builtin / grep.con commit Merge branch 'jn/maint-sequencer-fixes' into maint (770dd00)
   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
  22static char const * const grep_usage[] = {
  23        "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
  24        NULL
  25};
  26
  27static int use_threads = 1;
  28
  29#ifndef NO_PTHREADS
  30#define THREADS 8
  31static pthread_t threads[THREADS];
  32
  33static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  34                       const char *name);
  35static void *load_file(const char *filename, size_t *sz);
  36
  37enum 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 */
  43struct 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
  66static struct work_item todo[TODO_SIZE];
  67static int todo_start;
  68static int todo_end;
  69static int todo_done;
  70
  71/* Has all work items been added? */
  72static int all_work_added;
  73
  74/* This lock protects all the variables above. */
  75static pthread_mutex_t grep_mutex;
  76
  77static inline void grep_lock(void)
  78{
  79        if (use_threads)
  80                pthread_mutex_lock(&grep_mutex);
  81}
  82
  83static inline void grep_unlock(void)
  84{
  85        if (use_threads)
  86                pthread_mutex_unlock(&grep_mutex);
  87}
  88
  89/* Used to serialize calls to read_sha1_file. */
  90static pthread_mutex_t read_sha1_mutex;
  91
  92static inline void read_sha1_lock(void)
  93{
  94        if (use_threads)
  95                pthread_mutex_lock(&read_sha1_mutex);
  96}
  97
  98static inline void read_sha1_unlock(void)
  99{
 100        if (use_threads)
 101                pthread_mutex_unlock(&read_sha1_mutex);
 102}
 103
 104/* Signalled when a new work_item is added to todo. */
 105static pthread_cond_t cond_add;
 106
 107/* Signalled when the result from one work_item is written to
 108 * stdout.
 109 */
 110static pthread_cond_t cond_write;
 111
 112/* Signalled when we are finished with everything. */
 113static pthread_cond_t cond_result;
 114
 115static int skip_first_line;
 116
 117static void add_work(enum work_type type, char *name, void *id)
 118{
 119        grep_lock();
 120
 121        while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 122                pthread_cond_wait(&cond_write, &grep_mutex);
 123        }
 124
 125        todo[todo_end].type = type;
 126        todo[todo_end].name = name;
 127        todo[todo_end].identifier = id;
 128        todo[todo_end].done = 0;
 129        strbuf_reset(&todo[todo_end].out);
 130        todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 131
 132        pthread_cond_signal(&cond_add);
 133        grep_unlock();
 134}
 135
 136static struct work_item *get_work(void)
 137{
 138        struct work_item *ret;
 139
 140        grep_lock();
 141        while (todo_start == todo_end && !all_work_added) {
 142                pthread_cond_wait(&cond_add, &grep_mutex);
 143        }
 144
 145        if (todo_start == todo_end && all_work_added) {
 146                ret = NULL;
 147        } else {
 148                ret = &todo[todo_start];
 149                todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 150        }
 151        grep_unlock();
 152        return ret;
 153}
 154
 155static void grep_sha1_async(struct grep_opt *opt, char *name,
 156                            const unsigned char *sha1)
 157{
 158        unsigned char *s;
 159        s = xmalloc(20);
 160        memcpy(s, sha1, 20);
 161        add_work(WORK_SHA1, name, s);
 162}
 163
 164static void grep_file_async(struct grep_opt *opt, char *name,
 165                            const char *filename)
 166{
 167        add_work(WORK_FILE, name, xstrdup(filename));
 168}
 169
 170static void work_done(struct work_item *w)
 171{
 172        int old_done;
 173
 174        grep_lock();
 175        w->done = 1;
 176        old_done = todo_done;
 177        for(; todo[todo_done].done && todo_done != todo_start;
 178            todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 179                w = &todo[todo_done];
 180                if (w->out.len) {
 181                        const char *p = w->out.buf;
 182                        size_t len = w->out.len;
 183
 184                        /* Skip the leading hunk mark of the first file. */
 185                        if (skip_first_line) {
 186                                while (len) {
 187                                        len--;
 188                                        if (*p++ == '\n')
 189                                                break;
 190                                }
 191                                skip_first_line = 0;
 192                        }
 193
 194                        write_or_die(1, p, len);
 195                }
 196                free(w->name);
 197                free(w->identifier);
 198        }
 199
 200        if (old_done != todo_done)
 201                pthread_cond_signal(&cond_write);
 202
 203        if (all_work_added && todo_done == todo_end)
 204                pthread_cond_signal(&cond_result);
 205
 206        grep_unlock();
 207}
 208
 209static void *run(void *arg)
 210{
 211        int hit = 0;
 212        struct grep_opt *opt = arg;
 213
 214        while (1) {
 215                struct work_item *w = get_work();
 216                if (!w)
 217                        break;
 218
 219                opt->output_priv = w;
 220                if (w->type == WORK_SHA1) {
 221                        unsigned long sz;
 222                        void* data = load_sha1(w->identifier, &sz, w->name);
 223
 224                        if (data) {
 225                                hit |= grep_buffer(opt, w->name, data, sz);
 226                                free(data);
 227                        }
 228                } else if (w->type == WORK_FILE) {
 229                        size_t sz;
 230                        void* data = load_file(w->identifier, &sz);
 231                        if (data) {
 232                                hit |= grep_buffer(opt, w->name, data, sz);
 233                                free(data);
 234                        }
 235                } else {
 236                        assert(0);
 237                }
 238
 239                work_done(w);
 240        }
 241        free_grep_patterns(arg);
 242        free(arg);
 243
 244        return (void*) (intptr_t) hit;
 245}
 246
 247static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 248{
 249        struct work_item *w = opt->output_priv;
 250        strbuf_add(&w->out, buf, size);
 251}
 252
 253static void start_threads(struct grep_opt *opt)
 254{
 255        int i;
 256
 257        pthread_mutex_init(&grep_mutex, NULL);
 258        pthread_mutex_init(&read_sha1_mutex, NULL);
 259        pthread_cond_init(&cond_add, NULL);
 260        pthread_cond_init(&cond_write, NULL);
 261        pthread_cond_init(&cond_result, NULL);
 262
 263        for (i = 0; i < ARRAY_SIZE(todo); i++) {
 264                strbuf_init(&todo[i].out, 0);
 265        }
 266
 267        for (i = 0; i < ARRAY_SIZE(threads); i++) {
 268                int err;
 269                struct grep_opt *o = grep_opt_dup(opt);
 270                o->output = strbuf_out;
 271                compile_grep_patterns(o);
 272                err = pthread_create(&threads[i], NULL, run, o);
 273
 274                if (err)
 275                        die(_("grep: failed to create thread: %s"),
 276                            strerror(err));
 277        }
 278}
 279
 280static int wait_all(void)
 281{
 282        int hit = 0;
 283        int i;
 284
 285        grep_lock();
 286        all_work_added = 1;
 287
 288        /* Wait until all work is done. */
 289        while (todo_done != todo_end)
 290                pthread_cond_wait(&cond_result, &grep_mutex);
 291
 292        /* Wake up all the consumer threads so they can see that there
 293         * is no more work to do.
 294         */
 295        pthread_cond_broadcast(&cond_add);
 296        grep_unlock();
 297
 298        for (i = 0; i < ARRAY_SIZE(threads); i++) {
 299                void *h;
 300                pthread_join(threads[i], &h);
 301                hit |= (int) (intptr_t) h;
 302        }
 303
 304        pthread_mutex_destroy(&grep_mutex);
 305        pthread_mutex_destroy(&read_sha1_mutex);
 306        pthread_cond_destroy(&cond_add);
 307        pthread_cond_destroy(&cond_write);
 308        pthread_cond_destroy(&cond_result);
 309
 310        return hit;
 311}
 312#else /* !NO_PTHREADS */
 313#define read_sha1_lock()
 314#define read_sha1_unlock()
 315
 316static int wait_all(void)
 317{
 318        return 0;
 319}
 320#endif
 321
 322static int grep_config(const char *var, const char *value, void *cb)
 323{
 324        struct grep_opt *opt = cb;
 325        char *color = NULL;
 326
 327        switch (userdiff_config(var, value)) {
 328        case 0: break;
 329        case -1: return -1;
 330        default: return 0;
 331        }
 332
 333        if (!strcmp(var, "grep.extendedregexp")) {
 334                if (git_config_bool(var, value))
 335                        opt->regflags |= REG_EXTENDED;
 336                else
 337                        opt->regflags &= ~REG_EXTENDED;
 338                return 0;
 339        }
 340
 341        if (!strcmp(var, "grep.linenumber")) {
 342                opt->linenum = git_config_bool(var, value);
 343                return 0;
 344        }
 345
 346        if (!strcmp(var, "color.grep"))
 347                opt->color = git_config_colorbool(var, value);
 348        else if (!strcmp(var, "color.grep.context"))
 349                color = opt->color_context;
 350        else if (!strcmp(var, "color.grep.filename"))
 351                color = opt->color_filename;
 352        else if (!strcmp(var, "color.grep.function"))
 353                color = opt->color_function;
 354        else if (!strcmp(var, "color.grep.linenumber"))
 355                color = opt->color_lineno;
 356        else if (!strcmp(var, "color.grep.match"))
 357                color = opt->color_match;
 358        else if (!strcmp(var, "color.grep.selected"))
 359                color = opt->color_selected;
 360        else if (!strcmp(var, "color.grep.separator"))
 361                color = opt->color_sep;
 362        else
 363                return git_color_default_config(var, value, cb);
 364        if (color) {
 365                if (!value)
 366                        return config_error_nonbool(var);
 367                color_parse(value, var, color);
 368        }
 369        return 0;
 370}
 371
 372static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 373{
 374        void *data;
 375
 376        read_sha1_lock();
 377        data = read_sha1_file(sha1, type, size);
 378        read_sha1_unlock();
 379        return data;
 380}
 381
 382static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 383                       const char *name)
 384{
 385        enum object_type type;
 386        void *data = lock_and_read_sha1_file(sha1, &type, size);
 387
 388        if (!data)
 389                error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
 390
 391        return data;
 392}
 393
 394static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 395                     const char *filename, int tree_name_len)
 396{
 397        struct strbuf pathbuf = STRBUF_INIT;
 398        char *name;
 399
 400        if (opt->relative && opt->prefix_length) {
 401                quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 402                                    opt->prefix);
 403                strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 404        } else {
 405                strbuf_addstr(&pathbuf, filename);
 406        }
 407
 408        name = strbuf_detach(&pathbuf, NULL);
 409
 410#ifndef NO_PTHREADS
 411        if (use_threads) {
 412                grep_sha1_async(opt, name, sha1);
 413                return 0;
 414        } else
 415#endif
 416        {
 417                int hit;
 418                unsigned long sz;
 419                void *data = load_sha1(sha1, &sz, name);
 420                if (!data)
 421                        hit = 0;
 422                else
 423                        hit = grep_buffer(opt, name, data, sz);
 424
 425                free(data);
 426                free(name);
 427                return hit;
 428        }
 429}
 430
 431static void *load_file(const char *filename, size_t *sz)
 432{
 433        struct stat st;
 434        char *data;
 435        int i;
 436
 437        if (lstat(filename, &st) < 0) {
 438        err_ret:
 439                if (errno != ENOENT)
 440                        error(_("'%s': %s"), filename, strerror(errno));
 441                return NULL;
 442        }
 443        if (!S_ISREG(st.st_mode))
 444                return NULL;
 445        *sz = xsize_t(st.st_size);
 446        i = open(filename, O_RDONLY);
 447        if (i < 0)
 448                goto err_ret;
 449        data = xmalloc(*sz + 1);
 450        if (st.st_size != read_in_full(i, data, *sz)) {
 451                error(_("'%s': short read %s"), filename, strerror(errno));
 452                close(i);
 453                free(data);
 454                return NULL;
 455        }
 456        close(i);
 457        data[*sz] = 0;
 458        return data;
 459}
 460
 461static int grep_file(struct grep_opt *opt, const char *filename)
 462{
 463        struct strbuf buf = STRBUF_INIT;
 464        char *name;
 465
 466        if (opt->relative && opt->prefix_length)
 467                quote_path_relative(filename, -1, &buf, opt->prefix);
 468        else
 469                strbuf_addstr(&buf, filename);
 470        name = strbuf_detach(&buf, NULL);
 471
 472#ifndef NO_PTHREADS
 473        if (use_threads) {
 474                grep_file_async(opt, name, filename);
 475                return 0;
 476        } else
 477#endif
 478        {
 479                int hit;
 480                size_t sz;
 481                void *data = load_file(filename, &sz);
 482                if (!data)
 483                        hit = 0;
 484                else
 485                        hit = grep_buffer(opt, name, data, sz);
 486
 487                free(data);
 488                free(name);
 489                return hit;
 490        }
 491}
 492
 493static void append_path(struct grep_opt *opt, const void *data, size_t len)
 494{
 495        struct string_list *path_list = opt->output_priv;
 496
 497        if (len == 1 && *(const char *)data == '\0')
 498                return;
 499        string_list_append(path_list, xstrndup(data, len));
 500}
 501
 502static void run_pager(struct grep_opt *opt, const char *prefix)
 503{
 504        struct string_list *path_list = opt->output_priv;
 505        const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 506        int i, status;
 507
 508        for (i = 0; i < path_list->nr; i++)
 509                argv[i] = path_list->items[i].string;
 510        argv[path_list->nr] = NULL;
 511
 512        if (prefix && chdir(prefix))
 513                die(_("Failed to chdir: %s"), prefix);
 514        status = run_command_v_opt(argv, RUN_USING_SHELL);
 515        if (status)
 516                exit(status);
 517        free(argv);
 518}
 519
 520static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 521{
 522        int hit = 0;
 523        int nr;
 524        read_cache();
 525
 526        for (nr = 0; nr < active_nr; nr++) {
 527                struct cache_entry *ce = active_cache[nr];
 528                if (!S_ISREG(ce->ce_mode))
 529                        continue;
 530                if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 531                        continue;
 532                /*
 533                 * If CE_VALID is on, we assume worktree file and its cache entry
 534                 * are identical, even if worktree file has been modified, so use
 535                 * cache version instead
 536                 */
 537                if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 538                        if (ce_stage(ce))
 539                                continue;
 540                        hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 541                }
 542                else
 543                        hit |= grep_file(opt, ce->name);
 544                if (ce_stage(ce)) {
 545                        do {
 546                                nr++;
 547                        } while (nr < active_nr &&
 548                                 !strcmp(ce->name, active_cache[nr]->name));
 549                        nr--; /* compensate for loop control */
 550                }
 551                if (hit && opt->status_only)
 552                        break;
 553        }
 554        return hit;
 555}
 556
 557static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 558                     struct tree_desc *tree, struct strbuf *base, int tn_len)
 559{
 560        int hit = 0;
 561        enum interesting match = entry_not_interesting;
 562        struct name_entry entry;
 563        int old_baselen = base->len;
 564
 565        while (tree_entry(tree, &entry)) {
 566                int te_len = tree_entry_len(&entry);
 567
 568                if (match != all_entries_interesting) {
 569                        match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 570                        if (match == all_entries_not_interesting)
 571                                break;
 572                        if (match == entry_not_interesting)
 573                                continue;
 574                }
 575
 576                strbuf_add(base, entry.path, te_len);
 577
 578                if (S_ISREG(entry.mode)) {
 579                        hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 580                }
 581                else if (S_ISDIR(entry.mode)) {
 582                        enum object_type type;
 583                        struct tree_desc sub;
 584                        void *data;
 585                        unsigned long size;
 586
 587                        data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 588                        if (!data)
 589                                die(_("unable to read tree (%s)"),
 590                                    sha1_to_hex(entry.sha1));
 591
 592                        strbuf_addch(base, '/');
 593                        init_tree_desc(&sub, data, size);
 594                        hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 595                        free(data);
 596                }
 597                strbuf_setlen(base, old_baselen);
 598
 599                if (hit && opt->status_only)
 600                        break;
 601        }
 602        return hit;
 603}
 604
 605static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 606                       struct object *obj, const char *name)
 607{
 608        if (obj->type == OBJ_BLOB)
 609                return grep_sha1(opt, obj->sha1, name, 0);
 610        if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 611                struct tree_desc tree;
 612                void *data;
 613                unsigned long size;
 614                struct strbuf base;
 615                int hit, len;
 616
 617                read_sha1_lock();
 618                data = read_object_with_reference(obj->sha1, tree_type,
 619                                                  &size, NULL);
 620                read_sha1_unlock();
 621
 622                if (!data)
 623                        die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 624
 625                len = name ? strlen(name) : 0;
 626                strbuf_init(&base, PATH_MAX + len + 1);
 627                if (len) {
 628                        strbuf_add(&base, name, len);
 629                        strbuf_addch(&base, ':');
 630                }
 631                init_tree_desc(&tree, data, size);
 632                hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 633                strbuf_release(&base);
 634                free(data);
 635                return hit;
 636        }
 637        die(_("unable to grep from object of type %s"), typename(obj->type));
 638}
 639
 640static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 641                        const struct object_array *list)
 642{
 643        unsigned int i;
 644        int hit = 0;
 645        const unsigned int nr = list->nr;
 646
 647        for (i = 0; i < nr; i++) {
 648                struct object *real_obj;
 649                real_obj = deref_tag(list->objects[i].item, NULL, 0);
 650                if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 651                        hit = 1;
 652                        if (opt->status_only)
 653                                break;
 654                }
 655        }
 656        return hit;
 657}
 658
 659static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 660                          int exc_std)
 661{
 662        struct dir_struct dir;
 663        int i, hit = 0;
 664
 665        memset(&dir, 0, sizeof(dir));
 666        if (exc_std)
 667                setup_standard_excludes(&dir);
 668
 669        fill_directory(&dir, pathspec->raw);
 670        for (i = 0; i < dir.nr; i++) {
 671                const char *name = dir.entries[i]->name;
 672                int namelen = strlen(name);
 673                if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 674                        continue;
 675                hit |= grep_file(opt, dir.entries[i]->name);
 676                if (hit && opt->status_only)
 677                        break;
 678        }
 679        return hit;
 680}
 681
 682static int context_callback(const struct option *opt, const char *arg,
 683                            int unset)
 684{
 685        struct grep_opt *grep_opt = opt->value;
 686        int value;
 687        const char *endp;
 688
 689        if (unset) {
 690                grep_opt->pre_context = grep_opt->post_context = 0;
 691                return 0;
 692        }
 693        value = strtol(arg, (char **)&endp, 10);
 694        if (*endp) {
 695                return error(_("switch `%c' expects a numerical value"),
 696                             opt->short_name);
 697        }
 698        grep_opt->pre_context = grep_opt->post_context = value;
 699        return 0;
 700}
 701
 702static int file_callback(const struct option *opt, const char *arg, int unset)
 703{
 704        struct grep_opt *grep_opt = opt->value;
 705        int from_stdin = !strcmp(arg, "-");
 706        FILE *patterns;
 707        int lno = 0;
 708        struct strbuf sb = STRBUF_INIT;
 709
 710        patterns = from_stdin ? stdin : fopen(arg, "r");
 711        if (!patterns)
 712                die_errno(_("cannot open '%s'"), arg);
 713        while (strbuf_getline(&sb, patterns, '\n') == 0) {
 714                char *s;
 715                size_t len;
 716
 717                /* ignore empty line like grep does */
 718                if (sb.len == 0)
 719                        continue;
 720
 721                s = strbuf_detach(&sb, &len);
 722                append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 723        }
 724        if (!from_stdin)
 725                fclose(patterns);
 726        strbuf_release(&sb);
 727        return 0;
 728}
 729
 730static int not_callback(const struct option *opt, const char *arg, int unset)
 731{
 732        struct grep_opt *grep_opt = opt->value;
 733        append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 734        return 0;
 735}
 736
 737static int and_callback(const struct option *opt, const char *arg, int unset)
 738{
 739        struct grep_opt *grep_opt = opt->value;
 740        append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 741        return 0;
 742}
 743
 744static int open_callback(const struct option *opt, const char *arg, int unset)
 745{
 746        struct grep_opt *grep_opt = opt->value;
 747        append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 748        return 0;
 749}
 750
 751static int close_callback(const struct option *opt, const char *arg, int unset)
 752{
 753        struct grep_opt *grep_opt = opt->value;
 754        append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 755        return 0;
 756}
 757
 758static int pattern_callback(const struct option *opt, const char *arg,
 759                            int unset)
 760{
 761        struct grep_opt *grep_opt = opt->value;
 762        append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 763        return 0;
 764}
 765
 766static int help_callback(const struct option *opt, const char *arg, int unset)
 767{
 768        return -1;
 769}
 770
 771int cmd_grep(int argc, const char **argv, const char *prefix)
 772{
 773        int hit = 0;
 774        int cached = 0, untracked = 0, opt_exclude = -1;
 775        int seen_dashdash = 0;
 776        int external_grep_allowed__ignored;
 777        const char *show_in_pager = NULL, *default_pager = "dummy";
 778        struct grep_opt opt;
 779        struct object_array list = OBJECT_ARRAY_INIT;
 780        const char **paths = NULL;
 781        struct pathspec pathspec;
 782        struct string_list path_list = STRING_LIST_INIT_NODUP;
 783        int i;
 784        int dummy;
 785        int use_index = 1;
 786        enum {
 787                pattern_type_unspecified = 0,
 788                pattern_type_bre,
 789                pattern_type_ere,
 790                pattern_type_fixed,
 791                pattern_type_pcre,
 792        };
 793        int pattern_type = pattern_type_unspecified;
 794
 795        struct option options[] = {
 796                OPT_BOOLEAN(0, "cached", &cached,
 797                        "search in index instead of in the work tree"),
 798                { OPTION_BOOLEAN, 0, "index", &use_index, NULL,
 799                        "finds in contents not managed by git",
 800                        PARSE_OPT_NOARG | PARSE_OPT_NEGHELP },
 801                OPT_BOOLEAN(0, "untracked", &untracked,
 802                        "search in both tracked and untracked files"),
 803                OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 804                            "search also in ignored files", 1),
 805                OPT_GROUP(""),
 806                OPT_BOOLEAN('v', "invert-match", &opt.invert,
 807                        "show non-matching lines"),
 808                OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 809                        "case insensitive matching"),
 810                OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 811                        "match patterns only at word boundaries"),
 812                OPT_SET_INT('a', "text", &opt.binary,
 813                        "process binary files as text", GREP_BINARY_TEXT),
 814                OPT_SET_INT('I', NULL, &opt.binary,
 815                        "don't match patterns in binary files",
 816                        GREP_BINARY_NOMATCH),
 817                { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 818                        "descend at most <depth> levels", PARSE_OPT_NONEG,
 819                        NULL, 1 },
 820                OPT_GROUP(""),
 821                OPT_SET_INT('E', "extended-regexp", &pattern_type,
 822                            "use extended POSIX regular expressions",
 823                            pattern_type_ere),
 824                OPT_SET_INT('G', "basic-regexp", &pattern_type,
 825                            "use basic POSIX regular expressions (default)",
 826                            pattern_type_bre),
 827                OPT_SET_INT('F', "fixed-strings", &pattern_type,
 828                            "interpret patterns as fixed strings",
 829                            pattern_type_fixed),
 830                OPT_SET_INT('P', "perl-regexp", &pattern_type,
 831                            "use Perl-compatible regular expressions",
 832                            pattern_type_pcre),
 833                OPT_GROUP(""),
 834                OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
 835                OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 836                OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 837                OPT_NEGBIT(0, "full-name", &opt.relative,
 838                        "show filenames relative to top directory", 1),
 839                OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 840                        "show only filenames instead of matching lines"),
 841                OPT_BOOLEAN(0, "name-only", &opt.name_only,
 842                        "synonym for --files-with-matches"),
 843                OPT_BOOLEAN('L', "files-without-match",
 844                        &opt.unmatch_name_only,
 845                        "show only the names of files without match"),
 846                OPT_BOOLEAN('z', "null", &opt.null_following_name,
 847                        "print NUL after filenames"),
 848                OPT_BOOLEAN('c', "count", &opt.count,
 849                        "show the number of matches instead of matching lines"),
 850                OPT__COLOR(&opt.color, "highlight matches"),
 851                OPT_BOOLEAN(0, "break", &opt.file_break,
 852                        "print empty line between matches from different files"),
 853                OPT_BOOLEAN(0, "heading", &opt.heading,
 854                        "show filename only once above matches from same file"),
 855                OPT_GROUP(""),
 856                OPT_CALLBACK('C', "context", &opt, "n",
 857                        "show <n> context lines before and after matches",
 858                        context_callback),
 859                OPT_INTEGER('B', "before-context", &opt.pre_context,
 860                        "show <n> context lines before matches"),
 861                OPT_INTEGER('A', "after-context", &opt.post_context,
 862                        "show <n> context lines after matches"),
 863                OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 864                        context_callback),
 865                OPT_BOOLEAN('p', "show-function", &opt.funcname,
 866                        "show a line with the function name before matches"),
 867                OPT_BOOLEAN('W', "function-context", &opt.funcbody,
 868                        "show the surrounding function"),
 869                OPT_GROUP(""),
 870                OPT_CALLBACK('f', NULL, &opt, "file",
 871                        "read patterns from file", file_callback),
 872                { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 873                        "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 874                { OPTION_CALLBACK, 0, "and", &opt, NULL,
 875                  "combine patterns specified with -e",
 876                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 877                OPT_BOOLEAN(0, "or", &dummy, ""),
 878                { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 879                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 880                { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 881                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 882                  open_callback },
 883                { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 884                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 885                  close_callback },
 886                OPT__QUIET(&opt.status_only,
 887                           "indicate hit with exit status without output"),
 888                OPT_BOOLEAN(0, "all-match", &opt.all_match,
 889                        "show only matches from files that match all patterns"),
 890                OPT_GROUP(""),
 891                { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 892                        "pager", "show matching files in the pager",
 893                        PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 894                OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 895                            "allow calling of grep(1) (ignored by this build)"),
 896                { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 897                  PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 898                OPT_END()
 899        };
 900
 901        /*
 902         * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 903         * to show usage information and exit.
 904         */
 905        if (argc == 2 && !strcmp(argv[1], "-h"))
 906                usage_with_options(grep_usage, options);
 907
 908        memset(&opt, 0, sizeof(opt));
 909        opt.prefix = prefix;
 910        opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 911        opt.relative = 1;
 912        opt.pathname = 1;
 913        opt.pattern_tail = &opt.pattern_list;
 914        opt.header_tail = &opt.header_list;
 915        opt.regflags = REG_NEWLINE;
 916        opt.max_depth = -1;
 917
 918        strcpy(opt.color_context, "");
 919        strcpy(opt.color_filename, "");
 920        strcpy(opt.color_function, "");
 921        strcpy(opt.color_lineno, "");
 922        strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 923        strcpy(opt.color_selected, "");
 924        strcpy(opt.color_sep, GIT_COLOR_CYAN);
 925        opt.color = -1;
 926        git_config(grep_config, &opt);
 927
 928        /*
 929         * If there is no -- then the paths must exist in the working
 930         * tree.  If there is no explicit pattern specified with -e or
 931         * -f, we take the first unrecognized non option to be the
 932         * pattern, but then what follows it must be zero or more
 933         * valid refs up to the -- (if exists), and then existing
 934         * paths.  If there is an explicit pattern, then the first
 935         * unrecognized non option is the beginning of the refs list
 936         * that continues up to the -- (if exists), and then paths.
 937         */
 938        argc = parse_options(argc, argv, prefix, options, grep_usage,
 939                             PARSE_OPT_KEEP_DASHDASH |
 940                             PARSE_OPT_STOP_AT_NON_OPTION |
 941                             PARSE_OPT_NO_INTERNAL_HELP);
 942        switch (pattern_type) {
 943        case pattern_type_fixed:
 944                opt.fixed = 1;
 945                opt.pcre = 0;
 946                break;
 947        case pattern_type_bre:
 948                opt.fixed = 0;
 949                opt.pcre = 0;
 950                opt.regflags &= ~REG_EXTENDED;
 951                break;
 952        case pattern_type_ere:
 953                opt.fixed = 0;
 954                opt.pcre = 0;
 955                opt.regflags |= REG_EXTENDED;
 956                break;
 957        case pattern_type_pcre:
 958                opt.fixed = 0;
 959                opt.pcre = 1;
 960                break;
 961        default:
 962                break; /* nothing */
 963        }
 964
 965        if (use_index && !startup_info->have_repository)
 966                /* die the same way as if we did it at the beginning */
 967                setup_git_directory();
 968
 969        /*
 970         * skip a -- separator; we know it cannot be
 971         * separating revisions from pathnames if
 972         * we haven't even had any patterns yet
 973         */
 974        if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 975                argv++;
 976                argc--;
 977        }
 978
 979        /* First unrecognized non-option token */
 980        if (argc > 0 && !opt.pattern_list) {
 981                append_grep_pattern(&opt, argv[0], "command line", 0,
 982                                    GREP_PATTERN);
 983                argv++;
 984                argc--;
 985        }
 986
 987        if (show_in_pager == default_pager)
 988                show_in_pager = git_pager(1);
 989        if (show_in_pager) {
 990                opt.color = 0;
 991                opt.name_only = 1;
 992                opt.null_following_name = 1;
 993                opt.output_priv = &path_list;
 994                opt.output = append_path;
 995                string_list_append(&path_list, show_in_pager);
 996                use_threads = 0;
 997        }
 998
 999        if (!opt.pattern_list)
1000                die(_("no pattern given."));
1001        if (!opt.fixed && opt.ignore_case)
1002                opt.regflags |= REG_ICASE;
1003
1004#ifndef NO_PTHREADS
1005        if (online_cpus() == 1 || !grep_threads_ok(&opt))
1006                use_threads = 0;
1007
1008        if (use_threads) {
1009                if (opt.pre_context || opt.post_context || opt.file_break ||
1010                    opt.funcbody)
1011                        skip_first_line = 1;
1012                start_threads(&opt);
1013        }
1014#else
1015        use_threads = 0;
1016#endif
1017
1018        compile_grep_patterns(&opt);
1019
1020        /* Check revs and then paths */
1021        for (i = 0; i < argc; i++) {
1022                const char *arg = argv[i];
1023                unsigned char sha1[20];
1024                /* Is it a rev? */
1025                if (!get_sha1(arg, sha1)) {
1026                        struct object *object = parse_object(sha1);
1027                        if (!object)
1028                                die(_("bad object %s"), arg);
1029                        add_object_array(object, arg, &list);
1030                        continue;
1031                }
1032                if (!strcmp(arg, "--")) {
1033                        i++;
1034                        seen_dashdash = 1;
1035                }
1036                break;
1037        }
1038
1039        /* The rest are paths */
1040        if (!seen_dashdash) {
1041                int j;
1042                for (j = i; j < argc; j++)
1043                        verify_filename(prefix, argv[j]);
1044        }
1045
1046        paths = get_pathspec(prefix, argv + i);
1047        init_pathspec(&pathspec, paths);
1048        pathspec.max_depth = opt.max_depth;
1049        pathspec.recursive = 1;
1050
1051        if (show_in_pager && (cached || list.nr))
1052                die(_("--open-files-in-pager only works on the worktree"));
1053
1054        if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1055                const char *pager = path_list.items[0].string;
1056                int len = strlen(pager);
1057
1058                if (len > 4 && is_dir_sep(pager[len - 5]))
1059                        pager += len - 4;
1060
1061                if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1062                        struct strbuf buf = STRBUF_INIT;
1063                        strbuf_addf(&buf, "+/%s%s",
1064                                        strcmp("less", pager) ? "" : "*",
1065                                        opt.pattern_list->pattern);
1066                        string_list_append(&path_list, buf.buf);
1067                        strbuf_detach(&buf, NULL);
1068                }
1069        }
1070
1071        if (!show_in_pager)
1072                setup_pager();
1073
1074        if (!use_index && (untracked || cached))
1075                die(_("--cached or --untracked cannot be used with --no-index."));
1076
1077        if (!use_index || untracked) {
1078                int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1079                if (list.nr)
1080                        die(_("--no-index or --untracked cannot be used with revs."));
1081                hit = grep_directory(&opt, &pathspec, use_exclude);
1082        } else if (0 <= opt_exclude) {
1083                die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1084        } else if (!list.nr) {
1085                if (!cached)
1086                        setup_work_tree();
1087
1088                hit = grep_cache(&opt, &pathspec, cached);
1089        } else {
1090                if (cached)
1091                        die(_("both --cached and trees are given."));
1092                hit = grep_objects(&opt, &pathspec, &list);
1093        }
1094
1095        if (use_threads)
1096                hit |= wait_all();
1097        if (hit && show_in_pager)
1098                run_pager(&opt, prefix);
1099        free_grep_patterns(&opt);
1100        return !hit;
1101}