builtin-grep.con commit Make index file locking code reusable to others. (021b6e4)
   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 <regex.h>
  14#include <fnmatch.h>
  15#include <sys/wait.h>
  16
  17/*
  18 * git grep pathspecs are somewhat different from diff-tree pathspecs;
  19 * pathname wildcards are allowed.
  20 */
  21static int pathspec_matches(const char **paths, const char *name)
  22{
  23        int namelen, i;
  24        if (!paths || !*paths)
  25                return 1;
  26        namelen = strlen(name);
  27        for (i = 0; paths[i]; i++) {
  28                const char *match = paths[i];
  29                int matchlen = strlen(match);
  30                const char *cp, *meta;
  31
  32                if ((matchlen <= namelen) &&
  33                    !strncmp(name, match, matchlen) &&
  34                    (match[matchlen-1] == '/' ||
  35                     name[matchlen] == '\0' || name[matchlen] == '/'))
  36                        return 1;
  37                if (!fnmatch(match, name, 0))
  38                        return 1;
  39                if (name[namelen-1] != '/')
  40                        continue;
  41
  42                /* We are being asked if the directory ("name") is worth
  43                 * descending into.
  44                 *
  45                 * Find the longest leading directory name that does
  46                 * not have metacharacter in the pathspec; the name
  47                 * we are looking at must overlap with that directory.
  48                 */
  49                for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
  50                        char ch = *cp;
  51                        if (ch == '*' || ch == '[' || ch == '?') {
  52                                meta = cp;
  53                                break;
  54                        }
  55                }
  56                if (!meta)
  57                        meta = cp; /* fully literal */
  58
  59                if (namelen <= meta - match) {
  60                        /* Looking at "Documentation/" and
  61                         * the pattern says "Documentation/howto/", or
  62                         * "Documentation/diff*.txt".  The name we
  63                         * have should match prefix.
  64                         */
  65                        if (!memcmp(match, name, namelen))
  66                                return 1;
  67                        continue;
  68                }
  69
  70                if (meta - match < namelen) {
  71                        /* Looking at "Documentation/howto/" and
  72                         * the pattern says "Documentation/h*";
  73                         * match up to "Do.../h"; this avoids descending
  74                         * into "Documentation/technical/".
  75                         */
  76                        if (!memcmp(match, name, meta - match))
  77                                return 1;
  78                        continue;
  79                }
  80        }
  81        return 0;
  82}
  83
  84struct grep_pat {
  85        struct grep_pat *next;
  86        const char *origin;
  87        int no;
  88        const char *pattern;
  89        regex_t regexp;
  90};
  91
  92struct grep_opt {
  93        struct grep_pat *pattern_list;
  94        struct grep_pat **pattern_tail;
  95        regex_t regexp;
  96        unsigned linenum:1;
  97        unsigned invert:1;
  98        unsigned name_only:1;
  99        unsigned unmatch_name_only:1;
 100        unsigned count:1;
 101        unsigned word_regexp:1;
 102        unsigned fixed:1;
 103#define GREP_BINARY_DEFAULT     0
 104#define GREP_BINARY_NOMATCH     1
 105#define GREP_BINARY_TEXT        2
 106        unsigned binary:2;
 107        int regflags;
 108        unsigned pre_context;
 109        unsigned post_context;
 110};
 111
 112static void add_pattern(struct grep_opt *opt, const char *pat,
 113                        const char *origin, int no)
 114{
 115        struct grep_pat *p = xcalloc(1, sizeof(*p));
 116        p->pattern = pat;
 117        p->origin = origin;
 118        p->no = no;
 119        *opt->pattern_tail = p;
 120        opt->pattern_tail = &p->next;
 121        p->next = NULL;
 122}
 123
 124static void compile_patterns(struct grep_opt *opt)
 125{
 126        struct grep_pat *p;
 127        for (p = opt->pattern_list; p; p = p->next) {
 128                int err = regcomp(&p->regexp, p->pattern, opt->regflags);
 129                if (err) {
 130                        char errbuf[1024];
 131                        char where[1024];
 132                        if (p->no)
 133                                sprintf(where, "In '%s' at %d, ",
 134                                        p->origin, p->no);
 135                        else if (p->origin)
 136                                sprintf(where, "%s, ", p->origin);
 137                        else
 138                                where[0] = 0;
 139                        regerror(err, &p->regexp, errbuf, 1024);
 140                        regfree(&p->regexp);
 141                        die("%s'%s': %s", where, p->pattern, errbuf);
 142                }
 143        }
 144}
 145
 146static char *end_of_line(char *cp, unsigned long *left)
 147{
 148        unsigned long l = *left;
 149        while (l && *cp != '\n') {
 150                l--;
 151                cp++;
 152        }
 153        *left = l;
 154        return cp;
 155}
 156
 157static int word_char(char ch)
 158{
 159        return isalnum(ch) || ch == '_';
 160}
 161
 162static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 163                      const char *name, unsigned lno, char sign)
 164{
 165        printf("%s%c", name, sign);
 166        if (opt->linenum)
 167                printf("%d%c", lno, sign);
 168        printf("%.*s\n", (int)(eol-bol), bol);
 169}
 170
 171/*
 172 * NEEDSWORK: share code with diff.c
 173 */
 174#define FIRST_FEW_BYTES 8000
 175static int buffer_is_binary(const char *ptr, unsigned long size)
 176{
 177        if (FIRST_FEW_BYTES < size)
 178                size = FIRST_FEW_BYTES;
 179        if (memchr(ptr, 0, size))
 180                return 1;
 181        return 0;
 182}
 183
 184static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 185{
 186        char *hit = strstr(line, pattern);
 187        if (!hit) {
 188                match->rm_so = match->rm_eo = -1;
 189                return REG_NOMATCH;
 190        }
 191        else {
 192                match->rm_so = hit - line;
 193                match->rm_eo = match->rm_so + strlen(pattern);
 194                return 0;
 195        }
 196}
 197
 198static int grep_buffer(struct grep_opt *opt, const char *name,
 199                       char *buf, unsigned long size)
 200{
 201        char *bol = buf;
 202        unsigned long left = size;
 203        unsigned lno = 1;
 204        struct pre_context_line {
 205                char *bol;
 206                char *eol;
 207        } *prev = NULL, *pcl;
 208        unsigned last_hit = 0;
 209        unsigned last_shown = 0;
 210        int binary_match_only = 0;
 211        const char *hunk_mark = "";
 212        unsigned count = 0;
 213
 214        if (buffer_is_binary(buf, size)) {
 215                switch (opt->binary) {
 216                case GREP_BINARY_DEFAULT:
 217                        binary_match_only = 1;
 218                        break;
 219                case GREP_BINARY_NOMATCH:
 220                        return 0; /* Assume unmatch */
 221                        break;
 222                default:
 223                        break;
 224                }
 225        }
 226
 227        if (opt->pre_context)
 228                prev = xcalloc(opt->pre_context, sizeof(*prev));
 229        if (opt->pre_context || opt->post_context)
 230                hunk_mark = "--\n";
 231
 232        while (left) {
 233                regmatch_t pmatch[10];
 234                char *eol, ch;
 235                int hit = 0;
 236                struct grep_pat *p;
 237
 238                eol = end_of_line(bol, &left);
 239                ch = *eol;
 240                *eol = 0;
 241
 242                for (p = opt->pattern_list; p; p = p->next) {
 243                        if (!opt->fixed) {
 244                                regex_t *exp = &p->regexp;
 245                                hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 246                                               pmatch, 0);
 247                        }
 248                        else {
 249                                hit = !fixmatch(p->pattern, bol, pmatch);
 250                        }
 251
 252                        if (hit && opt->word_regexp) {
 253                                /* Match beginning must be either
 254                                 * beginning of the line, or at word
 255                                 * boundary (i.e. the last char must
 256                                 * not be alnum or underscore).
 257                                 */
 258                                if ((pmatch[0].rm_so < 0) ||
 259                                    (eol - bol) <= pmatch[0].rm_so ||
 260                                    (pmatch[0].rm_eo < 0) ||
 261                                    (eol - bol) < pmatch[0].rm_eo)
 262                                        die("regexp returned nonsense");
 263                                if (pmatch[0].rm_so != 0 &&
 264                                    word_char(bol[pmatch[0].rm_so-1]))
 265                                        hit = 0;
 266                                if (pmatch[0].rm_eo != (eol-bol) &&
 267                                    word_char(bol[pmatch[0].rm_eo]))
 268                                        hit = 0;
 269                        }
 270                        if (hit)
 271                                break;
 272                }
 273                /* "grep -v -e foo -e bla" should list lines
 274                 * that do not have either, so inversion should
 275                 * be done outside.
 276                 */
 277                if (opt->invert)
 278                        hit = !hit;
 279                if (opt->unmatch_name_only) {
 280                        if (hit)
 281                                return 0;
 282                        goto next_line;
 283                }
 284                if (hit) {
 285                        count++;
 286                        if (binary_match_only) {
 287                                printf("Binary file %s matches\n", name);
 288                                return 1;
 289                        }
 290                        if (opt->name_only) {
 291                                printf("%s\n", name);
 292                                return 1;
 293                        }
 294                        /* Hit at this line.  If we haven't shown the
 295                         * pre-context lines, we would need to show them.
 296                         * When asked to do "count", this still show
 297                         * the context which is nonsense, but the user
 298                         * deserves to get that ;-).
 299                         */
 300                        if (opt->pre_context) {
 301                                unsigned from;
 302                                if (opt->pre_context < lno)
 303                                        from = lno - opt->pre_context;
 304                                else
 305                                        from = 1;
 306                                if (from <= last_shown)
 307                                        from = last_shown + 1;
 308                                if (last_shown && from != last_shown + 1)
 309                                        printf(hunk_mark);
 310                                while (from < lno) {
 311                                        pcl = &prev[lno-from-1];
 312                                        show_line(opt, pcl->bol, pcl->eol,
 313                                                  name, from, '-');
 314                                        from++;
 315                                }
 316                                last_shown = lno-1;
 317                        }
 318                        if (last_shown && lno != last_shown + 1)
 319                                printf(hunk_mark);
 320                        if (!opt->count)
 321                                show_line(opt, bol, eol, name, lno, ':');
 322                        last_shown = last_hit = lno;
 323                }
 324                else if (last_hit &&
 325                         lno <= last_hit + opt->post_context) {
 326                        /* If the last hit is within the post context,
 327                         * we need to show this line.
 328                         */
 329                        if (last_shown && lno != last_shown + 1)
 330                                printf(hunk_mark);
 331                        show_line(opt, bol, eol, name, lno, '-');
 332                        last_shown = lno;
 333                }
 334                if (opt->pre_context) {
 335                        memmove(prev+1, prev,
 336                                (opt->pre_context-1) * sizeof(*prev));
 337                        prev->bol = bol;
 338                        prev->eol = eol;
 339                }
 340
 341        next_line:
 342                *eol = ch;
 343                bol = eol + 1;
 344                if (!left)
 345                        break;
 346                left--;
 347                lno++;
 348        }
 349
 350        if (opt->unmatch_name_only) {
 351                /* We did not see any hit, so we want to show this */
 352                printf("%s\n", name);
 353                return 1;
 354        }
 355
 356        /* NEEDSWORK:
 357         * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 358         * which feels mostly useless but sometimes useful.  Maybe
 359         * make it another option?  For now suppress them.
 360         */
 361        if (opt->count && count)
 362                printf("%s:%u\n", name, count);
 363        return !!last_hit;
 364}
 365
 366static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name)
 367{
 368        unsigned long size;
 369        char *data;
 370        char type[20];
 371        int hit;
 372        data = read_sha1_file(sha1, type, &size);
 373        if (!data) {
 374                error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 375                return 0;
 376        }
 377        hit = grep_buffer(opt, name, data, size);
 378        free(data);
 379        return hit;
 380}
 381
 382static int grep_file(struct grep_opt *opt, const char *filename)
 383{
 384        struct stat st;
 385        int i;
 386        char *data;
 387        if (lstat(filename, &st) < 0) {
 388        err_ret:
 389                if (errno != ENOENT)
 390                        error("'%s': %s", filename, strerror(errno));
 391                return 0;
 392        }
 393        if (!st.st_size)
 394                return 0; /* empty file -- no grep hit */
 395        if (!S_ISREG(st.st_mode))
 396                return 0;
 397        i = open(filename, O_RDONLY);
 398        if (i < 0)
 399                goto err_ret;
 400        data = xmalloc(st.st_size + 1);
 401        if (st.st_size != xread(i, data, st.st_size)) {
 402                error("'%s': short read %s", filename, strerror(errno));
 403                close(i);
 404                free(data);
 405                return 0;
 406        }
 407        close(i);
 408        i = grep_buffer(opt, filename, data, st.st_size);
 409        free(data);
 410        return i;
 411}
 412
 413static int exec_grep(int argc, const char **argv)
 414{
 415        pid_t pid;
 416        int status;
 417
 418        argv[argc] = NULL;
 419        pid = fork();
 420        if (pid < 0)
 421                return pid;
 422        if (!pid) {
 423                execvp("grep", (char **) argv);
 424                exit(255);
 425        }
 426        while (waitpid(pid, &status, 0) < 0) {
 427                if (errno == EINTR)
 428                        continue;
 429                return -1;
 430        }
 431        if (WIFEXITED(status)) {
 432                if (!WEXITSTATUS(status))
 433                        return 1;
 434                return 0;
 435        }
 436        return -1;
 437}
 438
 439#define MAXARGS 1000
 440#define ARGBUF 4096
 441#define push_arg(a) do { \
 442        if (nr < MAXARGS) argv[nr++] = (a); \
 443        else die("maximum number of args exceeded"); \
 444        } while (0)
 445
 446static int external_grep(struct grep_opt *opt, const char **paths, int cached)
 447{
 448        int i, nr, argc, hit, len;
 449        const char *argv[MAXARGS+1];
 450        char randarg[ARGBUF];
 451        char *argptr = randarg;
 452        struct grep_pat *p;
 453
 454        len = nr = 0;
 455        push_arg("grep");
 456        if (opt->fixed)
 457                push_arg("-F");
 458        if (opt->linenum)
 459                push_arg("-n");
 460        if (opt->regflags & REG_EXTENDED)
 461                push_arg("-E");
 462        if (opt->word_regexp)
 463                push_arg("-w");
 464        if (opt->name_only)
 465                push_arg("-l");
 466        if (opt->unmatch_name_only)
 467                push_arg("-L");
 468        if (opt->count)
 469                push_arg("-c");
 470        if (opt->post_context || opt->pre_context) {
 471                if (opt->post_context != opt->pre_context) {
 472                        if (opt->pre_context) {
 473                                push_arg("-B");
 474                                len += snprintf(argptr, sizeof(randarg)-len,
 475                                                "%u", opt->pre_context);
 476                                if (sizeof(randarg) <= len)
 477                                        die("maximum length of args exceeded");
 478                                push_arg(argptr);
 479                                argptr += len;
 480                        }
 481                        if (opt->post_context) {
 482                                push_arg("-A");
 483                                len += snprintf(argptr, sizeof(randarg)-len,
 484                                                "%u", opt->post_context);
 485                                if (sizeof(randarg) <= len)
 486                                        die("maximum length of args exceeded");
 487                                push_arg(argptr);
 488                                argptr += len;
 489                        }
 490                }
 491                else {
 492                        push_arg("-C");
 493                        len += snprintf(argptr, sizeof(randarg)-len,
 494                                        "%u", opt->post_context);
 495                        if (sizeof(randarg) <= len)
 496                                die("maximum length of args exceeded");
 497                        push_arg(argptr);
 498                        argptr += len;
 499                }
 500        }
 501        for (p = opt->pattern_list; p; p = p->next) {
 502                push_arg("-e");
 503                push_arg(p->pattern);
 504        }
 505
 506        /*
 507         * To make sure we get the header printed out when we want it,
 508         * add /dev/null to the paths to grep.  This is unnecessary
 509         * (and wrong) with "-l" or "-L", which always print out the
 510         * name anyway.
 511         *
 512         * GNU grep has "-H", but this is portable.
 513         */
 514        if (!opt->name_only && !opt->unmatch_name_only)
 515                push_arg("/dev/null");
 516
 517        hit = 0;
 518        argc = nr;
 519        for (i = 0; i < active_nr; i++) {
 520                struct cache_entry *ce = active_cache[i];
 521                char *name;
 522                if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 523                        continue;
 524                if (!pathspec_matches(paths, ce->name))
 525                        continue;
 526                name = ce->name;
 527                if (name[0] == '-') {
 528                        int len = ce_namelen(ce);
 529                        name = xmalloc(len + 3);
 530                        memcpy(name, "./", 2);
 531                        memcpy(name + 2, ce->name, len + 1);
 532                }
 533                argv[argc++] = name;
 534                if (argc < MAXARGS)
 535                        continue;
 536                hit += exec_grep(argc, argv);
 537                argc = nr;
 538        }
 539        if (argc > nr)
 540                hit += exec_grep(argc, argv);
 541        return 0;
 542}
 543
 544static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 545{
 546        int hit = 0;
 547        int nr;
 548        read_cache();
 549
 550#ifdef __unix__
 551        /*
 552         * Use the external "grep" command for the case where
 553         * we grep through the checked-out files. It tends to
 554         * be a lot more optimized
 555         */
 556        if (!cached) {
 557                hit = external_grep(opt, paths, cached);
 558                if (hit >= 0)
 559                        return hit;
 560        }
 561#endif
 562
 563        for (nr = 0; nr < active_nr; nr++) {
 564                struct cache_entry *ce = active_cache[nr];
 565                if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 566                        continue;
 567                if (!pathspec_matches(paths, ce->name))
 568                        continue;
 569                if (cached)
 570                        hit |= grep_sha1(opt, ce->sha1, ce->name);
 571                else
 572                        hit |= grep_file(opt, ce->name);
 573        }
 574        return hit;
 575}
 576
 577static int grep_tree(struct grep_opt *opt, const char **paths,
 578                     struct tree_desc *tree,
 579                     const char *tree_name, const char *base)
 580{
 581        int len;
 582        int hit = 0;
 583        struct name_entry entry;
 584        char *down;
 585        char *path_buf = xmalloc(PATH_MAX + strlen(tree_name) + 100);
 586
 587        if (tree_name[0]) {
 588                int offset = sprintf(path_buf, "%s:", tree_name);
 589                down = path_buf + offset;
 590                strcat(down, base);
 591        }
 592        else {
 593                down = path_buf;
 594                strcpy(down, base);
 595        }
 596        len = strlen(path_buf);
 597
 598        while (tree_entry(tree, &entry)) {
 599                strcpy(path_buf + len, entry.path);
 600
 601                if (S_ISDIR(entry.mode))
 602                        /* Match "abc/" against pathspec to
 603                         * decide if we want to descend into "abc"
 604                         * directory.
 605                         */
 606                        strcpy(path_buf + len + entry.pathlen, "/");
 607
 608                if (!pathspec_matches(paths, down))
 609                        ;
 610                else if (S_ISREG(entry.mode))
 611                        hit |= grep_sha1(opt, entry.sha1, path_buf);
 612                else if (S_ISDIR(entry.mode)) {
 613                        char type[20];
 614                        struct tree_desc sub;
 615                        void *data;
 616                        data = read_sha1_file(entry.sha1, type, &sub.size);
 617                        if (!data)
 618                                die("unable to read tree (%s)",
 619                                    sha1_to_hex(entry.sha1));
 620                        sub.buf = data;
 621                        hit |= grep_tree(opt, paths, &sub, tree_name, down);
 622                        free(data);
 623                }
 624        }
 625        return hit;
 626}
 627
 628static int grep_object(struct grep_opt *opt, const char **paths,
 629                       struct object *obj, const char *name)
 630{
 631        if (!strcmp(obj->type, blob_type))
 632                return grep_sha1(opt, obj->sha1, name);
 633        if (!strcmp(obj->type, commit_type) ||
 634            !strcmp(obj->type, tree_type)) {
 635                struct tree_desc tree;
 636                void *data;
 637                int hit;
 638                data = read_object_with_reference(obj->sha1, tree_type,
 639                                                  &tree.size, NULL);
 640                if (!data)
 641                        die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 642                tree.buf = data;
 643                hit = grep_tree(opt, paths, &tree, name, "");
 644                free(data);
 645                return hit;
 646        }
 647        die("unable to grep from object of type %s", obj->type);
 648}
 649
 650static const char builtin_grep_usage[] =
 651"git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 652
 653int cmd_grep(int argc, const char **argv, char **envp)
 654{
 655        int hit = 0;
 656        int cached = 0;
 657        int seen_dashdash = 0;
 658        struct grep_opt opt;
 659        struct object_list *list, **tail, *object_list = NULL;
 660        const char *prefix = setup_git_directory();
 661        const char **paths = NULL;
 662        int i;
 663
 664        memset(&opt, 0, sizeof(opt));
 665        opt.pattern_tail = &opt.pattern_list;
 666        opt.regflags = REG_NEWLINE;
 667
 668        /*
 669         * If there is no -- then the paths must exist in the working
 670         * tree.  If there is no explicit pattern specified with -e or
 671         * -f, we take the first unrecognized non option to be the
 672         * pattern, but then what follows it must be zero or more
 673         * valid refs up to the -- (if exists), and then existing
 674         * paths.  If there is an explicit pattern, then the first
 675         * unrecocnized non option is the beginning of the refs list
 676         * that continues up to the -- (if exists), and then paths.
 677         */
 678
 679        tail = &object_list;
 680        while (1 < argc) {
 681                const char *arg = argv[1];
 682                argc--; argv++;
 683                if (!strcmp("--cached", arg)) {
 684                        cached = 1;
 685                        continue;
 686                }
 687                if (!strcmp("-a", arg) ||
 688                    !strcmp("--text", arg)) {
 689                        opt.binary = GREP_BINARY_TEXT;
 690                        continue;
 691                }
 692                if (!strcmp("-i", arg) ||
 693                    !strcmp("--ignore-case", arg)) {
 694                        opt.regflags |= REG_ICASE;
 695                        continue;
 696                }
 697                if (!strcmp("-I", arg)) {
 698                        opt.binary = GREP_BINARY_NOMATCH;
 699                        continue;
 700                }
 701                if (!strcmp("-v", arg) ||
 702                    !strcmp("--invert-match", arg)) {
 703                        opt.invert = 1;
 704                        continue;
 705                }
 706                if (!strcmp("-E", arg) ||
 707                    !strcmp("--extended-regexp", arg)) {
 708                        opt.regflags |= REG_EXTENDED;
 709                        continue;
 710                }
 711                if (!strcmp("-F", arg) ||
 712                    !strcmp("--fixed-strings", arg)) {
 713                        opt.fixed = 1;
 714                        continue;
 715                }
 716                if (!strcmp("-G", arg) ||
 717                    !strcmp("--basic-regexp", arg)) {
 718                        opt.regflags &= ~REG_EXTENDED;
 719                        continue;
 720                }
 721                if (!strcmp("-n", arg)) {
 722                        opt.linenum = 1;
 723                        continue;
 724                }
 725                if (!strcmp("-H", arg)) {
 726                        /* We always show the pathname, so this
 727                         * is a noop.
 728                         */
 729                        continue;
 730                }
 731                if (!strcmp("-l", arg) ||
 732                    !strcmp("--files-with-matches", arg)) {
 733                        opt.name_only = 1;
 734                        continue;
 735                }
 736                if (!strcmp("-L", arg) ||
 737                    !strcmp("--files-without-match", arg)) {
 738                        opt.unmatch_name_only = 1;
 739                        continue;
 740                }
 741                if (!strcmp("-c", arg) ||
 742                    !strcmp("--count", arg)) {
 743                        opt.count = 1;
 744                        continue;
 745                }
 746                if (!strcmp("-w", arg) ||
 747                    !strcmp("--word-regexp", arg)) {
 748                        opt.word_regexp = 1;
 749                        continue;
 750                }
 751                if (!strncmp("-A", arg, 2) ||
 752                    !strncmp("-B", arg, 2) ||
 753                    !strncmp("-C", arg, 2) ||
 754                    (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
 755                        unsigned num;
 756                        const char *scan;
 757                        switch (arg[1]) {
 758                        case 'A': case 'B': case 'C':
 759                                if (!arg[2]) {
 760                                        if (argc <= 1)
 761                                                usage(builtin_grep_usage);
 762                                        scan = *++argv;
 763                                        argc--;
 764                                }
 765                                else
 766                                        scan = arg + 2;
 767                                break;
 768                        default:
 769                                scan = arg + 1;
 770                                break;
 771                        }
 772                        if (sscanf(scan, "%u", &num) != 1)
 773                                usage(builtin_grep_usage);
 774                        switch (arg[1]) {
 775                        case 'A':
 776                                opt.post_context = num;
 777                                break;
 778                        default:
 779                        case 'C':
 780                                opt.post_context = num;
 781                        case 'B':
 782                                opt.pre_context = num;
 783                                break;
 784                        }
 785                        continue;
 786                }
 787                if (!strcmp("-f", arg)) {
 788                        FILE *patterns;
 789                        int lno = 0;
 790                        char buf[1024];
 791                        if (argc <= 1)
 792                                usage(builtin_grep_usage);
 793                        patterns = fopen(argv[1], "r");
 794                        if (!patterns)
 795                                die("'%s': %s", argv[1], strerror(errno));
 796                        while (fgets(buf, sizeof(buf), patterns)) {
 797                                int len = strlen(buf);
 798                                if (buf[len-1] == '\n')
 799                                        buf[len-1] = 0;
 800                                /* ignore empty line like grep does */
 801                                if (!buf[0])
 802                                        continue;
 803                                add_pattern(&opt, strdup(buf), argv[1], ++lno);
 804                        }
 805                        fclose(patterns);
 806                        argv++;
 807                        argc--;
 808                        continue;
 809                }
 810                if (!strcmp("-e", arg)) {
 811                        if (1 < argc) {
 812                                add_pattern(&opt, argv[1], "-e option", 0);
 813                                argv++;
 814                                argc--;
 815                                continue;
 816                        }
 817                        usage(builtin_grep_usage);
 818                }
 819                if (!strcmp("--", arg))
 820                        break;
 821                if (*arg == '-')
 822                        usage(builtin_grep_usage);
 823
 824                /* First unrecognized non-option token */
 825                if (!opt.pattern_list) {
 826                        add_pattern(&opt, arg, "command line", 0);
 827                        break;
 828                }
 829                else {
 830                        /* We are looking at the first path or rev;
 831                         * it is found at argv[1] after leaving the
 832                         * loop.
 833                         */
 834                        argc++; argv--;
 835                        break;
 836                }
 837        }
 838
 839        if (!opt.pattern_list)
 840                die("no pattern given.");
 841        if ((opt.regflags != REG_NEWLINE) && opt.fixed)
 842                die("cannot mix --fixed-strings and regexp");
 843        if (!opt.fixed)
 844                compile_patterns(&opt);
 845
 846        /* Check revs and then paths */
 847        for (i = 1; i < argc; i++) {
 848                const char *arg = argv[i];
 849                unsigned char sha1[20];
 850                /* Is it a rev? */
 851                if (!get_sha1(arg, sha1)) {
 852                        struct object *object = parse_object(sha1);
 853                        struct object_list *elem;
 854                        if (!object)
 855                                die("bad object %s", arg);
 856                        elem = object_list_insert(object, tail);
 857                        elem->name = arg;
 858                        tail = &elem->next;
 859                        continue;
 860                }
 861                if (!strcmp(arg, "--")) {
 862                        i++;
 863                        seen_dashdash = 1;
 864                }
 865                break;
 866        }
 867
 868        /* The rest are paths */
 869        if (!seen_dashdash) {
 870                int j;
 871                for (j = i; j < argc; j++)
 872                        verify_filename(prefix, argv[j]);
 873        }
 874
 875        if (i < argc)
 876                paths = get_pathspec(prefix, argv + i);
 877        else if (prefix) {
 878                paths = xcalloc(2, sizeof(const char *));
 879                paths[0] = prefix;
 880                paths[1] = NULL;
 881        }
 882
 883        if (!object_list)
 884                return !grep_cache(&opt, paths, cached);
 885
 886        if (cached)
 887                die("both --cached and trees are given.");
 888
 889        for (list = object_list; list; list = list->next) {
 890                struct object *real_obj;
 891                real_obj = deref_tag(list->item, NULL, 0);
 892                if (grep_object(&opt, paths, real_obj, list->name))
 893                        hit = 1;
 894        }
 895        return !hit;
 896}