Andrew's git / gitweb.git / builtin/grep.c
?
builtin / grep.con commit Merge branch 'jk/bug-to-abort' (220c6a7)
   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 "pathspec.h"
  21#include "submodule.h"
  22#include "submodule-config.h"
  23
  24static char const * const grep_usage[] = {
  25        N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  26        NULL
  27};
  28
  29static const char *super_prefix;
  30static int recurse_submodules;
  31static struct argv_array submodule_options = ARGV_ARRAY_INIT;
  32static const char *parent_basename;
  33
  34static int grep_submodule_launch(struct grep_opt *opt,
  35                                 const struct grep_source *gs);
  36
  37#define GREP_NUM_THREADS_DEFAULT 8
  38static int num_threads;
  39
  40#ifndef NO_PTHREADS
  41static pthread_t *threads;
  42
  43/* We use one producer thread and THREADS consumer
  44 * threads. The producer adds struct work_items to 'todo' and the
  45 * consumers pick work items from the same array.
  46 */
  47struct work_item {
  48        struct grep_source source;
  49        char done;
  50        struct strbuf out;
  51};
  52
  53/* In the range [todo_done, todo_start) in 'todo' we have work_items
  54 * that have been or are processed by a consumer thread. We haven't
  55 * written the result for these to stdout yet.
  56 *
  57 * The work_items in [todo_start, todo_end) are waiting to be picked
  58 * up by a consumer thread.
  59 *
  60 * The ranges are modulo TODO_SIZE.
  61 */
  62#define TODO_SIZE 128
  63static struct work_item todo[TODO_SIZE];
  64static int todo_start;
  65static int todo_end;
  66static int todo_done;
  67
  68/* Has all work items been added? */
  69static int all_work_added;
  70
  71/* This lock protects all the variables above. */
  72static pthread_mutex_t grep_mutex;
  73
  74static inline void grep_lock(void)
  75{
  76        if (num_threads)
  77                pthread_mutex_lock(&grep_mutex);
  78}
  79
  80static inline void grep_unlock(void)
  81{
  82        if (num_threads)
  83                pthread_mutex_unlock(&grep_mutex);
  84}
  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 skip_first_line;
  98
  99static void add_work(struct grep_opt *opt, enum grep_source_type type,
 100                     const char *name, const char *path, const 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        grep_source_init(&todo[todo_end].source, type, name, path, id);
 109        if (opt->binary != GREP_BINARY_TEXT)
 110                grep_source_load_driver(&todo[todo_end].source);
 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 work_done(struct work_item *w)
 139{
 140        int old_done;
 141
 142        grep_lock();
 143        w->done = 1;
 144        old_done = todo_done;
 145        for(; todo[todo_done].done && todo_done != todo_start;
 146            todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 147                w = &todo[todo_done];
 148                if (w->out.len) {
 149                        const char *p = w->out.buf;
 150                        size_t len = w->out.len;
 151
 152                        /* Skip the leading hunk mark of the first file. */
 153                        if (skip_first_line) {
 154                                while (len) {
 155                                        len--;
 156                                        if (*p++ == '\n')
 157                                                break;
 158                                }
 159                                skip_first_line = 0;
 160                        }
 161
 162                        write_or_die(1, p, len);
 163                }
 164                grep_source_clear(&w->source);
 165        }
 166
 167        if (old_done != todo_done)
 168                pthread_cond_signal(&cond_write);
 169
 170        if (all_work_added && todo_done == todo_end)
 171                pthread_cond_signal(&cond_result);
 172
 173        grep_unlock();
 174}
 175
 176static void *run(void *arg)
 177{
 178        int hit = 0;
 179        struct grep_opt *opt = arg;
 180
 181        while (1) {
 182                struct work_item *w = get_work();
 183                if (!w)
 184                        break;
 185
 186                opt->output_priv = w;
 187                if (w->source.type == GREP_SOURCE_SUBMODULE)
 188                        hit |= grep_submodule_launch(opt, &w->source);
 189                else
 190                        hit |= grep_source(opt, &w->source);
 191                grep_source_clear_data(&w->source);
 192                work_done(w);
 193        }
 194        free_grep_patterns(arg);
 195        free(arg);
 196
 197        return (void*) (intptr_t) hit;
 198}
 199
 200static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 201{
 202        struct work_item *w = opt->output_priv;
 203        strbuf_add(&w->out, buf, size);
 204}
 205
 206static void start_threads(struct grep_opt *opt)
 207{
 208        int i;
 209
 210        pthread_mutex_init(&grep_mutex, NULL);
 211        pthread_mutex_init(&grep_read_mutex, NULL);
 212        pthread_mutex_init(&grep_attr_mutex, NULL);
 213        pthread_cond_init(&cond_add, NULL);
 214        pthread_cond_init(&cond_write, NULL);
 215        pthread_cond_init(&cond_result, NULL);
 216        grep_use_locks = 1;
 217
 218        for (i = 0; i < ARRAY_SIZE(todo); i++) {
 219                strbuf_init(&todo[i].out, 0);
 220        }
 221
 222        threads = xcalloc(num_threads, sizeof(*threads));
 223        for (i = 0; i < num_threads; i++) {
 224                int err;
 225                struct grep_opt *o = grep_opt_dup(opt);
 226                o->output = strbuf_out;
 227                o->debug = 0;
 228                compile_grep_patterns(o);
 229                err = pthread_create(&threads[i], NULL, run, o);
 230
 231                if (err)
 232                        die(_("grep: failed to create thread: %s"),
 233                            strerror(err));
 234        }
 235}
 236
 237static int wait_all(void)
 238{
 239        int hit = 0;
 240        int i;
 241
 242        grep_lock();
 243        all_work_added = 1;
 244
 245        /* Wait until all work is done. */
 246        while (todo_done != todo_end)
 247                pthread_cond_wait(&cond_result, &grep_mutex);
 248
 249        /* Wake up all the consumer threads so they can see that there
 250         * is no more work to do.
 251         */
 252        pthread_cond_broadcast(&cond_add);
 253        grep_unlock();
 254
 255        for (i = 0; i < num_threads; i++) {
 256                void *h;
 257                pthread_join(threads[i], &h);
 258                hit |= (int) (intptr_t) h;
 259        }
 260
 261        free(threads);
 262
 263        pthread_mutex_destroy(&grep_mutex);
 264        pthread_mutex_destroy(&grep_read_mutex);
 265        pthread_mutex_destroy(&grep_attr_mutex);
 266        pthread_cond_destroy(&cond_add);
 267        pthread_cond_destroy(&cond_write);
 268        pthread_cond_destroy(&cond_result);
 269        grep_use_locks = 0;
 270
 271        return hit;
 272}
 273#else /* !NO_PTHREADS */
 274
 275static int wait_all(void)
 276{
 277        return 0;
 278}
 279#endif
 280
 281static int grep_cmd_config(const char *var, const char *value, void *cb)
 282{
 283        int st = grep_config(var, value, cb);
 284        if (git_color_default_config(var, value, cb) < 0)
 285                st = -1;
 286
 287        if (!strcmp(var, "grep.threads")) {
 288                num_threads = git_config_int(var, value);
 289                if (num_threads < 0)
 290                        die(_("invalid number of threads specified (%d) for %s"),
 291                            num_threads, var);
 292        }
 293
 294        return st;
 295}
 296
 297static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
 298{
 299        void *data;
 300
 301        grep_read_lock();
 302        data = read_sha1_file(oid->hash, type, size);
 303        grep_read_unlock();
 304        return data;
 305}
 306
 307static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
 308                     const char *filename, int tree_name_len,
 309                     const char *path)
 310{
 311        struct strbuf pathbuf = STRBUF_INIT;
 312
 313        if (super_prefix) {
 314                strbuf_add(&pathbuf, filename, tree_name_len);
 315                strbuf_addstr(&pathbuf, super_prefix);
 316                strbuf_addstr(&pathbuf, filename + tree_name_len);
 317        } else {
 318                strbuf_addstr(&pathbuf, filename);
 319        }
 320
 321        if (opt->relative && opt->prefix_length) {
 322                char *name = strbuf_detach(&pathbuf, NULL);
 323                quote_path_relative(name + tree_name_len, opt->prefix, &pathbuf);
 324                strbuf_insert(&pathbuf, 0, name, tree_name_len);
 325                free(name);
 326        }
 327
 328#ifndef NO_PTHREADS
 329        if (num_threads) {
 330                add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
 331                strbuf_release(&pathbuf);
 332                return 0;
 333        } else
 334#endif
 335        {
 336                struct grep_source gs;
 337                int hit;
 338
 339                grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
 340                strbuf_release(&pathbuf);
 341                hit = grep_source(opt, &gs);
 342
 343                grep_source_clear(&gs);
 344                return hit;
 345        }
 346}
 347
 348static int grep_file(struct grep_opt *opt, const char *filename)
 349{
 350        struct strbuf buf = STRBUF_INIT;
 351
 352        if (super_prefix)
 353                strbuf_addstr(&buf, super_prefix);
 354        strbuf_addstr(&buf, filename);
 355
 356        if (opt->relative && opt->prefix_length) {
 357                char *name = strbuf_detach(&buf, NULL);
 358                quote_path_relative(name, opt->prefix, &buf);
 359                free(name);
 360        }
 361
 362#ifndef NO_PTHREADS
 363        if (num_threads) {
 364                add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
 365                strbuf_release(&buf);
 366                return 0;
 367        } else
 368#endif
 369        {
 370                struct grep_source gs;
 371                int hit;
 372
 373                grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 374                strbuf_release(&buf);
 375                hit = grep_source(opt, &gs);
 376
 377                grep_source_clear(&gs);
 378                return hit;
 379        }
 380}
 381
 382static void append_path(struct grep_opt *opt, const void *data, size_t len)
 383{
 384        struct string_list *path_list = opt->output_priv;
 385
 386        if (len == 1 && *(const char *)data == '\0')
 387                return;
 388        string_list_append(path_list, xstrndup(data, len));
 389}
 390
 391static void run_pager(struct grep_opt *opt, const char *prefix)
 392{
 393        struct string_list *path_list = opt->output_priv;
 394        struct child_process child = CHILD_PROCESS_INIT;
 395        int i, status;
 396
 397        for (i = 0; i < path_list->nr; i++)
 398                argv_array_push(&child.args, path_list->items[i].string);
 399        child.dir = prefix;
 400        child.use_shell = 1;
 401
 402        status = run_command(&child);
 403        if (status)
 404                exit(status);
 405}
 406
 407static void compile_submodule_options(const struct grep_opt *opt,
 408                                      const char **argv,
 409                                      int cached, int untracked,
 410                                      int opt_exclude, int use_index,
 411                                      int pattern_type_arg)
 412{
 413        struct grep_pat *pattern;
 414
 415        if (recurse_submodules)
 416                argv_array_push(&submodule_options, "--recurse-submodules");
 417
 418        if (cached)
 419                argv_array_push(&submodule_options, "--cached");
 420        if (!use_index)
 421                argv_array_push(&submodule_options, "--no-index");
 422        if (untracked)
 423                argv_array_push(&submodule_options, "--untracked");
 424        if (opt_exclude > 0)
 425                argv_array_push(&submodule_options, "--exclude-standard");
 426
 427        if (opt->invert)
 428                argv_array_push(&submodule_options, "-v");
 429        if (opt->ignore_case)
 430                argv_array_push(&submodule_options, "-i");
 431        if (opt->word_regexp)
 432                argv_array_push(&submodule_options, "-w");
 433        switch (opt->binary) {
 434        case GREP_BINARY_NOMATCH:
 435                argv_array_push(&submodule_options, "-I");
 436                break;
 437        case GREP_BINARY_TEXT:
 438                argv_array_push(&submodule_options, "-a");
 439                break;
 440        default:
 441                break;
 442        }
 443        if (opt->allow_textconv)
 444                argv_array_push(&submodule_options, "--textconv");
 445        if (opt->max_depth != -1)
 446                argv_array_pushf(&submodule_options, "--max-depth=%d",
 447                                 opt->max_depth);
 448        if (opt->linenum)
 449                argv_array_push(&submodule_options, "-n");
 450        if (!opt->pathname)
 451                argv_array_push(&submodule_options, "-h");
 452        if (!opt->relative)
 453                argv_array_push(&submodule_options, "--full-name");
 454        if (opt->name_only)
 455                argv_array_push(&submodule_options, "-l");
 456        if (opt->unmatch_name_only)
 457                argv_array_push(&submodule_options, "-L");
 458        if (opt->null_following_name)
 459                argv_array_push(&submodule_options, "-z");
 460        if (opt->count)
 461                argv_array_push(&submodule_options, "-c");
 462        if (opt->file_break)
 463                argv_array_push(&submodule_options, "--break");
 464        if (opt->heading)
 465                argv_array_push(&submodule_options, "--heading");
 466        if (opt->pre_context)
 467                argv_array_pushf(&submodule_options, "--before-context=%d",
 468                                 opt->pre_context);
 469        if (opt->post_context)
 470                argv_array_pushf(&submodule_options, "--after-context=%d",
 471                                 opt->post_context);
 472        if (opt->funcname)
 473                argv_array_push(&submodule_options, "-p");
 474        if (opt->funcbody)
 475                argv_array_push(&submodule_options, "-W");
 476        if (opt->all_match)
 477                argv_array_push(&submodule_options, "--all-match");
 478        if (opt->debug)
 479                argv_array_push(&submodule_options, "--debug");
 480        if (opt->status_only)
 481                argv_array_push(&submodule_options, "-q");
 482
 483        switch (pattern_type_arg) {
 484        case GREP_PATTERN_TYPE_BRE:
 485                argv_array_push(&submodule_options, "-G");
 486                break;
 487        case GREP_PATTERN_TYPE_ERE:
 488                argv_array_push(&submodule_options, "-E");
 489                break;
 490        case GREP_PATTERN_TYPE_FIXED:
 491                argv_array_push(&submodule_options, "-F");
 492                break;
 493        case GREP_PATTERN_TYPE_PCRE:
 494                argv_array_push(&submodule_options, "-P");
 495                break;
 496        case GREP_PATTERN_TYPE_UNSPECIFIED:
 497                break;
 498        }
 499
 500        for (pattern = opt->pattern_list; pattern != NULL;
 501             pattern = pattern->next) {
 502                switch (pattern->token) {
 503                case GREP_PATTERN:
 504                        argv_array_pushf(&submodule_options, "-e%s",
 505                                         pattern->pattern);
 506                        break;
 507                case GREP_AND:
 508                case GREP_OPEN_PAREN:
 509                case GREP_CLOSE_PAREN:
 510                case GREP_NOT:
 511                case GREP_OR:
 512                        argv_array_push(&submodule_options, pattern->pattern);
 513                        break;
 514                /* BODY and HEAD are not used by git-grep */
 515                case GREP_PATTERN_BODY:
 516                case GREP_PATTERN_HEAD:
 517                        break;
 518                }
 519        }
 520
 521        /*
 522         * Limit number of threads for child process to use.
 523         * This is to prevent potential fork-bomb behavior of git-grep as each
 524         * submodule process has its own thread pool.
 525         */
 526        argv_array_pushf(&submodule_options, "--threads=%d",
 527                         (num_threads + 1) / 2);
 528
 529        /* Add Pathspecs */
 530        argv_array_push(&submodule_options, "--");
 531        for (; *argv; argv++)
 532                argv_array_push(&submodule_options, *argv);
 533}
 534
 535/*
 536 * Launch child process to grep contents of a submodule
 537 */
 538static int grep_submodule_launch(struct grep_opt *opt,
 539                                 const struct grep_source *gs)
 540{
 541        struct child_process cp = CHILD_PROCESS_INIT;
 542        int status, i;
 543        const char *end_of_base;
 544        const char *name;
 545        struct strbuf child_output = STRBUF_INIT;
 546
 547        end_of_base = strchr(gs->name, ':');
 548        if (gs->identifier && end_of_base)
 549                name = end_of_base + 1;
 550        else
 551                name = gs->name;
 552
 553        prepare_submodule_repo_env(&cp.env_array);
 554        argv_array_push(&cp.env_array, GIT_DIR_ENVIRONMENT);
 555
 556        if (opt->relative && opt->prefix_length)
 557                argv_array_pushf(&cp.env_array, "%s=%s",
 558                                 GIT_TOPLEVEL_PREFIX_ENVIRONMENT,
 559                                 opt->prefix);
 560
 561        /* Add super prefix */
 562        argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
 563                         super_prefix ? super_prefix : "",
 564                         name);
 565        argv_array_push(&cp.args, "grep");
 566
 567        /*
 568         * Add basename of parent project
 569         * When performing grep on a tree object the filename is prefixed
 570         * with the object's name: 'tree-name:filename'.  In order to
 571         * provide uniformity of output we want to pass the name of the
 572         * parent project's object name to the submodule so the submodule can
 573         * prefix its output with the parent's name and not its own SHA1.
 574         */
 575        if (gs->identifier && end_of_base)
 576                argv_array_pushf(&cp.args, "--parent-basename=%.*s",
 577                                 (int) (end_of_base - gs->name),
 578                                 gs->name);
 579
 580        /* Add options */
 581        for (i = 0; i < submodule_options.argc; i++) {
 582                /*
 583                 * If there is a tree identifier for the submodule, add the
 584                 * rev after adding the submodule options but before the
 585                 * pathspecs.  To do this we listen for the '--' and insert the
 586                 * sha1 before pushing the '--' onto the child process argv
 587                 * array.
 588                 */
 589                if (gs->identifier &&
 590                    !strcmp("--", submodule_options.argv[i])) {
 591                        argv_array_push(&cp.args, sha1_to_hex(gs->identifier));
 592                }
 593
 594                argv_array_push(&cp.args, submodule_options.argv[i]);
 595        }
 596
 597        cp.git_cmd = 1;
 598        cp.dir = gs->path;
 599
 600        /*
 601         * Capture output to output buffer and check the return code from the
 602         * child process.  A '0' indicates a hit, a '1' indicates no hit and
 603         * anything else is an error.
 604         */
 605        status = capture_command(&cp, &child_output, 0);
 606        if (status && (status != 1)) {
 607                /* flush the buffer */
 608                write_or_die(1, child_output.buf, child_output.len);
 609                die("process for submodule '%s' failed with exit code: %d",
 610                    gs->name, status);
 611        }
 612
 613        opt->output(opt, child_output.buf, child_output.len);
 614        strbuf_release(&child_output);
 615        /* invert the return code to make a hit equal to 1 */
 616        return !status;
 617}
 618
 619/*
 620 * Prep grep structures for a submodule grep
 621 * sha1: the sha1 of the submodule or NULL if using the working tree
 622 * filename: name of the submodule including tree name of parent
 623 * path: location of the submodule
 624 */
 625static int grep_submodule(struct grep_opt *opt, const unsigned char *sha1,
 626                          const char *filename, const char *path)
 627{
 628        if (!is_submodule_initialized(path))
 629                return 0;
 630        if (!is_submodule_populated_gently(path, NULL)) {
 631                /*
 632                 * If searching history, check for the presense of the
 633                 * submodule's gitdir before skipping the submodule.
 634                 */
 635                if (sha1) {
 636                        const struct submodule *sub =
 637                                        submodule_from_path(null_sha1, path);
 638                        if (sub)
 639                                path = git_path("modules/%s", sub->name);
 640
 641                        if (!(is_directory(path) && is_git_directory(path)))
 642                                return 0;
 643                } else {
 644                        return 0;
 645                }
 646        }
 647
 648#ifndef NO_PTHREADS
 649        if (num_threads) {
 650                add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, sha1);
 651                return 0;
 652        } else
 653#endif
 654        {
 655                struct grep_source gs;
 656                int hit;
 657
 658                grep_source_init(&gs, GREP_SOURCE_SUBMODULE,
 659                                 filename, path, sha1);
 660                hit = grep_submodule_launch(opt, &gs);
 661
 662                grep_source_clear(&gs);
 663                return hit;
 664        }
 665}
 666
 667static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
 668                      int cached)
 669{
 670        int hit = 0;
 671        int nr;
 672        struct strbuf name = STRBUF_INIT;
 673        int name_base_len = 0;
 674        if (super_prefix) {
 675                name_base_len = strlen(super_prefix);
 676                strbuf_addstr(&name, super_prefix);
 677        }
 678
 679        read_cache();
 680
 681        for (nr = 0; nr < active_nr; nr++) {
 682                const struct cache_entry *ce = active_cache[nr];
 683                strbuf_setlen(&name, name_base_len);
 684                strbuf_addstr(&name, ce->name);
 685
 686                if (S_ISREG(ce->ce_mode) &&
 687                    match_pathspec(pathspec, name.buf, name.len, 0, NULL,
 688                                   S_ISDIR(ce->ce_mode) ||
 689                                   S_ISGITLINK(ce->ce_mode))) {
 690                        /*
 691                         * If CE_VALID is on, we assume worktree file and its
 692                         * cache entry are identical, even if worktree file has
 693                         * been modified, so use cache version instead
 694                         */
 695                        if (cached || (ce->ce_flags & CE_VALID) ||
 696                            ce_skip_worktree(ce)) {
 697                                if (ce_stage(ce) || ce_intent_to_add(ce))
 698                                        continue;
 699                                hit |= grep_oid(opt, &ce->oid, ce->name,
 700                                                 0, ce->name);
 701                        } else {
 702                                hit |= grep_file(opt, ce->name);
 703                        }
 704                } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 705                           submodule_path_match(pathspec, name.buf, NULL)) {
 706                        hit |= grep_submodule(opt, NULL, ce->name, ce->name);
 707                } else {
 708                        continue;
 709                }
 710
 711                if (ce_stage(ce)) {
 712                        do {
 713                                nr++;
 714                        } while (nr < active_nr &&
 715                                 !strcmp(ce->name, active_cache[nr]->name));
 716                        nr--; /* compensate for loop control */
 717                }
 718                if (hit && opt->status_only)
 719                        break;
 720        }
 721
 722        strbuf_release(&name);
 723        return hit;
 724}
 725
 726static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 727                     struct tree_desc *tree, struct strbuf *base, int tn_len,
 728                     int check_attr)
 729{
 730        int hit = 0;
 731        enum interesting match = entry_not_interesting;
 732        struct name_entry entry;
 733        int old_baselen = base->len;
 734        struct strbuf name = STRBUF_INIT;
 735        int name_base_len = 0;
 736        if (super_prefix) {
 737                strbuf_addstr(&name, super_prefix);
 738                name_base_len = name.len;
 739        }
 740
 741        while (tree_entry(tree, &entry)) {
 742                int te_len = tree_entry_len(&entry);
 743
 744                if (match != all_entries_interesting) {
 745                        strbuf_addstr(&name, base->buf + tn_len);
 746                        match = tree_entry_interesting(&entry, &name,
 747                                                       0, pathspec);
 748                        strbuf_setlen(&name, name_base_len);
 749
 750                        if (match == all_entries_not_interesting)
 751                                break;
 752                        if (match == entry_not_interesting)
 753                                continue;
 754                }
 755
 756                strbuf_add(base, entry.path, te_len);
 757
 758                if (S_ISREG(entry.mode)) {
 759                        hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
 760                                         check_attr ? base->buf + tn_len : NULL);
 761                } else if (S_ISDIR(entry.mode)) {
 762                        enum object_type type;
 763                        struct tree_desc sub;
 764                        void *data;
 765                        unsigned long size;
 766
 767                        data = lock_and_read_oid_file(entry.oid, &type, &size);
 768                        if (!data)
 769                                die(_("unable to read tree (%s)"),
 770                                    oid_to_hex(entry.oid));
 771
 772                        strbuf_addch(base, '/');
 773                        init_tree_desc(&sub, data, size);
 774                        hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 775                                         check_attr);
 776                        free(data);
 777                } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 778                        hit |= grep_submodule(opt, entry.oid->hash, base->buf,
 779                                              base->buf + tn_len);
 780                }
 781
 782                strbuf_setlen(base, old_baselen);
 783
 784                if (hit && opt->status_only)
 785                        break;
 786        }
 787
 788        strbuf_release(&name);
 789        return hit;
 790}
 791
 792static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 793                       struct object *obj, const char *name, const char *path)
 794{
 795        if (obj->type == OBJ_BLOB)
 796                return grep_oid(opt, &obj->oid, name, 0, path);
 797        if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 798                struct tree_desc tree;
 799                void *data;
 800                unsigned long size;
 801                struct strbuf base;
 802                int hit, len;
 803
 804                grep_read_lock();
 805                data = read_object_with_reference(obj->oid.hash, tree_type,
 806                                                  &size, NULL);
 807                grep_read_unlock();
 808
 809                if (!data)
 810                        die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 811
 812                /* Use parent's name as base when recursing submodules */
 813                if (recurse_submodules && parent_basename)
 814                        name = parent_basename;
 815
 816                len = name ? strlen(name) : 0;
 817                strbuf_init(&base, PATH_MAX + len + 1);
 818                if (len) {
 819                        strbuf_add(&base, name, len);
 820                        strbuf_addch(&base, ':');
 821                }
 822                init_tree_desc(&tree, data, size);
 823                hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 824                                obj->type == OBJ_COMMIT);
 825                strbuf_release(&base);
 826                free(data);
 827                return hit;
 828        }
 829        die(_("unable to grep from object of type %s"), typename(obj->type));
 830}
 831
 832static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 833                        const struct object_array *list)
 834{
 835        unsigned int i;
 836        int hit = 0;
 837        const unsigned int nr = list->nr;
 838
 839        for (i = 0; i < nr; i++) {
 840                struct object *real_obj;
 841                real_obj = deref_tag(list->objects[i].item, NULL, 0);
 842
 843                /* load the gitmodules file for this rev */
 844                if (recurse_submodules) {
 845                        submodule_free();
 846                        gitmodules_config_sha1(real_obj->oid.hash);
 847                }
 848                if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
 849                        hit = 1;
 850                        if (opt->status_only)
 851                                break;
 852                }
 853        }
 854        return hit;
 855}
 856
 857static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 858                          int exc_std, int use_index)
 859{
 860        struct dir_struct dir;
 861        int i, hit = 0;
 862
 863        memset(&dir, 0, sizeof(dir));
 864        if (!use_index)
 865                dir.flags |= DIR_NO_GITLINKS;
 866        if (exc_std)
 867                setup_standard_excludes(&dir);
 868
 869        fill_directory(&dir, &the_index, pathspec);
 870        for (i = 0; i < dir.nr; i++) {
 871                if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
 872                        continue;
 873                hit |= grep_file(opt, dir.entries[i]->name);
 874                if (hit && opt->status_only)
 875                        break;
 876        }
 877        return hit;
 878}
 879
 880static int context_callback(const struct option *opt, const char *arg,
 881                            int unset)
 882{
 883        struct grep_opt *grep_opt = opt->value;
 884        int value;
 885        const char *endp;
 886
 887        if (unset) {
 888                grep_opt->pre_context = grep_opt->post_context = 0;
 889                return 0;
 890        }
 891        value = strtol(arg, (char **)&endp, 10);
 892        if (*endp) {
 893                return error(_("switch `%c' expects a numerical value"),
 894                             opt->short_name);
 895        }
 896        grep_opt->pre_context = grep_opt->post_context = value;
 897        return 0;
 898}
 899
 900static int file_callback(const struct option *opt, const char *arg, int unset)
 901{
 902        struct grep_opt *grep_opt = opt->value;
 903        int from_stdin = !strcmp(arg, "-");
 904        FILE *patterns;
 905        int lno = 0;
 906        struct strbuf sb = STRBUF_INIT;
 907
 908        patterns = from_stdin ? stdin : fopen(arg, "r");
 909        if (!patterns)
 910                die_errno(_("cannot open '%s'"), arg);
 911        while (strbuf_getline(&sb, patterns) == 0) {
 912                /* ignore empty line like grep does */
 913                if (sb.len == 0)
 914                        continue;
 915
 916                append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 917                                GREP_PATTERN);
 918        }
 919        if (!from_stdin)
 920                fclose(patterns);
 921        strbuf_release(&sb);
 922        return 0;
 923}
 924
 925static int not_callback(const struct option *opt, const char *arg, int unset)
 926{
 927        struct grep_opt *grep_opt = opt->value;
 928        append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 929        return 0;
 930}
 931
 932static int and_callback(const struct option *opt, const char *arg, int unset)
 933{
 934        struct grep_opt *grep_opt = opt->value;
 935        append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 936        return 0;
 937}
 938
 939static int open_callback(const struct option *opt, const char *arg, int unset)
 940{
 941        struct grep_opt *grep_opt = opt->value;
 942        append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 943        return 0;
 944}
 945
 946static int close_callback(const struct option *opt, const char *arg, int unset)
 947{
 948        struct grep_opt *grep_opt = opt->value;
 949        append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 950        return 0;
 951}
 952
 953static int pattern_callback(const struct option *opt, const char *arg,
 954                            int unset)
 955{
 956        struct grep_opt *grep_opt = opt->value;
 957        append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 958        return 0;
 959}
 960
 961int cmd_grep(int argc, const char **argv, const char *prefix)
 962{
 963        int hit = 0;
 964        int cached = 0, untracked = 0, opt_exclude = -1;
 965        int seen_dashdash = 0;
 966        int external_grep_allowed__ignored;
 967        const char *show_in_pager = NULL, *default_pager = "dummy";
 968        struct grep_opt opt;
 969        struct object_array list = OBJECT_ARRAY_INIT;
 970        struct pathspec pathspec;
 971        struct string_list path_list = STRING_LIST_INIT_NODUP;
 972        int i;
 973        int dummy;
 974        int use_index = 1;
 975        int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 976        int allow_revs;
 977
 978        struct option options[] = {
 979                OPT_BOOL(0, "cached", &cached,
 980                        N_("search in index instead of in the work tree")),
 981                OPT_NEGBIT(0, "no-index", &use_index,
 982                         N_("find in contents not managed by git"), 1),
 983                OPT_BOOL(0, "untracked", &untracked,
 984                        N_("search in both tracked and untracked files")),
 985                OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 986                            N_("ignore files specified via '.gitignore'"), 1),
 987                OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 988                         N_("recursively search in each submodule")),
 989                OPT_STRING(0, "parent-basename", &parent_basename,
 990                           N_("basename"),
 991                           N_("prepend parent project's basename to output")),
 992                OPT_GROUP(""),
 993                OPT_BOOL('v', "invert-match", &opt.invert,
 994                        N_("show non-matching lines")),
 995                OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 996                        N_("case insensitive matching")),
 997                OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 998                        N_("match patterns only at word boundaries")),
 999                OPT_SET_INT('a', "text", &opt.binary,
1000                        N_("process binary files as text"), GREP_BINARY_TEXT),
1001                OPT_SET_INT('I', NULL, &opt.binary,
1002                        N_("don't match patterns in binary files"),
1003                        GREP_BINARY_NOMATCH),
1004                OPT_BOOL(0, "textconv", &opt.allow_textconv,
1005                         N_("process binary files with textconv filters")),
1006                { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
1007                        N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
1008                        NULL, 1 },
1009                OPT_GROUP(""),
1010                OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
1011                            N_("use extended POSIX regular expressions"),
1012                            GREP_PATTERN_TYPE_ERE),
1013                OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
1014                            N_("use basic POSIX regular expressions (default)"),
1015                            GREP_PATTERN_TYPE_BRE),
1016                OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
1017                            N_("interpret patterns as fixed strings"),
1018                            GREP_PATTERN_TYPE_FIXED),
1019                OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
1020                            N_("use Perl-compatible regular expressions"),
1021                            GREP_PATTERN_TYPE_PCRE),
1022                OPT_GROUP(""),
1023                OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1024                OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1025                OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1026                OPT_NEGBIT(0, "full-name", &opt.relative,
1027                        N_("show filenames relative to top directory"), 1),
1028                OPT_BOOL('l', "files-with-matches", &opt.name_only,
1029                        N_("show only filenames instead of matching lines")),
1030                OPT_BOOL(0, "name-only", &opt.name_only,
1031                        N_("synonym for --files-with-matches")),
1032                OPT_BOOL('L', "files-without-match",
1033                        &opt.unmatch_name_only,
1034                        N_("show only the names of files without match")),
1035                OPT_BOOL('z', "null", &opt.null_following_name,
1036                        N_("print NUL after filenames")),
1037                OPT_BOOL('c', "count", &opt.count,
1038                        N_("show the number of matches instead of matching lines")),
1039                OPT__COLOR(&opt.color, N_("highlight matches")),
1040                OPT_BOOL(0, "break", &opt.file_break,
1041                        N_("print empty line between matches from different files")),
1042                OPT_BOOL(0, "heading", &opt.heading,
1043                        N_("show filename only once above matches from same file")),
1044                OPT_GROUP(""),
1045                OPT_CALLBACK('C', "context", &opt, N_("n"),
1046                        N_("show <n> context lines before and after matches"),
1047                        context_callback),
1048                OPT_INTEGER('B', "before-context", &opt.pre_context,
1049                        N_("show <n> context lines before matches")),
1050                OPT_INTEGER('A', "after-context", &opt.post_context,
1051                        N_("show <n> context lines after matches")),
1052                OPT_INTEGER(0, "threads", &num_threads,
1053                        N_("use <n> worker threads")),
1054                OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1055                        context_callback),
1056                OPT_BOOL('p', "show-function", &opt.funcname,
1057                        N_("show a line with the function name before matches")),
1058                OPT_BOOL('W', "function-context", &opt.funcbody,
1059                        N_("show the surrounding function")),
1060                OPT_GROUP(""),
1061                OPT_CALLBACK('f', NULL, &opt, N_("file"),
1062                        N_("read patterns from file"), file_callback),
1063                { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1064                        N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1065                { OPTION_CALLBACK, 0, "and", &opt, NULL,
1066                  N_("combine patterns specified with -e"),
1067                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1068                OPT_BOOL(0, "or", &dummy, ""),
1069                { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1070                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1071                { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1072                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1073                  open_callback },
1074                { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1075                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1076                  close_callback },
1077                OPT__QUIET(&opt.status_only,
1078                           N_("indicate hit with exit status without output")),
1079                OPT_BOOL(0, "all-match", &opt.all_match,
1080                        N_("show only matches from files that match all patterns")),
1081                { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1082                  N_("show parse tree for grep expression"),
1083                  PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1084                OPT_GROUP(""),
1085                { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1086                        N_("pager"), N_("show matching files in the pager"),
1087                        PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1088                OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1089                         N_("allow calling of grep(1) (ignored by this build)")),
1090                OPT_END()
1091        };
1092
1093        init_grep_defaults();
1094        git_config(grep_cmd_config, NULL);
1095        grep_init(&opt, prefix);
1096        super_prefix = get_super_prefix();
1097
1098        /*
1099         * If there is no -- then the paths must exist in the working
1100         * tree.  If there is no explicit pattern specified with -e or
1101         * -f, we take the first unrecognized non option to be the
1102         * pattern, but then what follows it must be zero or more
1103         * valid refs up to the -- (if exists), and then existing
1104         * paths.  If there is an explicit pattern, then the first
1105         * unrecognized non option is the beginning of the refs list
1106         * that continues up to the -- (if exists), and then paths.
1107         */
1108        argc = parse_options(argc, argv, prefix, options, grep_usage,
1109                             PARSE_OPT_KEEP_DASHDASH |
1110                             PARSE_OPT_STOP_AT_NON_OPTION);
1111        grep_commit_pattern_type(pattern_type_arg, &opt);
1112
1113        if (use_index && !startup_info->have_repository) {
1114                int fallback = 0;
1115                git_config_get_bool("grep.fallbacktonoindex", &fallback);
1116                if (fallback)
1117                        use_index = 0;
1118                else
1119                        /* die the same way as if we did it at the beginning */
1120                        setup_git_directory();
1121        }
1122
1123        /*
1124         * skip a -- separator; we know it cannot be
1125         * separating revisions from pathnames if
1126         * we haven't even had any patterns yet
1127         */
1128        if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1129                argv++;
1130                argc--;
1131        }
1132
1133        /* First unrecognized non-option token */
1134        if (argc > 0 && !opt.pattern_list) {
1135                append_grep_pattern(&opt, argv[0], "command line", 0,
1136                                    GREP_PATTERN);
1137                argv++;
1138                argc--;
1139        }
1140
1141        if (show_in_pager == default_pager)
1142                show_in_pager = git_pager(1);
1143        if (show_in_pager) {
1144                opt.color = 0;
1145                opt.name_only = 1;
1146                opt.null_following_name = 1;
1147                opt.output_priv = &path_list;
1148                opt.output = append_path;
1149                string_list_append(&path_list, show_in_pager);
1150        }
1151
1152        if (!opt.pattern_list)
1153                die(_("no pattern given."));
1154        if (!opt.fixed && opt.ignore_case)
1155                opt.regflags |= REG_ICASE;
1156
1157        compile_grep_patterns(&opt);
1158
1159        /*
1160         * We have to find "--" in a separate pass, because its presence
1161         * influences how we will parse arguments that come before it.
1162         */
1163        for (i = 0; i < argc; i++) {
1164                if (!strcmp(argv[i], "--")) {
1165                        seen_dashdash = 1;
1166                        break;
1167                }
1168        }
1169
1170        /*
1171         * Resolve any rev arguments. If we have a dashdash, then everything up
1172         * to it must resolve as a rev. If not, then we stop at the first
1173         * non-rev and assume everything else is a path.
1174         */
1175        allow_revs = use_index && !untracked;
1176        for (i = 0; i < argc; i++) {
1177                const char *arg = argv[i];
1178                struct object_id oid;
1179                struct object_context oc;
1180                struct object *object;
1181
1182                if (!strcmp(arg, "--")) {
1183                        i++;
1184                        break;
1185                }
1186
1187                if (!allow_revs) {
1188                        if (seen_dashdash)
1189                                die(_("--no-index or --untracked cannot be used with revs"));
1190                        break;
1191                }
1192
1193                if (get_sha1_with_context(arg, 0, oid.hash, &oc)) {
1194                        if (seen_dashdash)
1195                                die(_("unable to resolve revision: %s"), arg);
1196                        break;
1197                }
1198
1199                object = parse_object_or_die(&oid, arg);
1200                if (!seen_dashdash)
1201                        verify_non_filename(prefix, arg);
1202                add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1203        }
1204
1205        /*
1206         * Anything left over is presumed to be a path. But in the non-dashdash
1207         * "do what I mean" case, we verify and complain when that isn't true.
1208         */
1209        if (!seen_dashdash) {
1210                int j;
1211                for (j = i; j < argc; j++)
1212                        verify_filename(prefix, argv[j], j == i && allow_revs);
1213        }
1214
1215        parse_pathspec(&pathspec, 0,
1216                       PATHSPEC_PREFER_CWD |
1217                       (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1218                       prefix, argv + i);
1219        pathspec.max_depth = opt.max_depth;
1220        pathspec.recursive = 1;
1221
1222#ifndef NO_PTHREADS
1223        if (list.nr || cached || show_in_pager)
1224                num_threads = 0;
1225        else if (num_threads == 0)
1226                num_threads = GREP_NUM_THREADS_DEFAULT;
1227        else if (num_threads < 0)
1228                die(_("invalid number of threads specified (%d)"), num_threads);
1229#else
1230        num_threads = 0;
1231#endif
1232
1233#ifndef NO_PTHREADS
1234        if (num_threads) {
1235                if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1236                    && (opt.pre_context || opt.post_context ||
1237                        opt.file_break || opt.funcbody))
1238                        skip_first_line = 1;
1239                start_threads(&opt);
1240        }
1241#endif
1242
1243        if (recurse_submodules) {
1244                gitmodules_config();
1245                compile_submodule_options(&opt, argv + i, cached, untracked,
1246                                          opt_exclude, use_index,
1247                                          pattern_type_arg);
1248        }
1249
1250        if (show_in_pager && (cached || list.nr))
1251                die(_("--open-files-in-pager only works on the worktree"));
1252
1253        if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1254                const char *pager = path_list.items[0].string;
1255                int len = strlen(pager);
1256
1257                if (len > 4 && is_dir_sep(pager[len - 5]))
1258                        pager += len - 4;
1259
1260                if (opt.ignore_case && !strcmp("less", pager))
1261                        string_list_append(&path_list, "-I");
1262
1263                if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1264                        struct strbuf buf = STRBUF_INIT;
1265                        strbuf_addf(&buf, "+/%s%s",
1266                                        strcmp("less", pager) ? "" : "*",
1267                                        opt.pattern_list->pattern);
1268                        string_list_append(&path_list, buf.buf);
1269                        strbuf_detach(&buf, NULL);
1270                }
1271        }
1272
1273        if (recurse_submodules && (!use_index || untracked))
1274                die(_("option not supported with --recurse-submodules."));
1275
1276        if (!show_in_pager && !opt.status_only)
1277                setup_pager();
1278
1279        if (!use_index && (untracked || cached))
1280                die(_("--cached or --untracked cannot be used with --no-index."));
1281
1282        if (!use_index || untracked) {
1283                int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1284                hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1285        } else if (0 <= opt_exclude) {
1286                die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1287        } else if (!list.nr) {
1288                if (!cached)
1289                        setup_work_tree();
1290
1291                hit = grep_cache(&opt, &pathspec, cached);
1292        } else {
1293                if (cached)
1294                        die(_("both --cached and trees are given."));
1295                hit = grep_objects(&opt, &pathspec, &list);
1296        }
1297
1298        if (num_threads)
1299                hit |= wait_all();
1300        if (hit && show_in_pager)
1301                run_pager(&opt, prefix);
1302        clear_pathspec(&pathspec);
1303        free_grep_patterns(&opt);
1304        return !hit;
1305}