore_case)
 966                opt.regflags |= REG_ICASE;
 967        if ((opt.regflags != REG_NEWLINE) && opt.fixed)
 968                die("cannot mix --fixed-strings and regexp");
 969
 970#ifndef NO_PTHREADS
 971        if (online_cpus() == 1 || !grep_threads_ok(&opt))
 972                use_threads = 0;
 973
 974        if (use_threads) {
 975                if (opt.pre_context || opt.post_context)
 976                        print_hunk_marks_between_files = 1;
 977                start_threads(&opt);
 978        }
 979#else
 980        use_threads = 0;
 981#endif
 982
 983        compile_grep_patterns(&opt);
 984
 985        /* Check revs and then paths */
 986        for (i = 0; i < argc; i++) {
 987                const char *arg = argv[i];
 988                unsigned char sha1[20];
 989                /* Is it a rev? */
 990                if (!get_sha1(arg, sha1)) {
 991                        struct object *object = parse_object(sha1);
 992                        if (!object)
 993                                die("bad object %s", arg);
 994                        add_object_array(object, arg, &list);
 995                        continue;
 996                }
 997                if (!strcmp(arg, "--")) {
 998                        i++;
 999                        seen_dashdash = 1;
1000                }
1001                break;
1002        }
1003
1004        /* The rest are paths */
1005        if (!seen_dashdash) {
1006                int j;
1007                for (j = i; j < argc; j++)
1008                        verify_filename(prefix, argv[j]);
1009        }
1010
1011        if (i < argc)
1012                paths = get_pathspec(prefix, argv + i);
1013        else if (prefix) {
1014                paths = xcalloc(2, sizeof(const char *));
1015                paths[0] = prefix;
1016                paths[1] = NULL;
1017        }
1018
1019        if (!use_index) {
1020                if (cached)
1021                        die("--cached cannot be used with --no-index.");
1022                if (list.nr)
1023                        die("--no-index cannot be used with revs.");
1024                hit = grep_directory(&opt, paths);
1025        } else if (!list.nr) {
1026                if (!cached)
1027                        setup_work_tree();
1028
1029                hit = grep_cache(&opt, paths, cached);
1030        } else {
1031                if (cached)
1032                        die("both --cached and trees are given.");
1033                hit = grep_objects(&opt, paths, &list);
1034        }
1035
1036        if (use_threads)
1037                hit |= wait_all();
1038        free_grep_patterns(&opt);
1039        return !hit;
1040}