grep.c

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