builtin / grep.con commit grep: use writable strbuf from caller for grep_tree() (e5e062b)
   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{
  45        enum work_type type;
  46        char *name;
  47
  48        /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  49         * otherwise type == WORK_FILE, and 'identifier' is a NUL
  50         * terminated filename.
  51         */
  52        void *identifier;
  53        char done;
  54        struct strbuf out;
  55};
  56
  57/* In the range [todo_done, todo_start) in 'todo' we have work_items
  58 * that have been or are processed by a consumer thread. We haven't
  59 * written the result for these to stdout yet.
  60 *
  61 * The work_items in [todo_start, todo_end) are waiting to be picked
  62 * up by a consumer thread.
  63 *
  64 * The ranges are modulo TODO_SIZE.
  65 */
  66#define TODO_SIZE 128
  67static struct work_item todo[TODO_SIZE];
  68static int todo_start;
  69static int todo_end;
  70static int todo_done;
  71
  72/* Has all work items been added? */
  73static int all_work_added;
  74
  75/* This lock protects all the variables above. */
  76static pthread_mutex_t grep_mutex;
  77
  78/* Used to serialize calls to read_sha1_file. */
  79static pthread_mutex_t read_sha1_mutex;
  80
  81#define grep_lock() pthread_mutex_lock(&grep_mutex)
  82#define grep_unlock() pthread_mutex_unlock(&grep_mutex)
  83#define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
  84#define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
  85
  86/* Signalled when a new work_item is added to todo. */
  87static pthread_cond_t cond_add;
  88
  89/* Signalled when the result from one work_item is written to
  90 * stdout.
  91 */
  92static pthread_cond_t cond_write;
  93
  94/* Signalled when we are finished with everything. */
  95static pthread_cond_t cond_result;
  96
  97static int print_hunk_marks_between_files;
  98static int printed_something;
  99
 100static void add_work(enum work_type type, char *name, void *id)
 101{
 102        grep_lock();
 103
 104        while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 105                pthread_cond_wait(&cond_write, &grep_mutex);
 106        }
 107
 108        todo[todo_end].type = type;
 109        todo[todo_end].name = name;
 110        todo[todo_end].identifier = id;
 111        todo[todo_end].done = 0;
 112        strbuf_reset(&todo[todo_end].out);
 113        todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 114
 115        pthread_cond_signal(&cond_add);
 116        grep_unlock();
 117}
 118
 119static struct work_item *get_work(void)
 120{
 121        struct work_item *ret;
 122
 123        grep_lock();
 124        while (todo_start == todo_end && !all_work_added) {
 125                pthread_cond_wait(&cond_add, &grep_mutex);
 126        }
 127
 128        if (todo_start == todo_end && all_work_added) {
 129                ret = NULL;
 130        } else {
 131                ret = &todo[todo_start];
 132                todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 133        }
 134        grep_unlock();
 135        return ret;
 136}
 137
 138static void grep_sha1_async(struct grep_opt *opt, char *name,
 139                            const unsigned char *sha1)
 140{
 141        unsigned char *s;
 142        s = xmalloc(20);
 143        memcpy(s, sha1, 20);
 144        add_work(WORK_SHA1, name, s);
 145}
 146
 147static void grep_file_async(struct grep_opt *opt, char *name,
 148                            const char *filename)
 149{
 150        add_work(WORK_FILE, name, xstrdup(filename));
 151}
 152
 153static void work_done(struct work_item *w)
 154{
 155        int old_done;
 156
 157        grep_lock();
 158        w->done = 1;
 159        old_done = todo_done;
 160        for(; todo[todo_done].done && todo_done != todo_start;
 161            todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 162                w = &todo[todo_done];
 163                if (w->out.len) {
 164                        if (print_hunk_marks_between_files && printed_something)
 165                                write_or_die(1, "--\n", 3);
 166                        write_or_die(1, w->out.buf, w->out.len);
 167                        printed_something = 1;
 168                }
 169                free(w->name);
 170                free(w->identifier);
 171        }
 172
 173        if (old_done != todo_done)
 174                pthread_cond_signal(&cond_write);
 175
 176        if (all_work_added && todo_done == todo_end)
 177                pthread_cond_signal(&cond_result);
 178
 179        grep_unlock();
 180}
 181
 182static void *run(void *arg)
 183{
 184        int hit = 0;
 185        struct grep_opt *opt = arg;
 186
 187        while (1) {
 188                struct work_item *w = get_work();
 189                if (!w)
 190                        break;
 191
 192                opt->output_priv = w;
 193                if (w->type == WORK_SHA1) {
 194                        unsigned long sz;
 195                        void* data = load_sha1(w->identifier, &sz, w->name);
 196
 197                        if (data) {
 198                                hit |= grep_buffer(opt, w->name, data, sz);
 199                                free(data);
 200                        }
 201                } else if (w->type == WORK_FILE) {
 202                        size_t sz;
 203                        void* data = load_file(w->identifier, &sz);
 204                        if (data) {
 205                                hit |= grep_buffer(opt, w->name, data, sz);
 206                                free(data);
 207                        }
 208                } else {
 209                        assert(0);
 210                }
 211
 212                work_done(w);
 213        }
 214        free_grep_patterns(arg);
 215        free(arg);
 216
 217        return (void*) (intptr_t) hit;
 218}
 219
 220static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 221{
 222        struct work_item *w = opt->output_priv;
 223        strbuf_add(&w->out, buf, size);
 224}
 225
 226static void start_threads(struct grep_opt *opt)
 227{
 228        int i;
 229
 230        pthread_mutex_init(&grep_mutex, NULL);
 231        pthread_mutex_init(&read_sha1_mutex, NULL);
 232        pthread_cond_init(&cond_add, NULL);
 233        pthread_cond_init(&cond_write, NULL);
 234        pthread_cond_init(&cond_result, NULL);
 235
 236        for (i = 0; i < ARRAY_SIZE(todo); i++) {
 237                strbuf_init(&todo[i].out, 0);
 238        }
 239
 240        for (i = 0; i < ARRAY_SIZE(threads); i++) {
 241                int err;
 242                struct grep_opt *o = grep_opt_dup(opt);
 243                o->output = strbuf_out;
 244                compile_grep_patterns(o);
 245                err = pthread_create(&threads[i], NULL, run, o);
 246
 247                if (err)
 248                        die("grep: failed to create thread: %s",
 249                            strerror(err));
 250        }
 251}
 252
 253static int wait_all(void)
 254{
 255        int hit = 0;
 256        int i;
 257
 258        grep_lock();
 259        all_work_added = 1;
 260
 261        /* Wait until all work is done. */
 262        while (todo_done != todo_end)
 263                pthread_cond_wait(&cond_result, &grep_mutex);
 264
 265        /* Wake up all the consumer threads so they can see that there
 266         * is no more work to do.
 267         */
 268        pthread_cond_broadcast(&cond_add);
 269        grep_unlock();
 270
 271        for (i = 0; i < ARRAY_SIZE(threads); i++) {
 272                void *h;
 273                pthread_join(threads[i], &h);
 274                hit |= (int) (intptr_t) h;
 275        }
 276
 277        pthread_mutex_destroy(&grep_mutex);
 278        pthread_mutex_destroy(&read_sha1_mutex);
 279        pthread_cond_destroy(&cond_add);
 280        pthread_cond_destroy(&cond_write);
 281        pthread_cond_destroy(&cond_result);
 282
 283        return hit;
 284}
 285#else /* !NO_PTHREADS */
 286#define read_sha1_lock()
 287#define read_sha1_unlock()
 288
 289static int wait_all(void)
 290{
 291        return 0;
 292}
 293#endif
 294
 295static int grep_config(const char *var, const char *value, void *cb)
 296{
 297        struct grep_opt *opt = cb;
 298        char *color = NULL;
 299
 300        switch (userdiff_config(var, value)) {
 301        case 0: break;
 302        case -1: return -1;
 303        default: return 0;
 304        }
 305
 306        if (!strcmp(var, "color.grep"))
 307                opt->color = git_config_colorbool(var, value, -1);
 308        else if (!strcmp(var, "color.grep.context"))
 309                color = opt->color_context;
 310        else if (!strcmp(var, "color.grep.filename"))
 311                color = opt->color_filename;
 312        else if (!strcmp(var, "color.grep.function"))
 313                color = opt->color_function;
 314        else if (!strcmp(var, "color.grep.linenumber"))
 315                color = opt->color_lineno;
 316        else if (!strcmp(var, "color.grep.match"))
 317                color = opt->color_match;
 318        else if (!strcmp(var, "color.grep.selected"))
 319                color = opt->color_selected;
 320        else if (!strcmp(var, "color.grep.separator"))
 321                color = opt->color_sep;
 322        else
 323                return git_color_default_config(var, value, cb);
 324        if (color) {
 325                if (!value)
 326                        return config_error_nonbool(var);
 327                color_parse(value, var, color);
 328        }
 329        return 0;
 330}
 331
 332/*
 333 * Return non-zero if max_depth is negative or path has no more then max_depth
 334 * slashes.
 335 */
 336static int accept_subdir(const char *path, int max_depth)
 337{
 338        if (max_depth < 0)
 339                return 1;
 340
 341        while ((path = strchr(path, '/')) != NULL) {
 342                max_depth--;
 343                if (max_depth < 0)
 344                        return 0;
 345                path++;
 346        }
 347        return 1;
 348}
 349
 350/*
 351 * Return non-zero if name is a subdirectory of match and is not too deep.
 352 */
 353static int is_subdir(const char *name, int namelen,
 354                const char *match, int matchlen, int max_depth)
 355{
 356        if (matchlen > namelen || strncmp(name, match, matchlen))
 357                return 0;
 358
 359        if (name[matchlen] == '\0') /* exact match */
 360                return 1;
 361
 362        if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
 363                return accept_subdir(name + matchlen + 1, max_depth);
 364
 365        return 0;
 366}
 367
 368/*
 369 * git grep pathspecs are somewhat different from diff-tree pathspecs;
 370 * pathname wildcards are allowed.
 371 */
 372static int pathspec_matches(const char **paths, const char *name, int max_depth)
 373{
 374        int namelen, i;
 375        if (!paths || !*paths)
 376                return accept_subdir(name, max_depth);
 377        namelen = strlen(name);
 378        for (i = 0; paths[i]; i++) {
 379                const char *match = paths[i];
 380                int matchlen = strlen(match);
 381                const char *cp, *meta;
 382
 383                if (is_subdir(name, namelen, match, matchlen, max_depth))
 384                        return 1;
 385                if (!fnmatch(match, name, 0))
 386                        return 1;
 387                if (name[namelen-1] != '/')
 388                        continue;
 389
 390                /* We are being asked if the directory ("name") is worth
 391                 * descending into.
 392                 *
 393                 * Find the longest leading directory name that does
 394                 * not have metacharacter in the pathspec; the name
 395                 * we are looking at must overlap with that directory.
 396                 */
 397                for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
 398                        char ch = *cp;
 399                        if (ch == '*' || ch == '[' || ch == '?') {
 400                                meta = cp;
 401                                break;
 402                        }
 403                }
 404                if (!meta)
 405                        meta = cp; /* fully literal */
 406
 407                if (namelen <= meta - match) {
 408                        /* Looking at "Documentation/" and
 409                         * the pattern says "Documentation/howto/", or
 410                         * "Documentation/diff*.txt".  The name we
 411                         * have should match prefix.
 412                         */
 413                        if (!memcmp(match, name, namelen))
 414                                return 1;
 415                        continue;
 416                }
 417
 418                if (meta - match < namelen) {
 419                        /* Looking at "Documentation/howto/" and
 420                         * the pattern says "Documentation/h*";
 421                         * match up to "Do.../h"; this avoids descending
 422                         * into "Documentation/technical/".
 423                         */
 424                        if (!memcmp(match, name, meta - match))
 425                                return 1;
 426                        continue;
 427                }
 428        }
 429        return 0;
 430}
 431
 432static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 433{
 434        void *data;
 435
 436        if (use_threads) {
 437                read_sha1_lock();
 438                data = read_sha1_file(sha1, type, size);
 439                read_sha1_unlock();
 440        } else {
 441                data = read_sha1_file(sha1, type, size);
 442        }
 443        return data;
 444}
 445
 446static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 447                       const char *name)
 448{
 449        enum object_type type;
 450        void *data = lock_and_read_sha1_file(sha1, &type, size);
 451
 452        if (!data)
 453                error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 454
 455        return data;
 456}
 457
 458static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 459                     const char *filename, int tree_name_len)
 460{
 461        struct strbuf pathbuf = STRBUF_INIT;
 462        char *name;
 463
 464        if (opt->relative && opt->prefix_length) {
 465                quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 466                                    opt->prefix);
 467                strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 468        } else {
 469                strbuf_addstr(&pathbuf, filename);
 470        }
 471
 472        name = strbuf_detach(&pathbuf, NULL);
 473
 474#ifndef NO_PTHREADS
 475        if (use_threads) {
 476                grep_sha1_async(opt, name, sha1);
 477                return 0;
 478        } else
 479#endif
 480        {
 481                int hit;
 482                unsigned long sz;
 483                void *data = load_sha1(sha1, &sz, name);
 484                if (!data)
 485                        hit = 0;
 486                else
 487                        hit = grep_buffer(opt, name, data, sz);
 488
 489                free(data);
 490                free(name);
 491                return hit;
 492        }
 493}
 494
 495static void *load_file(const char *filename, size_t *sz)
 496{
 497        struct stat st;
 498        char *data;
 499        int i;
 500
 501        if (lstat(filename, &st) < 0) {
 502        err_ret:
 503                if (errno != ENOENT)
 504                        error("'%s': %s", filename, strerror(errno));
 505                return 0;
 506        }
 507        if (!S_ISREG(st.st_mode))
 508                return 0;
 509        *sz = xsize_t(st.st_size);
 510        i = open(filename, O_RDONLY);
 511        if (i < 0)
 512                goto err_ret;
 513        data = xmalloc(*sz + 1);
 514        if (st.st_size != read_in_full(i, data, *sz)) {
 515                error("'%s': short read %s", filename, strerror(errno));
 516                close(i);
 517                free(data);
 518                return 0;
 519        }
 520        close(i);
 521        data[*sz] = 0;
 522        return data;
 523}
 524
 525static int grep_file(struct grep_opt *opt, const char *filename)
 526{
 527        struct strbuf buf = STRBUF_INIT;
 528        char *name;
 529
 530        if (opt->relative && opt->prefix_length)
 531                quote_path_relative(filename, -1, &buf, opt->prefix);
 532        else
 533                strbuf_addstr(&buf, filename);
 534        name = strbuf_detach(&buf, NULL);
 535
 536#ifndef NO_PTHREADS
 537        if (use_threads) {
 538                grep_file_async(opt, name, filename);
 539                return 0;
 540        } else
 541#endif
 542        {
 543                int hit;
 544                size_t sz;
 545                void *data = load_file(filename, &sz);
 546                if (!data)
 547                        hit = 0;
 548                else
 549                        hit = grep_buffer(opt, name, data, sz);
 550
 551                free(data);
 552                free(name);
 553                return hit;
 554        }
 555}
 556
 557static void append_path(struct grep_opt *opt, const void *data, size_t len)
 558{
 559        struct string_list *path_list = opt->output_priv;
 560
 561        if (len == 1 && *(const char *)data == '\0')
 562                return;
 563        string_list_append(path_list, xstrndup(data, len));
 564}
 565
 566static void run_pager(struct grep_opt *opt, const char *prefix)
 567{
 568        struct string_list *path_list = opt->output_priv;
 569        const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 570        int i, status;
 571
 572        for (i = 0; i < path_list->nr; i++)
 573                argv[i] = path_list->items[i].string;
 574        argv[path_list->nr] = NULL;
 575
 576        if (prefix && chdir(prefix))
 577                die("Failed to chdir: %s", prefix);
 578        status = run_command_v_opt(argv, RUN_USING_SHELL);
 579        if (status)
 580                exit(status);
 581        free(argv);
 582}
 583
 584static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 585{
 586        int hit = 0;
 587        int nr;
 588        read_cache();
 589
 590        for (nr = 0; nr < active_nr; nr++) {
 591                struct cache_entry *ce = active_cache[nr];
 592                if (!S_ISREG(ce->ce_mode))
 593                        continue;
 594                if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 595                        continue;
 596                /*
 597                 * If CE_VALID is on, we assume worktree file and its cache entry
 598                 * are identical, even if worktree file has been modified, so use
 599                 * cache version instead
 600                 */
 601                if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 602                        if (ce_stage(ce))
 603                                continue;
 604                        hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 605                }
 606                else
 607                        hit |= grep_file(opt, ce->name);
 608                if (ce_stage(ce)) {
 609                        do {
 610                                nr++;
 611                        } while (nr < active_nr &&
 612                                 !strcmp(ce->name, active_cache[nr]->name));
 613                        nr--; /* compensate for loop control */
 614                }
 615                if (hit && opt->status_only)
 616                        break;
 617        }
 618        return hit;
 619}
 620
 621static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 622                     struct tree_desc *tree, struct strbuf *base, int tn_len)
 623{
 624        int hit = 0;
 625        struct name_entry entry;
 626        int old_baselen = base->len;
 627
 628        while (tree_entry(tree, &entry)) {
 629                int te_len = tree_entry_len(entry.path, entry.sha1);
 630
 631                strbuf_add(base, entry.path, te_len);
 632
 633                if (S_ISDIR(entry.mode))
 634                        /* Match "abc/" against pathspec to
 635                         * decide if we want to descend into "abc"
 636                         * directory.
 637                         */
 638                        strbuf_addch(base, '/');
 639
 640                if (!pathspec_matches(pathspec->raw, base->buf + tn_len, opt->max_depth))
 641                        ;
 642                else if (S_ISREG(entry.mode)) {
 643                        hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 644                }
 645                else if (S_ISDIR(entry.mode)) {
 646                        enum object_type type;
 647                        struct tree_desc sub;
 648                        void *data;
 649                        unsigned long size;
 650
 651                        data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 652                        if (!data)
 653                                die("unable to read tree (%s)",
 654                                    sha1_to_hex(entry.sha1));
 655                        init_tree_desc(&sub, data, size);
 656                        hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 657                        free(data);
 658                }
 659                strbuf_setlen(base, old_baselen);
 660
 661                if (hit && opt->status_only)
 662                        break;
 663        }
 664        return hit;
 665}
 666
 667static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 668                       struct object *obj, const char *name)
 669{
 670        if (obj->type == OBJ_BLOB)
 671                return grep_sha1(opt, obj->sha1, name, 0);
 672        if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 673                struct tree_desc tree;
 674                void *data;
 675                unsigned long size;
 676                struct strbuf base;
 677                int hit, len;
 678
 679                data = read_object_with_reference(obj->sha1, tree_type,
 680                                                  &size, NULL);
 681                if (!data)
 682                        die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 683
 684                len = name ? strlen(name) : 0;
 685                strbuf_init(&base, PATH_MAX + len + 1);
 686                if (len) {
 687                        strbuf_add(&base, name, len);
 688                        strbuf_addch(&base, ':');
 689                }
 690                init_tree_desc(&tree, data, size);
 691                hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 692                strbuf_release(&base);
 693                free(data);
 694                return hit;
 695        }
 696        die("unable to grep from object of type %s", typename(obj->type));
 697}
 698
 699static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 700                        const struct object_array *list)
 701{
 702        unsigned int i;
 703        int hit = 0;
 704        const unsigned int nr = list->nr;
 705
 706        for (i = 0; i < nr; i++) {
 707                struct object *real_obj;
 708                real_obj = deref_tag(list->objects[i].item, NULL, 0);
 709                if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 710                        hit = 1;
 711                        if (opt->status_only)
 712                                break;
 713                }
 714        }
 715        return hit;
 716}
 717
 718static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec)
 719{
 720        struct dir_struct dir;
 721        int i, hit = 0;
 722
 723        memset(&dir, 0, sizeof(dir));
 724        setup_standard_excludes(&dir);
 725
 726        fill_directory(&dir, pathspec->raw);
 727        for (i = 0; i < dir.nr; i++) {
 728                hit |= grep_file(opt, dir.entries[i]->name);
 729                if (hit && opt->status_only)
 730                        break;
 731        }
 732        return hit;
 733}
 734
 735static int context_callback(const struct option *opt, const char *arg,
 736                            int unset)
 737{
 738        struct grep_opt *grep_opt = opt->value;
 739        int value;
 740        const char *endp;
 741
 742        if (unset) {
 743                grep_opt->pre_context = grep_opt->post_context = 0;
 744                return 0;
 745        }
 746        value = strtol(arg, (char **)&endp, 10);
 747        if (*endp) {
 748                return error("switch `%c' expects a numerical value",
 749                             opt->short_name);
 750        }
 751        grep_opt->pre_context = grep_opt->post_context = value;
 752        return 0;
 753}
 754
 755static int file_callback(const struct option *opt, const char *arg, int unset)
 756{
 757        struct grep_opt *grep_opt = opt->value;
 758        FILE *patterns;
 759        int lno = 0;
 760        struct strbuf sb = STRBUF_INIT;
 761
 762        patterns = fopen(arg, "r");
 763        if (!patterns)
 764                die_errno("cannot open '%s'", arg);
 765        while (strbuf_getline(&sb, patterns, '\n') == 0) {
 766                char *s;
 767                size_t len;
 768
 769                /* ignore empty line like grep does */
 770                if (sb.len == 0)
 771                        continue;
 772
 773                s = strbuf_detach(&sb, &len);
 774                append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 775        }
 776        fclose(patterns);
 777        strbuf_release(&sb);
 778        return 0;
 779}
 780
 781static int not_callback(const struct option *opt, const char *arg, int unset)
 782{
 783        struct grep_opt *grep_opt = opt->value;
 784        append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 785        return 0;
 786}
 787
 788static int and_callback(const struct option *opt, const char *arg, int unset)
 789{
 790        struct grep_opt *grep_opt = opt->value;
 791        append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 792        return 0;
 793}
 794
 795static int open_callback(const struct option *opt, const char *arg, int unset)
 796{
 797        struct grep_opt *grep_opt = opt->value;
 798        append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 799        return 0;
 800}
 801
 802static int close_callback(const struct option *opt, const char *arg, int unset)
 803{
 804        struct grep_opt *grep_opt = opt->value;
 805        append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 806        return 0;
 807}
 808
 809static int pattern_callback(const struct option *opt, const char *arg,
 810                            int unset)
 811{
 812        struct grep_opt *grep_opt = opt->value;
 813        append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 814        return 0;
 815}
 816
 817static int help_callback(const struct option *opt, const char *arg, int unset)
 818{
 819        return -1;
 820}
 821
 822int cmd_grep(int argc, const char **argv, const char *prefix)
 823{
 824        int hit = 0;
 825        int cached = 0;
 826        int seen_dashdash = 0;
 827        int external_grep_allowed__ignored;
 828        const char *show_in_pager = NULL, *default_pager = "dummy";
 829        struct grep_opt opt;
 830        struct object_array list = OBJECT_ARRAY_INIT;
 831        const char **paths = NULL;
 832        struct pathspec pathspec;
 833        struct string_list path_list = STRING_LIST_INIT_NODUP;
 834        int i;
 835        int dummy;
 836        int use_index = 1;
 837        struct option options[] = {
 838                OPT_BOOLEAN(0, "cached", &cached,
 839                        "search in index instead of in the work tree"),
 840                OPT_BOOLEAN(0, "index", &use_index,
 841                        "--no-index finds in contents not managed by git"),
 842                OPT_GROUP(""),
 843                OPT_BOOLEAN('v', "invert-match", &opt.invert,
 844                        "show non-matching lines"),
 845                OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 846                        "case insensitive matching"),
 847                OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 848                        "match patterns only at word boundaries"),
 849                OPT_SET_INT('a', "text", &opt.binary,
 850                        "process binary files as text", GREP_BINARY_TEXT),
 851                OPT_SET_INT('I', NULL, &opt.binary,
 852                        "don't match patterns in binary files",
 853                        GREP_BINARY_NOMATCH),
 854                { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 855                        "descend at most <depth> levels", PARSE_OPT_NONEG,
 856                        NULL, 1 },
 857                OPT_GROUP(""),
 858                OPT_BIT('E', "extended-regexp", &opt.regflags,
 859                        "use extended POSIX regular expressions", REG_EXTENDED),
 860                OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
 861                        "use basic POSIX regular expressions (default)",
 862                        REG_EXTENDED),
 863                OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
 864                        "interpret patterns as fixed strings"),
 865                OPT_GROUP(""),
 866                OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
 867                OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 868                OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 869                OPT_NEGBIT(0, "full-name", &opt.relative,
 870                        "show filenames relative to top directory", 1),
 871                OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 872                        "show only filenames instead of matching lines"),
 873                OPT_BOOLEAN(0, "name-only", &opt.name_only,
 874                        "synonym for --files-with-matches"),
 875                OPT_BOOLEAN('L', "files-without-match",
 876                        &opt.unmatch_name_only,
 877                        "show only the names of files without match"),
 878                OPT_BOOLEAN('z', "null", &opt.null_following_name,
 879                        "print NUL after filenames"),
 880                OPT_BOOLEAN('c', "count", &opt.count,
 881                        "show the number of matches instead of matching lines"),
 882                OPT__COLOR(&opt.color, "highlight matches"),
 883                OPT_GROUP(""),
 884                OPT_CALLBACK('C', NULL, &opt, "n",
 885                        "show <n> context lines before and after matches",
 886                        context_callback),
 887                OPT_INTEGER('B', NULL, &opt.pre_context,
 888                        "show <n> context lines before matches"),
 889                OPT_INTEGER('A', NULL, &opt.post_context,
 890                        "show <n> context lines after matches"),
 891                OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 892                        context_callback),
 893                OPT_BOOLEAN('p', "show-function", &opt.funcname,
 894                        "show a line with the function name before matches"),
 895                OPT_GROUP(""),
 896                OPT_CALLBACK('f', NULL, &opt, "file",
 897                        "read patterns from file", file_callback),
 898                { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 899                        "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 900                { OPTION_CALLBACK, 0, "and", &opt, NULL,
 901                  "combine patterns specified with -e",
 902                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 903                OPT_BOOLEAN(0, "or", &dummy, ""),
 904                { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 905                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 906                { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 907                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 908                  open_callback },
 909                { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 910                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 911                  close_callback },
 912                OPT__QUIET(&opt.status_only,
 913                           "indicate hit with exit status without output"),
 914                OPT_BOOLEAN(0, "all-match", &opt.all_match,
 915                        "show only matches from files that match all patterns"),
 916                OPT_GROUP(""),
 917                { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 918                        "pager", "show matching files in the pager",
 919                        PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 920                OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 921                            "allow calling of grep(1) (ignored by this build)"),
 922                { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 923                  PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 924                OPT_END()
 925        };
 926
 927        /*
 928         * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 929         * to show usage information and exit.
 930         */
 931        if (argc == 2 && !strcmp(argv[1], "-h"))
 932                usage_with_options(grep_usage, options);
 933
 934        memset(&opt, 0, sizeof(opt));
 935        opt.prefix = prefix;
 936        opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 937        opt.relative = 1;
 938        opt.pathname = 1;
 939        opt.pattern_tail = &opt.pattern_list;
 940        opt.header_tail = &opt.header_list;
 941        opt.regflags = REG_NEWLINE;
 942        opt.max_depth = -1;
 943
 944        strcpy(opt.color_context, "");
 945        strcpy(opt.color_filename, "");
 946        strcpy(opt.color_function, "");
 947        strcpy(opt.color_lineno, "");
 948        strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 949        strcpy(opt.color_selected, "");
 950        strcpy(opt.color_sep, GIT_COLOR_CYAN);
 951        opt.color = -1;
 952        git_config(grep_config, &opt);
 953        if (opt.color == -1)
 954                opt.color = git_use_color_default;
 955
 956        /*
 957         * If there is no -- then the paths must exist in the working
 958         * tree.  If there is no explicit pattern specified with -e or
 959         * -f, we take the first unrecognized non option to be the
 960         * pattern, but then what follows it must be zero or more
 961         * valid refs up to the -- (if exists), and then existing
 962         * paths.  If there is an explicit pattern, then the first
 963         * unrecognized non option is the beginning of the refs list
 964         * that continues up to the -- (if exists), and then paths.
 965         */
 966        argc = parse_options(argc, argv, prefix, options, grep_usage,
 967                             PARSE_OPT_KEEP_DASHDASH |
 968                             PARSE_OPT_STOP_AT_NON_OPTION |
 969                             PARSE_OPT_NO_INTERNAL_HELP);
 970
 971        if (use_index && !startup_info->have_repository)
 972                /* die the same way as if we did it at the beginning */
 973                setup_git_directory();
 974
 975        /*
 976         * skip a -- separator; we know it cannot be
 977         * separating revisions from pathnames if
 978         * we haven't even had any patterns yet
 979         */
 980        if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 981                argv++;
 982                argc--;
 983        }
 984
 985        /* First unrecognized non-option token */
 986        if (argc > 0 && !opt.pattern_list) {
 987                append_grep_pattern(&opt, argv[0], "command line", 0,
 988                                    GREP_PATTERN);
 989                argv++;
 990                argc--;
 991        }
 992
 993        if (show_in_pager == default_pager)
 994                show_in_pager = git_pager(1);
 995        if (show_in_pager) {
 996                opt.color = 0;
 997                opt.name_only = 1;
 998                opt.null_following_name = 1;
 999                opt.output_priv = &path_list;
1000                opt.output = append_path;
1001                string_list_append(&path_list, show_in_pager);
1002                use_threads = 0;
1003        }
1004
1005        if (!opt.pattern_list)
1006                die("no pattern given.");
1007        if (!opt.fixed && opt.ignore_case)
1008                opt.regflags |= REG_ICASE;
1009        if ((opt.regflags != REG_NEWLINE) && opt.fixed)
1010                die("cannot mix --fixed-strings and regexp");
1011
1012#ifndef NO_PTHREADS
1013        if (online_cpus() == 1 || !grep_threads_ok(&opt))
1014                use_threads = 0;
1015
1016        if (use_threads) {
1017                if (opt.pre_context || opt.post_context)
1018                        print_hunk_marks_between_files = 1;
1019                start_threads(&opt);
1020        }
1021#else
1022        use_threads = 0;
1023#endif
1024
1025        compile_grep_patterns(&opt);
1026
1027        /* Check revs and then paths */
1028        for (i = 0; i < argc; i++) {
1029                const char *arg = argv[i];
1030                unsigned char sha1[20];
1031                /* Is it a rev? */
1032                if (!get_sha1(arg, sha1)) {
1033                        struct object *object = parse_object(sha1);
1034                        if (!object)
1035                                die("bad object %s", arg);
1036                        add_object_array(object, arg, &list);
1037                        continue;
1038                }
1039                if (!strcmp(arg, "--")) {
1040                        i++;
1041                        seen_dashdash = 1;
1042                }
1043                break;
1044        }
1045
1046        /* The rest are paths */
1047        if (!seen_dashdash) {
1048                int j;
1049                for (j = i; j < argc; j++)
1050                        verify_filename(prefix, argv[j]);
1051        }
1052
1053        if (i < argc)
1054                paths = get_pathspec(prefix, argv + i);
1055        else if (prefix) {
1056                paths = xcalloc(2, sizeof(const char *));
1057                paths[0] = prefix;
1058                paths[1] = NULL;
1059        }
1060        init_pathspec(&pathspec, paths);
1061
1062        if (show_in_pager && (cached || list.nr))
1063                die("--open-files-in-pager only works on the worktree");
1064
1065        if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1066                const char *pager = path_list.items[0].string;
1067                int len = strlen(pager);
1068
1069                if (len > 4 && is_dir_sep(pager[len - 5]))
1070                        pager += len - 4;
1071
1072                if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1073                        struct strbuf buf = STRBUF_INIT;
1074                        strbuf_addf(&buf, "+/%s%s",
1075                                        strcmp("less", pager) ? "" : "*",
1076                                        opt.pattern_list->pattern);
1077                        string_list_append(&path_list, buf.buf);
1078                        strbuf_detach(&buf, NULL);
1079                }
1080        }
1081
1082        if (!show_in_pager)
1083                setup_pager();
1084
1085
1086        if (!use_index) {
1087                if (cached)
1088                        die("--cached cannot be used with --no-index.");
1089                if (list.nr)
1090                        die("--no-index cannot be used with revs.");
1091                hit = grep_directory(&opt, &pathspec);
1092        } else if (!list.nr) {
1093                if (!cached)
1094                        setup_work_tree();
1095
1096                hit = grep_cache(&opt, &pathspec, cached);
1097        } else {
1098                if (cached)
1099                        die("both --cached and trees are given.");
1100                hit = grep_objects(&opt, &pathspec, &list);
1101        }
1102
1103        if (use_threads)
1104                hit |= wait_all();
1105        if (hit && show_in_pager)
1106                run_pager(&opt, prefix);
1107        free_grep_patterns(&opt);
1108        return !hit;
1109}