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 #if DEBUG
 179 static inline void indent(int in)
 180 {
 181         int i;
 182         for (i = 0; i < in; i++) putchar(' ');
 183 }
 184
 185 static void dump_pattern_exp(struct grep_expr *x, int in)
 186 {
 187         switch (x->node) {
 188         case GREP_NODE_ATOM:
 189                 indent(in);
 190                 puts(x->u.atom->pattern);
 191                 break;
 192         case GREP_NODE_NOT:
 193                 indent(in);
 194                 puts("--not");
 195                 dump_pattern_exp(x->u.unary, in+1);
 196                 break;
 197         case GREP_NODE_AND:
 198                 dump_pattern_exp(x->u.binary.left, in+1);
 199                 indent(in);
 200                 puts("--and");
 201                 dump_pattern_exp(x->u.binary.right, in+1);
 202                 break;
 203         case GREP_NODE_OR:
 204                 dump_pattern_exp(x->u.binary.left, in+1);
 205                 indent(in);
 206                 puts("--or");
 207                 dump_pattern_exp(x->u.binary.right, in+1);
 208                 break;
 209         }
 210 }
 211
 212 static void looking_at(const char *msg, struct grep_pat **list)
 213 {
 214         struct grep_pat *p = *list;
 215         fprintf(stderr, "%s: looking at ", msg);
 216         if (!p)
 217                 fprintf(stderr, "empty\n");
 218         else
 219                 fprintf(stderr, "<%s>\n", p->pattern);
 220 }
 221 #else
 222 #define looking_at(a,b) do {} while(0)
 223 #endif
 224
 225 static struct grep_expr *compile_pattern_expr(struct grep_pat **);
 226 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 227 {
 228         struct grep_pat *p;
 229         struct grep_expr *x;
 230
 231         looking_at("atom", list);
 232
 233         p = *list;
 234         switch (p->token) {
 235         case GREP_PATTERN: /* atom */
 236                 x = xcalloc(1, sizeof (struct grep_expr));
 237                 x->node = GREP_NODE_ATOM;
 238                 x->u.atom = p;
 239                 *list = p->next;
 240                 return x;
 241         case GREP_OPEN_PAREN:
 242                 *list = p->next;
 243                 x = compile_pattern_expr(list);
 244                 if (!x)
 245                         return NULL;
 246                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 247                         die("unmatched parenthesis");
 248                 *list = (*list)->next;
 249                 return x;
 250         default:
 251                 return NULL;
 252         }
 253 }
 254
 255 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 256 {
 257         struct grep_pat *p;
 258         struct grep_expr *x;
 259
 260         looking_at("not", list);
 261
 262         p = *list;
 263         switch (p->token) {
 264         case GREP_NOT:
 265                 if (!p->next)
 266                         die("--not not followed by pattern expression");
 267                 *list = p->next;
 268                 x = xcalloc(1, sizeof (struct grep_expr));
 269                 x->node = GREP_NODE_NOT;
 270                 x->u.unary = compile_pattern_not(list);
 271                 if (!x->u.unary)
 272                         die("--not followed by non pattern expression");
 273                 return x;
 274         default:
 275                 return compile_pattern_atom(list);
 276         }
 277 }
 278
 279 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 280 {
 281         struct grep_pat *p;
 282         struct grep_expr *x, *y, *z;
 283
 284         looking_at("and", list);
 285
 286         x = compile_pattern_not(list);
 287         p = *list;
 288         if (p && p->token == GREP_AND) {
 289                 if (!p->next)
 290                         die("--and not followed by pattern expression");
 291                 *list = p->next;
 292                 y = compile_pattern_and(list);
 293                 if (!y)
 294                         die("--and not followed by pattern expression");
 295                 z = xcalloc(1, sizeof (struct grep_expr));
 296                 z->node = GREP_NODE_AND;
 297                 z->u.binary.left = x;
 298                 z->u.binary.right = y;
 299                 return z;
 300         }
 301         return x;
 302 }
 303
 304 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 305 {
 306         struct grep_pat *p;
 307         struct grep_expr *x, *y, *z;
 308
 309         looking_at("or", list);
 310
 311         x = compile_pattern_and(list);
 312         p = *list;
 313         if (x && p && p->token != GREP_CLOSE_PAREN) {
 314                 y = compile_pattern_or(list);
 315                 if (!y)
 316                         die("not a pattern expression %s", p->pattern);
 317                 z = xcalloc(1, sizeof (struct grep_expr));
 318                 z->node = GREP_NODE_OR;
 319                 z->u.binary.left = x;
 320                 z->u.binary.right = y;
 321                 return z;
 322         }
 323         return x;
 324 }
 325
 326 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 327 {
 328         looking_at("expr", list);
 329
 330         return compile_pattern_or(list);
 331 }
 332
 333 static void compile_patterns(struct grep_opt *opt)
 334 {
 335         struct grep_pat *p;
 336
 337         /* First compile regexps */
 338         for (p = opt->pattern_list; p; p = p->next) {
 339                 if (p->token == GREP_PATTERN)
 340                         compile_regexp(p, opt);
 341                 else
 342                         opt->extended = 1;
 343         }
 344
 345         if (!opt->extended)
 346                 return;
 347
 348         /* Then bundle them up in an expression.
 349          * A classic recursive descent parser would do.
 350          */
 351         p = opt->pattern_list;
 352         opt->pattern_expression = compile_pattern_expr(&p);
 353 #if DEBUG
 354         dump_pattern_exp(opt->pattern_expression, 0);
 355 #endif
 356         if (p)
 357                 die("incomplete pattern expression: %s", p->pattern);
 358 }
 359
 360 static char *end_of_line(char *cp, unsigned long *left)
 361 {
 362         unsigned long l = *left;
 363         while (l && *cp != '\n') {
 364                 l--;
 365                 cp++;
 366         }
 367         *left = l;
 368         return cp;
 369 }
 370
 371 static int word_char(char ch)
 372 {
 373         return isalnum(ch) || ch == '_';
 374 }
 375
 376 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 377                       const char *name, unsigned lno, char sign)
 378 {
 379         printf("%s%c", name, sign);
 380         if (opt->linenum)
 381                 printf("%d%c", lno, sign);
 382         printf("%.*s\n", (int)(eol-bol), bol);
 383 }
 384
 385 /*
 386  * NEEDSWORK: share code with diff.c
 387  */
 388 #define FIRST_FEW_BYTES 8000
 389 static int buffer_is_binary(const char *ptr, unsigned long size)
 390 {
 391         if (FIRST_FEW_BYTES < size)
 392                 size = FIRST_FEW_BYTES;
 393         if (memchr(ptr, 0, size))
 394                 return 1;
 395         return 0;
 396 }
 397
 398 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 399 {
 400         char *hit = strstr(line, pattern);
 401         if (!hit) {
 402                 match->rm_so = match->rm_eo = -1;
 403                 return REG_NOMATCH;
 404         }
 405         else {
 406                 match->rm_so = hit - line;
 407                 match->rm_eo = match->rm_so + strlen(pattern);
 408                 return 0;
 409         }
 410 }
 411
 412 static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol, char *eol)
 413 {
 414         int hit = 0;
 415         int at_true_bol = 1;
 416         regmatch_t pmatch[10];
 417
 418  again:
 419         if (!opt->fixed) {
 420                 regex_t *exp = &p->regexp;
 421                 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 422                                pmatch, 0);
 423         }
 424         else {
 425                 hit = !fixmatch(p->pattern, bol, pmatch);
 426         }
 427
 428         if (hit && opt->word_regexp) {
 429                 if ((pmatch[0].rm_so < 0) ||
 430                     (eol - bol) <= pmatch[0].rm_so ||
 431                     (pmatch[0].rm_eo < 0) ||
 432                     (eol - bol) < pmatch[0].rm_eo)
 433                         die("regexp returned nonsense");
 434
 435                 /* Match beginning must be either beginning of the
 436                  * line, or at word boundary (i.e. the last char must
 437                  * not be a word char).  Similarly, match end must be
 438                  * either end of the line, or at word boundary
 439                  * (i.e. the next char must not be a word char).
 440                  */
 441                 if ( ((pmatch[0].rm_so == 0 && at_true_bol) ||
 442                       !word_char(bol[pmatch[0].rm_so-1])) &&
 443                      ((pmatch[0].rm_eo == (eol-bol)) ||
 444                       !word_char(bol[pmatch[0].rm_eo])) )
 445                         ;
 446                 else
 447                         hit = 0;
 448
 449                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 450                         /* There could be more than one match on the
 451                          * line, and the first match might not be
 452                          * strict word match.  But later ones could be!
 453                          */
 454                         bol = pmatch[0].rm_so + bol + 1;
 455                         at_true_bol = 0;
 456                         goto again;
 457                 }
 458         }
 459         return hit;
 460 }
 461
 462 static int match_expr_eval(struct grep_opt *opt,
 463                            struct grep_expr *x,
 464                            char *bol, char *eol)
 465 {
 466         switch (x->node) {
 467         case GREP_NODE_ATOM:
 468                 return match_one_pattern(opt, x->u.atom, bol, eol);
 469                 break;
 470         case GREP_NODE_NOT:
 471                 return !match_expr_eval(opt, x->u.unary, bol, eol);
 472         case GREP_NODE_AND:
 473                 return (match_expr_eval(opt, x->u.binary.left, bol, eol) &&
 474                         match_expr_eval(opt, x->u.binary.right, bol, eol));
 475         case GREP_NODE_OR:
 476                 return (match_expr_eval(opt, x->u.binary.left, bol, eol) ||
 477                         match_expr_eval(opt, x->u.binary.right, bol, eol));
 478         }
 479         die("Unexpected node type (internal error) %d\n", x->node);
 480 }
 481
 482 static int match_expr(struct grep_opt *opt, char *bol, char *eol)
 483 {
 484         struct grep_expr *x = opt->pattern_expression;
 485         return match_expr_eval(opt, x, bol, eol);
 486 }
 487
 488 static int match_line(struct grep_opt *opt, char *bol, char *eol)
 489 {
 490         struct grep_pat *p;
 491         if (opt->extended)
 492                 return match_expr(opt, bol, eol);
 493         for (p = opt->pattern_list; p; p = p->next) {
 494                 if (match_one_pattern(opt, p, bol, eol))
 495                         return 1;
 496         }
 497         return 0;
 498 }
 499
 500 static int grep_buffer(struct grep_opt *opt, const char *name,
 501                        char *buf, unsigned long size)
 502 {
 503         char *bol = buf;
 504         unsigned long left = size;
 505         unsigned lno = 1;
 506         struct pre_context_line {
 507                 char *bol;
 508                 char *eol;
 509         } *prev = NULL, *pcl;
 510         unsigned last_hit = 0;
 511         unsigned last_shown = 0;
 512         int binary_match_only = 0;
 513         const char *hunk_mark = "";
 514         unsigned count = 0;
 515
 516         if (buffer_is_binary(buf, size)) {
 517                 switch (opt->binary) {
 518                 case GREP_BINARY_DEFAULT:
 519                         binary_match_only = 1;
 520                         break;
 521                 case GREP_BINARY_NOMATCH:
 522                         return 0; /* Assume unmatch */
 523                         break;
 524                 default:
 525                         break;
 526                 }
 527         }
 528
 529         if (opt->pre_context)
 530                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 531         if (opt->pre_context || opt->post_context)
 532                 hunk_mark = "--\n";
 533
 534         while (left) {
 535                 char *eol, ch;
 536                 int hit = 0;
 537
 538                 eol = end_of_line(bol, &left);
 539                 ch = *eol;
 540                 *eol = 0;
 541
 542                 hit = match_line(opt, bol, eol);
 543
 544                 /* "grep -v -e foo -e bla" should list lines
 545                  * that do not have either, so inversion should
 546                  * be done outside.
 547                  */
 548                 if (opt->invert)
 549                         hit = !hit;
 550                 if (opt->unmatch_name_only) {
 551                         if (hit)
 552                                 return 0;
 553                         goto next_line;
 554                 }
 555                 if (hit) {
 556                         count++;
 557                         if (binary_match_only) {
 558                                 printf("Binary file %s matches\n", name);
 559                                 return 1;
 560                         }
 561                         if (opt->name_only) {
 562                                 printf("%s\n", name);
 563                                 return 1;
 564                         }
 565                         /* Hit at this line.  If we haven't shown the
 566                          * pre-context lines, we would need to show them.
 567                          * When asked to do "count", this still show
 568                          * the context which is nonsense, but the user
 569                          * deserves to get that ;-).
 570                          */
 571                         if (opt->pre_context) {
 572                                 unsigned from;
 573                                 if (opt->pre_context < lno)
 574                                         from = lno - opt->pre_context;
 575                                 else
 576                                         from = 1;
 577                                 if (from <= last_shown)
 578                                         from = last_shown + 1;
 579                                 if (last_shown && from != last_shown + 1)
 580                                         printf(hunk_mark);
 581                                 while (from < lno) {
 582                                         pcl = &prev[lno-from-1];
 583                                         show_line(opt, pcl->bol, pcl->eol,
 584                                                   name, from, '-');
 585                                         from++;
 586                                 }
 587                                 last_shown = lno-1;
 588                         }
 589                         if (last_shown && lno != last_shown + 1)
 590                                 printf(hunk_mark);
 591                         if (!opt->count)
 592                                 show_line(opt, bol, eol, name, lno, ':');
 593                         last_shown = last_hit = lno;
 594                 }
 595                 else if (last_hit &&
 596                          lno <= last_hit + opt->post_context) {
 597                         /* If the last hit is within the post context,
 598                          * we need to show this line.
 599                          */
 600                         if (last_shown && lno != last_shown + 1)
 601                                 printf(hunk_mark);
 602                         show_line(opt, bol, eol, name, lno, '-');
 603                         last_shown = lno;
 604                 }
 605                 if (opt->pre_context) {
 606                         memmove(prev+1, prev,
 607                                 (opt->pre_context-1) * sizeof(*prev));
 608                         prev->bol = bol;
 609                         prev->eol = eol;
 610                 }
 611
 612         next_line:
 613                 *eol = ch;
 614                 bol = eol + 1;
 615                 if (!left)
 616                         break;
 617                 left--;
 618                 lno++;
 619         }
 620
 621         if (opt->unmatch_name_only) {
 622                 /* We did not see any hit, so we want to show this */
 623                 printf("%s\n", name);
 624                 return 1;
 625         }
 626
 627         /* NEEDSWORK:
 628          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 629          * which feels mostly useless but sometimes useful.  Maybe
 630          * make it another option?  For now suppress them.
 631          */
 632         if (opt->count && count)
 633                 printf("%s:%u\n", name, count);
 634         return !!last_hit;
 635 }
 636
 637 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
 638 {
 639         unsigned long size;
 640         char *data;
 641         char type[20];
 642         char *to_free = NULL;
 643         int hit;
 644
 645         data = read_sha1_file(sha1, type, &size);
 646         if (!data) {
 647                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 648                 return 0;
 649         }
 650         if (opt->relative && opt->prefix_length) {
 651                 static char name_buf[PATH_MAX];
 652                 char *cp;
 653                 int name_len = strlen(name) - opt->prefix_length + 1;
 654
 655                 if (!tree_name_len)
 656                         name += opt->prefix_length;
 657                 else {
 658                         if (ARRAY_SIZE(name_buf) <= name_len)
 659                                 cp = to_free = xmalloc(name_len);
 660                         else
 661                                 cp = name_buf;
 662                         memcpy(cp, name, tree_name_len);
 663                         strcpy(cp + tree_name_len,
 664                                name + tree_name_len + opt->prefix_length);
 665                         name = cp;
 666                 }
 667         }
 668         hit = grep_buffer(opt, name, data, size);
 669         free(data);
 670         free(to_free);
 671         return hit;
 672 }
 673
 674 static int grep_file(struct grep_opt *opt, const char *filename)
 675 {
 676         struct stat st;
 677         int i;
 678         char *data;
 679         if (lstat(filename, &st) < 0) {
 680         err_ret:
 681                 if (errno != ENOENT)
 682                         error("'%s': %s", filename, strerror(errno));
 683                 return 0;
 684         }
 685         if (!st.st_size)
 686                 return 0; /* empty file -- no grep hit */
 687         if (!S_ISREG(st.st_mode))
 688                 return 0;
 689         i = open(filename, O_RDONLY);
 690         if (i < 0)
 691                 goto err_ret;
 692         data = xmalloc(st.st_size + 1);
 693         if (st.st_size != xread(i, data, st.st_size)) {
 694                 error("'%s': short read %s", filename, strerror(errno));
 695                 close(i);
 696                 free(data);
 697                 return 0;
 698         }
 699         close(i);
 700         if (opt->relative && opt->prefix_length)
 701                 filename += opt->prefix_length;
 702         i = grep_buffer(opt, filename, data, st.st_size);
 703         free(data);
 704         return i;
 705 }
 706
 707 static int exec_grep(int argc, const char **argv)
 708 {
 709         pid_t pid;
 710         int status;
 711
 712         argv[argc] = NULL;
 713         pid = fork();
 714         if (pid < 0)
 715                 return pid;
 716         if (!pid) {
 717                 execvp("grep", (char **) argv);
 718                 exit(255);
 719         }
 720         while (waitpid(pid, &status, 0) < 0) {
 721                 if (errno == EINTR)
 722                         continue;
 723                 return -1;
 724         }
 725         if (WIFEXITED(status)) {
 726                 if (!WEXITSTATUS(status))
 727                         return 1;
 728                 return 0;
 729         }
 730         return -1;
 731 }
 732
 733 #define MAXARGS 1000
 734 #define ARGBUF 4096
 735 #define push_arg(a) do { \
 736         if (nr < MAXARGS) argv[nr++] = (a); \
 737         else die("maximum number of args exceeded"); \
 738         } while (0)
 739
 740 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
 741 {
 742         int i, nr, argc, hit, len, status;
 743         const char *argv[MAXARGS+1];
 744         char randarg[ARGBUF];
 745         char *argptr = randarg;
 746         struct grep_pat *p;
 747
 748         if (opt->extended || (opt->relative && opt->prefix_length))
 749                 return -1;
 750         len = nr = 0;
 751         push_arg("grep");
 752         if (opt->fixed)
 753                 push_arg("-F");
 754         if (opt->linenum)
 755                 push_arg("-n");
 756         if (opt->regflags & REG_EXTENDED)
 757                 push_arg("-E");
 758         if (opt->regflags & REG_ICASE)
 759                 push_arg("-i");
 760         if (opt->word_regexp)
 761                 push_arg("-w");
 762         if (opt->name_only)
 763                 push_arg("-l");
 764         if (opt->unmatch_name_only)
 765                 push_arg("-L");
 766         if (opt->count)
 767                 push_arg("-c");
 768         if (opt->post_context || opt->pre_context) {
 769                 if (opt->post_context != opt->pre_context) {
 770                         if (opt->pre_context) {
 771                                 push_arg("-B");
 772                                 len += snprintf(argptr, sizeof(randarg)-len,
 773                                                 "%u", opt->pre_context);
 774                                 if (sizeof(randarg) <= len)
 775                                         die("maximum length of args exceeded");
 776                                 push_arg(argptr);
 777                                 argptr += len;
 778                         }
 779                         if (opt->post_context) {
 780                                 push_arg("-A");
 781                                 len += snprintf(argptr, sizeof(randarg)-len,
 782                                                 "%u", opt->post_context);
 783                                 if (sizeof(randarg) <= len)
 784                                         die("maximum length of args exceeded");
 785                                 push_arg(argptr);
 786                                 argptr += len;
 787                         }
 788                 }
 789                 else {
 790                         push_arg("-C");
 791                         len += snprintf(argptr, sizeof(randarg)-len,
 792                                         "%u", opt->post_context);
 793                         if (sizeof(randarg) <= len)
 794                                 die("maximum length of args exceeded");
 795                         push_arg(argptr);
 796                         argptr += len;
 797                 }
 798         }
 799         for (p = opt->pattern_list; p; p = p->next) {
 800                 push_arg("-e");
 801                 push_arg(p->pattern);
 802         }
 803
 804         /*
 805          * To make sure we get the header printed out when we want it,
 806          * add /dev/null to the paths to grep.  This is unnecessary
 807          * (and wrong) with "-l" or "-L", which always print out the
 808          * name anyway.
 809          *
 810          * GNU grep has "-H", but this is portable.
 811          */
 812         if (!opt->name_only && !opt->unmatch_name_only)
 813                 push_arg("/dev/null");
 814
 815         hit = 0;
 816         argc = nr;
 817         for (i = 0; i < active_nr; i++) {
 818                 struct cache_entry *ce = active_cache[i];
 819                 char *name;
 820                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 821                         continue;
 822                 if (!pathspec_matches(paths, ce->name))
 823                         continue;
 824                 name = ce->name;
 825                 if (name[0] == '-') {
 826                         int len = ce_namelen(ce);
 827                         name = xmalloc(len + 3);
 828                         memcpy(name, "./", 2);
 829                         memcpy(name + 2, ce->name, len + 1);
 830                 }
 831                 argv[argc++] = name;
 832                 if (argc < MAXARGS)
 833                         continue;
 834                 status = exec_grep(argc, argv);
 835                 if (0 < status)
 836                         hit = 1;
 837                 argc = nr;
 838         }
 839         if (argc > nr) {
 840                 status = exec_grep(argc, argv);
 841                 if (0 < status)
 842                         hit = 1;
 843         }
 844         return hit;
 845 }
 846
 847 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 848 {
 849         int hit = 0;
 850         int nr;
 851         read_cache();
 852
 853 #ifdef __unix__
 854         /*
 855          * Use the external "grep" command for the case where
 856          * we grep through the checked-out files. It tends to
 857          * be a lot more optimized
 858          */
 859         if (!cached) {
 860                 hit = external_grep(opt, paths, cached);
 861                 if (hit >= 0)
 862                         return hit;
 863         }
 864 #endif
 865
 866         for (nr = 0; nr < active_nr; nr++) {
 867                 struct cache_entry *ce = active_cache[nr];
 868                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 869                         continue;
 870                 if (!pathspec_matches(paths, ce->name))
 871                         continue;
 872                 if (cached)
 873                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 874                 else
 875                         hit |= grep_file(opt, ce->name);
 876         }
 877         return hit;
 878 }
 879
 880 static int grep_tree(struct grep_opt *opt, const char **paths,
 881                      struct tree_desc *tree,
 882                      const char *tree_name, const char *base)
 883 {
 884         int len;
 885         int hit = 0;
 886         struct name_entry entry;
 887         char *down;
 888         int tn_len = strlen(tree_name);
 889         char *path_buf = xmalloc(PATH_MAX + tn_len + 100);
 890
 891         if (tn_len) {
 892                 tn_len = sprintf(path_buf, "%s:", tree_name);
 893                 down = path_buf + tn_len;
 894                 strcat(down, base);
 895         }
 896         else {
 897                 down = path_buf;
 898                 strcpy(down, base);
 899         }
 900         len = strlen(path_buf);
 901
 902         while (tree_entry(tree, &entry)) {
 903                 strcpy(path_buf + len, entry.path);
 904
 905                 if (S_ISDIR(entry.mode))
 906                         /* Match "abc/" against pathspec to
 907                          * decide if we want to descend into "abc"
 908                          * directory.
 909                          */
 910                         strcpy(path_buf + len + entry.pathlen, "/");
 911
 912                 if (!pathspec_matches(paths, down))
 913                         ;
 914                 else if (S_ISREG(entry.mode))
 915                         hit |= grep_sha1(opt, entry.sha1, path_buf, tn_len);
 916                 else if (S_ISDIR(entry.mode)) {
 917                         char type[20];
 918                         struct tree_desc sub;
 919                         void *data;
 920                         data = read_sha1_file(entry.sha1, type, &sub.size);
 921                         if (!data)
 922                                 die("unable to read tree (%s)",
 923                                     sha1_to_hex(entry.sha1));
 924                         sub.buf = data;
 925                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 926                         free(data);
 927                 }
 928         }
 929         return hit;
 930 }
 931
 932 static int grep_object(struct grep_opt *opt, const char **paths,
 933                        struct object *obj, const char *name)
 934 {
 935         if (obj->type == OBJ_BLOB)
 936                 return grep_sha1(opt, obj->sha1, name, 0);
 937         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 938                 struct tree_desc tree;
 939                 void *data;
 940                 int hit;
 941                 data = read_object_with_reference(obj->sha1, tree_type,
 942                                                   &tree.size, NULL);
 943                 if (!data)
 944                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 945                 tree.buf = data;
 946                 hit = grep_tree(opt, paths, &tree, name, "");
 947                 free(data);
 948                 return hit;
 949         }
 950         die("unable to grep from object of type %s", typename(obj->type));
 951 }
 952
 953 static const char builtin_grep_usage[] =
 954 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 955
 956 static const char emsg_invalid_context_len[] =
 957 "%s: invalid context length argument";
 958 static const char emsg_missing_context_len[] =
 959 "missing context length argument";
 960 static const char emsg_missing_argument[] =
 961 "option requires an argument -%s";
 962
 963 int cmd_grep(int argc, const char **argv, const char *prefix)
 964 {
 965         int hit = 0;
 966         int cached = 0;
 967         int seen_dashdash = 0;
 968         struct grep_opt opt;
 969         struct object_array list = { 0, 0, NULL };
 970         const char **paths = NULL;
 971         int i;
 972
 973         memset(&opt, 0, sizeof(opt));
 974         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 975         opt.relative = 1;
 976         opt.pattern_tail = &opt.pattern_list;
 977         opt.regflags = REG_NEWLINE;
 978
 979         /*
 980          * If there is no -- then the paths must exist in the working
 981          * tree.  If there is no explicit pattern specified with -e or
 982          * -f, we take the first unrecognized non option to be the
 983          * pattern, but then what follows it must be zero or more
 984          * valid refs up to the -- (if exists), and then existing
 985          * paths.  If there is an explicit pattern, then the first
 986          * unrecognized non option is the beginning of the refs list
 987          * that continues up to the -- (if exists), and then paths.
 988          */
 989
 990         while (1 < argc) {
 991                 const char *arg = argv[1];
 992                 argc--; argv++;
 993                 if (!strcmp("--cached", arg)) {
 994                         cached = 1;
 995                         continue;
 996                 }
 997                 if (!strcmp("-a", arg) ||
 998                     !strcmp("--text", arg)) {
 999                         opt.binary = GREP_BINARY_TEXT;
1000                         continue;
1001                 }
1002                 if (!strcmp("-i", arg) ||
1003                     !strcmp("--ignore-case", arg)) {
1004                         opt.regflags |= REG_ICASE;
1005                         continue;
1006                 }
1007                 if (!strcmp("-I", arg)) {
1008                         opt.binary = GREP_BINARY_NOMATCH;
1009                         continue;
1010                 }
1011                 if (!strcmp("-v", arg) ||
1012                     !strcmp("--invert-match", arg)) {
1013                         opt.invert = 1;
1014                         continue;
1015                 }
1016                 if (!strcmp("-E", arg) ||
1017                     !strcmp("--extended-regexp", arg)) {
1018                         opt.regflags |= REG_EXTENDED;
1019                         continue;
1020                 }
1021                 if (!strcmp("-F", arg) ||
1022                     !strcmp("--fixed-strings", arg)) {
1023                         opt.fixed = 1;
1024                         continue;
1025                 }
1026                 if (!strcmp("-G", arg) ||
1027                     !strcmp("--basic-regexp", arg)) {
1028                         opt.regflags &= ~REG_EXTENDED;
1029                         continue;
1030                 }
1031                 if (!strcmp("-n", arg)) {
1032                         opt.linenum = 1;
1033                         continue;
1034                 }
1035                 if (!strcmp("-H", arg)) {
1036                         /* We always show the pathname, so this
1037                          * is a noop.
1038                          */
1039                         continue;
1040                 }
1041                 if (!strcmp("-l", arg) ||
1042                     !strcmp("--files-with-matches", arg)) {
1043                         opt.name_only = 1;
1044                         continue;
1045                 }
1046                 if (!strcmp("-L", arg) ||
1047                     !strcmp("--files-without-match", arg)) {
1048                         opt.unmatch_name_only = 1;
1049                         continue;
1050                 }
1051                 if (!strcmp("-c", arg) ||
1052                     !strcmp("--count", arg)) {
1053                         opt.count = 1;
1054                         continue;
1055                 }
1056                 if (!strcmp("-w", arg) ||
1057                     !strcmp("--word-regexp", arg)) {
1058                         opt.word_regexp = 1;
1059                         continue;
1060                 }
1061                 if (!strncmp("-A", arg, 2) ||
1062                     !strncmp("-B", arg, 2) ||
1063                     !strncmp("-C", arg, 2) ||
1064                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
1065                         unsigned num;
1066                         const char *scan;
1067                         switch (arg[1]) {
1068                         case 'A': case 'B': case 'C':
1069                                 if (!arg[2]) {
1070                                         if (argc <= 1)
1071                                                 die(emsg_missing_context_len);
1072                                         scan = *++argv;
1073                                         argc--;
1074                                 }
1075                                 else
1076                                         scan = arg + 2;
1077                                 break;
1078                         default:
1079                                 scan = arg + 1;
1080                                 break;
1081                         }
1082                         if (sscanf(scan, "%u", &num) != 1)
1083                                 die(emsg_invalid_context_len, scan);
1084                         switch (arg[1]) {
1085                         case 'A':
1086                                 opt.post_context = num;
1087                                 break;
1088                         default:
1089                         case 'C':
1090                                 opt.post_context = num;
1091                         case 'B':
1092                                 opt.pre_context = num;
1093                                 break;
1094                         }
1095                         continue;
1096                 }
1097                 if (!strcmp("-f", arg)) {
1098                         FILE *patterns;
1099                         int lno = 0;
1100                         char buf[1024];
1101                         if (argc <= 1)
1102                                 die(emsg_missing_argument, arg);
1103                         patterns = fopen(argv[1], "r");
1104                         if (!patterns)
1105                                 die("'%s': %s", argv[1], strerror(errno));
1106                         while (fgets(buf, sizeof(buf), patterns)) {
1107                                 int len = strlen(buf);
1108                                 if (buf[len-1] == '\n')
1109                                         buf[len-1] = 0;
1110                                 /* ignore empty line like grep does */
1111                                 if (!buf[0])
1112                                         continue;
1113                                 add_pattern(&opt, strdup(buf), argv[1], ++lno,
1114                                             GREP_PATTERN);
1115                         }
1116                         fclose(patterns);
1117                         argv++;
1118                         argc--;
1119                         continue;
1120                 }
1121                 if (!strcmp("--not", arg)) {
1122                         add_pattern(&opt, arg, "command line", 0, GREP_NOT);
1123                         continue;
1124                 }
1125                 if (!strcmp("--and", arg)) {
1126                         add_pattern(&opt, arg, "command line", 0, GREP_AND);
1127                         continue;
1128                 }
1129                 if (!strcmp("--or", arg))
1130                         continue; /* no-op */
1131                 if (!strcmp("(", arg)) {
1132                         add_pattern(&opt, arg, "command line", 0, GREP_OPEN_PAREN);
1133                         continue;
1134                 }
1135                 if (!strcmp(")", arg)) {
1136                         add_pattern(&opt, arg, "command line", 0, GREP_CLOSE_PAREN);
1137                         continue;
1138                 }
1139                 if (!strcmp("-e", arg)) {
1140                         if (1 < argc) {
1141                                 add_pattern(&opt, argv[1], "-e option", 0,
1142                                             GREP_PATTERN);
1143                                 argv++;
1144                                 argc--;
1145                                 continue;
1146                         }
1147                         die(emsg_missing_argument, arg);
1148                 }
1149                 if (!strcmp("--full-name", arg)) {
1150                         opt.relative = 0;
1151                         continue;
1152                 }
1153                 if (!strcmp("--", arg)) {
1154                         /* later processing wants to have this at argv[1] */
1155                         argv--;
1156                         argc++;
1157                         break;
1158                 }
1159                 if (*arg == '-')
1160                         usage(builtin_grep_usage);
1161
1162                 /* First unrecognized non-option token */
1163                 if (!opt.pattern_list) {
1164                         add_pattern(&opt, arg, "command line", 0,
1165                                     GREP_PATTERN);
1166                         break;
1167                 }
1168                 else {
1169                         /* We are looking at the first path or rev;
1170                          * it is found at argv[1] after leaving the
1171                          * loop.
1172                          */
1173                         argc++; argv--;
1174                         break;
1175                 }
1176         }
1177
1178         if (!opt.pattern_list)
1179                 die("no pattern given.");
1180         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
1181                 die("cannot mix --fixed-strings and regexp");
1182         if (!opt.fixed)
1183                 compile_patterns(&opt);
1184
1185         /* Check revs and then paths */
1186         for (i = 1; i < argc; i++) {
1187                 const char *arg = argv[i];
1188                 unsigned char sha1[20];
1189                 /* Is it a rev? */
1190                 if (!get_sha1(arg, sha1)) {
1191                         struct object *object = parse_object(sha1);
1192                         if (!object)
1193                                 die("bad object %s", arg);
1194                         add_object_array(object, arg, &list);
1195                         continue;
1196                 }
1197                 if (!strcmp(arg, "--")) {
1198                         i++;
1199                         seen_dashdash = 1;
1200                 }
1201                 break;
1202         }
1203
1204         /* The rest are paths */
1205         if (!seen_dashdash) {
1206                 int j;
1207                 for (j = i; j < argc; j++)
1208                         verify_filename(prefix, argv[j]);
1209         }
1210
1211         if (i < argc) {
1212                 paths = get_pathspec(prefix, argv + i);
1213                 if (opt.prefix_length && opt.relative) {
1214                         /* Make sure we do not get outside of paths */
1215                         for (i = 0; paths[i]; i++)
1216                                 if (strncmp(prefix, paths[i], opt.prefix_length))
1217                                         die("git-grep: cannot generate relative filenames containing '..'");
1218                 }
1219         }
1220         else if (prefix) {
1221                 paths = xcalloc(2, sizeof(const char *));
1222                 paths[0] = prefix;
1223                 paths[1] = NULL;
1224         }
1225
1226         if (!list.nr)
1227                 return !grep_cache(&opt, paths, cached);
1228
1229         if (cached)
1230                 die("both --cached and trees are given.");
1231
1232         for (i = 0; i < list.nr; i++) {
1233                 struct object *real_obj;
1234                 real_obj = deref_tag(list.objects[i].item, NULL, 0);
1235                 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
1236                         hit = 1;
1237         }
1238         return !hit;
1239 }