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         unsigned pathname:1;
 142         int regflags;
 143         unsigned pre_context;
 144         unsigned post_context;
 145 };
 146
 147 static void add_pattern(struct grep_opt *opt, const char *pat,
 148                         const char *origin, int no, enum grep_pat_token t)
 149 {
 150         struct grep_pat *p = xcalloc(1, sizeof(*p));
 151         p->pattern = pat;
 152         p->origin = origin;
 153         p->no = no;
 154         p->token = t;
 155         *opt->pattern_tail = p;
 156         opt->pattern_tail = &p->next;
 157         p->next = NULL;
 158 }
 159
 160 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 161 {
 162         int err = regcomp(&p->regexp, p->pattern, opt->regflags);
 163         if (err) {
 164                 char errbuf[1024];
 165                 char where[1024];
 166                 if (p->no)
 167                         sprintf(where, "In '%s' at %d, ",
 168                                 p->origin, p->no);
 169                 else if (p->origin)
 170                         sprintf(where, "%s, ", p->origin);
 171                 else
 172                         where[0] = 0;
 173                 regerror(err, &p->regexp, errbuf, 1024);
 174                 regfree(&p->regexp);
 175                 die("%s'%s': %s", where, p->pattern, errbuf);
 176         }
 177 }
 178
 179 static struct grep_expr *compile_pattern_expr(struct grep_pat **);
 180 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 181 {
 182         struct grep_pat *p;
 183         struct grep_expr *x;
 184
 185         p = *list;
 186         switch (p->token) {
 187         case GREP_PATTERN: /* atom */
 188                 x = xcalloc(1, sizeof (struct grep_expr));
 189                 x->node = GREP_NODE_ATOM;
 190                 x->u.atom = p;
 191                 *list = p->next;
 192                 return x;
 193         case GREP_OPEN_PAREN:
 194                 *list = p->next;
 195                 x = compile_pattern_expr(list);
 196                 if (!x)
 197                         return NULL;
 198                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 199                         die("unmatched parenthesis");
 200                 *list = (*list)->next;
 201                 return x;
 202         default:
 203                 return NULL;
 204         }
 205 }
 206
 207 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 208 {
 209         struct grep_pat *p;
 210         struct grep_expr *x;
 211
 212         p = *list;
 213         switch (p->token) {
 214         case GREP_NOT:
 215                 if (!p->next)
 216                         die("--not not followed by pattern expression");
 217                 *list = p->next;
 218                 x = xcalloc(1, sizeof (struct grep_expr));
 219                 x->node = GREP_NODE_NOT;
 220                 x->u.unary = compile_pattern_not(list);
 221                 if (!x->u.unary)
 222                         die("--not followed by non pattern expression");
 223                 return x;
 224         default:
 225                 return compile_pattern_atom(list);
 226         }
 227 }
 228
 229 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 230 {
 231         struct grep_pat *p;
 232         struct grep_expr *x, *y, *z;
 233
 234         x = compile_pattern_not(list);
 235         p = *list;
 236         if (p && p->token == GREP_AND) {
 237                 if (!p->next)
 238                         die("--and not followed by pattern expression");
 239                 *list = p->next;
 240                 y = compile_pattern_and(list);
 241                 if (!y)
 242                         die("--and not followed by pattern expression");
 243                 z = xcalloc(1, sizeof (struct grep_expr));
 244                 z->node = GREP_NODE_AND;
 245                 z->u.binary.left = x;
 246                 z->u.binary.right = y;
 247                 return z;
 248         }
 249         return x;
 250 }
 251
 252 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 253 {
 254         struct grep_pat *p;
 255         struct grep_expr *x, *y, *z;
 256
 257         x = compile_pattern_and(list);
 258         p = *list;
 259         if (x && p && p->token != GREP_CLOSE_PAREN) {
 260                 y = compile_pattern_or(list);
 261                 if (!y)
 262                         die("not a pattern expression %s", p->pattern);
 263                 z = xcalloc(1, sizeof (struct grep_expr));
 264                 z->node = GREP_NODE_OR;
 265                 z->u.binary.left = x;
 266                 z->u.binary.right = y;
 267                 return z;
 268         }
 269         return x;
 270 }
 271
 272 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 273 {
 274         return compile_pattern_or(list);
 275 }
 276
 277 static void compile_patterns(struct grep_opt *opt)
 278 {
 279         struct grep_pat *p;
 280
 281         /* First compile regexps */
 282         for (p = opt->pattern_list; p; p = p->next) {
 283                 if (p->token == GREP_PATTERN)
 284                         compile_regexp(p, opt);
 285                 else
 286                         opt->extended = 1;
 287         }
 288
 289         if (!opt->extended)
 290                 return;
 291
 292         /* Then bundle them up in an expression.
 293          * A classic recursive descent parser would do.
 294          */
 295         p = opt->pattern_list;
 296         opt->pattern_expression = compile_pattern_expr(&p);
 297         if (p)
 298                 die("incomplete pattern expression: %s", p->pattern);
 299 }
 300
 301 static char *end_of_line(char *cp, unsigned long *left)
 302 {
 303         unsigned long l = *left;
 304         while (l && *cp != '\n') {
 305                 l--;
 306                 cp++;
 307         }
 308         *left = l;
 309         return cp;
 310 }
 311
 312 static int word_char(char ch)
 313 {
 314         return isalnum(ch) || ch == '_';
 315 }
 316
 317 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 318                       const char *name, unsigned lno, char sign)
 319 {
 320         if (opt->pathname)
 321                 printf("%s%c", name, sign);
 322         if (opt->linenum)
 323                 printf("%d%c", lno, sign);
 324         printf("%.*s\n", (int)(eol-bol), bol);
 325 }
 326
 327 /*
 328  * NEEDSWORK: share code with diff.c
 329  */
 330 #define FIRST_FEW_BYTES 8000
 331 static int buffer_is_binary(const char *ptr, unsigned long size)
 332 {
 333         if (FIRST_FEW_BYTES < size)
 334                 size = FIRST_FEW_BYTES;
 335         return !!memchr(ptr, 0, size);
 336 }
 337
 338 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 339 {
 340         char *hit = strstr(line, pattern);
 341         if (!hit) {
 342                 match->rm_so = match->rm_eo = -1;
 343                 return REG_NOMATCH;
 344         }
 345         else {
 346                 match->rm_so = hit - line;
 347                 match->rm_eo = match->rm_so + strlen(pattern);
 348                 return 0;
 349         }
 350 }
 351
 352 static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol, char *eol)
 353 {
 354         int hit = 0;
 355         int at_true_bol = 1;
 356         regmatch_t pmatch[10];
 357
 358  again:
 359         if (!opt->fixed) {
 360                 regex_t *exp = &p->regexp;
 361                 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 362                                pmatch, 0);
 363         }
 364         else {
 365                 hit = !fixmatch(p->pattern, bol, pmatch);
 366         }
 367
 368         if (hit && opt->word_regexp) {
 369                 if ((pmatch[0].rm_so < 0) ||
 370                     (eol - bol) <= pmatch[0].rm_so ||
 371                     (pmatch[0].rm_eo < 0) ||
 372                     (eol - bol) < pmatch[0].rm_eo)
 373                         die("regexp returned nonsense");
 374
 375                 /* Match beginning must be either beginning of the
 376                  * line, or at word boundary (i.e. the last char must
 377                  * not be a word char).  Similarly, match end must be
 378                  * either end of the line, or at word boundary
 379                  * (i.e. the next char must not be a word char).
 380                  */
 381                 if ( ((pmatch[0].rm_so == 0 && at_true_bol) ||
 382                       !word_char(bol[pmatch[0].rm_so-1])) &&
 383                      ((pmatch[0].rm_eo == (eol-bol)) ||
 384                       !word_char(bol[pmatch[0].rm_eo])) )
 385                         ;
 386                 else
 387                         hit = 0;
 388
 389                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 390                         /* There could be more than one match on the
 391                          * line, and the first match might not be
 392                          * strict word match.  But later ones could be!
 393                          */
 394                         bol = pmatch[0].rm_so + bol + 1;
 395                         at_true_bol = 0;
 396                         goto again;
 397                 }
 398         }
 399         return hit;
 400 }
 401
 402 static int match_expr_eval(struct grep_opt *opt,
 403                            struct grep_expr *x,
 404                            char *bol, char *eol)
 405 {
 406         switch (x->node) {
 407         case GREP_NODE_ATOM:
 408                 return match_one_pattern(opt, x->u.atom, bol, eol);
 409                 break;
 410         case GREP_NODE_NOT:
 411                 return !match_expr_eval(opt, x->u.unary, bol, eol);
 412         case GREP_NODE_AND:
 413                 return (match_expr_eval(opt, x->u.binary.left, bol, eol) &&
 414                         match_expr_eval(opt, x->u.binary.right, bol, eol));
 415         case GREP_NODE_OR:
 416                 return (match_expr_eval(opt, x->u.binary.left, bol, eol) ||
 417                         match_expr_eval(opt, x->u.binary.right, bol, eol));
 418         }
 419         die("Unexpected node type (internal error) %d\n", x->node);
 420 }
 421
 422 static int match_expr(struct grep_opt *opt, char *bol, char *eol)
 423 {
 424         struct grep_expr *x = opt->pattern_expression;
 425         return match_expr_eval(opt, x, bol, eol);
 426 }
 427
 428 static int match_line(struct grep_opt *opt, char *bol, char *eol)
 429 {
 430         struct grep_pat *p;
 431         if (opt->extended)
 432                 return match_expr(opt, bol, eol);
 433         for (p = opt->pattern_list; p; p = p->next) {
 434                 if (match_one_pattern(opt, p, bol, eol))
 435                         return 1;
 436         }
 437         return 0;
 438 }
 439
 440 static int grep_buffer(struct grep_opt *opt, const char *name,
 441                        char *buf, unsigned long size)
 442 {
 443         char *bol = buf;
 444         unsigned long left = size;
 445         unsigned lno = 1;
 446         struct pre_context_line {
 447                 char *bol;
 448                 char *eol;
 449         } *prev = NULL, *pcl;
 450         unsigned last_hit = 0;
 451         unsigned last_shown = 0;
 452         int binary_match_only = 0;
 453         const char *hunk_mark = "";
 454         unsigned count = 0;
 455
 456         if (buffer_is_binary(buf, size)) {
 457                 switch (opt->binary) {
 458                 case GREP_BINARY_DEFAULT:
 459                         binary_match_only = 1;
 460                         break;
 461                 case GREP_BINARY_NOMATCH:
 462                         return 0; /* Assume unmatch */
 463                         break;
 464                 default:
 465                         break;
 466                 }
 467         }
 468
 469         if (opt->pre_context)
 470                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 471         if (opt->pre_context || opt->post_context)
 472                 hunk_mark = "--\n";
 473
 474         while (left) {
 475                 char *eol, ch;
 476                 int hit = 0;
 477
 478                 eol = end_of_line(bol, &left);
 479                 ch = *eol;
 480                 *eol = 0;
 481
 482                 hit = match_line(opt, bol, eol);
 483
 484                 /* "grep -v -e foo -e bla" should list lines
 485                  * that do not have either, so inversion should
 486                  * be done outside.
 487                  */
 488                 if (opt->invert)
 489                         hit = !hit;
 490                 if (opt->unmatch_name_only) {
 491                         if (hit)
 492                                 return 0;
 493                         goto next_line;
 494                 }
 495                 if (hit) {
 496                         count++;
 497                         if (binary_match_only) {
 498                                 printf("Binary file %s matches\n", name);
 499                                 return 1;
 500                         }
 501                         if (opt->name_only) {
 502                                 printf("%s\n", name);
 503                                 return 1;
 504                         }
 505                         /* Hit at this line.  If we haven't shown the
 506                          * pre-context lines, we would need to show them.
 507                          * When asked to do "count", this still show
 508                          * the context which is nonsense, but the user
 509                          * deserves to get that ;-).
 510                          */
 511                         if (opt->pre_context) {
 512                                 unsigned from;
 513                                 if (opt->pre_context < lno)
 514                                         from = lno - opt->pre_context;
 515                                 else
 516                                         from = 1;
 517                                 if (from <= last_shown)
 518                                         from = last_shown + 1;
 519                                 if (last_shown && from != last_shown + 1)
 520                                         printf(hunk_mark);
 521                                 while (from < lno) {
 522                                         pcl = &prev[lno-from-1];
 523                                         show_line(opt, pcl->bol, pcl->eol,
 524                                                   name, from, '-');
 525                                         from++;
 526                                 }
 527                                 last_shown = lno-1;
 528                         }
 529                         if (last_shown && lno != last_shown + 1)
 530                                 printf(hunk_mark);
 531                         if (!opt->count)
 532                                 show_line(opt, bol, eol, name, lno, ':');
 533                         last_shown = last_hit = lno;
 534                 }
 535                 else if (last_hit &&
 536                          lno <= last_hit + opt->post_context) {
 537                         /* If the last hit is within the post context,
 538                          * we need to show this line.
 539                          */
 540                         if (last_shown && lno != last_shown + 1)
 541                                 printf(hunk_mark);
 542                         show_line(opt, bol, eol, name, lno, '-');
 543                         last_shown = lno;
 544                 }
 545                 if (opt->pre_context) {
 546                         memmove(prev+1, prev,
 547                                 (opt->pre_context-1) * sizeof(*prev));
 548                         prev->bol = bol;
 549                         prev->eol = eol;
 550                 }
 551
 552         next_line:
 553                 *eol = ch;
 554                 bol = eol + 1;
 555                 if (!left)
 556                         break;
 557                 left--;
 558                 lno++;
 559         }
 560
 561         if (opt->unmatch_name_only) {
 562                 /* We did not see any hit, so we want to show this */
 563                 printf("%s\n", name);
 564                 return 1;
 565         }
 566
 567         /* NEEDSWORK:
 568          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 569          * which feels mostly useless but sometimes useful.  Maybe
 570          * make it another option?  For now suppress them.
 571          */
 572         if (opt->count && count)
 573                 printf("%s:%u\n", name, count);
 574         return !!last_hit;
 575 }
 576
 577 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
 578 {
 579         unsigned long size;
 580         char *data;
 581         char type[20];
 582         char *to_free = NULL;
 583         int hit;
 584
 585         data = read_sha1_file(sha1, type, &size);
 586         if (!data) {
 587                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 588                 return 0;
 589         }
 590         if (opt->relative && opt->prefix_length) {
 591                 static char name_buf[PATH_MAX];
 592                 char *cp;
 593                 int name_len = strlen(name) - opt->prefix_length + 1;
 594
 595                 if (!tree_name_len)
 596                         name += opt->prefix_length;
 597                 else {
 598                         if (ARRAY_SIZE(name_buf) <= name_len)
 599                                 cp = to_free = xmalloc(name_len);
 600                         else
 601                                 cp = name_buf;
 602                         memcpy(cp, name, tree_name_len);
 603                         strcpy(cp + tree_name_len,
 604                                name + tree_name_len + opt->prefix_length);
 605                         name = cp;
 606                 }
 607         }
 608         hit = grep_buffer(opt, name, data, size);
 609         free(data);
 610         free(to_free);
 611         return hit;
 612 }
 613
 614 static int grep_file(struct grep_opt *opt, const char *filename)
 615 {
 616         struct stat st;
 617         int i;
 618         char *data;
 619         if (lstat(filename, &st) < 0) {
 620         err_ret:
 621                 if (errno != ENOENT)
 622                         error("'%s': %s", filename, strerror(errno));
 623                 return 0;
 624         }
 625         if (!st.st_size)
 626                 return 0; /* empty file -- no grep hit */
 627         if (!S_ISREG(st.st_mode))
 628                 return 0;
 629         i = open(filename, O_RDONLY);
 630         if (i < 0)
 631                 goto err_ret;
 632         data = xmalloc(st.st_size + 1);
 633         if (st.st_size != xread(i, data, st.st_size)) {
 634                 error("'%s': short read %s", filename, strerror(errno));
 635                 close(i);
 636                 free(data);
 637                 return 0;
 638         }
 639         close(i);
 640         if (opt->relative && opt->prefix_length)
 641                 filename += opt->prefix_length;
 642         i = grep_buffer(opt, filename, data, st.st_size);
 643         free(data);
 644         return i;
 645 }
 646
 647 static int exec_grep(int argc, const char **argv)
 648 {
 649         pid_t pid;
 650         int status;
 651
 652         argv[argc] = NULL;
 653         pid = fork();
 654         if (pid < 0)
 655                 return pid;
 656         if (!pid) {
 657                 execvp("grep", (char **) argv);
 658                 exit(255);
 659         }
 660         while (waitpid(pid, &status, 0) < 0) {
 661                 if (errno == EINTR)
 662                         continue;
 663                 return -1;
 664         }
 665         if (WIFEXITED(status)) {
 666                 if (!WEXITSTATUS(status))
 667                         return 1;
 668                 return 0;
 669         }
 670         return -1;
 671 }
 672
 673 #define MAXARGS 1000
 674 #define ARGBUF 4096
 675 #define push_arg(a) do { \
 676         if (nr < MAXARGS) argv[nr++] = (a); \
 677         else die("maximum number of args exceeded"); \
 678         } while (0)
 679
 680 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
 681 {
 682         int i, nr, argc, hit, len, status;
 683         const char *argv[MAXARGS+1];
 684         char randarg[ARGBUF];
 685         char *argptr = randarg;
 686         struct grep_pat *p;
 687
 688         if (opt->extended || (opt->relative && opt->prefix_length))
 689                 return -1;
 690         len = nr = 0;
 691         push_arg("grep");
 692         if (opt->fixed)
 693                 push_arg("-F");
 694         if (opt->linenum)
 695                 push_arg("-n");
 696         if (!opt->pathname)
 697                 push_arg("-h");
 698         if (opt->regflags & REG_EXTENDED)
 699                 push_arg("-E");
 700         if (opt->regflags & REG_ICASE)
 701                 push_arg("-i");
 702         if (opt->word_regexp)
 703                 push_arg("-w");
 704         if (opt->name_only)
 705                 push_arg("-l");
 706         if (opt->unmatch_name_only)
 707                 push_arg("-L");
 708         if (opt->count)
 709                 push_arg("-c");
 710         if (opt->post_context || opt->pre_context) {
 711                 if (opt->post_context != opt->pre_context) {
 712                         if (opt->pre_context) {
 713                                 push_arg("-B");
 714                                 len += snprintf(argptr, sizeof(randarg)-len,
 715                                                 "%u", opt->pre_context);
 716                                 if (sizeof(randarg) <= len)
 717                                         die("maximum length of args exceeded");
 718                                 push_arg(argptr);
 719                                 argptr += len;
 720                         }
 721                         if (opt->post_context) {
 722                                 push_arg("-A");
 723                                 len += snprintf(argptr, sizeof(randarg)-len,
 724                                                 "%u", opt->post_context);
 725                                 if (sizeof(randarg) <= len)
 726                                         die("maximum length of args exceeded");
 727                                 push_arg(argptr);
 728                                 argptr += len;
 729                         }
 730                 }
 731                 else {
 732                         push_arg("-C");
 733                         len += snprintf(argptr, sizeof(randarg)-len,
 734                                         "%u", opt->post_context);
 735                         if (sizeof(randarg) <= len)
 736                                 die("maximum length of args exceeded");
 737                         push_arg(argptr);
 738                         argptr += len;
 739                 }
 740         }
 741         for (p = opt->pattern_list; p; p = p->next) {
 742                 push_arg("-e");
 743                 push_arg(p->pattern);
 744         }
 745
 746         /*
 747          * To make sure we get the header printed out when we want it,
 748          * add /dev/null to the paths to grep.  This is unnecessary
 749          * (and wrong) with "-l" or "-L", which always print out the
 750          * name anyway.
 751          *
 752          * GNU grep has "-H", but this is portable.
 753          */
 754         if (!opt->name_only && !opt->unmatch_name_only)
 755                 push_arg("/dev/null");
 756
 757         hit = 0;
 758         argc = nr;
 759         for (i = 0; i < active_nr; i++) {
 760                 struct cache_entry *ce = active_cache[i];
 761                 char *name;
 762                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 763                         continue;
 764                 if (!pathspec_matches(paths, ce->name))
 765                         continue;
 766                 name = ce->name;
 767                 if (name[0] == '-') {
 768                         int len = ce_namelen(ce);
 769                         name = xmalloc(len + 3);
 770                         memcpy(name, "./", 2);
 771                         memcpy(name + 2, ce->name, len + 1);
 772                 }
 773                 argv[argc++] = name;
 774                 if (argc < MAXARGS)
 775                         continue;
 776                 status = exec_grep(argc, argv);
 777                 if (0 < status)
 778                         hit = 1;
 779                 argc = nr;
 780         }
 781         if (argc > nr) {
 782                 status = exec_grep(argc, argv);
 783                 if (0 < status)
 784                         hit = 1;
 785         }
 786         return hit;
 787 }
 788
 789 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 790 {
 791         int hit = 0;
 792         int nr;
 793         read_cache();
 794
 795 #ifdef __unix__
 796         /*
 797          * Use the external "grep" command for the case where
 798          * we grep through the checked-out files. It tends to
 799          * be a lot more optimized
 800          */
 801         if (!cached) {
 802                 hit = external_grep(opt, paths, cached);
 803                 if (hit >= 0)
 804                         return hit;
 805         }
 806 #endif
 807
 808         for (nr = 0; nr < active_nr; nr++) {
 809                 struct cache_entry *ce = active_cache[nr];
 810                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 811                         continue;
 812                 if (!pathspec_matches(paths, ce->name))
 813                         continue;
 814                 if (cached)
 815                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 816                 else
 817                         hit |= grep_file(opt, ce->name);
 818         }
 819         return hit;
 820 }
 821
 822 static int grep_tree(struct grep_opt *opt, const char **paths,
 823                      struct tree_desc *tree,
 824                      const char *tree_name, const char *base)
 825 {
 826         int len;
 827         int hit = 0;
 828         struct name_entry entry;
 829         char *down;
 830         int tn_len = strlen(tree_name);
 831         char *path_buf = xmalloc(PATH_MAX + tn_len + 100);
 832
 833         if (tn_len) {
 834                 tn_len = sprintf(path_buf, "%s:", tree_name);
 835                 down = path_buf + tn_len;
 836                 strcat(down, base);
 837         }
 838         else {
 839                 down = path_buf;
 840                 strcpy(down, base);
 841         }
 842         len = strlen(path_buf);
 843
 844         while (tree_entry(tree, &entry)) {
 845                 strcpy(path_buf + len, entry.path);
 846
 847                 if (S_ISDIR(entry.mode))
 848                         /* Match "abc/" against pathspec to
 849                          * decide if we want to descend into "abc"
 850                          * directory.
 851                          */
 852                         strcpy(path_buf + len + entry.pathlen, "/");
 853
 854                 if (!pathspec_matches(paths, down))
 855                         ;
 856                 else if (S_ISREG(entry.mode))
 857                         hit |= grep_sha1(opt, entry.sha1, path_buf, tn_len);
 858                 else if (S_ISDIR(entry.mode)) {
 859                         char type[20];
 860                         struct tree_desc sub;
 861                         void *data;
 862                         data = read_sha1_file(entry.sha1, type, &sub.size);
 863                         if (!data)
 864                                 die("unable to read tree (%s)",
 865                                     sha1_to_hex(entry.sha1));
 866                         sub.buf = data;
 867                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 868                         free(data);
 869                 }
 870         }
 871         return hit;
 872 }
 873
 874 static int grep_object(struct grep_opt *opt, const char **paths,
 875                        struct object *obj, const char *name)
 876 {
 877         if (obj->type == OBJ_BLOB)
 878                 return grep_sha1(opt, obj->sha1, name, 0);
 879         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 880                 struct tree_desc tree;
 881                 void *data;
 882                 int hit;
 883                 data = read_object_with_reference(obj->sha1, tree_type,
 884                                                   &tree.size, NULL);
 885                 if (!data)
 886                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 887                 tree.buf = data;
 888                 hit = grep_tree(opt, paths, &tree, name, "");
 889                 free(data);
 890                 return hit;
 891         }
 892         die("unable to grep from object of type %s", typename(obj->type));
 893 }
 894
 895 static const char builtin_grep_usage[] =
 896 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 897
 898 static const char emsg_invalid_context_len[] =
 899 "%s: invalid context length argument";
 900 static const char emsg_missing_context_len[] =
 901 "missing context length argument";
 902 static const char emsg_missing_argument[] =
 903 "option requires an argument -%s";
 904
 905 int cmd_grep(int argc, const char **argv, const char *prefix)
 906 {
 907         int hit = 0;
 908         int cached = 0;
 909         int seen_dashdash = 0;
 910         struct grep_opt opt;
 911         struct object_array list = { 0, 0, NULL };
 912         const char **paths = NULL;
 913         int i;
 914
 915         memset(&opt, 0, sizeof(opt));
 916         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 917         opt.relative = 1;
 918         opt.pathname = 1;
 919         opt.pattern_tail = &opt.pattern_list;
 920         opt.regflags = REG_NEWLINE;
 921
 922         /*
 923          * If there is no -- then the paths must exist in the working
 924          * tree.  If there is no explicit pattern specified with -e or
 925          * -f, we take the first unrecognized non option to be the
 926          * pattern, but then what follows it must be zero or more
 927          * valid refs up to the -- (if exists), and then existing
 928          * paths.  If there is an explicit pattern, then the first
 929          * unrecognized non option is the beginning of the refs list
 930          * that continues up to the -- (if exists), and then paths.
 931          */
 932
 933         while (1 < argc) {
 934                 const char *arg = argv[1];
 935                 argc--; argv++;
 936                 if (!strcmp("--cached", arg)) {
 937                         cached = 1;
 938                         continue;
 939                 }
 940                 if (!strcmp("-a", arg) ||
 941                     !strcmp("--text", arg)) {
 942                         opt.binary = GREP_BINARY_TEXT;
 943                         continue;
 944                 }
 945                 if (!strcmp("-i", arg) ||
 946                     !strcmp("--ignore-case", arg)) {
 947                         opt.regflags |= REG_ICASE;
 948                         continue;
 949                 }
 950                 if (!strcmp("-I", arg)) {
 951                         opt.binary = GREP_BINARY_NOMATCH;
 952                         continue;
 953                 }
 954                 if (!strcmp("-v", arg) ||
 955                     !strcmp("--invert-match", arg)) {
 956                         opt.invert = 1;
 957                         continue;
 958                 }
 959                 if (!strcmp("-E", arg) ||
 960                     !strcmp("--extended-regexp", arg)) {
 961                         opt.regflags |= REG_EXTENDED;
 962                         continue;
 963                 }
 964                 if (!strcmp("-F", arg) ||
 965                     !strcmp("--fixed-strings", arg)) {
 966                         opt.fixed = 1;
 967                         continue;
 968                 }
 969                 if (!strcmp("-G", arg) ||
 970                     !strcmp("--basic-regexp", arg)) {
 971                         opt.regflags &= ~REG_EXTENDED;
 972                         continue;
 973                 }
 974                 if (!strcmp("-n", arg)) {
 975                         opt.linenum = 1;
 976                         continue;
 977                 }
 978                 if (!strcmp("-h", arg)) {
 979                         opt.pathname = 0;
 980                         continue;
 981                 }
 982                 if (!strcmp("-H", arg)) {
 983                         opt.pathname = 1;
 984                         continue;
 985                 }
 986                 if (!strcmp("-l", arg) ||
 987                     !strcmp("--files-with-matches", arg)) {
 988                         opt.name_only = 1;
 989                         continue;
 990                 }
 991                 if (!strcmp("-L", arg) ||
 992                     !strcmp("--files-without-match", arg)) {
 993                         opt.unmatch_name_only = 1;
 994                         continue;
 995                 }
 996                 if (!strcmp("-c", arg) ||
 997                     !strcmp("--count", arg)) {
 998                         opt.count = 1;
 999                         continue;
1000                 }
1001                 if (!strcmp("-w", arg) ||
1002                     !strcmp("--word-regexp", arg)) {
1003                         opt.word_regexp = 1;
1004                         continue;
1005                 }
1006                 if (!strncmp("-A", arg, 2) ||
1007                     !strncmp("-B", arg, 2) ||
1008                     !strncmp("-C", arg, 2) ||
1009                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
1010                         unsigned num;
1011                         const char *scan;
1012                         switch (arg[1]) {
1013                         case 'A': case 'B': case 'C':
1014                                 if (!arg[2]) {
1015                                         if (argc <= 1)
1016                                                 die(emsg_missing_context_len);
1017                                         scan = *++argv;
1018                                         argc--;
1019                                 }
1020                                 else
1021                                         scan = arg + 2;
1022                                 break;
1023                         default:
1024                                 scan = arg + 1;
1025                                 break;
1026                         }
1027                         if (sscanf(scan, "%u", &num) != 1)
1028                                 die(emsg_invalid_context_len, scan);
1029                         switch (arg[1]) {
1030                         case 'A':
1031                                 opt.post_context = num;
1032                                 break;
1033                         default:
1034                         case 'C':
1035                                 opt.post_context = num;
1036                         case 'B':
1037                                 opt.pre_context = num;
1038                                 break;
1039                         }
1040                         continue;
1041                 }
1042                 if (!strcmp("-f", arg)) {
1043                         FILE *patterns;
1044                         int lno = 0;
1045                         char buf[1024];
1046                         if (argc <= 1)
1047                                 die(emsg_missing_argument, arg);
1048                         patterns = fopen(argv[1], "r");
1049                         if (!patterns)
1050                                 die("'%s': %s", argv[1], strerror(errno));
1051                         while (fgets(buf, sizeof(buf), patterns)) {
1052                                 int len = strlen(buf);
1053                                 if (buf[len-1] == '\n')
1054                                         buf[len-1] = 0;
1055                                 /* ignore empty line like grep does */
1056                                 if (!buf[0])
1057                                         continue;
1058                                 add_pattern(&opt, xstrdup(buf), argv[1], ++lno,
1059                                             GREP_PATTERN);
1060                         }
1061                         fclose(patterns);
1062                         argv++;
1063                         argc--;
1064                         continue;
1065                 }
1066                 if (!strcmp("--not", arg)) {
1067                         add_pattern(&opt, arg, "command line", 0, GREP_NOT);
1068                         continue;
1069                 }
1070                 if (!strcmp("--and", arg)) {
1071                         add_pattern(&opt, arg, "command line", 0, GREP_AND);
1072                         continue;
1073                 }
1074                 if (!strcmp("--or", arg))
1075                         continue; /* no-op */
1076                 if (!strcmp("(", arg)) {
1077                         add_pattern(&opt, arg, "command line", 0, GREP_OPEN_PAREN);
1078                         continue;
1079                 }
1080                 if (!strcmp(")", arg)) {
1081                         add_pattern(&opt, arg, "command line", 0, GREP_CLOSE_PAREN);
1082                         continue;
1083                 }
1084                 if (!strcmp("-e", arg)) {
1085                         if (1 < argc) {
1086                                 add_pattern(&opt, argv[1], "-e option", 0,
1087                                             GREP_PATTERN);
1088                                 argv++;
1089                                 argc--;
1090                                 continue;
1091                         }
1092                         die(emsg_missing_argument, arg);
1093                 }
1094                 if (!strcmp("--full-name", arg)) {
1095                         opt.relative = 0;
1096                         continue;
1097                 }
1098                 if (!strcmp("--", arg)) {
1099                         /* later processing wants to have this at argv[1] */
1100                         argv--;
1101                         argc++;
1102                         break;
1103                 }
1104                 if (*arg == '-')
1105                         usage(builtin_grep_usage);
1106
1107                 /* First unrecognized non-option token */
1108                 if (!opt.pattern_list) {
1109                         add_pattern(&opt, arg, "command line", 0,
1110                                     GREP_PATTERN);
1111                         break;
1112                 }
1113                 else {
1114                         /* We are looking at the first path or rev;
1115                          * it is found at argv[1] after leaving the
1116                          * loop.
1117                          */
1118                         argc++; argv--;
1119                         break;
1120                 }
1121         }
1122
1123         if (!opt.pattern_list)
1124                 die("no pattern given.");
1125         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
1126                 die("cannot mix --fixed-strings and regexp");
1127         if (!opt.fixed)
1128                 compile_patterns(&opt);
1129
1130         /* Check revs and then paths */
1131         for (i = 1; i < argc; i++) {
1132                 const char *arg = argv[i];
1133                 unsigned char sha1[20];
1134                 /* Is it a rev? */
1135                 if (!get_sha1(arg, sha1)) {
1136                         struct object *object = parse_object(sha1);
1137                         if (!object)
1138                                 die("bad object %s", arg);
1139                         add_object_array(object, arg, &list);
1140                         continue;
1141                 }
1142                 if (!strcmp(arg, "--")) {
1143                         i++;
1144                         seen_dashdash = 1;
1145                 }
1146                 break;
1147         }
1148
1149         /* The rest are paths */
1150         if (!seen_dashdash) {
1151                 int j;
1152                 for (j = i; j < argc; j++)
1153                         verify_filename(prefix, argv[j]);
1154         }
1155
1156         if (i < argc) {
1157                 paths = get_pathspec(prefix, argv + i);
1158                 if (opt.prefix_length && opt.relative) {
1159                         /* Make sure we do not get outside of paths */
1160                         for (i = 0; paths[i]; i++)
1161                                 if (strncmp(prefix, paths[i], opt.prefix_length))
1162                                         die("git-grep: cannot generate relative filenames containing '..'");
1163                 }
1164         }
1165         else if (prefix) {
1166                 paths = xcalloc(2, sizeof(const char *));
1167                 paths[0] = prefix;
1168                 paths[1] = NULL;
1169         }
1170
1171         if (!list.nr)
1172                 return !grep_cache(&opt, paths, cached);
1173
1174         if (cached)
1175                 die("both --cached and trees are given.");
1176
1177         for (i = 0; i < list.nr; i++) {
1178                 struct object *real_obj;
1179                 real_obj = deref_tag(list.objects[i].item, NULL, 0);
1180                 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
1181                         hit = 1;
1182         }
1183         return !hit;
1184 }