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 <= 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
  84 struct grep_pat {
  85         struct grep_pat *next;
  86         const char *origin;
  87         int no;
  88         const char *pattern;
  89         regex_t regexp;
  90 };
  91
  92 struct 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
 112 static 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
 124 static 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
 146 static 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
 157 static int word_char(char ch)
 158 {
 159         return isalnum(ch) || ch == '_';
 160 }
 161
 162 static 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
 175 static 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
 184 static 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
 198 static 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
 366 static 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
 382 static 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
 413 static 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
 446 static 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->regflags & REG_ICASE)
 463                 push_arg("-i");
 464         if (opt->word_regexp)
 465                 push_arg("-w");
 466         if (opt->name_only)
 467                 push_arg("-l");
 468         if (opt->unmatch_name_only)
 469                 push_arg("-L");
 470         if (opt->count)
 471                 push_arg("-c");
 472         if (opt->post_context || opt->pre_context) {
 473                 if (opt->post_context != opt->pre_context) {
 474                         if (opt->pre_context) {
 475                                 push_arg("-B");
 476                                 len += snprintf(argptr, sizeof(randarg)-len,
 477                                                 "%u", opt->pre_context);
 478                                 if (sizeof(randarg) <= len)
 479                                         die("maximum length of args exceeded");
 480                                 push_arg(argptr);
 481                                 argptr += len;
 482                         }
 483                         if (opt->post_context) {
 484                                 push_arg("-A");
 485                                 len += snprintf(argptr, sizeof(randarg)-len,
 486                                                 "%u", opt->post_context);
 487                                 if (sizeof(randarg) <= len)
 488                                         die("maximum length of args exceeded");
 489                                 push_arg(argptr);
 490                                 argptr += len;
 491                         }
 492                 }
 493                 else {
 494                         push_arg("-C");
 495                         len += snprintf(argptr, sizeof(randarg)-len,
 496                                         "%u", opt->post_context);
 497                         if (sizeof(randarg) <= len)
 498                                 die("maximum length of args exceeded");
 499                         push_arg(argptr);
 500                         argptr += len;
 501                 }
 502         }
 503         for (p = opt->pattern_list; p; p = p->next) {
 504                 push_arg("-e");
 505                 push_arg(p->pattern);
 506         }
 507
 508         /*
 509          * To make sure we get the header printed out when we want it,
 510          * add /dev/null to the paths to grep.  This is unnecessary
 511          * (and wrong) with "-l" or "-L", which always print out the
 512          * name anyway.
 513          *
 514          * GNU grep has "-H", but this is portable.
 515          */
 516         if (!opt->name_only && !opt->unmatch_name_only)
 517                 push_arg("/dev/null");
 518
 519         hit = 0;
 520         argc = nr;
 521         for (i = 0; i < active_nr; i++) {
 522                 struct cache_entry *ce = active_cache[i];
 523                 char *name;
 524                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 525                         continue;
 526                 if (!pathspec_matches(paths, ce->name))
 527                         continue;
 528                 name = ce->name;
 529                 if (name[0] == '-') {
 530                         int len = ce_namelen(ce);
 531                         name = xmalloc(len + 3);
 532                         memcpy(name, "./", 2);
 533                         memcpy(name + 2, ce->name, len + 1);
 534                 }
 535                 argv[argc++] = name;
 536                 if (argc < MAXARGS)
 537                         continue;
 538                 hit += exec_grep(argc, argv);
 539                 argc = nr;
 540         }
 541         if (argc > nr)
 542                 hit += exec_grep(argc, argv);
 543         return 0;
 544 }
 545
 546 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 547 {
 548         int hit = 0;
 549         int nr;
 550         read_cache();
 551
 552 #ifdef __unix__
 553         /*
 554          * Use the external "grep" command for the case where
 555          * we grep through the checked-out files. It tends to
 556          * be a lot more optimized
 557          */
 558         if (!cached) {
 559                 hit = external_grep(opt, paths, cached);
 560                 if (hit >= 0)
 561                         return hit;
 562         }
 563 #endif
 564
 565         for (nr = 0; nr < active_nr; nr++) {
 566                 struct cache_entry *ce = active_cache[nr];
 567                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 568                         continue;
 569                 if (!pathspec_matches(paths, ce->name))
 570                         continue;
 571                 if (cached)
 572                         hit |= grep_sha1(opt, ce->sha1, ce->name);
 573                 else
 574                         hit |= grep_file(opt, ce->name);
 575         }
 576         return hit;
 577 }
 578
 579 static int grep_tree(struct grep_opt *opt, const char **paths,
 580                      struct tree_desc *tree,
 581                      const char *tree_name, const char *base)
 582 {
 583         int len;
 584         int hit = 0;
 585         struct name_entry entry;
 586         char *down;
 587         char *path_buf = xmalloc(PATH_MAX + strlen(tree_name) + 100);
 588
 589         if (tree_name[0]) {
 590                 int offset = sprintf(path_buf, "%s:", tree_name);
 591                 down = path_buf + offset;
 592                 strcat(down, base);
 593         }
 594         else {
 595                 down = path_buf;
 596                 strcpy(down, base);
 597         }
 598         len = strlen(path_buf);
 599
 600         while (tree_entry(tree, &entry)) {
 601                 strcpy(path_buf + len, entry.path);
 602
 603                 if (S_ISDIR(entry.mode))
 604                         /* Match "abc/" against pathspec to
 605                          * decide if we want to descend into "abc"
 606                          * directory.
 607                          */
 608                         strcpy(path_buf + len + entry.pathlen, "/");
 609
 610                 if (!pathspec_matches(paths, down))
 611                         ;
 612                 else if (S_ISREG(entry.mode))
 613                         hit |= grep_sha1(opt, entry.sha1, path_buf);
 614                 else if (S_ISDIR(entry.mode)) {
 615                         char type[20];
 616                         struct tree_desc sub;
 617                         void *data;
 618                         data = read_sha1_file(entry.sha1, type, &sub.size);
 619                         if (!data)
 620                                 die("unable to read tree (%s)",
 621                                     sha1_to_hex(entry.sha1));
 622                         sub.buf = data;
 623                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 624                         free(data);
 625                 }
 626         }
 627         return hit;
 628 }
 629
 630 static int grep_object(struct grep_opt *opt, const char **paths,
 631                        struct object *obj, const char *name)
 632 {
 633         if (obj->type == TYPE_BLOB)
 634                 return grep_sha1(opt, obj->sha1, name);
 635         if (obj->type == TYPE_COMMIT || obj->type == TYPE_TREE) {
 636                 struct tree_desc tree;
 637                 void *data;
 638                 int hit;
 639                 data = read_object_with_reference(obj->sha1, tree_type,
 640                                                   &tree.size, NULL);
 641                 if (!data)
 642                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 643                 tree.buf = data;
 644                 hit = grep_tree(opt, paths, &tree, name, "");
 645                 free(data);
 646                 return hit;
 647         }
 648         die("unable to grep from object of type %s", typename(obj->type));
 649 }
 650
 651 static const char builtin_grep_usage[] =
 652 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 653
 654 int cmd_grep(int argc, const char **argv, char **envp)
 655 {
 656         int hit = 0;
 657         int cached = 0;
 658         int seen_dashdash = 0;
 659         struct grep_opt opt;
 660         struct object_list *list, **tail, *object_list = NULL;
 661         const char *prefix = setup_git_directory();
 662         const char **paths = NULL;
 663         int i;
 664
 665         memset(&opt, 0, sizeof(opt));
 666         opt.pattern_tail = &opt.pattern_list;
 667         opt.regflags = REG_NEWLINE;
 668
 669         /*
 670          * If there is no -- then the paths must exist in the working
 671          * tree.  If there is no explicit pattern specified with -e or
 672          * -f, we take the first unrecognized non option to be the
 673          * pattern, but then what follows it must be zero or more
 674          * valid refs up to the -- (if exists), and then existing
 675          * paths.  If there is an explicit pattern, then the first
 676          * unrecocnized non option is the beginning of the refs list
 677          * that continues up to the -- (if exists), and then paths.
 678          */
 679
 680         tail = &object_list;
 681         while (1 < argc) {
 682                 const char *arg = argv[1];
 683                 argc--; argv++;
 684                 if (!strcmp("--cached", arg)) {
 685                         cached = 1;
 686                         continue;
 687                 }
 688                 if (!strcmp("-a", arg) ||
 689                     !strcmp("--text", arg)) {
 690                         opt.binary = GREP_BINARY_TEXT;
 691                         continue;
 692                 }
 693                 if (!strcmp("-i", arg) ||
 694                     !strcmp("--ignore-case", arg)) {
 695                         opt.regflags |= REG_ICASE;
 696                         continue;
 697                 }
 698                 if (!strcmp("-I", arg)) {
 699                         opt.binary = GREP_BINARY_NOMATCH;
 700                         continue;
 701                 }
 702                 if (!strcmp("-v", arg) ||
 703                     !strcmp("--invert-match", arg)) {
 704                         opt.invert = 1;
 705                         continue;
 706                 }
 707                 if (!strcmp("-E", arg) ||
 708                     !strcmp("--extended-regexp", arg)) {
 709                         opt.regflags |= REG_EXTENDED;
 710                         continue;
 711                 }
 712                 if (!strcmp("-F", arg) ||
 713                     !strcmp("--fixed-strings", arg)) {
 714                         opt.fixed = 1;
 715                         continue;
 716                 }
 717                 if (!strcmp("-G", arg) ||
 718                     !strcmp("--basic-regexp", arg)) {
 719                         opt.regflags &= ~REG_EXTENDED;
 720                         continue;
 721                 }
 722                 if (!strcmp("-n", arg)) {
 723                         opt.linenum = 1;
 724                         continue;
 725                 }
 726                 if (!strcmp("-H", arg)) {
 727                         /* We always show the pathname, so this
 728                          * is a noop.
 729                          */
 730                         continue;
 731                 }
 732                 if (!strcmp("-l", arg) ||
 733                     !strcmp("--files-with-matches", arg)) {
 734                         opt.name_only = 1;
 735                         continue;
 736                 }
 737                 if (!strcmp("-L", arg) ||
 738                     !strcmp("--files-without-match", arg)) {
 739                         opt.unmatch_name_only = 1;
 740                         continue;
 741                 }
 742                 if (!strcmp("-c", arg) ||
 743                     !strcmp("--count", arg)) {
 744                         opt.count = 1;
 745                         continue;
 746                 }
 747                 if (!strcmp("-w", arg) ||
 748                     !strcmp("--word-regexp", arg)) {
 749                         opt.word_regexp = 1;
 750                         continue;
 751                 }
 752                 if (!strncmp("-A", arg, 2) ||
 753                     !strncmp("-B", arg, 2) ||
 754                     !strncmp("-C", arg, 2) ||
 755                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
 756                         unsigned num;
 757                         const char *scan;
 758                         switch (arg[1]) {
 759                         case 'A': case 'B': case 'C':
 760                                 if (!arg[2]) {
 761                                         if (argc <= 1)
 762                                                 usage(builtin_grep_usage);
 763                                         scan = *++argv;
 764                                         argc--;
 765                                 }
 766                                 else
 767                                         scan = arg + 2;
 768                                 break;
 769                         default:
 770                                 scan = arg + 1;
 771                                 break;
 772                         }
 773                         if (sscanf(scan, "%u", &num) != 1)
 774                                 usage(builtin_grep_usage);
 775                         switch (arg[1]) {
 776                         case 'A':
 777                                 opt.post_context = num;
 778                                 break;
 779                         default:
 780                         case 'C':
 781                                 opt.post_context = num;
 782                         case 'B':
 783                                 opt.pre_context = num;
 784                                 break;
 785                         }
 786                         continue;
 787                 }
 788                 if (!strcmp("-f", arg)) {
 789                         FILE *patterns;
 790                         int lno = 0;
 791                         char buf[1024];
 792                         if (argc <= 1)
 793                                 usage(builtin_grep_usage);
 794                         patterns = fopen(argv[1], "r");
 795                         if (!patterns)
 796                                 die("'%s': %s", argv[1], strerror(errno));
 797                         while (fgets(buf, sizeof(buf), patterns)) {
 798                                 int len = strlen(buf);
 799                                 if (buf[len-1] == '\n')
 800                                         buf[len-1] = 0;
 801                                 /* ignore empty line like grep does */
 802                                 if (!buf[0])
 803                                         continue;
 804                                 add_pattern(&opt, strdup(buf), argv[1], ++lno);
 805                         }
 806                         fclose(patterns);
 807                         argv++;
 808                         argc--;
 809                         continue;
 810                 }
 811                 if (!strcmp("-e", arg)) {
 812                         if (1 < argc) {
 813                                 add_pattern(&opt, argv[1], "-e option", 0);
 814                                 argv++;
 815                                 argc--;
 816                                 continue;
 817                         }
 818                         usage(builtin_grep_usage);
 819                 }
 820                 if (!strcmp("--", arg))
 821                         break;
 822                 if (*arg == '-')
 823                         usage(builtin_grep_usage);
 824
 825                 /* First unrecognized non-option token */
 826                 if (!opt.pattern_list) {
 827                         add_pattern(&opt, arg, "command line", 0);
 828                         break;
 829                 }
 830                 else {
 831                         /* We are looking at the first path or rev;
 832                          * it is found at argv[1] after leaving the
 833                          * loop.
 834                          */
 835                         argc++; argv--;
 836                         break;
 837                 }
 838         }
 839
 840         if (!opt.pattern_list)
 841                 die("no pattern given.");
 842         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
 843                 die("cannot mix --fixed-strings and regexp");
 844         if (!opt.fixed)
 845                 compile_patterns(&opt);
 846
 847         /* Check revs and then paths */
 848         for (i = 1; i < argc; i++) {
 849                 const char *arg = argv[i];
 850                 unsigned char sha1[20];
 851                 /* Is it a rev? */
 852                 if (!get_sha1(arg, sha1)) {
 853                         struct object *object = parse_object(sha1);
 854                         struct object_list *elem;
 855                         if (!object)
 856                                 die("bad object %s", arg);
 857                         elem = object_list_insert(object, tail);
 858                         elem->name = arg;
 859                         tail = &elem->next;
 860                         continue;
 861                 }
 862                 if (!strcmp(arg, "--")) {
 863                         i++;
 864                         seen_dashdash = 1;
 865                 }
 866                 break;
 867         }
 868
 869         /* The rest are paths */
 870         if (!seen_dashdash) {
 871                 int j;
 872                 for (j = i; j < argc; j++)
 873                         verify_filename(prefix, argv[j]);
 874         }
 875
 876         if (i < argc)
 877                 paths = get_pathspec(prefix, argv + i);
 878         else if (prefix) {
 879                 paths = xcalloc(2, sizeof(const char *));
 880                 paths[0] = prefix;
 881                 paths[1] = NULL;
 882         }
 883
 884         if (!object_list)
 885                 return !grep_cache(&opt, paths, cached);
 886
 887         if (cached)
 888                 die("both --cached and trees are given.");
 889
 890         for (list = object_list; list; list = list->next) {
 891                 struct object *real_obj;
 892                 real_obj = deref_tag(list->item, NULL, 0);
 893                 if (grep_object(&opt, paths, real_obj, list->name))
 894                         hit = 1;
 895         }
 896         return !hit;
 897 }