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