builtin-grep.c

   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  */
  21 static 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 ||
  33                     ((matchlen <= namelen) &&
  34                      !strncmp(name, match, matchlen) &&
  35                      (match[matchlen-1] == '/' ||
  36                       name[matchlen] == '\0' || name[matchlen] == '/')))
  37                         return 1;
  38                 if (!fnmatch(match, name, 0))
  39                         return 1;
  40                 if (name[namelen-1] != '/')
  41                         continue;
  42
  43                 /* We are being asked if the directory ("name") is worth
  44                  * descending into.
  45                  *
  46                  * Find the longest leading directory name that does
  47                  * not have metacharacter in the pathspec; the name
  48                  * we are looking at must overlap with that directory.
  49                  */
  50                 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
  51                         char ch = *cp;
  52                         if (ch == '*' || ch == '[' || ch == '?') {
  53                                 meta = cp;
  54                                 break;
  55                         }
  56                 }
  57                 if (!meta)
  58                         meta = cp; /* fully literal */
  59
  60                 if (namelen <= meta - match) {
  61                         /* Looking at "Documentation/" and
  62                          * the pattern says "Documentation/howto/", or
  63                          * "Documentation/diff*.txt".  The name we
  64                          * have should match prefix.
  65                          */
  66                         if (!memcmp(match, name, namelen))
  67                                 return 1;
  68                         continue;
  69                 }
  70
  71                 if (meta - match < namelen) {
  72                         /* Looking at "Documentation/howto/" and
  73                          * the pattern says "Documentation/h*";
  74                          * match up to "Do.../h"; this avoids descending
  75                          * into "Documentation/technical/".
  76                          */
  77                         if (!memcmp(match, name, meta - match))
  78                                 return 1;
  79                         continue;
  80                 }
  81         }
  82         return 0;
  83 }
  84
  85 struct grep_pat {
  86         struct grep_pat *next;
  87         const char *origin;
  88         int no;
  89         const char *pattern;
  90         regex_t regexp;
  91 };
  92
  93 struct grep_opt {
  94         struct grep_pat *pattern_list;
  95         struct grep_pat **pattern_tail;
  96         regex_t regexp;
  97         unsigned linenum:1;
  98         unsigned invert:1;
  99         unsigned name_only:1;
 100         unsigned unmatch_name_only:1;
 101         unsigned count:1;
 102         unsigned word_regexp:1;
 103         unsigned fixed:1;
 104 #define GREP_BINARY_DEFAULT     0
 105 #define GREP_BINARY_NOMATCH     1
 106 #define GREP_BINARY_TEXT        2
 107         unsigned binary:2;
 108         int regflags;
 109         unsigned pre_context;
 110         unsigned post_context;
 111 };
 112
 113 static void add_pattern(struct grep_opt *opt, const char *pat,
 114                         const char *origin, int no)
 115 {
 116         struct grep_pat *p = xcalloc(1, sizeof(*p));
 117         p->pattern = pat;
 118         p->origin = origin;
 119         p->no = no;
 120         *opt->pattern_tail = p;
 121         opt->pattern_tail = &p->next;
 122         p->next = NULL;
 123 }
 124
 125 static void compile_patterns(struct grep_opt *opt)
 126 {
 127         struct grep_pat *p;
 128         for (p = opt->pattern_list; p; p = p->next) {
 129                 int err = regcomp(&p->regexp, p->pattern, opt->regflags);
 130                 if (err) {
 131                         char errbuf[1024];
 132                         char where[1024];
 133                         if (p->no)
 134                                 sprintf(where, "In '%s' at %d, ",
 135                                         p->origin, p->no);
 136                         else if (p->origin)
 137                                 sprintf(where, "%s, ", p->origin);
 138                         else
 139                                 where[0] = 0;
 140                         regerror(err, &p->regexp, errbuf, 1024);
 141                         regfree(&p->regexp);
 142                         die("%s'%s': %s", where, p->pattern, errbuf);
 143                 }
 144         }
 145 }
 146
 147 static char *end_of_line(char *cp, unsigned long *left)
 148 {
 149         unsigned long l = *left;
 150         while (l && *cp != '\n') {
 151                 l--;
 152                 cp++;
 153         }
 154         *left = l;
 155         return cp;
 156 }
 157
 158 static int word_char(char ch)
 159 {
 160         return isalnum(ch) || ch == '_';
 161 }
 162
 163 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 164                       const char *name, unsigned lno, char sign)
 165 {
 166         printf("%s%c", name, sign);
 167         if (opt->linenum)
 168                 printf("%d%c", lno, sign);
 169         printf("%.*s\n", (int)(eol-bol), bol);
 170 }
 171
 172 /*
 173  * NEEDSWORK: share code with diff.c
 174  */
 175 #define FIRST_FEW_BYTES 8000
 176 static int buffer_is_binary(const char *ptr, unsigned long size)
 177 {
 178         if (FIRST_FEW_BYTES < size)
 179                 size = FIRST_FEW_BYTES;
 180         if (memchr(ptr, 0, size))
 181                 return 1;
 182         return 0;
 183 }
 184
 185 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 186 {
 187         char *hit = strstr(line, pattern);
 188         if (!hit) {
 189                 match->rm_so = match->rm_eo = -1;
 190                 return REG_NOMATCH;
 191         }
 192         else {
 193                 match->rm_so = hit - line;
 194                 match->rm_eo = match->rm_so + strlen(pattern);
 195                 return 0;
 196         }
 197 }
 198
 199 static int grep_buffer(struct grep_opt *opt, const char *name,
 200                        char *buf, unsigned long size)
 201 {
 202         char *bol = buf;
 203         unsigned long left = size;
 204         unsigned lno = 1;
 205         struct pre_context_line {
 206                 char *bol;
 207                 char *eol;
 208         } *prev = NULL, *pcl;
 209         unsigned last_hit = 0;
 210         unsigned last_shown = 0;
 211         int binary_match_only = 0;
 212         const char *hunk_mark = "";
 213         unsigned count = 0;
 214
 215         if (buffer_is_binary(buf, size)) {
 216                 switch (opt->binary) {
 217                 case GREP_BINARY_DEFAULT:
 218                         binary_match_only = 1;
 219                         break;
 220                 case GREP_BINARY_NOMATCH:
 221                         return 0; /* Assume unmatch */
 222                         break;
 223                 default:
 224                         break;
 225                 }
 226         }
 227
 228         if (opt->pre_context)
 229                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 230         if (opt->pre_context || opt->post_context)
 231                 hunk_mark = "--\n";
 232
 233         while (left) {
 234                 regmatch_t pmatch[10];
 235                 char *eol, ch;
 236                 int hit = 0;
 237                 struct grep_pat *p;
 238
 239                 eol = end_of_line(bol, &left);
 240                 ch = *eol;
 241                 *eol = 0;
 242
 243                 for (p = opt->pattern_list; p; p = p->next) {
 244                         if (!opt->fixed) {
 245                                 regex_t *exp = &p->regexp;
 246                                 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 247                                                pmatch, 0);
 248                         }
 249                         else {
 250                                 hit = !fixmatch(p->pattern, bol, pmatch);
 251                         }
 252
 253                         if (hit && opt->word_regexp) {
 254                                 /* Match beginning must be either
 255                                  * beginning of the line, or at word
 256                                  * boundary (i.e. the last char must
 257                                  * not be alnum or underscore).
 258                                  */
 259                                 if ((pmatch[0].rm_so < 0) ||
 260                                     (eol - bol) <= pmatch[0].rm_so ||
 261                                     (pmatch[0].rm_eo < 0) ||
 262                                     (eol - bol) < pmatch[0].rm_eo)
 263                                         die("regexp returned nonsense");
 264                                 if (pmatch[0].rm_so != 0 &&
 265                                     word_char(bol[pmatch[0].rm_so-1]))
 266                                         hit = 0;
 267                                 if (pmatch[0].rm_eo != (eol-bol) &&
 268                                     word_char(bol[pmatch[0].rm_eo]))
 269                                         hit = 0;
 270                         }
 271                         if (hit)
 272                                 break;
 273                 }
 274                 /* "grep -v -e foo -e bla" should list lines
 275                  * that do not have either, so inversion should
 276                  * be done outside.
 277                  */
 278                 if (opt->invert)
 279                         hit = !hit;
 280                 if (opt->unmatch_name_only) {
 281                         if (hit)
 282                                 return 0;
 283                         goto next_line;
 284                 }
 285                 if (hit) {
 286                         count++;
 287                         if (binary_match_only) {
 288                                 printf("Binary file %s matches\n", name);
 289                                 return 1;
 290                         }
 291                         if (opt->name_only) {
 292                                 printf("%s\n", name);
 293                                 return 1;
 294                         }
 295                         /* Hit at this line.  If we haven't shown the
 296                          * pre-context lines, we would need to show them.
 297                          * When asked to do "count", this still show
 298                          * the context which is nonsense, but the user
 299                          * deserves to get that ;-).
 300                          */
 301                         if (opt->pre_context) {
 302                                 unsigned from;
 303                                 if (opt->pre_context < lno)
 304                                         from = lno - opt->pre_context;
 305                                 else
 306                                         from = 1;
 307                                 if (from <= last_shown)
 308                                         from = last_shown + 1;
 309                                 if (last_shown && from != last_shown + 1)
 310                                         printf(hunk_mark);
 311                                 while (from < lno) {
 312                                         pcl = &prev[lno-from-1];
 313                                         show_line(opt, pcl->bol, pcl->eol,
 314                                                   name, from, '-');
 315                                         from++;
 316                                 }
 317                                 last_shown = lno-1;
 318                         }
 319                         if (last_shown && lno != last_shown + 1)
 320                                 printf(hunk_mark);
 321                         if (!opt->count)
 322                                 show_line(opt, bol, eol, name, lno, ':');
 323                         last_shown = last_hit = lno;
 324                 }
 325                 else if (last_hit &&
 326                          lno <= last_hit + opt->post_context) {
 327                         /* If the last hit is within the post context,
 328                          * we need to show this line.
 329                          */
 330                         if (last_shown && lno != last_shown + 1)
 331                                 printf(hunk_mark);
 332                         show_line(opt, bol, eol, name, lno, '-');
 333                         last_shown = lno;
 334                 }
 335                 if (opt->pre_context) {
 336                         memmove(prev+1, prev,
 337                                 (opt->pre_context-1) * sizeof(*prev));
 338                         prev->bol = bol;
 339                         prev->eol = eol;
 340                 }
 341
 342         next_line:
 343                 *eol = ch;
 344                 bol = eol + 1;
 345                 if (!left)
 346                         break;
 347                 left--;
 348                 lno++;
 349         }
 350
 351         if (opt->unmatch_name_only) {
 352                 /* We did not see any hit, so we want to show this */
 353                 printf("%s\n", name);
 354                 return 1;
 355         }
 356
 357         /* NEEDSWORK:
 358          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 359          * which feels mostly useless but sometimes useful.  Maybe
 360          * make it another option?  For now suppress them.
 361          */
 362         if (opt->count && count)
 363                 printf("%s:%u\n", name, count);
 364         return !!last_hit;
 365 }
 366
 367 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name)
 368 {
 369         unsigned long size;
 370         char *data;
 371         char type[20];
 372         int hit;
 373         data = read_sha1_file(sha1, type, &size);
 374         if (!data) {
 375                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 376                 return 0;
 377         }
 378         hit = grep_buffer(opt, name, data, size);
 379         free(data);
 380         return hit;
 381 }
 382
 383 static int grep_file(struct grep_opt *opt, const char *filename)
 384 {
 385         struct stat st;
 386         int i;
 387         char *data;
 388         if (lstat(filename, &st) < 0) {
 389         err_ret:
 390                 if (errno != ENOENT)
 391                         error("'%s': %s", filename, strerror(errno));
 392                 return 0;
 393         }
 394         if (!st.st_size)
 395                 return 0; /* empty file -- no grep hit */
 396         if (!S_ISREG(st.st_mode))
 397                 return 0;
 398         i = open(filename, O_RDONLY);
 399         if (i < 0)
 400                 goto err_ret;
 401         data = xmalloc(st.st_size + 1);
 402         if (st.st_size != xread(i, data, st.st_size)) {
 403                 error("'%s': short read %s", filename, strerror(errno));
 404                 close(i);
 405                 free(data);
 406                 return 0;
 407         }
 408         close(i);
 409         i = grep_buffer(opt, filename, data, st.st_size);
 410         free(data);
 411         return i;
 412 }
 413
 414 static int exec_grep(int argc, const char **argv)
 415 {
 416         pid_t pid;
 417         int status;
 418
 419         argv[argc] = NULL;
 420         pid = fork();
 421         if (pid < 0)
 422                 return pid;
 423         if (!pid) {
 424                 execvp("grep", (char **) argv);
 425                 exit(255);
 426         }
 427         while (waitpid(pid, &status, 0) < 0) {
 428                 if (errno == EINTR)
 429                         continue;
 430                 return -1;
 431         }
 432         if (WIFEXITED(status)) {
 433                 if (!WEXITSTATUS(status))
 434                         return 1;
 435                 return 0;
 436         }
 437         return -1;
 438 }
 439
 440 #define MAXARGS 1000
 441 #define ARGBUF 4096
 442 #define push_arg(a) do { \
 443         if (nr < MAXARGS) argv[nr++] = (a); \
 444         else die("maximum number of args exceeded"); \
 445         } while (0)
 446
 447 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
 448 {
 449         int i, nr, argc, hit, len, status;
 450         const char *argv[MAXARGS+1];
 451         char randarg[ARGBUF];
 452         char *argptr = randarg;
 453         struct grep_pat *p;
 454
 455         len = nr = 0;
 456         push_arg("grep");
 457         if (opt->fixed)
 458                 push_arg("-F");
 459         if (opt->linenum)
 460                 push_arg("-n");
 461         if (opt->regflags & REG_EXTENDED)
 462                 push_arg("-E");
 463         if (opt->regflags & REG_ICASE)
 464                 push_arg("-i");
 465         if (opt->word_regexp)
 466                 push_arg("-w");
 467         if (opt->name_only)
 468                 push_arg("-l");
 469         if (opt->unmatch_name_only)
 470                 push_arg("-L");
 471         if (opt->count)
 472                 push_arg("-c");
 473         if (opt->post_context || opt->pre_context) {
 474                 if (opt->post_context != opt->pre_context) {
 475                         if (opt->pre_context) {
 476                                 push_arg("-B");
 477                                 len += snprintf(argptr, sizeof(randarg)-len,
 478                                                 "%u", opt->pre_context);
 479                                 if (sizeof(randarg) <= len)
 480                                         die("maximum length of args exceeded");
 481                                 push_arg(argptr);
 482                                 argptr += len;
 483                         }
 484                         if (opt->post_context) {
 485                                 push_arg("-A");
 486                                 len += snprintf(argptr, sizeof(randarg)-len,
 487                                                 "%u", opt->post_context);
 488                                 if (sizeof(randarg) <= len)
 489                                         die("maximum length of args exceeded");
 490                                 push_arg(argptr);
 491                                 argptr += len;
 492                         }
 493                 }
 494                 else {
 495                         push_arg("-C");
 496                         len += snprintf(argptr, sizeof(randarg)-len,
 497                                         "%u", opt->post_context);
 498                         if (sizeof(randarg) <= len)
 499                                 die("maximum length of args exceeded");
 500                         push_arg(argptr);
 501                         argptr += len;
 502                 }
 503         }
 504         for (p = opt->pattern_list; p; p = p->next) {
 505                 push_arg("-e");
 506                 push_arg(p->pattern);
 507         }
 508
 509         /*
 510          * To make sure we get the header printed out when we want it,
 511          * add /dev/null to the paths to grep.  This is unnecessary
 512          * (and wrong) with "-l" or "-L", which always print out the
 513          * name anyway.
 514          *
 515          * GNU grep has "-H", but this is portable.
 516          */
 517         if (!opt->name_only && !opt->unmatch_name_only)
 518                 push_arg("/dev/null");
 519
 520         hit = 0;
 521         argc = nr;
 522         for (i = 0; i < active_nr; i++) {
 523                 struct cache_entry *ce = active_cache[i];
 524                 char *name;
 525                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 526                         continue;
 527                 if (!pathspec_matches(paths, ce->name))
 528                         continue;
 529                 name = ce->name;
 530                 if (name[0] == '-') {
 531                         int len = ce_namelen(ce);
 532                         name = xmalloc(len + 3);
 533                         memcpy(name, "./", 2);
 534                         memcpy(name + 2, ce->name, len + 1);
 535                 }
 536                 argv[argc++] = name;
 537                 if (argc < MAXARGS)
 538                         continue;
 539                 status = exec_grep(argc, argv);
 540                 if (0 < status)
 541                         hit = 1;
 542                 argc = nr;
 543         }
 544         if (argc > nr) {
 545                 status = exec_grep(argc, argv);
 546                 if (0 < status)
 547                         hit = 1;
 548         }
 549         return hit;
 550 }
 551
 552 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 553 {
 554         int hit = 0;
 555         int nr;
 556         read_cache();
 557
 558 #ifdef __unix__
 559         /*
 560          * Use the external "grep" command for the case where
 561          * we grep through the checked-out files. It tends to
 562          * be a lot more optimized
 563          */
 564         if (!cached) {
 565                 hit = external_grep(opt, paths, cached);
 566                 if (hit >= 0)
 567                         return hit;
 568         }
 569 #endif
 570
 571         for (nr = 0; nr < active_nr; nr++) {
 572                 struct cache_entry *ce = active_cache[nr];
 573                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 574                         continue;
 575                 if (!pathspec_matches(paths, ce->name))
 576                         continue;
 577                 if (cached)
 578                         hit |= grep_sha1(opt, ce->sha1, ce->name);
 579                 else
 580                         hit |= grep_file(opt, ce->name);
 581         }
 582         return hit;
 583 }
 584
 585 static int grep_tree(struct grep_opt *opt, const char **paths,
 586                      struct tree_desc *tree,
 587                      const char *tree_name, const char *base)
 588 {
 589         int len;
 590         int hit = 0;
 591         struct name_entry entry;
 592         char *down;
 593         char *path_buf = xmalloc(PATH_MAX + strlen(tree_name) + 100);
 594
 595         if (tree_name[0]) {
 596                 int offset = sprintf(path_buf, "%s:", tree_name);
 597                 down = path_buf + offset;
 598                 strcat(down, base);
 599         }
 600         else {
 601                 down = path_buf;
 602                 strcpy(down, base);
 603         }
 604         len = strlen(path_buf);
 605
 606         while (tree_entry(tree, &entry)) {
 607                 strcpy(path_buf + len, entry.path);
 608
 609                 if (S_ISDIR(entry.mode))
 610                         /* Match "abc/" against pathspec to
 611                          * decide if we want to descend into "abc"
 612                          * directory.
 613                          */
 614                         strcpy(path_buf + len + entry.pathlen, "/");
 615
 616                 if (!pathspec_matches(paths, down))
 617                         ;
 618                 else if (S_ISREG(entry.mode))
 619                         hit |= grep_sha1(opt, entry.sha1, path_buf);
 620                 else if (S_ISDIR(entry.mode)) {
 621                         char type[20];
 622                         struct tree_desc sub;
 623                         void *data;
 624                         data = read_sha1_file(entry.sha1, type, &sub.size);
 625                         if (!data)
 626                                 die("unable to read tree (%s)",
 627                                     sha1_to_hex(entry.sha1));
 628                         sub.buf = data;
 629                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 630                         free(data);
 631                 }
 632         }
 633         return hit;
 634 }
 635
 636 static int grep_object(struct grep_opt *opt, const char **paths,
 637                        struct object *obj, const char *name)
 638 {
 639         if (obj->type == TYPE_BLOB)
 640                 return grep_sha1(opt, obj->sha1, name);
 641         if (obj->type == TYPE_COMMIT || obj->type == TYPE_TREE) {
 642                 struct tree_desc tree;
 643                 void *data;
 644                 int hit;
 645                 data = read_object_with_reference(obj->sha1, tree_type,
 646                                                   &tree.size, NULL);
 647                 if (!data)
 648                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 649                 tree.buf = data;
 650                 hit = grep_tree(opt, paths, &tree, name, "");
 651                 free(data);
 652                 return hit;
 653         }
 654         die("unable to grep from object of type %s", typename(obj->type));
 655 }
 656
 657 static const char builtin_grep_usage[] =
 658 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 659
 660 static const char emsg_invalid_context_len[] =
 661 "%s: invalid context length argument";
 662 static const char emsg_missing_context_len[] =
 663 "missing context length argument";
 664 static const char emsg_missing_argument[] =
 665 "option requires an argument -%s";
 666
 667 int cmd_grep(int argc, const char **argv, char **envp)
 668 {
 669         int hit = 0;
 670         int cached = 0;
 671         int seen_dashdash = 0;
 672         struct grep_opt opt;
 673         struct object_array list = { 0, 0, NULL };
 674         const char *prefix = setup_git_directory();
 675         const char **paths = NULL;
 676         int i;
 677
 678         memset(&opt, 0, sizeof(opt));
 679         opt.pattern_tail = &opt.pattern_list;
 680         opt.regflags = REG_NEWLINE;
 681
 682         /*
 683          * If there is no -- then the paths must exist in the working
 684          * tree.  If there is no explicit pattern specified with -e or
 685          * -f, we take the first unrecognized non option to be the
 686          * pattern, but then what follows it must be zero or more
 687          * valid refs up to the -- (if exists), and then existing
 688          * paths.  If there is an explicit pattern, then the first
 689          * unrecocnized non option is the beginning of the refs list
 690          * that continues up to the -- (if exists), and then paths.
 691          */
 692
 693         while (1 < argc) {
 694                 const char *arg = argv[1];
 695                 argc--; argv++;
 696                 if (!strcmp("--cached", arg)) {
 697                         cached = 1;
 698                         continue;
 699                 }
 700                 if (!strcmp("-a", arg) ||
 701                     !strcmp("--text", arg)) {
 702                         opt.binary = GREP_BINARY_TEXT;
 703                         continue;
 704                 }
 705                 if (!strcmp("-i", arg) ||
 706                     !strcmp("--ignore-case", arg)) {
 707                         opt.regflags |= REG_ICASE;
 708                         continue;
 709                 }
 710                 if (!strcmp("-I", arg)) {
 711                         opt.binary = GREP_BINARY_NOMATCH;
 712                         continue;
 713                 }
 714                 if (!strcmp("-v", arg) ||
 715                     !strcmp("--invert-match", arg)) {
 716                         opt.invert = 1;
 717                         continue;
 718                 }
 719                 if (!strcmp("-E", arg) ||
 720                     !strcmp("--extended-regexp", arg)) {
 721                         opt.regflags |= REG_EXTENDED;
 722                         continue;
 723                 }
 724                 if (!strcmp("-F", arg) ||
 725                     !strcmp("--fixed-strings", arg)) {
 726                         opt.fixed = 1;
 727                         continue;
 728                 }
 729                 if (!strcmp("-G", arg) ||
 730                     !strcmp("--basic-regexp", arg)) {
 731                         opt.regflags &= ~REG_EXTENDED;
 732                         continue;
 733                 }
 734                 if (!strcmp("-n", arg)) {
 735                         opt.linenum = 1;
 736                         continue;
 737                 }
 738                 if (!strcmp("-H", arg)) {
 739                         /* We always show the pathname, so this
 740                          * is a noop.
 741                          */
 742                         continue;
 743                 }
 744                 if (!strcmp("-l", arg) ||
 745                     !strcmp("--files-with-matches", arg)) {
 746                         opt.name_only = 1;
 747                         continue;
 748                 }
 749                 if (!strcmp("-L", arg) ||
 750                     !strcmp("--files-without-match", arg)) {
 751                         opt.unmatch_name_only = 1;
 752                         continue;
 753                 }
 754                 if (!strcmp("-c", arg) ||
 755                     !strcmp("--count", arg)) {
 756                         opt.count = 1;
 757                         continue;
 758                 }
 759                 if (!strcmp("-w", arg) ||
 760                     !strcmp("--word-regexp", arg)) {
 761                         opt.word_regexp = 1;
 762                         continue;
 763                 }
 764                 if (!strncmp("-A", arg, 2) ||
 765                     !strncmp("-B", arg, 2) ||
 766                     !strncmp("-C", arg, 2) ||
 767                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
 768                         unsigned num;
 769                         const char *scan;
 770                         switch (arg[1]) {
 771                         case 'A': case 'B': case 'C':
 772                                 if (!arg[2]) {
 773                                         if (argc <= 1)
 774                                                 die(emsg_missing_context_len);
 775                                         scan = *++argv;
 776                                         argc--;
 777                                 }
 778                                 else
 779                                         scan = arg + 2;
 780                                 break;
 781                         default:
 782                                 scan = arg + 1;
 783                                 break;
 784                         }
 785                         if (sscanf(scan, "%u", &num) != 1)
 786                                 die(emsg_invalid_context_len, scan);
 787                         switch (arg[1]) {
 788                         case 'A':
 789                                 opt.post_context = num;
 790                                 break;
 791                         default:
 792                         case 'C':
 793                                 opt.post_context = num;
 794                         case 'B':
 795                                 opt.pre_context = num;
 796                                 break;
 797                         }
 798                         continue;
 799                 }
 800                 if (!strcmp("-f", arg)) {
 801                         FILE *patterns;
 802                         int lno = 0;
 803                         char buf[1024];
 804                         if (argc <= 1)
 805                                 die(emsg_missing_argument, arg);
 806                         patterns = fopen(argv[1], "r");
 807                         if (!patterns)
 808                                 die("'%s': %s", argv[1], strerror(errno));
 809                         while (fgets(buf, sizeof(buf), patterns)) {
 810                                 int len = strlen(buf);
 811                                 if (buf[len-1] == '\n')
 812                                         buf[len-1] = 0;
 813                                 /* ignore empty line like grep does */
 814                                 if (!buf[0])
 815                                         continue;
 816                                 add_pattern(&opt, strdup(buf), argv[1], ++lno);
 817                         }
 818                         fclose(patterns);
 819                         argv++;
 820                         argc--;
 821                         continue;
 822                 }
 823                 if (!strcmp("-e", arg)) {
 824                         if (1 < argc) {
 825                                 add_pattern(&opt, argv[1], "-e option", 0);
 826                                 argv++;
 827                                 argc--;
 828                                 continue;
 829                         }
 830                         die(emsg_missing_argument, arg);
 831                 }
 832                 if (!strcmp("--", arg)) {
 833                         /* later processing wants to have this at argv[1] */
 834                         argv--;
 835                         argc++;
 836                         break;
 837                 }
 838                 if (*arg == '-')
 839                         usage(builtin_grep_usage);
 840
 841                 /* First unrecognized non-option token */
 842                 if (!opt.pattern_list) {
 843                         add_pattern(&opt, arg, "command line", 0);
 844                         break;
 845                 }
 846                 else {
 847                         /* We are looking at the first path or rev;
 848                          * it is found at argv[1] after leaving the
 849                          * loop.
 850                          */
 851                         argc++; argv--;
 852                         break;
 853                 }
 854         }
 855
 856         if (!opt.pattern_list)
 857                 die("no pattern given.");
 858         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
 859                 die("cannot mix --fixed-strings and regexp");
 860         if (!opt.fixed)
 861                 compile_patterns(&opt);
 862
 863         /* Check revs and then paths */
 864         for (i = 1; i < argc; i++) {
 865                 const char *arg = argv[i];
 866                 unsigned char sha1[20];
 867                 /* Is it a rev? */
 868                 if (!get_sha1(arg, sha1)) {
 869                         struct object *object = parse_object(sha1);
 870                         if (!object)
 871                                 die("bad object %s", arg);
 872                         add_object_array(object, arg, &list);
 873                         continue;
 874                 }
 875                 if (!strcmp(arg, "--")) {
 876                         i++;
 877                         seen_dashdash = 1;
 878                 }
 879                 break;
 880         }
 881
 882         /* The rest are paths */
 883         if (!seen_dashdash) {
 884                 int j;
 885                 for (j = i; j < argc; j++)
 886                         verify_filename(prefix, argv[j]);
 887         }
 888
 889         if (i < argc)
 890                 paths = get_pathspec(prefix, argv + i);
 891         else if (prefix) {
 892                 paths = xcalloc(2, sizeof(const char *));
 893                 paths[0] = prefix;
 894                 paths[1] = NULL;
 895         }
 896
 897         if (!list.nr)
 898                 return !grep_cache(&opt, paths, cached);
 899
 900         if (cached)
 901                 die("both --cached and trees are given.");
 902
 903         for (i = 0; i < list.nr; i++) {
 904                 struct object *real_obj;
 905                 real_obj = deref_tag(list.objects[i].item, NULL, 0);
 906                 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
 907                         hit = 1;
 908         }
 909         return !hit;
 910 }