grep.c

   1 #include "cache.h"
   2 #include "grep.h"
   3 #include "userdiff.h"
   4 #include "xdiff-interface.h"
   5
   6 static int grep_source_load(struct grep_source *gs);
   7 static int grep_source_is_binary(struct grep_source *gs);
   8
   9
  10 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
  11                                         const char *origin, int no,
  12                                         enum grep_pat_token t,
  13                                         enum grep_header_field field)
  14 {
  15         struct grep_pat *p = xcalloc(1, sizeof(*p));
  16         p->pattern = xmemdupz(pat, patlen);
  17         p->patternlen = patlen;
  18         p->origin = origin;
  19         p->no = no;
  20         p->token = t;
  21         p->field = field;
  22         return p;
  23 }
  24
  25 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
  26 {
  27         **tail = p;
  28         *tail = &p->next;
  29         p->next = NULL;
  30
  31         switch (p->token) {
  32         case GREP_PATTERN: /* atom */
  33         case GREP_PATTERN_HEAD:
  34         case GREP_PATTERN_BODY:
  35                 for (;;) {
  36                         struct grep_pat *new_pat;
  37                         size_t len = 0;
  38                         char *cp = p->pattern + p->patternlen, *nl = NULL;
  39                         while (++len <= p->patternlen) {
  40                                 if (*(--cp) == '\n') {
  41                                         nl = cp;
  42                                         break;
  43                                 }
  44                         }
  45                         if (!nl)
  46                                 break;
  47                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
  48                                                   p->no, p->token, p->field);
  49                         new_pat->next = p->next;
  50                         if (!p->next)
  51                                 *tail = &new_pat->next;
  52                         p->next = new_pat;
  53                         *nl = '\0';
  54                         p->patternlen -= len;
  55                 }
  56                 break;
  57         default:
  58                 break;
  59         }
  60 }
  61
  62 void append_header_grep_pattern(struct grep_opt *opt,
  63                                 enum grep_header_field field, const char *pat)
  64 {
  65         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
  66                                              GREP_PATTERN_HEAD, field);
  67         do_append_grep_pat(&opt->header_tail, p);
  68 }
  69
  70 void append_grep_pattern(struct grep_opt *opt, const char *pat,
  71                          const char *origin, int no, enum grep_pat_token t)
  72 {
  73         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
  74 }
  75
  76 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
  77                      const char *origin, int no, enum grep_pat_token t)
  78 {
  79         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
  80         do_append_grep_pat(&opt->pattern_tail, p);
  81 }
  82
  83 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
  84 {
  85         struct grep_pat *pat;
  86         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
  87         *ret = *opt;
  88
  89         ret->pattern_list = NULL;
  90         ret->pattern_tail = &ret->pattern_list;
  91
  92         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
  93         {
  94                 if(pat->token == GREP_PATTERN_HEAD)
  95                         append_header_grep_pattern(ret, pat->field,
  96                                                    pat->pattern);
  97                 else
  98                         append_grep_pat(ret, pat->pattern, pat->patternlen,
  99                                         pat->origin, pat->no, pat->token);
 100         }
 101
 102         return ret;
 103 }
 104
 105 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 106                 const char *error)
 107 {
 108         char where[1024];
 109
 110         if (p->no)
 111                 sprintf(where, "In '%s' at %d, ", p->origin, p->no);
 112         else if (p->origin)
 113                 sprintf(where, "%s, ", p->origin);
 114         else
 115                 where[0] = 0;
 116
 117         die("%s'%s': %s", where, p->pattern, error);
 118 }
 119
 120 #ifdef USE_LIBPCRE
 121 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 122 {
 123         const char *error;
 124         int erroffset;
 125         int options = PCRE_MULTILINE;
 126
 127         if (opt->ignore_case)
 128                 options |= PCRE_CASELESS;
 129
 130         p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 131                         NULL);
 132         if (!p->pcre_regexp)
 133                 compile_regexp_failed(p, error);
 134
 135         p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 136         if (!p->pcre_extra_info && error)
 137                 die("%s", error);
 138 }
 139
 140 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 141                 regmatch_t *match, int eflags)
 142 {
 143         int ovector[30], ret, flags = 0;
 144
 145         if (eflags & REG_NOTBOL)
 146                 flags |= PCRE_NOTBOL;
 147
 148         ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 149                         0, flags, ovector, ARRAY_SIZE(ovector));
 150         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 151                 die("pcre_exec failed with error code %d", ret);
 152         if (ret > 0) {
 153                 ret = 0;
 154                 match->rm_so = ovector[0];
 155                 match->rm_eo = ovector[1];
 156         }
 157
 158         return ret;
 159 }
 160
 161 static void free_pcre_regexp(struct grep_pat *p)
 162 {
 163         pcre_free(p->pcre_regexp);
 164         pcre_free(p->pcre_extra_info);
 165 }
 166 #else /* !USE_LIBPCRE */
 167 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 168 {
 169         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 170 }
 171
 172 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 173                 regmatch_t *match, int eflags)
 174 {
 175         return 1;
 176 }
 177
 178 static void free_pcre_regexp(struct grep_pat *p)
 179 {
 180 }
 181 #endif /* !USE_LIBPCRE */
 182
 183 static int is_fixed(const char *s, size_t len)
 184 {
 185         size_t i;
 186
 187         /* regcomp cannot accept patterns with NULs so we
 188          * consider any pattern containing a NUL fixed.
 189          */
 190         if (memchr(s, 0, len))
 191                 return 1;
 192
 193         for (i = 0; i < len; i++) {
 194                 if (is_regex_special(s[i]))
 195                         return 0;
 196         }
 197
 198         return 1;
 199 }
 200
 201 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 202 {
 203         int err;
 204
 205         p->word_regexp = opt->word_regexp;
 206         p->ignore_case = opt->ignore_case;
 207
 208         if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 209                 p->fixed = 1;
 210         else
 211                 p->fixed = 0;
 212
 213         if (p->fixed) {
 214                 if (opt->regflags & REG_ICASE || p->ignore_case)
 215                         p->kws = kwsalloc(tolower_trans_tbl);
 216                 else
 217                         p->kws = kwsalloc(NULL);
 218                 kwsincr(p->kws, p->pattern, p->patternlen);
 219                 kwsprep(p->kws);
 220                 return;
 221         }
 222
 223         if (opt->pcre) {
 224                 compile_pcre_regexp(p, opt);
 225                 return;
 226         }
 227
 228         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 229         if (err) {
 230                 char errbuf[1024];
 231                 regerror(err, &p->regexp, errbuf, 1024);
 232                 regfree(&p->regexp);
 233                 compile_regexp_failed(p, errbuf);
 234         }
 235 }
 236
 237 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 238 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 239 {
 240         struct grep_pat *p;
 241         struct grep_expr *x;
 242
 243         p = *list;
 244         if (!p)
 245                 return NULL;
 246         switch (p->token) {
 247         case GREP_PATTERN: /* atom */
 248         case GREP_PATTERN_HEAD:
 249         case GREP_PATTERN_BODY:
 250                 x = xcalloc(1, sizeof (struct grep_expr));
 251                 x->node = GREP_NODE_ATOM;
 252                 x->u.atom = p;
 253                 *list = p->next;
 254                 return x;
 255         case GREP_OPEN_PAREN:
 256                 *list = p->next;
 257                 x = compile_pattern_or(list);
 258                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 259                         die("unmatched parenthesis");
 260                 *list = (*list)->next;
 261                 return x;
 262         default:
 263                 return NULL;
 264         }
 265 }
 266
 267 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 268 {
 269         struct grep_pat *p;
 270         struct grep_expr *x;
 271
 272         p = *list;
 273         if (!p)
 274                 return NULL;
 275         switch (p->token) {
 276         case GREP_NOT:
 277                 if (!p->next)
 278                         die("--not not followed by pattern expression");
 279                 *list = p->next;
 280                 x = xcalloc(1, sizeof (struct grep_expr));
 281                 x->node = GREP_NODE_NOT;
 282                 x->u.unary = compile_pattern_not(list);
 283                 if (!x->u.unary)
 284                         die("--not followed by non pattern expression");
 285                 return x;
 286         default:
 287                 return compile_pattern_atom(list);
 288         }
 289 }
 290
 291 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 292 {
 293         struct grep_pat *p;
 294         struct grep_expr *x, *y, *z;
 295
 296         x = compile_pattern_not(list);
 297         p = *list;
 298         if (p && p->token == GREP_AND) {
 299                 if (!p->next)
 300                         die("--and not followed by pattern expression");
 301                 *list = p->next;
 302                 y = compile_pattern_and(list);
 303                 if (!y)
 304                         die("--and not followed by pattern expression");
 305                 z = xcalloc(1, sizeof (struct grep_expr));
 306                 z->node = GREP_NODE_AND;
 307                 z->u.binary.left = x;
 308                 z->u.binary.right = y;
 309                 return z;
 310         }
 311         return x;
 312 }
 313
 314 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 315 {
 316         struct grep_pat *p;
 317         struct grep_expr *x, *y, *z;
 318
 319         x = compile_pattern_and(list);
 320         p = *list;
 321         if (x && p && p->token != GREP_CLOSE_PAREN) {
 322                 y = compile_pattern_or(list);
 323                 if (!y)
 324                         die("not a pattern expression %s", p->pattern);
 325                 z = xcalloc(1, sizeof (struct grep_expr));
 326                 z->node = GREP_NODE_OR;
 327                 z->u.binary.left = x;
 328                 z->u.binary.right = y;
 329                 return z;
 330         }
 331         return x;
 332 }
 333
 334 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 335 {
 336         return compile_pattern_or(list);
 337 }
 338
 339 static void indent(int in)
 340 {
 341         while (in-- > 0)
 342                 fputc(' ', stderr);
 343 }
 344
 345 static void dump_grep_pat(struct grep_pat *p)
 346 {
 347         switch (p->token) {
 348         case GREP_AND: fprintf(stderr, "*and*"); break;
 349         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 350         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 351         case GREP_NOT: fprintf(stderr, "*not*"); break;
 352         case GREP_OR: fprintf(stderr, "*or*"); break;
 353
 354         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 355         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 356         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 357         }
 358
 359         switch (p->token) {
 360         default: break;
 361         case GREP_PATTERN_HEAD:
 362                 fprintf(stderr, "<head %d>", p->field); break;
 363         case GREP_PATTERN_BODY:
 364                 fprintf(stderr, "<body>"); break;
 365         }
 366         switch (p->token) {
 367         default: break;
 368         case GREP_PATTERN_HEAD:
 369         case GREP_PATTERN_BODY:
 370         case GREP_PATTERN:
 371                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 372                 break;
 373         }
 374         fputc('\n', stderr);
 375 }
 376
 377 static void dump_grep_expression_1(struct grep_expr *x, int in)
 378 {
 379         indent(in);
 380         switch (x->node) {
 381         case GREP_NODE_TRUE:
 382                 fprintf(stderr, "true\n");
 383                 break;
 384         case GREP_NODE_ATOM:
 385                 dump_grep_pat(x->u.atom);
 386                 break;
 387         case GREP_NODE_NOT:
 388                 fprintf(stderr, "(not\n");
 389                 dump_grep_expression_1(x->u.unary, in+1);
 390                 indent(in);
 391                 fprintf(stderr, ")\n");
 392                 break;
 393         case GREP_NODE_AND:
 394                 fprintf(stderr, "(and\n");
 395                 dump_grep_expression_1(x->u.binary.left, in+1);
 396                 dump_grep_expression_1(x->u.binary.right, in+1);
 397                 indent(in);
 398                 fprintf(stderr, ")\n");
 399                 break;
 400         case GREP_NODE_OR:
 401                 fprintf(stderr, "(or\n");
 402                 dump_grep_expression_1(x->u.binary.left, in+1);
 403                 dump_grep_expression_1(x->u.binary.right, in+1);
 404                 indent(in);
 405                 fprintf(stderr, ")\n");
 406                 break;
 407         }
 408 }
 409
 410 static void dump_grep_expression(struct grep_opt *opt)
 411 {
 412         struct grep_expr *x = opt->pattern_expression;
 413
 414         if (opt->all_match)
 415                 fprintf(stderr, "[all-match]\n");
 416         dump_grep_expression_1(x, 0);
 417         fflush(NULL);
 418 }
 419
 420 static struct grep_expr *grep_true_expr(void)
 421 {
 422         struct grep_expr *z = xcalloc(1, sizeof(*z));
 423         z->node = GREP_NODE_TRUE;
 424         return z;
 425 }
 426
 427 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 428 {
 429         struct grep_expr *z = xcalloc(1, sizeof(*z));
 430         z->node = GREP_NODE_OR;
 431         z->u.binary.left = left;
 432         z->u.binary.right = right;
 433         return z;
 434 }
 435
 436 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 437 {
 438         struct grep_pat *p;
 439         struct grep_expr *header_expr;
 440         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 441         enum grep_header_field fld;
 442
 443         if (!opt->header_list)
 444                 return NULL;
 445
 446         for (p = opt->header_list; p; p = p->next) {
 447                 if (p->token != GREP_PATTERN_HEAD)
 448                         die("bug: a non-header pattern in grep header list.");
 449                 if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
 450                         die("bug: unknown header field %d", p->field);
 451                 compile_regexp(p, opt);
 452         }
 453
 454         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 455                 header_group[fld] = NULL;
 456
 457         for (p = opt->header_list; p; p = p->next) {
 458                 struct grep_expr *h;
 459                 struct grep_pat *pp = p;
 460
 461                 h = compile_pattern_atom(&pp);
 462                 if (!h || pp != p->next)
 463                         die("bug: malformed header expr");
 464                 if (!header_group[p->field]) {
 465                         header_group[p->field] = h;
 466                         continue;
 467                 }
 468                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 469         }
 470
 471         header_expr = NULL;
 472
 473         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 474                 if (!header_group[fld])
 475                         continue;
 476                 if (!header_expr)
 477                         header_expr = grep_true_expr();
 478                 header_expr = grep_or_expr(header_group[fld], header_expr);
 479         }
 480         return header_expr;
 481 }
 482
 483 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 484 {
 485         struct grep_expr *z = x;
 486
 487         while (x) {
 488                 assert(x->node == GREP_NODE_OR);
 489                 if (x->u.binary.right &&
 490                     x->u.binary.right->node == GREP_NODE_TRUE) {
 491                         x->u.binary.right = y;
 492                         break;
 493                 }
 494                 x = x->u.binary.right;
 495         }
 496         return z;
 497 }
 498
 499 static void compile_grep_patterns_real(struct grep_opt *opt)
 500 {
 501         struct grep_pat *p;
 502         struct grep_expr *header_expr = prep_header_patterns(opt);
 503
 504         for (p = opt->pattern_list; p; p = p->next) {
 505                 switch (p->token) {
 506                 case GREP_PATTERN: /* atom */
 507                 case GREP_PATTERN_HEAD:
 508                 case GREP_PATTERN_BODY:
 509                         compile_regexp(p, opt);
 510                         break;
 511                 default:
 512                         opt->extended = 1;
 513                         break;
 514                 }
 515         }
 516
 517         if (opt->all_match || header_expr)
 518                 opt->extended = 1;
 519         else if (!opt->extended && !opt->debug)
 520                 return;
 521
 522         p = opt->pattern_list;
 523         if (p)
 524                 opt->pattern_expression = compile_pattern_expr(&p);
 525         if (p)
 526                 die("incomplete pattern expression: %s", p->pattern);
 527
 528         if (!header_expr)
 529                 return;
 530
 531         if (!opt->pattern_expression)
 532                 opt->pattern_expression = header_expr;
 533         else if (opt->all_match)
 534                 opt->pattern_expression = grep_splice_or(header_expr,
 535                                                          opt->pattern_expression);
 536         else
 537                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 538                                                        header_expr);
 539         opt->all_match = 1;
 540 }
 541
 542 void compile_grep_patterns(struct grep_opt *opt)
 543 {
 544         compile_grep_patterns_real(opt);
 545         if (opt->debug)
 546                 dump_grep_expression(opt);
 547 }
 548
 549 static void free_pattern_expr(struct grep_expr *x)
 550 {
 551         switch (x->node) {
 552         case GREP_NODE_TRUE:
 553         case GREP_NODE_ATOM:
 554                 break;
 555         case GREP_NODE_NOT:
 556                 free_pattern_expr(x->u.unary);
 557                 break;
 558         case GREP_NODE_AND:
 559         case GREP_NODE_OR:
 560                 free_pattern_expr(x->u.binary.left);
 561                 free_pattern_expr(x->u.binary.right);
 562                 break;
 563         }
 564         free(x);
 565 }
 566
 567 void free_grep_patterns(struct grep_opt *opt)
 568 {
 569         struct grep_pat *p, *n;
 570
 571         for (p = opt->pattern_list; p; p = n) {
 572                 n = p->next;
 573                 switch (p->token) {
 574                 case GREP_PATTERN: /* atom */
 575                 case GREP_PATTERN_HEAD:
 576                 case GREP_PATTERN_BODY:
 577                         if (p->kws)
 578                                 kwsfree(p->kws);
 579                         else if (p->pcre_regexp)
 580                                 free_pcre_regexp(p);
 581                         else
 582                                 regfree(&p->regexp);
 583                         free(p->pattern);
 584                         break;
 585                 default:
 586                         break;
 587                 }
 588                 free(p);
 589         }
 590
 591         if (!opt->extended)
 592                 return;
 593         free_pattern_expr(opt->pattern_expression);
 594 }
 595
 596 static char *end_of_line(char *cp, unsigned long *left)
 597 {
 598         unsigned long l = *left;
 599         while (l && *cp != '\n') {
 600                 l--;
 601                 cp++;
 602         }
 603         *left = l;
 604         return cp;
 605 }
 606
 607 static int word_char(char ch)
 608 {
 609         return isalnum(ch) || ch == '_';
 610 }
 611
 612 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 613                          const char *color)
 614 {
 615         if (want_color(opt->color) && color && color[0]) {
 616                 opt->output(opt, color, strlen(color));
 617                 opt->output(opt, data, size);
 618                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 619         } else
 620                 opt->output(opt, data, size);
 621 }
 622
 623 static void output_sep(struct grep_opt *opt, char sign)
 624 {
 625         if (opt->null_following_name)
 626                 opt->output(opt, "\0", 1);
 627         else
 628                 output_color(opt, &sign, 1, opt->color_sep);
 629 }
 630
 631 static void show_name(struct grep_opt *opt, const char *name)
 632 {
 633         output_color(opt, name, strlen(name), opt->color_filename);
 634         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 635 }
 636
 637 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 638                     regmatch_t *match)
 639 {
 640         struct kwsmatch kwsm;
 641         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 642         if (offset == -1) {
 643                 match->rm_so = match->rm_eo = -1;
 644                 return REG_NOMATCH;
 645         } else {
 646                 match->rm_so = offset;
 647                 match->rm_eo = match->rm_so + kwsm.size[0];
 648                 return 0;
 649         }
 650 }
 651
 652 static int regmatch(const regex_t *preg, char *line, char *eol,
 653                     regmatch_t *match, int eflags)
 654 {
 655 #ifdef REG_STARTEND
 656         match->rm_so = 0;
 657         match->rm_eo = eol - line;
 658         eflags |= REG_STARTEND;
 659 #endif
 660         return regexec(preg, line, 1, match, eflags);
 661 }
 662
 663 static int patmatch(struct grep_pat *p, char *line, char *eol,
 664                     regmatch_t *match, int eflags)
 665 {
 666         int hit;
 667
 668         if (p->fixed)
 669                 hit = !fixmatch(p, line, eol, match);
 670         else if (p->pcre_regexp)
 671                 hit = !pcrematch(p, line, eol, match, eflags);
 672         else
 673                 hit = !regmatch(&p->regexp, line, eol, match, eflags);
 674
 675         return hit;
 676 }
 677
 678 static int strip_timestamp(char *bol, char **eol_p)
 679 {
 680         char *eol = *eol_p;
 681         int ch;
 682
 683         while (bol < --eol) {
 684                 if (*eol != '>')
 685                         continue;
 686                 *eol_p = ++eol;
 687                 ch = *eol;
 688                 *eol = '\0';
 689                 return ch;
 690         }
 691         return 0;
 692 }
 693
 694 static struct {
 695         const char *field;
 696         size_t len;
 697 } header_field[] = {
 698         { "author ", 7 },
 699         { "committer ", 10 },
 700         { "reflog ", 7 },
 701 };
 702
 703 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 704                              enum grep_context ctx,
 705                              regmatch_t *pmatch, int eflags)
 706 {
 707         int hit = 0;
 708         int saved_ch = 0;
 709         const char *start = bol;
 710
 711         if ((p->token != GREP_PATTERN) &&
 712             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 713                 return 0;
 714
 715         if (p->token == GREP_PATTERN_HEAD) {
 716                 const char *field;
 717                 size_t len;
 718                 assert(p->field < ARRAY_SIZE(header_field));
 719                 field = header_field[p->field].field;
 720                 len = header_field[p->field].len;
 721                 if (strncmp(bol, field, len))
 722                         return 0;
 723                 bol += len;
 724                 switch (p->field) {
 725                 case GREP_HEADER_AUTHOR:
 726                 case GREP_HEADER_COMMITTER:
 727                         saved_ch = strip_timestamp(bol, &eol);
 728                         break;
 729                 default:
 730                         break;
 731                 }
 732         }
 733
 734  again:
 735         hit = patmatch(p, bol, eol, pmatch, eflags);
 736
 737         if (hit && p->word_regexp) {
 738                 if ((pmatch[0].rm_so < 0) ||
 739                     (eol - bol) < pmatch[0].rm_so ||
 740                     (pmatch[0].rm_eo < 0) ||
 741                     (eol - bol) < pmatch[0].rm_eo)
 742                         die("regexp returned nonsense");
 743
 744                 /* Match beginning must be either beginning of the
 745                  * line, or at word boundary (i.e. the last char must
 746                  * not be a word char).  Similarly, match end must be
 747                  * either end of the line, or at word boundary
 748                  * (i.e. the next char must not be a word char).
 749                  */
 750                 if ( ((pmatch[0].rm_so == 0) ||
 751                       !word_char(bol[pmatch[0].rm_so-1])) &&
 752                      ((pmatch[0].rm_eo == (eol-bol)) ||
 753                       !word_char(bol[pmatch[0].rm_eo])) )
 754                         ;
 755                 else
 756                         hit = 0;
 757
 758                 /* Words consist of at least one character. */
 759                 if (pmatch->rm_so == pmatch->rm_eo)
 760                         hit = 0;
 761
 762                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 763                         /* There could be more than one match on the
 764                          * line, and the first match might not be
 765                          * strict word match.  But later ones could be!
 766                          * Forward to the next possible start, i.e. the
 767                          * next position following a non-word char.
 768                          */
 769                         bol = pmatch[0].rm_so + bol + 1;
 770                         while (word_char(bol[-1]) && bol < eol)
 771                                 bol++;
 772                         eflags |= REG_NOTBOL;
 773                         if (bol < eol)
 774                                 goto again;
 775                 }
 776         }
 777         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 778                 *eol = saved_ch;
 779         if (hit) {
 780                 pmatch[0].rm_so += bol - start;
 781                 pmatch[0].rm_eo += bol - start;
 782         }
 783         return hit;
 784 }
 785
 786 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 787                            enum grep_context ctx, int collect_hits)
 788 {
 789         int h = 0;
 790         regmatch_t match;
 791
 792         if (!x)
 793                 die("Not a valid grep expression");
 794         switch (x->node) {
 795         case GREP_NODE_TRUE:
 796                 h = 1;
 797                 break;
 798         case GREP_NODE_ATOM:
 799                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 800                 break;
 801         case GREP_NODE_NOT:
 802                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 803                 break;
 804         case GREP_NODE_AND:
 805                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 806                         return 0;
 807                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 808                 break;
 809         case GREP_NODE_OR:
 810                 if (!collect_hits)
 811                         return (match_expr_eval(x->u.binary.left,
 812                                                 bol, eol, ctx, 0) ||
 813                                 match_expr_eval(x->u.binary.right,
 814                                                 bol, eol, ctx, 0));
 815                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 816                 x->u.binary.left->hit |= h;
 817                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 818                 break;
 819         default:
 820                 die("Unexpected node type (internal error) %d", x->node);
 821         }
 822         if (collect_hits)
 823                 x->hit |= h;
 824         return h;
 825 }
 826
 827 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 828                       enum grep_context ctx, int collect_hits)
 829 {
 830         struct grep_expr *x = opt->pattern_expression;
 831         return match_expr_eval(x, bol, eol, ctx, collect_hits);
 832 }
 833
 834 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 835                       enum grep_context ctx, int collect_hits)
 836 {
 837         struct grep_pat *p;
 838         regmatch_t match;
 839
 840         if (opt->extended)
 841                 return match_expr(opt, bol, eol, ctx, collect_hits);
 842
 843         /* we do not call with collect_hits without being extended */
 844         for (p = opt->pattern_list; p; p = p->next) {
 845                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 846                         return 1;
 847         }
 848         return 0;
 849 }
 850
 851 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 852                               enum grep_context ctx,
 853                               regmatch_t *pmatch, int eflags)
 854 {
 855         regmatch_t match;
 856
 857         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 858                 return 0;
 859         if (match.rm_so < 0 || match.rm_eo < 0)
 860                 return 0;
 861         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 862                 if (match.rm_so > pmatch->rm_so)
 863                         return 1;
 864                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 865                         return 1;
 866         }
 867         pmatch->rm_so = match.rm_so;
 868         pmatch->rm_eo = match.rm_eo;
 869         return 1;
 870 }
 871
 872 static int next_match(struct grep_opt *opt, char *bol, char *eol,
 873                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
 874 {
 875         struct grep_pat *p;
 876         int hit = 0;
 877
 878         pmatch->rm_so = pmatch->rm_eo = -1;
 879         if (bol < eol) {
 880                 for (p = opt->pattern_list; p; p = p->next) {
 881                         switch (p->token) {
 882                         case GREP_PATTERN: /* atom */
 883                         case GREP_PATTERN_HEAD:
 884                         case GREP_PATTERN_BODY:
 885                                 hit |= match_next_pattern(p, bol, eol, ctx,
 886                                                           pmatch, eflags);
 887                                 break;
 888                         default:
 889                                 break;
 890                         }
 891                 }
 892         }
 893         return hit;
 894 }
 895
 896 static void show_line(struct grep_opt *opt, char *bol, char *eol,
 897                       const char *name, unsigned lno, char sign)
 898 {
 899         int rest = eol - bol;
 900         char *line_color = NULL;
 901
 902         if (opt->file_break && opt->last_shown == 0) {
 903                 if (opt->show_hunk_mark)
 904                         opt->output(opt, "\n", 1);
 905         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
 906                 if (opt->last_shown == 0) {
 907                         if (opt->show_hunk_mark) {
 908                                 output_color(opt, "--", 2, opt->color_sep);
 909                                 opt->output(opt, "\n", 1);
 910                         }
 911                 } else if (lno > opt->last_shown + 1) {
 912                         output_color(opt, "--", 2, opt->color_sep);
 913                         opt->output(opt, "\n", 1);
 914                 }
 915         }
 916         if (opt->heading && opt->last_shown == 0) {
 917                 output_color(opt, name, strlen(name), opt->color_filename);
 918                 opt->output(opt, "\n", 1);
 919         }
 920         opt->last_shown = lno;
 921
 922         if (!opt->heading && opt->pathname) {
 923                 output_color(opt, name, strlen(name), opt->color_filename);
 924                 output_sep(opt, sign);
 925         }
 926         if (opt->linenum) {
 927                 char buf[32];
 928                 snprintf(buf, sizeof(buf), "%d", lno);
 929                 output_color(opt, buf, strlen(buf), opt->color_lineno);
 930                 output_sep(opt, sign);
 931         }
 932         if (opt->color) {
 933                 regmatch_t match;
 934                 enum grep_context ctx = GREP_CONTEXT_BODY;
 935                 int ch = *eol;
 936                 int eflags = 0;
 937
 938                 if (sign == ':')
 939                         line_color = opt->color_selected;
 940                 else if (sign == '-')
 941                         line_color = opt->color_context;
 942                 else if (sign == '=')
 943                         line_color = opt->color_function;
 944                 *eol = '\0';
 945                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
 946                         if (match.rm_so == match.rm_eo)
 947                                 break;
 948
 949                         output_color(opt, bol, match.rm_so, line_color);
 950                         output_color(opt, bol + match.rm_so,
 951                                      match.rm_eo - match.rm_so,
 952                                      opt->color_match);
 953                         bol += match.rm_eo;
 954                         rest -= match.rm_eo;
 955                         eflags = REG_NOTBOL;
 956                 }
 957                 *eol = ch;
 958         }
 959         output_color(opt, bol, rest, line_color);
 960         opt->output(opt, "\n", 1);
 961 }
 962
 963 #ifndef NO_PTHREADS
 964 int grep_use_locks;
 965
 966 /*
 967  * This lock protects access to the gitattributes machinery, which is
 968  * not thread-safe.
 969  */
 970 pthread_mutex_t grep_attr_mutex;
 971
 972 static inline void grep_attr_lock(void)
 973 {
 974         if (grep_use_locks)
 975                 pthread_mutex_lock(&grep_attr_mutex);
 976 }
 977
 978 static inline void grep_attr_unlock(void)
 979 {
 980         if (grep_use_locks)
 981                 pthread_mutex_unlock(&grep_attr_mutex);
 982 }
 983
 984 /*
 985  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
 986  */
 987 pthread_mutex_t grep_read_mutex;
 988
 989 #else
 990 #define grep_attr_lock()
 991 #define grep_attr_unlock()
 992 #endif
 993
 994 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
 995 {
 996         xdemitconf_t *xecfg = opt->priv;
 997         if (xecfg && !xecfg->find_func) {
 998                 grep_source_load_driver(gs);
 999                 if (gs->driver->funcname.pattern) {
1000                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1001                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1002                 } else {
1003                         xecfg = opt->priv = NULL;
1004                 }
1005         }
1006
1007         if (xecfg) {
1008                 char buf[1];
1009                 return xecfg->find_func(bol, eol - bol, buf, 1,
1010                                         xecfg->find_func_priv) >= 0;
1011         }
1012
1013         if (bol == eol)
1014                 return 0;
1015         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1016                 return 1;
1017         return 0;
1018 }
1019
1020 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1021                                char *bol, unsigned lno)
1022 {
1023         while (bol > gs->buf) {
1024                 char *eol = --bol;
1025
1026                 while (bol > gs->buf && bol[-1] != '\n')
1027                         bol--;
1028                 lno--;
1029
1030                 if (lno <= opt->last_shown)
1031                         break;
1032
1033                 if (match_funcname(opt, gs, bol, eol)) {
1034                         show_line(opt, bol, eol, gs->name, lno, '=');
1035                         break;
1036                 }
1037         }
1038 }
1039
1040 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1041                              char *bol, char *end, unsigned lno)
1042 {
1043         unsigned cur = lno, from = 1, funcname_lno = 0;
1044         int funcname_needed = !!opt->funcname;
1045
1046         if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1047                 funcname_needed = 2;
1048
1049         if (opt->pre_context < lno)
1050                 from = lno - opt->pre_context;
1051         if (from <= opt->last_shown)
1052                 from = opt->last_shown + 1;
1053
1054         /* Rewind. */
1055         while (bol > gs->buf &&
1056                cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1057                 char *eol = --bol;
1058
1059                 while (bol > gs->buf && bol[-1] != '\n')
1060                         bol--;
1061                 cur--;
1062                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1063                         funcname_lno = cur;
1064                         funcname_needed = 0;
1065                 }
1066         }
1067
1068         /* We need to look even further back to find a function signature. */
1069         if (opt->funcname && funcname_needed)
1070                 show_funcname_line(opt, gs, bol, cur);
1071
1072         /* Back forward. */
1073         while (cur < lno) {
1074                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1075
1076                 while (*eol != '\n')
1077                         eol++;
1078                 show_line(opt, bol, eol, gs->name, cur, sign);
1079                 bol = eol + 1;
1080                 cur++;
1081         }
1082 }
1083
1084 static int should_lookahead(struct grep_opt *opt)
1085 {
1086         struct grep_pat *p;
1087
1088         if (opt->extended)
1089                 return 0; /* punt for too complex stuff */
1090         if (opt->invert)
1091                 return 0;
1092         for (p = opt->pattern_list; p; p = p->next) {
1093                 if (p->token != GREP_PATTERN)
1094                         return 0; /* punt for "header only" and stuff */
1095         }
1096         return 1;
1097 }
1098
1099 static int look_ahead(struct grep_opt *opt,
1100                       unsigned long *left_p,
1101                       unsigned *lno_p,
1102                       char **bol_p)
1103 {
1104         unsigned lno = *lno_p;
1105         char *bol = *bol_p;
1106         struct grep_pat *p;
1107         char *sp, *last_bol;
1108         regoff_t earliest = -1;
1109
1110         for (p = opt->pattern_list; p; p = p->next) {
1111                 int hit;
1112                 regmatch_t m;
1113
1114                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1115                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1116                         continue;
1117                 if (earliest < 0 || m.rm_so < earliest)
1118                         earliest = m.rm_so;
1119         }
1120
1121         if (earliest < 0) {
1122                 *bol_p = bol + *left_p;
1123                 *left_p = 0;
1124                 return 1;
1125         }
1126         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1127                 ; /* find the beginning of the line */
1128         last_bol = sp;
1129
1130         for (sp = bol; sp < last_bol; sp++) {
1131                 if (*sp == '\n')
1132                         lno++;
1133         }
1134         *left_p -= last_bol - bol;
1135         *bol_p = last_bol;
1136         *lno_p = lno;
1137         return 0;
1138 }
1139
1140 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1141 {
1142         fwrite(buf, size, 1, stdout);
1143 }
1144
1145 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1146 {
1147         char *bol;
1148         unsigned long left;
1149         unsigned lno = 1;
1150         unsigned last_hit = 0;
1151         int binary_match_only = 0;
1152         unsigned count = 0;
1153         int try_lookahead = 0;
1154         int show_function = 0;
1155         enum grep_context ctx = GREP_CONTEXT_HEAD;
1156         xdemitconf_t xecfg;
1157
1158         if (!opt->output)
1159                 opt->output = std_output;
1160
1161         if (opt->pre_context || opt->post_context || opt->file_break ||
1162             opt->funcbody) {
1163                 /* Show hunk marks, except for the first file. */
1164                 if (opt->last_shown)
1165                         opt->show_hunk_mark = 1;
1166                 /*
1167                  * If we're using threads then we can't easily identify
1168                  * the first file.  Always put hunk marks in that case
1169                  * and skip the very first one later in work_done().
1170                  */
1171                 if (opt->output != std_output)
1172                         opt->show_hunk_mark = 1;
1173         }
1174         opt->last_shown = 0;
1175
1176         switch (opt->binary) {
1177         case GREP_BINARY_DEFAULT:
1178                 if (grep_source_is_binary(gs))
1179                         binary_match_only = 1;
1180                 break;
1181         case GREP_BINARY_NOMATCH:
1182                 if (grep_source_is_binary(gs))
1183                         return 0; /* Assume unmatch */
1184                 break;
1185         case GREP_BINARY_TEXT:
1186                 break;
1187         default:
1188                 die("bug: unknown binary handling mode");
1189         }
1190
1191         memset(&xecfg, 0, sizeof(xecfg));
1192         opt->priv = &xecfg;
1193
1194         try_lookahead = should_lookahead(opt);
1195
1196         if (grep_source_load(gs) < 0)
1197                 return 0;
1198
1199         bol = gs->buf;
1200         left = gs->size;
1201         while (left) {
1202                 char *eol, ch;
1203                 int hit;
1204
1205                 /*
1206                  * look_ahead() skips quickly to the line that possibly
1207                  * has the next hit; don't call it if we need to do
1208                  * something more than just skipping the current line
1209                  * in response to an unmatch for the current line.  E.g.
1210                  * inside a post-context window, we will show the current
1211                  * line as a context around the previous hit when it
1212                  * doesn't hit.
1213                  */
1214                 if (try_lookahead
1215                     && !(last_hit
1216                          && (show_function ||
1217                              lno <= last_hit + opt->post_context))
1218                     && look_ahead(opt, &left, &lno, &bol))
1219                         break;
1220                 eol = end_of_line(bol, &left);
1221                 ch = *eol;
1222                 *eol = 0;
1223
1224                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1225                         ctx = GREP_CONTEXT_BODY;
1226
1227                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1228                 *eol = ch;
1229
1230                 if (collect_hits)
1231                         goto next_line;
1232
1233                 /* "grep -v -e foo -e bla" should list lines
1234                  * that do not have either, so inversion should
1235                  * be done outside.
1236                  */
1237                 if (opt->invert)
1238                         hit = !hit;
1239                 if (opt->unmatch_name_only) {
1240                         if (hit)
1241                                 return 0;
1242                         goto next_line;
1243                 }
1244                 if (hit) {
1245                         count++;
1246                         if (opt->status_only)
1247                                 return 1;
1248                         if (opt->name_only) {
1249                                 show_name(opt, gs->name);
1250                                 return 1;
1251                         }
1252                         if (opt->count)
1253                                 goto next_line;
1254                         if (binary_match_only) {
1255                                 opt->output(opt, "Binary file ", 12);
1256                                 output_color(opt, gs->name, strlen(gs->name),
1257                                              opt->color_filename);
1258                                 opt->output(opt, " matches\n", 9);
1259                                 return 1;
1260                         }
1261                         /* Hit at this line.  If we haven't shown the
1262                          * pre-context lines, we would need to show them.
1263                          */
1264                         if (opt->pre_context || opt->funcbody)
1265                                 show_pre_context(opt, gs, bol, eol, lno);
1266                         else if (opt->funcname)
1267                                 show_funcname_line(opt, gs, bol, lno);
1268                         show_line(opt, bol, eol, gs->name, lno, ':');
1269                         last_hit = lno;
1270                         if (opt->funcbody)
1271                                 show_function = 1;
1272                         goto next_line;
1273                 }
1274                 if (show_function && match_funcname(opt, gs, bol, eol))
1275                         show_function = 0;
1276                 if (show_function ||
1277                     (last_hit && lno <= last_hit + opt->post_context)) {
1278                         /* If the last hit is within the post context,
1279                          * we need to show this line.
1280                          */
1281                         show_line(opt, bol, eol, gs->name, lno, '-');
1282                 }
1283
1284         next_line:
1285                 bol = eol + 1;
1286                 if (!left)
1287                         break;
1288                 left--;
1289                 lno++;
1290         }
1291
1292         if (collect_hits)
1293                 return 0;
1294
1295         if (opt->status_only)
1296                 return 0;
1297         if (opt->unmatch_name_only) {
1298                 /* We did not see any hit, so we want to show this */
1299                 show_name(opt, gs->name);
1300                 return 1;
1301         }
1302
1303         xdiff_clear_find_func(&xecfg);
1304         opt->priv = NULL;
1305
1306         /* NEEDSWORK:
1307          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1308          * which feels mostly useless but sometimes useful.  Maybe
1309          * make it another option?  For now suppress them.
1310          */
1311         if (opt->count && count) {
1312                 char buf[32];
1313                 output_color(opt, gs->name, strlen(gs->name), opt->color_filename);
1314                 output_sep(opt, ':');
1315                 snprintf(buf, sizeof(buf), "%u\n", count);
1316                 opt->output(opt, buf, strlen(buf));
1317                 return 1;
1318         }
1319         return !!last_hit;
1320 }
1321
1322 static void clr_hit_marker(struct grep_expr *x)
1323 {
1324         /* All-hit markers are meaningful only at the very top level
1325          * OR node.
1326          */
1327         while (1) {
1328                 x->hit = 0;
1329                 if (x->node != GREP_NODE_OR)
1330                         return;
1331                 x->u.binary.left->hit = 0;
1332                 x = x->u.binary.right;
1333         }
1334 }
1335
1336 static int chk_hit_marker(struct grep_expr *x)
1337 {
1338         /* Top level nodes have hit markers.  See if they all are hits */
1339         while (1) {
1340                 if (x->node != GREP_NODE_OR)
1341                         return x->hit;
1342                 if (!x->u.binary.left->hit)
1343                         return 0;
1344                 x = x->u.binary.right;
1345         }
1346 }
1347
1348 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1349 {
1350         /*
1351          * we do not have to do the two-pass grep when we do not check
1352          * buffer-wide "all-match".
1353          */
1354         if (!opt->all_match)
1355                 return grep_source_1(opt, gs, 0);
1356
1357         /* Otherwise the toplevel "or" terms hit a bit differently.
1358          * We first clear hit markers from them.
1359          */
1360         clr_hit_marker(opt->pattern_expression);
1361         grep_source_1(opt, gs, 1);
1362
1363         if (!chk_hit_marker(opt->pattern_expression))
1364                 return 0;
1365
1366         return grep_source_1(opt, gs, 0);
1367 }
1368
1369 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1370 {
1371         struct grep_source gs;
1372         int r;
1373
1374         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL);
1375         gs.buf = buf;
1376         gs.size = size;
1377
1378         r = grep_source(opt, &gs);
1379
1380         grep_source_clear(&gs);
1381         return r;
1382 }
1383
1384 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1385                       const char *name, const void *identifier)
1386 {
1387         gs->type = type;
1388         gs->name = name ? xstrdup(name) : NULL;
1389         gs->buf = NULL;
1390         gs->size = 0;
1391         gs->driver = NULL;
1392
1393         switch (type) {
1394         case GREP_SOURCE_FILE:
1395                 gs->identifier = xstrdup(identifier);
1396                 break;
1397         case GREP_SOURCE_SHA1:
1398                 gs->identifier = xmalloc(20);
1399                 memcpy(gs->identifier, identifier, 20);
1400                 break;
1401         case GREP_SOURCE_BUF:
1402                 gs->identifier = NULL;
1403         }
1404 }
1405
1406 void grep_source_clear(struct grep_source *gs)
1407 {
1408         free(gs->name);
1409         gs->name = NULL;
1410         free(gs->identifier);
1411         gs->identifier = NULL;
1412         grep_source_clear_data(gs);
1413 }
1414
1415 void grep_source_clear_data(struct grep_source *gs)
1416 {
1417         switch (gs->type) {
1418         case GREP_SOURCE_FILE:
1419         case GREP_SOURCE_SHA1:
1420                 free(gs->buf);
1421                 gs->buf = NULL;
1422                 gs->size = 0;
1423                 break;
1424         case GREP_SOURCE_BUF:
1425                 /* leave user-provided buf intact */
1426                 break;
1427         }
1428 }
1429
1430 static int grep_source_load_sha1(struct grep_source *gs)
1431 {
1432         enum object_type type;
1433
1434         grep_read_lock();
1435         gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1436         grep_read_unlock();
1437
1438         if (!gs->buf)
1439                 return error(_("'%s': unable to read %s"),
1440                              gs->name,
1441                              sha1_to_hex(gs->identifier));
1442         return 0;
1443 }
1444
1445 static int grep_source_load_file(struct grep_source *gs)
1446 {
1447         const char *filename = gs->identifier;
1448         struct stat st;
1449         char *data;
1450         size_t size;
1451         int i;
1452
1453         if (lstat(filename, &st) < 0) {
1454         err_ret:
1455                 if (errno != ENOENT)
1456                         error(_("'%s': %s"), filename, strerror(errno));
1457                 return -1;
1458         }
1459         if (!S_ISREG(st.st_mode))
1460                 return -1;
1461         size = xsize_t(st.st_size);
1462         i = open(filename, O_RDONLY);
1463         if (i < 0)
1464                 goto err_ret;
1465         data = xmalloc(size + 1);
1466         if (st.st_size != read_in_full(i, data, size)) {
1467                 error(_("'%s': short read %s"), filename, strerror(errno));
1468                 close(i);
1469                 free(data);
1470                 return -1;
1471         }
1472         close(i);
1473         data[size] = 0;
1474
1475         gs->buf = data;
1476         gs->size = size;
1477         return 0;
1478 }
1479
1480 static int grep_source_load(struct grep_source *gs)
1481 {
1482         if (gs->buf)
1483                 return 0;
1484
1485         switch (gs->type) {
1486         case GREP_SOURCE_FILE:
1487                 return grep_source_load_file(gs);
1488         case GREP_SOURCE_SHA1:
1489                 return grep_source_load_sha1(gs);
1490         case GREP_SOURCE_BUF:
1491                 return gs->buf ? 0 : -1;
1492         }
1493         die("BUG: invalid grep_source type");
1494 }
1495
1496 void grep_source_load_driver(struct grep_source *gs)
1497 {
1498         if (gs->driver)
1499                 return;
1500
1501         grep_attr_lock();
1502         gs->driver = userdiff_find_by_path(gs->name);
1503         if (!gs->driver)
1504                 gs->driver = userdiff_find_by_name("default");
1505         grep_attr_unlock();
1506 }
1507
1508 static int grep_source_is_binary(struct grep_source *gs)
1509 {
1510         grep_source_load_driver(gs);
1511         if (gs->driver->binary != -1)
1512                 return gs->driver->binary;
1513
1514         if (!grep_source_load(gs))
1515                 return buffer_is_binary(gs->buf, gs->size);
1516
1517         return 0;
1518 }