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 enum grep_pat_token {
  86         GREP_PATTERN,
  87         GREP_AND,
  88         GREP_OPEN_PAREN,
  89         GREP_CLOSE_PAREN,
  90         GREP_NOT,
  91         GREP_OR,
  92 };
  93
  94 struct grep_pat {
  95         struct grep_pat *next;
  96         const char *origin;
  97         int no;
  98         enum grep_pat_token token;
  99         const char *pattern;
 100         regex_t regexp;
 101 };
 102
 103 enum grep_expr_node {
 104         GREP_NODE_ATOM,
 105         GREP_NODE_NOT,
 106         GREP_NODE_AND,
 107         GREP_NODE_OR,
 108 };
 109
 110 struct grep_expr {
 111         enum grep_expr_node node;
 112         union {
 113                 struct grep_pat *atom;
 114                 struct grep_expr *unary;
 115                 struct {
 116                         struct grep_expr *left;
 117                         struct grep_expr *right;
 118                 } binary;
 119         } u;
 120 };
 121
 122 struct grep_opt {
 123         struct grep_pat *pattern_list;
 124         struct grep_pat **pattern_tail;
 125         struct grep_expr *pattern_expression;
 126         int prefix_length;
 127         regex_t regexp;
 128         unsigned linenum:1;
 129         unsigned invert:1;
 130         unsigned name_only:1;
 131         unsigned unmatch_name_only:1;
 132         unsigned count:1;
 133         unsigned word_regexp:1;
 134         unsigned fixed:1;
 135 #define GREP_BINARY_DEFAULT     0
 136 #define GREP_BINARY_NOMATCH     1
 137 #define GREP_BINARY_TEXT        2
 138         unsigned binary:2;
 139         unsigned extended:1;
 140         unsigned relative:1;
 141         int regflags;
 142         unsigned pre_context;
 143         unsigned post_context;
 144 };
 145
 146 static void add_pattern(struct grep_opt *opt, const char *pat,
 147                         const char *origin, int no, enum grep_pat_token t)
 148 {
 149         struct grep_pat *p = xcalloc(1, sizeof(*p));
 150         p->pattern = pat;
 151         p->origin = origin;
 152         p->no = no;
 153         p->token = t;
 154         *opt->pattern_tail = p;
 155         opt->pattern_tail = &p->next;
 156         p->next = NULL;
 157 }
 158
 159 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 160 {
 161         int err = regcomp(&p->regexp, p->pattern, opt->regflags);
 162         if (err) {
 163                 char errbuf[1024];
 164                 char where[1024];
 165                 if (p->no)
 166                         sprintf(where, "In '%s' at %d, ",
 167                                 p->origin, p->no);
 168                 else if (p->origin)
 169                         sprintf(where, "%s, ", p->origin);
 170                 else
 171                         where[0] = 0;
 172                 regerror(err, &p->regexp, errbuf, 1024);
 173                 regfree(&p->regexp);
 174                 die("%s'%s': %s", where, p->pattern, errbuf);
 175         }
 176 }
 177
 178 static struct grep_expr *compile_pattern_expr(struct grep_pat **);
 179 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 180 {
 181         struct grep_pat *p;
 182         struct grep_expr *x;
 183
 184         p = *list;
 185         switch (p->token) {
 186         case GREP_PATTERN: /* atom */
 187                 x = xcalloc(1, sizeof (struct grep_expr));
 188                 x->node = GREP_NODE_ATOM;
 189                 x->u.atom = p;
 190                 *list = p->next;
 191                 return x;
 192         case GREP_OPEN_PAREN:
 193                 *list = p->next;
 194                 x = compile_pattern_expr(list);
 195                 if (!x)
 196                         return NULL;
 197                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 198                         die("unmatched parenthesis");
 199                 *list = (*list)->next;
 200                 return x;
 201         default:
 202                 return NULL;
 203         }
 204 }
 205
 206 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 207 {
 208         struct grep_pat *p;
 209         struct grep_expr *x;
 210
 211         p = *list;
 212         switch (p->token) {
 213         case GREP_NOT:
 214                 if (!p->next)
 215                         die("--not not followed by pattern expression");
 216                 *list = p->next;
 217                 x = xcalloc(1, sizeof (struct grep_expr));
 218                 x->node = GREP_NODE_NOT;
 219                 x->u.unary = compile_pattern_not(list);
 220                 if (!x->u.unary)
 221                         die("--not followed by non pattern expression");
 222                 return x;
 223         default:
 224                 return compile_pattern_atom(list);
 225         }
 226 }
 227
 228 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 229 {
 230         struct grep_pat *p;
 231         struct grep_expr *x, *y, *z;
 232
 233         x = compile_pattern_not(list);
 234         p = *list;
 235         if (p && p->token == GREP_AND) {
 236                 if (!p->next)
 237                         die("--and not followed by pattern expression");
 238                 *list = p->next;
 239                 y = compile_pattern_and(list);
 240                 if (!y)
 241                         die("--and not followed by pattern expression");
 242                 z = xcalloc(1, sizeof (struct grep_expr));
 243                 z->node = GREP_NODE_AND;
 244                 z->u.binary.left = x;
 245                 z->u.binary.right = y;
 246                 return z;
 247         }
 248         return x;
 249 }
 250
 251 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 252 {
 253         struct grep_pat *p;
 254         struct grep_expr *x, *y, *z;
 255
 256         x = compile_pattern_and(list);
 257         p = *list;
 258         if (x && p && p->token != GREP_CLOSE_PAREN) {
 259                 y = compile_pattern_or(list);
 260                 if (!y)
 261                         die("not a pattern expression %s", p->pattern);
 262                 z = xcalloc(1, sizeof (struct grep_expr));
 263                 z->node = GREP_NODE_OR;
 264                 z->u.binary.left = x;
 265                 z->u.binary.right = y;
 266                 return z;
 267         }
 268         return x;
 269 }
 270
 271 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 272 {
 273         return compile_pattern_or(list);
 274 }
 275
 276 static void compile_patterns(struct grep_opt *opt)
 277 {
 278         struct grep_pat *p;
 279
 280         /* First compile regexps */
 281         for (p = opt->pattern_list; p; p = p->next) {
 282                 if (p->token == GREP_PATTERN)
 283                         compile_regexp(p, opt);
 284                 else
 285                         opt->extended = 1;
 286         }
 287
 288         if (!opt->extended)
 289                 return;
 290
 291         /* Then bundle them up in an expression.
 292          * A classic recursive descent parser would do.
 293          */
 294         p = opt->pattern_list;
 295         opt->pattern_expression = compile_pattern_expr(&p);
 296         if (p)
 297                 die("incomplete pattern expression: %s", p->pattern);
 298 }
 299
 300 static char *end_of_line(char *cp, unsigned long *left)
 301 {
 302         unsigned long l = *left;
 303         while (l && *cp != '\n') {
 304                 l--;
 305                 cp++;
 306         }
 307         *left = l;
 308         return cp;
 309 }
 310
 311 static int word_char(char ch)
 312 {
 313         return isalnum(ch) || ch == '_';
 314 }
 315
 316 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 317                       const char *name, unsigned lno, char sign)
 318 {
 319         printf("%s%c", name, sign);
 320         if (opt->linenum)
 321                 printf("%d%c", lno, sign);
 322         printf("%.*s\n", (int)(eol-bol), bol);
 323 }
 324
 325 /*
 326  * NEEDSWORK: share code with diff.c
 327  */
 328 #define FIRST_FEW_BYTES 8000
 329 static int buffer_is_binary(const char *ptr, unsigned long size)
 330 {
 331         if (FIRST_FEW_BYTES < size)
 332                 size = FIRST_FEW_BYTES;
 333         return !!memchr(ptr, 0, size);
 334 }
 335
 336 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 337 {
 338         char *hit = strstr(line, pattern);
 339         if (!hit) {
 340                 match->rm_so = match->rm_eo = -1;
 341                 return REG_NOMATCH;
 342         }
 343         else {
 344                 match->rm_so = hit - line;
 345                 match->rm_eo = match->rm_so + strlen(pattern);
 346                 return 0;
 347         }
 348 }
 349
 350 static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol, char *eol)
 351 {
 352         int hit = 0;
 353         int at_true_bol = 1;
 354         regmatch_t pmatch[10];
 355
 356  again:
 357         if (!opt->fixed) {
 358                 regex_t *exp = &p->regexp;
 359                 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 360                                pmatch, 0);
 361         }
 362         else {
 363                 hit = !fixmatch(p->pattern, bol, pmatch);
 364         }
 365
 366         if (hit && opt->word_regexp) {
 367                 if ((pmatch[0].rm_so < 0) ||
 368                     (eol - bol) <= pmatch[0].rm_so ||
 369                     (pmatch[0].rm_eo < 0) ||
 370                     (eol - bol) < pmatch[0].rm_eo)
 371                         die("regexp returned nonsense");
 372
 373                 /* Match beginning must be either beginning of the
 374                  * line, or at word boundary (i.e. the last char must
 375                  * not be a word char).  Similarly, match end must be
 376                  * either end of the line, or at word boundary
 377                  * (i.e. the next char must not be a word char).
 378                  */
 379                 if ( ((pmatch[0].rm_so == 0 && at_true_bol) ||
 380                       !word_char(bol[pmatch[0].rm_so-1])) &&
 381                      ((pmatch[0].rm_eo == (eol-bol)) ||
 382                       !word_char(bol[pmatch[0].rm_eo])) )
 383                         ;
 384                 else
 385                         hit = 0;
 386
 387                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 388                         /* There could be more than one match on the
 389                          * line, and the first match might not be
 390                          * strict word match.  But later ones could be!
 391                          */
 392                         bol = pmatch[0].rm_so + bol + 1;
 393                         at_true_bol = 0;
 394                         goto again;
 395                 }
 396         }
 397         return hit;
 398 }
 399
 400 static int match_expr_eval(struct grep_opt *opt,
 401                            struct grep_expr *x,
 402                            char *bol, char *eol)
 403 {
 404         switch (x->node) {
 405         case GREP_NODE_ATOM:
 406                 return match_one_pattern(opt, x->u.atom, bol, eol);
 407                 break;
 408         case GREP_NODE_NOT:
 409                 return !match_expr_eval(opt, x->u.unary, bol, eol);
 410         case GREP_NODE_AND:
 411                 return (match_expr_eval(opt, x->u.binary.left, bol, eol) &&
 412                         match_expr_eval(opt, x->u.binary.right, bol, eol));
 413         case GREP_NODE_OR:
 414                 return (match_expr_eval(opt, x->u.binary.left, bol, eol) ||
 415                         match_expr_eval(opt, x->u.binary.right, bol, eol));
 416         }
 417         die("Unexpected node type (internal error) %d\n", x->node);
 418 }
 419
 420 static int match_expr(struct grep_opt *opt, char *bol, char *eol)
 421 {
 422         struct grep_expr *x = opt->pattern_expression;
 423         return match_expr_eval(opt, x, bol, eol);
 424 }
 425
 426 static int match_line(struct grep_opt *opt, char *bol, char *eol)
 427 {
 428         struct grep_pat *p;
 429         if (opt->extended)
 430                 return match_expr(opt, bol, eol);
 431         for (p = opt->pattern_list; p; p = p->next) {
 432                 if (match_one_pattern(opt, p, bol, eol))
 433                         return 1;
 434         }
 435         return 0;
 436 }
 437
 438 static int grep_buffer(struct grep_opt *opt, const char *name,
 439                        char *buf, unsigned long size)
 440 {
 441         char *bol = buf;
 442         unsigned long left = size;
 443         unsigned lno = 1;
 444         struct pre_context_line {
 445                 char *bol;
 446                 char *eol;
 447         } *prev = NULL, *pcl;
 448         unsigned last_hit = 0;
 449         unsigned last_shown = 0;
 450         int binary_match_only = 0;
 451         const char *hunk_mark = "";
 452         unsigned count = 0;
 453
 454         if (buffer_is_binary(buf, size)) {
 455                 switch (opt->binary) {
 456                 case GREP_BINARY_DEFAULT:
 457                         binary_match_only = 1;
 458                         break;
 459                 case GREP_BINARY_NOMATCH:
 460                         return 0; /* Assume unmatch */
 461                         break;
 462                 default:
 463                         break;
 464                 }
 465         }
 466
 467         if (opt->pre_context)
 468                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 469         if (opt->pre_context || opt->post_context)
 470                 hunk_mark = "--\n";
 471
 472         while (left) {
 473                 char *eol, ch;
 474                 int hit = 0;
 475
 476                 eol = end_of_line(bol, &left);
 477                 ch = *eol;
 478                 *eol = 0;
 479
 480                 hit = match_line(opt, bol, eol);
 481
 482                 /* "grep -v -e foo -e bla" should list lines
 483                  * that do not have either, so inversion should
 484                  * be done outside.
 485                  */
 486                 if (opt->invert)
 487                         hit = !hit;
 488                 if (opt->unmatch_name_only) {
 489                         if (hit)
 490                                 return 0;
 491                         goto next_line;
 492                 }
 493                 if (hit) {
 494                         count++;
 495                         if (binary_match_only) {
 496                                 printf("Binary file %s matches\n", name);
 497                                 return 1;
 498                         }
 499                         if (opt->name_only) {
 500                                 printf("%s\n", name);
 501                                 return 1;
 502                         }
 503                         /* Hit at this line.  If we haven't shown the
 504                          * pre-context lines, we would need to show them.
 505                          * When asked to do "count", this still show
 506                          * the context which is nonsense, but the user
 507                          * deserves to get that ;-).
 508                          */
 509                         if (opt->pre_context) {
 510                                 unsigned from;
 511                                 if (opt->pre_context < lno)
 512                                         from = lno - opt->pre_context;
 513                                 else
 514                                         from = 1;
 515                                 if (from <= last_shown)
 516                                         from = last_shown + 1;
 517                                 if (last_shown && from != last_shown + 1)
 518                                         printf(hunk_mark);
 519                                 while (from < lno) {
 520                                         pcl = &prev[lno-from-1];
 521                                         show_line(opt, pcl->bol, pcl->eol,
 522                                                   name, from, '-');
 523                                         from++;
 524                                 }
 525                                 last_shown = lno-1;
 526                         }
 527                         if (last_shown && lno != last_shown + 1)
 528                                 printf(hunk_mark);
 529                         if (!opt->count)
 530                                 show_line(opt, bol, eol, name, lno, ':');
 531                         last_shown = last_hit = lno;
 532                 }
 533                 else if (last_hit &&
 534                          lno <= last_hit + opt->post_context) {
 535                         /* If the last hit is within the post context,
 536                          * we need to show this line.
 537                          */
 538                         if (last_shown && lno != last_shown + 1)
 539                                 printf(hunk_mark);
 540                         show_line(opt, bol, eol, name, lno, '-');
 541                         last_shown = lno;
 542                 }
 543                 if (opt->pre_context) {
 544                         memmove(prev+1, prev,
 545                                 (opt->pre_context-1) * sizeof(*prev));
 546                         prev->bol = bol;
 547                         prev->eol = eol;
 548                 }
 549
 550         next_line:
 551                 *eol = ch;
 552                 bol = eol + 1;
 553                 if (!left)
 554                         break;
 555                 left--;
 556                 lno++;
 557         }
 558
 559         if (opt->unmatch_name_only) {
 560                 /* We did not see any hit, so we want to show this */
 561                 printf("%s\n", name);
 562                 return 1;
 563         }
 564
 565         /* NEEDSWORK:
 566          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 567          * which feels mostly useless but sometimes useful.  Maybe
 568          * make it another option?  For now suppress them.
 569          */
 570         if (opt->count && count)
 571                 printf("%s:%u\n", name, count);
 572         return !!last_hit;
 573 }
 574
 575 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
 576 {
 577         unsigned long size;
 578         char *data;
 579         char type[20];
 580         char *to_free = NULL;
 581         int hit;
 582
 583         data = read_sha1_file(sha1, type, &size);
 584         if (!data) {
 585                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 586                 return 0;
 587         }
 588         if (opt->relative && opt->prefix_length) {
 589                 static char name_buf[PATH_MAX];
 590                 char *cp;
 591                 int name_len = strlen(name) - opt->prefix_length + 1;
 592
 593                 if (!tree_name_len)
 594                         name += opt->prefix_length;
 595                 else {
 596                         if (ARRAY_SIZE(name_buf) <= name_len)
 597                                 cp = to_free = xmalloc(name_len);
 598                         else
 599                                 cp = name_buf;
 600                         memcpy(cp, name, tree_name_len);
 601                         strcpy(cp + tree_name_len,
 602                                name + tree_name_len + opt->prefix_length);
 603                         name = cp;
 604                 }
 605         }
 606         hit = grep_buffer(opt, name, data, size);
 607         free(data);
 608         free(to_free);
 609         return hit;
 610 }
 611
 612 static int grep_file(struct grep_opt *opt, const char *filename)
 613 {
 614         struct stat st;
 615         int i;
 616         char *data;
 617         if (lstat(filename, &st) < 0) {
 618         err_ret:
 619                 if (errno != ENOENT)
 620                         error("'%s': %s", filename, strerror(errno));
 621                 return 0;
 622         }
 623         if (!st.st_size)
 624                 return 0; /* empty file -- no grep hit */
 625         if (!S_ISREG(st.st_mode))
 626                 return 0;
 627         i = open(filename, O_RDONLY);
 628         if (i < 0)
 629                 goto err_ret;
 630         data = xmalloc(st.st_size + 1);
 631         if (st.st_size != xread(i, data, st.st_size)) {
 632                 error("'%s': short read %s", filename, strerror(errno));
 633                 close(i);
 634                 free(data);
 635                 return 0;
 636         }
 637         close(i);
 638         if (opt->relative && opt->prefix_length)
 639                 filename += opt->prefix_length;
 640         i = grep_buffer(opt, filename, data, st.st_size);
 641         free(data);
 642         return i;
 643 }
 644
 645 static int exec_grep(int argc, const char **argv)
 646 {
 647         pid_t pid;
 648         int status;
 649
 650         argv[argc] = NULL;
 651         pid = fork();
 652         if (pid < 0)
 653                 return pid;
 654         if (!pid) {
 655                 execvp("grep", (char **) argv);
 656                 exit(255);
 657         }
 658         while (waitpid(pid, &status, 0) < 0) {
 659                 if (errno == EINTR)
 660                         continue;
 661                 return -1;
 662         }
 663         if (WIFEXITED(status)) {
 664                 if (!WEXITSTATUS(status))
 665                         return 1;
 666                 return 0;
 667         }
 668         return -1;
 669 }
 670
 671 #define MAXARGS 1000
 672 #define ARGBUF 4096
 673 #define push_arg(a) do { \
 674         if (nr < MAXARGS) argv[nr++] = (a); \
 675         else die("maximum number of args exceeded"); \
 676         } while (0)
 677
 678 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
 679 {
 680         int i, nr, argc, hit, len, status;
 681         const char *argv[MAXARGS+1];
 682         char randarg[ARGBUF];
 683         char *argptr = randarg;
 684         struct grep_pat *p;
 685
 686         if (opt->extended || (opt->relative && opt->prefix_length))
 687                 return -1;
 688         len = nr = 0;
 689         push_arg("grep");
 690         if (opt->fixed)
 691                 push_arg("-F");
 692         if (opt->linenum)
 693                 push_arg("-n");
 694         if (opt->regflags & REG_EXTENDED)
 695                 push_arg("-E");
 696         if (opt->regflags & REG_ICASE)
 697                 push_arg("-i");
 698         if (opt->word_regexp)
 699                 push_arg("-w");
 700         if (opt->name_only)
 701                 push_arg("-l");
 702         if (opt->unmatch_name_only)
 703                 push_arg("-L");
 704         if (opt->count)
 705                 push_arg("-c");
 706         if (opt->post_context || opt->pre_context) {
 707                 if (opt->post_context != opt->pre_context) {
 708                         if (opt->pre_context) {
 709                                 push_arg("-B");
 710                                 len += snprintf(argptr, sizeof(randarg)-len,
 711                                                 "%u", opt->pre_context);
 712                                 if (sizeof(randarg) <= len)
 713                                         die("maximum length of args exceeded");
 714                                 push_arg(argptr);
 715                                 argptr += len;
 716                         }
 717                         if (opt->post_context) {
 718                                 push_arg("-A");
 719                                 len += snprintf(argptr, sizeof(randarg)-len,
 720                                                 "%u", opt->post_context);
 721                                 if (sizeof(randarg) <= len)
 722                                         die("maximum length of args exceeded");
 723                                 push_arg(argptr);
 724                                 argptr += len;
 725                         }
 726                 }
 727                 else {
 728                         push_arg("-C");
 729                         len += snprintf(argptr, sizeof(randarg)-len,
 730                                         "%u", opt->post_context);
 731                         if (sizeof(randarg) <= len)
 732                                 die("maximum length of args exceeded");
 733                         push_arg(argptr);
 734                         argptr += len;
 735                 }
 736         }
 737         for (p = opt->pattern_list; p; p = p->next) {
 738                 push_arg("-e");
 739                 push_arg(p->pattern);
 740         }
 741
 742         /*
 743          * To make sure we get the header printed out when we want it,
 744          * add /dev/null to the paths to grep.  This is unnecessary
 745          * (and wrong) with "-l" or "-L", which always print out the
 746          * name anyway.
 747          *
 748          * GNU grep has "-H", but this is portable.
 749          */
 750         if (!opt->name_only && !opt->unmatch_name_only)
 751                 push_arg("/dev/null");
 752
 753         hit = 0;
 754         argc = nr;
 755         for (i = 0; i < active_nr; i++) {
 756                 struct cache_entry *ce = active_cache[i];
 757                 char *name;
 758                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 759                         continue;
 760                 if (!pathspec_matches(paths, ce->name))
 761                         continue;
 762                 name = ce->name;
 763                 if (name[0] == '-') {
 764                         int len = ce_namelen(ce);
 765                         name = xmalloc(len + 3);
 766                         memcpy(name, "./", 2);
 767                         memcpy(name + 2, ce->name, len + 1);
 768                 }
 769                 argv[argc++] = name;
 770                 if (argc < MAXARGS)
 771                         continue;
 772                 status = exec_grep(argc, argv);
 773                 if (0 < status)
 774                         hit = 1;
 775                 argc = nr;
 776         }
 777         if (argc > nr) {
 778                 status = exec_grep(argc, argv);
 779                 if (0 < status)
 780                         hit = 1;
 781         }
 782         return hit;
 783 }
 784
 785 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 786 {
 787         int hit = 0;
 788         int nr;
 789         read_cache();
 790
 791 #ifdef __unix__
 792         /*
 793          * Use the external "grep" command for the case where
 794          * we grep through the checked-out files. It tends to
 795          * be a lot more optimized
 796          */
 797         if (!cached) {
 798                 hit = external_grep(opt, paths, cached);
 799                 if (hit >= 0)
 800                         return hit;
 801         }
 802 #endif
 803
 804         for (nr = 0; nr < active_nr; nr++) {
 805                 struct cache_entry *ce = active_cache[nr];
 806                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 807                         continue;
 808                 if (!pathspec_matches(paths, ce->name))
 809                         continue;
 810                 if (cached)
 811                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 812                 else
 813                         hit |= grep_file(opt, ce->name);
 814         }
 815         return hit;
 816 }
 817
 818 static int grep_tree(struct grep_opt *opt, const char **paths,
 819                      struct tree_desc *tree,
 820                      const char *tree_name, const char *base)
 821 {
 822         int len;
 823         int hit = 0;
 824         struct name_entry entry;
 825         char *down;
 826         int tn_len = strlen(tree_name);
 827         char *path_buf = xmalloc(PATH_MAX + tn_len + 100);
 828
 829         if (tn_len) {
 830                 tn_len = sprintf(path_buf, "%s:", tree_name);
 831                 down = path_buf + tn_len;
 832                 strcat(down, base);
 833         }
 834         else {
 835                 down = path_buf;
 836                 strcpy(down, base);
 837         }
 838         len = strlen(path_buf);
 839
 840         while (tree_entry(tree, &entry)) {
 841                 strcpy(path_buf + len, entry.path);
 842
 843                 if (S_ISDIR(entry.mode))
 844                         /* Match "abc/" against pathspec to
 845                          * decide if we want to descend into "abc"
 846                          * directory.
 847                          */
 848                         strcpy(path_buf + len + entry.pathlen, "/");
 849
 850                 if (!pathspec_matches(paths, down))
 851                         ;
 852                 else if (S_ISREG(entry.mode))
 853                         hit |= grep_sha1(opt, entry.sha1, path_buf, tn_len);
 854                 else if (S_ISDIR(entry.mode)) {
 855                         char type[20];
 856                         struct tree_desc sub;
 857                         void *data;
 858                         data = read_sha1_file(entry.sha1, type, &sub.size);
 859                         if (!data)
 860                                 die("unable to read tree (%s)",
 861                                     sha1_to_hex(entry.sha1));
 862                         sub.buf = data;
 863                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 864                         free(data);
 865                 }
 866         }
 867         return hit;
 868 }
 869
 870 static int grep_object(struct grep_opt *opt, const char **paths,
 871                        struct object *obj, const char *name)
 872 {
 873         if (obj->type == OBJ_BLOB)
 874                 return grep_sha1(opt, obj->sha1, name, 0);
 875         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 876                 struct tree_desc tree;
 877                 void *data;
 878                 int hit;
 879                 data = read_object_with_reference(obj->sha1, tree_type,
 880                                                   &tree.size, NULL);
 881                 if (!data)
 882                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 883                 tree.buf = data;
 884                 hit = grep_tree(opt, paths, &tree, name, "");
 885                 free(data);
 886                 return hit;
 887         }
 888         die("unable to grep from object of type %s", typename(obj->type));
 889 }
 890
 891 static const char builtin_grep_usage[] =
 892 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 893
 894 static const char emsg_invalid_context_len[] =
 895 "%s: invalid context length argument";
 896 static const char emsg_missing_context_len[] =
 897 "missing context length argument";
 898 static const char emsg_missing_argument[] =
 899 "option requires an argument -%s";
 900
 901 int cmd_grep(int argc, const char **argv, const char *prefix)
 902 {
 903         int hit = 0;
 904         int cached = 0;
 905         int seen_dashdash = 0;
 906         struct grep_opt opt;
 907         struct object_array list = { 0, 0, NULL };
 908         const char **paths = NULL;
 909         int i;
 910
 911         memset(&opt, 0, sizeof(opt));
 912         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 913         opt.relative = 1;
 914         opt.pattern_tail = &opt.pattern_list;
 915         opt.regflags = REG_NEWLINE;
 916
 917         /*
 918          * If there is no -- then the paths must exist in the working
 919          * tree.  If there is no explicit pattern specified with -e or
 920          * -f, we take the first unrecognized non option to be the
 921          * pattern, but then what follows it must be zero or more
 922          * valid refs up to the -- (if exists), and then existing
 923          * paths.  If there is an explicit pattern, then the first
 924          * unrecognized non option is the beginning of the refs list
 925          * that continues up to the -- (if exists), and then paths.
 926          */
 927
 928         while (1 < argc) {
 929                 const char *arg = argv[1];
 930                 argc--; argv++;
 931                 if (!strcmp("--cached", arg)) {
 932                         cached = 1;
 933                         continue;
 934                 }
 935                 if (!strcmp("-a", arg) ||
 936                     !strcmp("--text", arg)) {
 937                         opt.binary = GREP_BINARY_TEXT;
 938                         continue;
 939                 }
 940                 if (!strcmp("-i", arg) ||
 941                     !strcmp("--ignore-case", arg)) {
 942                         opt.regflags |= REG_ICASE;
 943                         continue;
 944                 }
 945                 if (!strcmp("-I", arg)) {
 946                         opt.binary = GREP_BINARY_NOMATCH;
 947                         continue;
 948                 }
 949                 if (!strcmp("-v", arg) ||
 950                     !strcmp("--invert-match", arg)) {
 951                         opt.invert = 1;
 952                         continue;
 953                 }
 954                 if (!strcmp("-E", arg) ||
 955                     !strcmp("--extended-regexp", arg)) {
 956                         opt.regflags |= REG_EXTENDED;
 957                         continue;
 958                 }
 959                 if (!strcmp("-F", arg) ||
 960                     !strcmp("--fixed-strings", arg)) {
 961                         opt.fixed = 1;
 962                         continue;
 963                 }
 964                 if (!strcmp("-G", arg) ||
 965                     !strcmp("--basic-regexp", arg)) {
 966                         opt.regflags &= ~REG_EXTENDED;
 967                         continue;
 968                 }
 969                 if (!strcmp("-n", arg)) {
 970                         opt.linenum = 1;
 971                         continue;
 972                 }
 973                 if (!strcmp("-H", arg)) {
 974                         /* We always show the pathname, so this
 975                          * is a noop.
 976                          */
 977                         continue;
 978                 }
 979                 if (!strcmp("-l", arg) ||
 980                     !strcmp("--files-with-matches", arg)) {
 981                         opt.name_only = 1;
 982                         continue;
 983                 }
 984                 if (!strcmp("-L", arg) ||
 985                     !strcmp("--files-without-match", arg)) {
 986                         opt.unmatch_name_only = 1;
 987                         continue;
 988                 }
 989                 if (!strcmp("-c", arg) ||
 990                     !strcmp("--count", arg)) {
 991                         opt.count = 1;
 992                         continue;
 993                 }
 994                 if (!strcmp("-w", arg) ||
 995                     !strcmp("--word-regexp", arg)) {
 996                         opt.word_regexp = 1;
 997                         continue;
 998                 }
 999                 if (!strncmp("-A", arg, 2) ||
1000                     !strncmp("-B", arg, 2) ||
1001                     !strncmp("-C", arg, 2) ||
1002                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
1003                         unsigned num;
1004                         const char *scan;
1005                         switch (arg[1]) {
1006                         case 'A': case 'B': case 'C':
1007                                 if (!arg[2]) {
1008                                         if (argc <= 1)
1009                                                 die(emsg_missing_context_len);
1010                                         scan = *++argv;
1011                                         argc--;
1012                                 }
1013                                 else
1014                                         scan = arg + 2;
1015                                 break;
1016                         default:
1017                                 scan = arg + 1;
1018                                 break;
1019                         }
1020                         if (sscanf(scan, "%u", &num) != 1)
1021                                 die(emsg_invalid_context_len, scan);
1022                         switch (arg[1]) {
1023                         case 'A':
1024                                 opt.post_context = num;
1025                                 break;
1026                         default:
1027                         case 'C':
1028                                 opt.post_context = num;
1029                         case 'B':
1030                                 opt.pre_context = num;
1031                                 break;
1032                         }
1033                         continue;
1034                 }
1035                 if (!strcmp("-f", arg)) {
1036                         FILE *patterns;
1037                         int lno = 0;
1038                         char buf[1024];
1039                         if (argc <= 1)
1040                                 die(emsg_missing_argument, arg);
1041                         patterns = fopen(argv[1], "r");
1042                         if (!patterns)
1043                                 die("'%s': %s", argv[1], strerror(errno));
1044                         while (fgets(buf, sizeof(buf), patterns)) {
1045                                 int len = strlen(buf);
1046                                 if (buf[len-1] == '\n')
1047                                         buf[len-1] = 0;
1048                                 /* ignore empty line like grep does */
1049                                 if (!buf[0])
1050                                         continue;
1051                                 add_pattern(&opt, xstrdup(buf), argv[1], ++lno,
1052                                             GREP_PATTERN);
1053                         }
1054                         fclose(patterns);
1055                         argv++;
1056                         argc--;
1057                         continue;
1058                 }
1059                 if (!strcmp("--not", arg)) {
1060                         add_pattern(&opt, arg, "command line", 0, GREP_NOT);
1061                         continue;
1062                 }
1063                 if (!strcmp("--and", arg)) {
1064                         add_pattern(&opt, arg, "command line", 0, GREP_AND);
1065                         continue;
1066                 }
1067                 if (!strcmp("--or", arg))
1068                         continue; /* no-op */
1069                 if (!strcmp("(", arg)) {
1070                         add_pattern(&opt, arg, "command line", 0, GREP_OPEN_PAREN);
1071                         continue;
1072                 }
1073                 if (!strcmp(")", arg)) {
1074                         add_pattern(&opt, arg, "command line", 0, GREP_CLOSE_PAREN);
1075                         continue;
1076                 }
1077                 if (!strcmp("-e", arg)) {
1078                         if (1 < argc) {
1079                                 add_pattern(&opt, argv[1], "-e option", 0,
1080                                             GREP_PATTERN);
1081                                 argv++;
1082                                 argc--;
1083                                 continue;
1084                         }
1085                         die(emsg_missing_argument, arg);
1086                 }
1087                 if (!strcmp("--full-name", arg)) {
1088                         opt.relative = 0;
1089                         continue;
1090                 }
1091                 if (!strcmp("--", arg)) {
1092                         /* later processing wants to have this at argv[1] */
1093                         argv--;
1094                         argc++;
1095                         break;
1096                 }
1097                 if (*arg == '-')
1098                         usage(builtin_grep_usage);
1099
1100                 /* First unrecognized non-option token */
1101                 if (!opt.pattern_list) {
1102                         add_pattern(&opt, arg, "command line", 0,
1103                                     GREP_PATTERN);
1104                         break;
1105                 }
1106                 else {
1107                         /* We are looking at the first path or rev;
1108                          * it is found at argv[1] after leaving the
1109                          * loop.
1110                          */
1111                         argc++; argv--;
1112                         break;
1113                 }
1114         }
1115
1116         if (!opt.pattern_list)
1117                 die("no pattern given.");
1118         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
1119                 die("cannot mix --fixed-strings and regexp");
1120         if (!opt.fixed)
1121                 compile_patterns(&opt);
1122
1123         /* Check revs and then paths */
1124         for (i = 1; i < argc; i++) {
1125                 const char *arg = argv[i];
1126                 unsigned char sha1[20];
1127                 /* Is it a rev? */
1128                 if (!get_sha1(arg, sha1)) {
1129                         struct object *object = parse_object(sha1);
1130                         if (!object)
1131                                 die("bad object %s", arg);
1132                         add_object_array(object, arg, &list);
1133                         continue;
1134                 }
1135                 if (!strcmp(arg, "--")) {
1136                         i++;
1137                         seen_dashdash = 1;
1138                 }
1139                 break;
1140         }
1141
1142         /* The rest are paths */
1143         if (!seen_dashdash) {
1144                 int j;
1145                 for (j = i; j < argc; j++)
1146                         verify_filename(prefix, argv[j]);
1147         }
1148
1149         if (i < argc) {
1150                 paths = get_pathspec(prefix, argv + i);
1151                 if (opt.prefix_length && opt.relative) {
1152                         /* Make sure we do not get outside of paths */
1153                         for (i = 0; paths[i]; i++)
1154                                 if (strncmp(prefix, paths[i], opt.prefix_length))
1155                                         die("git-grep: cannot generate relative filenames containing '..'");
1156                 }
1157         }
1158         else if (prefix) {
1159                 paths = xcalloc(2, sizeof(const char *));
1160                 paths[0] = prefix;
1161                 paths[1] = NULL;
1162         }
1163
1164         if (!list.nr)
1165                 return !grep_cache(&opt, paths, cached);
1166
1167         if (cached)
1168                 die("both --cached and trees are given.");
1169
1170         for (i = 0; i < list.nr; i++) {
1171                 struct object *real_obj;
1172                 real_obj = deref_tag(list.objects[i].item, NULL, 0);
1173                 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
1174                         hit = 1;
1175         }
1176         return !hit;
1177 }