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 struct grep_expr *grep_true_expr(void)
 336 {
 337         struct grep_expr *z = xcalloc(1, sizeof(*z));
 338         z->node = GREP_NODE_TRUE;
 339         return z;
 340 }
 341
 342 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 343 {
 344         struct grep_expr *z = xcalloc(1, sizeof(*z));
 345         z->node = GREP_NODE_OR;
 346         z->u.binary.left = left;
 347         z->u.binary.right = right;
 348         return z;
 349 }
 350
 351 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 352 {
 353         struct grep_pat *p;
 354         struct grep_expr *header_expr;
 355         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 356         enum grep_header_field fld;
 357
 358         if (!opt->header_list)
 359                 return NULL;
 360
 361         for (p = opt->header_list; p; p = p->next) {
 362                 if (p->token != GREP_PATTERN_HEAD)
 363                         die("bug: a non-header pattern in grep header list.");
 364                 if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
 365                         die("bug: unknown header field %d", p->field);
 366                 compile_regexp(p, opt);
 367         }
 368
 369         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 370                 header_group[fld] = NULL;
 371
 372         for (p = opt->header_list; p; p = p->next) {
 373                 struct grep_expr *h;
 374                 struct grep_pat *pp = p;
 375
 376                 h = compile_pattern_atom(&pp);
 377                 if (!h || pp != p->next)
 378                         die("bug: malformed header expr");
 379                 if (!header_group[p->field]) {
 380                         header_group[p->field] = h;
 381                         continue;
 382                 }
 383                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 384         }
 385
 386         header_expr = NULL;
 387
 388         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 389                 if (!header_group[fld])
 390                         continue;
 391                 if (!header_expr)
 392                         header_expr = grep_true_expr();
 393                 header_expr = grep_or_expr(header_group[fld], header_expr);
 394         }
 395         return header_expr;
 396 }
 397
 398 void compile_grep_patterns(struct grep_opt *opt)
 399 {
 400         struct grep_pat *p;
 401         struct grep_expr *header_expr = prep_header_patterns(opt);
 402
 403         for (p = opt->pattern_list; p; p = p->next) {
 404                 switch (p->token) {
 405                 case GREP_PATTERN: /* atom */
 406                 case GREP_PATTERN_HEAD:
 407                 case GREP_PATTERN_BODY:
 408                         compile_regexp(p, opt);
 409                         break;
 410                 default:
 411                         opt->extended = 1;
 412                         break;
 413                 }
 414         }
 415
 416         if (opt->all_match || header_expr)
 417                 opt->extended = 1;
 418         else if (!opt->extended)
 419                 return;
 420
 421         p = opt->pattern_list;
 422         if (p)
 423                 opt->pattern_expression = compile_pattern_expr(&p);
 424         if (p)
 425                 die("incomplete pattern expression: %s", p->pattern);
 426
 427         if (!header_expr)
 428                 return;
 429
 430         if (!opt->pattern_expression)
 431                 opt->pattern_expression = header_expr;
 432         else
 433                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 434                                                        header_expr);
 435         opt->all_match = 1;
 436 }
 437
 438 static void free_pattern_expr(struct grep_expr *x)
 439 {
 440         switch (x->node) {
 441         case GREP_NODE_TRUE:
 442         case GREP_NODE_ATOM:
 443                 break;
 444         case GREP_NODE_NOT:
 445                 free_pattern_expr(x->u.unary);
 446                 break;
 447         case GREP_NODE_AND:
 448         case GREP_NODE_OR:
 449                 free_pattern_expr(x->u.binary.left);
 450                 free_pattern_expr(x->u.binary.right);
 451                 break;
 452         }
 453         free(x);
 454 }
 455
 456 void free_grep_patterns(struct grep_opt *opt)
 457 {
 458         struct grep_pat *p, *n;
 459
 460         for (p = opt->pattern_list; p; p = n) {
 461                 n = p->next;
 462                 switch (p->token) {
 463                 case GREP_PATTERN: /* atom */
 464                 case GREP_PATTERN_HEAD:
 465                 case GREP_PATTERN_BODY:
 466                         if (p->kws)
 467                                 kwsfree(p->kws);
 468                         else if (p->pcre_regexp)
 469                                 free_pcre_regexp(p);
 470                         else
 471                                 regfree(&p->regexp);
 472                         free(p->pattern);
 473                         break;
 474                 default:
 475                         break;
 476                 }
 477                 free(p);
 478         }
 479
 480         if (!opt->extended)
 481                 return;
 482         free_pattern_expr(opt->pattern_expression);
 483 }
 484
 485 static char *end_of_line(char *cp, unsigned long *left)
 486 {
 487         unsigned long l = *left;
 488         while (l && *cp != '\n') {
 489                 l--;
 490                 cp++;
 491         }
 492         *left = l;
 493         return cp;
 494 }
 495
 496 static int word_char(char ch)
 497 {
 498         return isalnum(ch) || ch == '_';
 499 }
 500
 501 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 502                          const char *color)
 503 {
 504         if (want_color(opt->color) && color && color[0]) {
 505                 opt->output(opt, color, strlen(color));
 506                 opt->output(opt, data, size);
 507                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 508         } else
 509                 opt->output(opt, data, size);
 510 }
 511
 512 static void output_sep(struct grep_opt *opt, char sign)
 513 {
 514         if (opt->null_following_name)
 515                 opt->output(opt, "\0", 1);
 516         else
 517                 output_color(opt, &sign, 1, opt->color_sep);
 518 }
 519
 520 static void show_name(struct grep_opt *opt, const char *name)
 521 {
 522         output_color(opt, name, strlen(name), opt->color_filename);
 523         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 524 }
 525
 526 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 527                     regmatch_t *match)
 528 {
 529         struct kwsmatch kwsm;
 530         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 531         if (offset == -1) {
 532                 match->rm_so = match->rm_eo = -1;
 533                 return REG_NOMATCH;
 534         } else {
 535                 match->rm_so = offset;
 536                 match->rm_eo = match->rm_so + kwsm.size[0];
 537                 return 0;
 538         }
 539 }
 540
 541 static int regmatch(const regex_t *preg, char *line, char *eol,
 542                     regmatch_t *match, int eflags)
 543 {
 544 #ifdef REG_STARTEND
 545         match->rm_so = 0;
 546         match->rm_eo = eol - line;
 547         eflags |= REG_STARTEND;
 548 #endif
 549         return regexec(preg, line, 1, match, eflags);
 550 }
 551
 552 static int patmatch(struct grep_pat *p, char *line, char *eol,
 553                     regmatch_t *match, int eflags)
 554 {
 555         int hit;
 556
 557         if (p->fixed)
 558                 hit = !fixmatch(p, line, eol, match);
 559         else if (p->pcre_regexp)
 560                 hit = !pcrematch(p, line, eol, match, eflags);
 561         else
 562                 hit = !regmatch(&p->regexp, line, eol, match, eflags);
 563
 564         return hit;
 565 }
 566
 567 static int strip_timestamp(char *bol, char **eol_p)
 568 {
 569         char *eol = *eol_p;
 570         int ch;
 571
 572         while (bol < --eol) {
 573                 if (*eol != '>')
 574                         continue;
 575                 *eol_p = ++eol;
 576                 ch = *eol;
 577                 *eol = '\0';
 578                 return ch;
 579         }
 580         return 0;
 581 }
 582
 583 static struct {
 584         const char *field;
 585         size_t len;
 586 } header_field[] = {
 587         { "author ", 7 },
 588         { "committer ", 10 },
 589 };
 590
 591 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 592                              enum grep_context ctx,
 593                              regmatch_t *pmatch, int eflags)
 594 {
 595         int hit = 0;
 596         int saved_ch = 0;
 597         const char *start = bol;
 598
 599         if ((p->token != GREP_PATTERN) &&
 600             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 601                 return 0;
 602
 603         if (p->token == GREP_PATTERN_HEAD) {
 604                 const char *field;
 605                 size_t len;
 606                 assert(p->field < ARRAY_SIZE(header_field));
 607                 field = header_field[p->field].field;
 608                 len = header_field[p->field].len;
 609                 if (strncmp(bol, field, len))
 610                         return 0;
 611                 bol += len;
 612                 saved_ch = strip_timestamp(bol, &eol);
 613         }
 614
 615  again:
 616         hit = patmatch(p, bol, eol, pmatch, eflags);
 617
 618         if (hit && p->word_regexp) {
 619                 if ((pmatch[0].rm_so < 0) ||
 620                     (eol - bol) < pmatch[0].rm_so ||
 621                     (pmatch[0].rm_eo < 0) ||
 622                     (eol - bol) < pmatch[0].rm_eo)
 623                         die("regexp returned nonsense");
 624
 625                 /* Match beginning must be either beginning of the
 626                  * line, or at word boundary (i.e. the last char must
 627                  * not be a word char).  Similarly, match end must be
 628                  * either end of the line, or at word boundary
 629                  * (i.e. the next char must not be a word char).
 630                  */
 631                 if ( ((pmatch[0].rm_so == 0) ||
 632                       !word_char(bol[pmatch[0].rm_so-1])) &&
 633                      ((pmatch[0].rm_eo == (eol-bol)) ||
 634                       !word_char(bol[pmatch[0].rm_eo])) )
 635                         ;
 636                 else
 637                         hit = 0;
 638
 639                 /* Words consist of at least one character. */
 640                 if (pmatch->rm_so == pmatch->rm_eo)
 641                         hit = 0;
 642
 643                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 644                         /* There could be more than one match on the
 645                          * line, and the first match might not be
 646                          * strict word match.  But later ones could be!
 647                          * Forward to the next possible start, i.e. the
 648                          * next position following a non-word char.
 649                          */
 650                         bol = pmatch[0].rm_so + bol + 1;
 651                         while (word_char(bol[-1]) && bol < eol)
 652                                 bol++;
 653                         eflags |= REG_NOTBOL;
 654                         if (bol < eol)
 655                                 goto again;
 656                 }
 657         }
 658         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 659                 *eol = saved_ch;
 660         if (hit) {
 661                 pmatch[0].rm_so += bol - start;
 662                 pmatch[0].rm_eo += bol - start;
 663         }
 664         return hit;
 665 }
 666
 667 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 668                            enum grep_context ctx, int collect_hits)
 669 {
 670         int h = 0;
 671         regmatch_t match;
 672
 673         if (!x)
 674                 die("Not a valid grep expression");
 675         switch (x->node) {
 676         case GREP_NODE_TRUE:
 677                 h = 1;
 678                 break;
 679         case GREP_NODE_ATOM:
 680                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 681                 break;
 682         case GREP_NODE_NOT:
 683                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 684                 break;
 685         case GREP_NODE_AND:
 686                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 687                         return 0;
 688                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 689                 break;
 690         case GREP_NODE_OR:
 691                 if (!collect_hits)
 692                         return (match_expr_eval(x->u.binary.left,
 693                                                 bol, eol, ctx, 0) ||
 694                                 match_expr_eval(x->u.binary.right,
 695                                                 bol, eol, ctx, 0));
 696                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 697                 x->u.binary.left->hit |= h;
 698                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 699                 break;
 700         default:
 701                 die("Unexpected node type (internal error) %d", x->node);
 702         }
 703         if (collect_hits)
 704                 x->hit |= h;
 705         return h;
 706 }
 707
 708 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 709                       enum grep_context ctx, int collect_hits)
 710 {
 711         struct grep_expr *x = opt->pattern_expression;
 712         return match_expr_eval(x, bol, eol, ctx, collect_hits);
 713 }
 714
 715 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 716                       enum grep_context ctx, int collect_hits)
 717 {
 718         struct grep_pat *p;
 719         regmatch_t match;
 720
 721         if (opt->extended)
 722                 return match_expr(opt, bol, eol, ctx, collect_hits);
 723
 724         /* we do not call with collect_hits without being extended */
 725         for (p = opt->pattern_list; p; p = p->next) {
 726                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 727                         return 1;
 728         }
 729         return 0;
 730 }
 731
 732 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 733                               enum grep_context ctx,
 734                               regmatch_t *pmatch, int eflags)
 735 {
 736         regmatch_t match;
 737
 738         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 739                 return 0;
 740         if (match.rm_so < 0 || match.rm_eo < 0)
 741                 return 0;
 742         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 743                 if (match.rm_so > pmatch->rm_so)
 744                         return 1;
 745                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 746                         return 1;
 747         }
 748         pmatch->rm_so = match.rm_so;
 749         pmatch->rm_eo = match.rm_eo;
 750         return 1;
 751 }
 752
 753 static int next_match(struct grep_opt *opt, char *bol, char *eol,
 754                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
 755 {
 756         struct grep_pat *p;
 757         int hit = 0;
 758
 759         pmatch->rm_so = pmatch->rm_eo = -1;
 760         if (bol < eol) {
 761                 for (p = opt->pattern_list; p; p = p->next) {
 762                         switch (p->token) {
 763                         case GREP_PATTERN: /* atom */
 764                         case GREP_PATTERN_HEAD:
 765                         case GREP_PATTERN_BODY:
 766                                 hit |= match_next_pattern(p, bol, eol, ctx,
 767                                                           pmatch, eflags);
 768                                 break;
 769                         default:
 770                                 break;
 771                         }
 772                 }
 773         }
 774         return hit;
 775 }
 776
 777 static void show_line(struct grep_opt *opt, char *bol, char *eol,
 778                       const char *name, unsigned lno, char sign)
 779 {
 780         int rest = eol - bol;
 781         char *line_color = NULL;
 782
 783         if (opt->file_break && opt->last_shown == 0) {
 784                 if (opt->show_hunk_mark)
 785                         opt->output(opt, "\n", 1);
 786         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
 787                 if (opt->last_shown == 0) {
 788                         if (opt->show_hunk_mark) {
 789                                 output_color(opt, "--", 2, opt->color_sep);
 790                                 opt->output(opt, "\n", 1);
 791                         }
 792                 } else if (lno > opt->last_shown + 1) {
 793                         output_color(opt, "--", 2, opt->color_sep);
 794                         opt->output(opt, "\n", 1);
 795                 }
 796         }
 797         if (opt->heading && opt->last_shown == 0) {
 798                 output_color(opt, name, strlen(name), opt->color_filename);
 799                 opt->output(opt, "\n", 1);
 800         }
 801         opt->last_shown = lno;
 802
 803         if (!opt->heading && opt->pathname) {
 804                 output_color(opt, name, strlen(name), opt->color_filename);
 805                 output_sep(opt, sign);
 806         }
 807         if (opt->linenum) {
 808                 char buf[32];
 809                 snprintf(buf, sizeof(buf), "%d", lno);
 810                 output_color(opt, buf, strlen(buf), opt->color_lineno);
 811                 output_sep(opt, sign);
 812         }
 813         if (opt->color) {
 814                 regmatch_t match;
 815                 enum grep_context ctx = GREP_CONTEXT_BODY;
 816                 int ch = *eol;
 817                 int eflags = 0;
 818
 819                 if (sign == ':')
 820                         line_color = opt->color_selected;
 821                 else if (sign == '-')
 822                         line_color = opt->color_context;
 823                 else if (sign == '=')
 824                         line_color = opt->color_function;
 825                 *eol = '\0';
 826                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
 827                         if (match.rm_so == match.rm_eo)
 828                                 break;
 829
 830                         output_color(opt, bol, match.rm_so, line_color);
 831                         output_color(opt, bol + match.rm_so,
 832                                      match.rm_eo - match.rm_so,
 833                                      opt->color_match);
 834                         bol += match.rm_eo;
 835                         rest -= match.rm_eo;
 836                         eflags = REG_NOTBOL;
 837                 }
 838                 *eol = ch;
 839         }
 840         output_color(opt, bol, rest, line_color);
 841         opt->output(opt, "\n", 1);
 842 }
 843
 844 #ifndef NO_PTHREADS
 845 int grep_use_locks;
 846
 847 /*
 848  * This lock protects access to the gitattributes machinery, which is
 849  * not thread-safe.
 850  */
 851 pthread_mutex_t grep_attr_mutex;
 852
 853 static inline void grep_attr_lock(void)
 854 {
 855         if (grep_use_locks)
 856                 pthread_mutex_lock(&grep_attr_mutex);
 857 }
 858
 859 static inline void grep_attr_unlock(void)
 860 {
 861         if (grep_use_locks)
 862                 pthread_mutex_unlock(&grep_attr_mutex);
 863 }
 864
 865 /*
 866  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
 867  */
 868 pthread_mutex_t grep_read_mutex;
 869
 870 #else
 871 #define grep_attr_lock()
 872 #define grep_attr_unlock()
 873 #endif
 874
 875 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
 876 {
 877         xdemitconf_t *xecfg = opt->priv;
 878         if (xecfg && !xecfg->find_func) {
 879                 grep_source_load_driver(gs);
 880                 if (gs->driver->funcname.pattern) {
 881                         const struct userdiff_funcname *pe = &gs->driver->funcname;
 882                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
 883                 } else {
 884                         xecfg = opt->priv = NULL;
 885                 }
 886         }
 887
 888         if (xecfg) {
 889                 char buf[1];
 890                 return xecfg->find_func(bol, eol - bol, buf, 1,
 891                                         xecfg->find_func_priv) >= 0;
 892         }
 893
 894         if (bol == eol)
 895                 return 0;
 896         if (isalpha(*bol) || *bol == '_' || *bol == '$')
 897                 return 1;
 898         return 0;
 899 }
 900
 901 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
 902                                char *bol, unsigned lno)
 903 {
 904         while (bol > gs->buf) {
 905                 char *eol = --bol;
 906
 907                 while (bol > gs->buf && bol[-1] != '\n')
 908                         bol--;
 909                 lno--;
 910
 911                 if (lno <= opt->last_shown)
 912                         break;
 913
 914                 if (match_funcname(opt, gs, bol, eol)) {
 915                         show_line(opt, bol, eol, gs->name, lno, '=');
 916                         break;
 917                 }
 918         }
 919 }
 920
 921 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
 922                              char *bol, char *end, unsigned lno)
 923 {
 924         unsigned cur = lno, from = 1, funcname_lno = 0;
 925         int funcname_needed = !!opt->funcname;
 926
 927         if (opt->funcbody && !match_funcname(opt, gs, bol, end))
 928                 funcname_needed = 2;
 929
 930         if (opt->pre_context < lno)
 931                 from = lno - opt->pre_context;
 932         if (from <= opt->last_shown)
 933                 from = opt->last_shown + 1;
 934
 935         /* Rewind. */
 936         while (bol > gs->buf &&
 937                cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
 938                 char *eol = --bol;
 939
 940                 while (bol > gs->buf && bol[-1] != '\n')
 941                         bol--;
 942                 cur--;
 943                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
 944                         funcname_lno = cur;
 945                         funcname_needed = 0;
 946                 }
 947         }
 948
 949         /* We need to look even further back to find a function signature. */
 950         if (opt->funcname && funcname_needed)
 951                 show_funcname_line(opt, gs, bol, cur);
 952
 953         /* Back forward. */
 954         while (cur < lno) {
 955                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
 956
 957                 while (*eol != '\n')
 958                         eol++;
 959                 show_line(opt, bol, eol, gs->name, cur, sign);
 960                 bol = eol + 1;
 961                 cur++;
 962         }
 963 }
 964
 965 static int should_lookahead(struct grep_opt *opt)
 966 {
 967         struct grep_pat *p;
 968
 969         if (opt->extended)
 970                 return 0; /* punt for too complex stuff */
 971         if (opt->invert)
 972                 return 0;
 973         for (p = opt->pattern_list; p; p = p->next) {
 974                 if (p->token != GREP_PATTERN)
 975                         return 0; /* punt for "header only" and stuff */
 976         }
 977         return 1;
 978 }
 979
 980 static int look_ahead(struct grep_opt *opt,
 981                       unsigned long *left_p,
 982                       unsigned *lno_p,
 983                       char **bol_p)
 984 {
 985         unsigned lno = *lno_p;
 986         char *bol = *bol_p;
 987         struct grep_pat *p;
 988         char *sp, *last_bol;
 989         regoff_t earliest = -1;
 990
 991         for (p = opt->pattern_list; p; p = p->next) {
 992                 int hit;
 993                 regmatch_t m;
 994
 995                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
 996                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
 997                         continue;
 998                 if (earliest < 0 || m.rm_so < earliest)
 999                         earliest = m.rm_so;
1000         }
1001
1002         if (earliest < 0) {
1003                 *bol_p = bol + *left_p;
1004                 *left_p = 0;
1005                 return 1;
1006         }
1007         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1008                 ; /* find the beginning of the line */
1009         last_bol = sp;
1010
1011         for (sp = bol; sp < last_bol; sp++) {
1012                 if (*sp == '\n')
1013                         lno++;
1014         }
1015         *left_p -= last_bol - bol;
1016         *bol_p = last_bol;
1017         *lno_p = lno;
1018         return 0;
1019 }
1020
1021 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1022 {
1023         fwrite(buf, size, 1, stdout);
1024 }
1025
1026 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1027 {
1028         char *bol;
1029         unsigned long left;
1030         unsigned lno = 1;
1031         unsigned last_hit = 0;
1032         int binary_match_only = 0;
1033         unsigned count = 0;
1034         int try_lookahead = 0;
1035         int show_function = 0;
1036         enum grep_context ctx = GREP_CONTEXT_HEAD;
1037         xdemitconf_t xecfg;
1038
1039         if (!opt->output)
1040                 opt->output = std_output;
1041
1042         if (opt->pre_context || opt->post_context || opt->file_break ||
1043             opt->funcbody) {
1044                 /* Show hunk marks, except for the first file. */
1045                 if (opt->last_shown)
1046                         opt->show_hunk_mark = 1;
1047                 /*
1048                  * If we're using threads then we can't easily identify
1049                  * the first file.  Always put hunk marks in that case
1050                  * and skip the very first one later in work_done().
1051                  */
1052                 if (opt->output != std_output)
1053                         opt->show_hunk_mark = 1;
1054         }
1055         opt->last_shown = 0;
1056
1057         switch (opt->binary) {
1058         case GREP_BINARY_DEFAULT:
1059                 if (grep_source_is_binary(gs))
1060                         binary_match_only = 1;
1061                 break;
1062         case GREP_BINARY_NOMATCH:
1063                 if (grep_source_is_binary(gs))
1064                         return 0; /* Assume unmatch */
1065                 break;
1066         case GREP_BINARY_TEXT:
1067                 break;
1068         default:
1069                 die("bug: unknown binary handling mode");
1070         }
1071
1072         memset(&xecfg, 0, sizeof(xecfg));
1073         opt->priv = &xecfg;
1074
1075         try_lookahead = should_lookahead(opt);
1076
1077         if (grep_source_load(gs) < 0)
1078                 return 0;
1079
1080         bol = gs->buf;
1081         left = gs->size;
1082         while (left) {
1083                 char *eol, ch;
1084                 int hit;
1085
1086                 /*
1087                  * look_ahead() skips quickly to the line that possibly
1088                  * has the next hit; don't call it if we need to do
1089                  * something more than just skipping the current line
1090                  * in response to an unmatch for the current line.  E.g.
1091                  * inside a post-context window, we will show the current
1092                  * line as a context around the previous hit when it
1093                  * doesn't hit.
1094                  */
1095                 if (try_lookahead
1096                     && !(last_hit
1097                          && (show_function ||
1098                              lno <= last_hit + opt->post_context))
1099                     && look_ahead(opt, &left, &lno, &bol))
1100                         break;
1101                 eol = end_of_line(bol, &left);
1102                 ch = *eol;
1103                 *eol = 0;
1104
1105                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1106                         ctx = GREP_CONTEXT_BODY;
1107
1108                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1109                 *eol = ch;
1110
1111                 if (collect_hits)
1112                         goto next_line;
1113
1114                 /* "grep -v -e foo -e bla" should list lines
1115                  * that do not have either, so inversion should
1116                  * be done outside.
1117                  */
1118                 if (opt->invert)
1119                         hit = !hit;
1120                 if (opt->unmatch_name_only) {
1121                         if (hit)
1122                                 return 0;
1123                         goto next_line;
1124                 }
1125                 if (hit) {
1126                         count++;
1127                         if (opt->status_only)
1128                                 return 1;
1129                         if (opt->name_only) {
1130                                 show_name(opt, gs->name);
1131                                 return 1;
1132                         }
1133                         if (opt->count)
1134                                 goto next_line;
1135                         if (binary_match_only) {
1136                                 opt->output(opt, "Binary file ", 12);
1137                                 output_color(opt, gs->name, strlen(gs->name),
1138                                              opt->color_filename);
1139                                 opt->output(opt, " matches\n", 9);
1140                                 return 1;
1141                         }
1142                         /* Hit at this line.  If we haven't shown the
1143                          * pre-context lines, we would need to show them.
1144                          */
1145                         if (opt->pre_context || opt->funcbody)
1146                                 show_pre_context(opt, gs, bol, eol, lno);
1147                         else if (opt->funcname)
1148                                 show_funcname_line(opt, gs, bol, lno);
1149                         show_line(opt, bol, eol, gs->name, lno, ':');
1150                         last_hit = lno;
1151                         if (opt->funcbody)
1152                                 show_function = 1;
1153                         goto next_line;
1154                 }
1155                 if (show_function && match_funcname(opt, gs, bol, eol))
1156                         show_function = 0;
1157                 if (show_function ||
1158                     (last_hit && lno <= last_hit + opt->post_context)) {
1159                         /* If the last hit is within the post context,
1160                          * we need to show this line.
1161                          */
1162                         show_line(opt, bol, eol, gs->name, lno, '-');
1163                 }
1164
1165         next_line:
1166                 bol = eol + 1;
1167                 if (!left)
1168                         break;
1169                 left--;
1170                 lno++;
1171         }
1172
1173         if (collect_hits)
1174                 return 0;
1175
1176         if (opt->status_only)
1177                 return 0;
1178         if (opt->unmatch_name_only) {
1179                 /* We did not see any hit, so we want to show this */
1180                 show_name(opt, gs->name);
1181                 return 1;
1182         }
1183
1184         xdiff_clear_find_func(&xecfg);
1185         opt->priv = NULL;
1186
1187         /* NEEDSWORK:
1188          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1189          * which feels mostly useless but sometimes useful.  Maybe
1190          * make it another option?  For now suppress them.
1191          */
1192         if (opt->count && count) {
1193                 char buf[32];
1194                 output_color(opt, gs->name, strlen(gs->name), opt->color_filename);
1195                 output_sep(opt, ':');
1196                 snprintf(buf, sizeof(buf), "%u\n", count);
1197                 opt->output(opt, buf, strlen(buf));
1198                 return 1;
1199         }
1200         return !!last_hit;
1201 }
1202
1203 static void clr_hit_marker(struct grep_expr *x)
1204 {
1205         /* All-hit markers are meaningful only at the very top level
1206          * OR node.
1207          */
1208         while (1) {
1209                 x->hit = 0;
1210                 if (x->node != GREP_NODE_OR)
1211                         return;
1212                 x->u.binary.left->hit = 0;
1213                 x = x->u.binary.right;
1214         }
1215 }
1216
1217 static int chk_hit_marker(struct grep_expr *x)
1218 {
1219         /* Top level nodes have hit markers.  See if they all are hits */
1220         while (1) {
1221                 if (x->node != GREP_NODE_OR)
1222                         return x->hit;
1223                 if (!x->u.binary.left->hit)
1224                         return 0;
1225                 x = x->u.binary.right;
1226         }
1227 }
1228
1229 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1230 {
1231         /*
1232          * we do not have to do the two-pass grep when we do not check
1233          * buffer-wide "all-match".
1234          */
1235         if (!opt->all_match)
1236                 return grep_source_1(opt, gs, 0);
1237
1238         /* Otherwise the toplevel "or" terms hit a bit differently.
1239          * We first clear hit markers from them.
1240          */
1241         clr_hit_marker(opt->pattern_expression);
1242         grep_source_1(opt, gs, 1);
1243
1244         if (!chk_hit_marker(opt->pattern_expression))
1245                 return 0;
1246
1247         return grep_source_1(opt, gs, 0);
1248 }
1249
1250 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1251 {
1252         struct grep_source gs;
1253         int r;
1254
1255         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL);
1256         gs.buf = buf;
1257         gs.size = size;
1258
1259         r = grep_source(opt, &gs);
1260
1261         grep_source_clear(&gs);
1262         return r;
1263 }
1264
1265 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1266                       const char *name, const void *identifier)
1267 {
1268         gs->type = type;
1269         gs->name = name ? xstrdup(name) : NULL;
1270         gs->buf = NULL;
1271         gs->size = 0;
1272         gs->driver = NULL;
1273
1274         switch (type) {
1275         case GREP_SOURCE_FILE:
1276                 gs->identifier = xstrdup(identifier);
1277                 break;
1278         case GREP_SOURCE_SHA1:
1279                 gs->identifier = xmalloc(20);
1280                 memcpy(gs->identifier, identifier, 20);
1281                 break;
1282         case GREP_SOURCE_BUF:
1283                 gs->identifier = NULL;
1284         }
1285 }
1286
1287 void grep_source_clear(struct grep_source *gs)
1288 {
1289         free(gs->name);
1290         gs->name = NULL;
1291         free(gs->identifier);
1292         gs->identifier = NULL;
1293         grep_source_clear_data(gs);
1294 }
1295
1296 void grep_source_clear_data(struct grep_source *gs)
1297 {
1298         switch (gs->type) {
1299         case GREP_SOURCE_FILE:
1300         case GREP_SOURCE_SHA1:
1301                 free(gs->buf);
1302                 gs->buf = NULL;
1303                 gs->size = 0;
1304                 break;
1305         case GREP_SOURCE_BUF:
1306                 /* leave user-provided buf intact */
1307                 break;
1308         }
1309 }
1310
1311 static int grep_source_load_sha1(struct grep_source *gs)
1312 {
1313         enum object_type type;
1314
1315         grep_read_lock();
1316         gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1317         grep_read_unlock();
1318
1319         if (!gs->buf)
1320                 return error(_("'%s': unable to read %s"),
1321                              gs->name,
1322                              sha1_to_hex(gs->identifier));
1323         return 0;
1324 }
1325
1326 static int grep_source_load_file(struct grep_source *gs)
1327 {
1328         const char *filename = gs->identifier;
1329         struct stat st;
1330         char *data;
1331         size_t size;
1332         int i;
1333
1334         if (lstat(filename, &st) < 0) {
1335         err_ret:
1336                 if (errno != ENOENT)
1337                         error(_("'%s': %s"), filename, strerror(errno));
1338                 return -1;
1339         }
1340         if (!S_ISREG(st.st_mode))
1341                 return -1;
1342         size = xsize_t(st.st_size);
1343         i = open(filename, O_RDONLY);
1344         if (i < 0)
1345                 goto err_ret;
1346         data = xmalloc(size + 1);
1347         if (st.st_size != read_in_full(i, data, size)) {
1348                 error(_("'%s': short read %s"), filename, strerror(errno));
1349                 close(i);
1350                 free(data);
1351                 return -1;
1352         }
1353         close(i);
1354         data[size] = 0;
1355
1356         gs->buf = data;
1357         gs->size = size;
1358         return 0;
1359 }
1360
1361 int grep_source_load(struct grep_source *gs)
1362 {
1363         if (gs->buf)
1364                 return 0;
1365
1366         switch (gs->type) {
1367         case GREP_SOURCE_FILE:
1368                 return grep_source_load_file(gs);
1369         case GREP_SOURCE_SHA1:
1370                 return grep_source_load_sha1(gs);
1371         case GREP_SOURCE_BUF:
1372                 return gs->buf ? 0 : -1;
1373         }
1374         die("BUG: invalid grep_source type");
1375 }
1376
1377 void grep_source_load_driver(struct grep_source *gs)
1378 {
1379         if (gs->driver)
1380                 return;
1381
1382         grep_attr_lock();
1383         gs->driver = userdiff_find_by_path(gs->name);
1384         if (!gs->driver)
1385                 gs->driver = userdiff_find_by_name("default");
1386         grep_attr_unlock();
1387 }
1388
1389 int grep_source_is_binary(struct grep_source *gs)
1390 {
1391         grep_source_load_driver(gs);
1392         if (gs->driver->binary != -1)
1393                 return gs->driver->binary;
1394
1395         if (!grep_source_load(gs))
1396                 return buffer_is_binary(gs->buf, gs->size);
1397
1398         return 0;
1399 }