grep.c

   1 #include "cache.h"
   2 #include "grep.h"
   3 #include "userdiff.h"
   4 #include "xdiff-interface.h"
   5
   6 void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat)
   7 {
   8         struct grep_pat *p = xcalloc(1, sizeof(*p));
   9         p->pattern = pat;
  10         p->patternlen = strlen(pat);
  11         p->origin = "header";
  12         p->no = 0;
  13         p->token = GREP_PATTERN_HEAD;
  14         p->field = field;
  15         *opt->header_tail = p;
  16         opt->header_tail = &p->next;
  17         p->next = NULL;
  18 }
  19
  20 void append_grep_pattern(struct grep_opt *opt, const char *pat,
  21                          const char *origin, int no, enum grep_pat_token t)
  22 {
  23         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
  24 }
  25
  26 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
  27                      const char *origin, int no, enum grep_pat_token t)
  28 {
  29         struct grep_pat *p = xcalloc(1, sizeof(*p));
  30         p->pattern = pat;
  31         p->patternlen = patlen;
  32         p->origin = origin;
  33         p->no = no;
  34         p->token = t;
  35         *opt->pattern_tail = p;
  36         opt->pattern_tail = &p->next;
  37         p->next = NULL;
  38 }
  39
  40 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
  41 {
  42         struct grep_pat *pat;
  43         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
  44         *ret = *opt;
  45
  46         ret->pattern_list = NULL;
  47         ret->pattern_tail = &ret->pattern_list;
  48
  49         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
  50         {
  51                 if(pat->token == GREP_PATTERN_HEAD)
  52                         append_header_grep_pattern(ret, pat->field,
  53                                                    pat->pattern);
  54                 else
  55                         append_grep_pat(ret, pat->pattern, pat->patternlen,
  56                                         pat->origin, pat->no, pat->token);
  57         }
  58
  59         return ret;
  60 }
  61
  62 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
  63 {
  64         int err;
  65
  66         p->word_regexp = opt->word_regexp;
  67         p->ignore_case = opt->ignore_case;
  68         p->fixed = opt->fixed;
  69
  70         if (p->fixed)
  71                 return;
  72
  73         err = regcomp(&p->regexp, p->pattern, opt->regflags);
  74         if (err) {
  75                 char errbuf[1024];
  76                 char where[1024];
  77                 if (p->no)
  78                         sprintf(where, "In '%s' at %d, ",
  79                                 p->origin, p->no);
  80                 else if (p->origin)
  81                         sprintf(where, "%s, ", p->origin);
  82                 else
  83                         where[0] = 0;
  84                 regerror(err, &p->regexp, errbuf, 1024);
  85                 regfree(&p->regexp);
  86                 die("%s'%s': %s", where, p->pattern, errbuf);
  87         }
  88 }
  89
  90 static struct grep_expr *compile_pattern_or(struct grep_pat **);
  91 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
  92 {
  93         struct grep_pat *p;
  94         struct grep_expr *x;
  95
  96         p = *list;
  97         if (!p)
  98                 return NULL;
  99         switch (p->token) {
 100         case GREP_PATTERN: /* atom */
 101         case GREP_PATTERN_HEAD:
 102         case GREP_PATTERN_BODY:
 103                 x = xcalloc(1, sizeof (struct grep_expr));
 104                 x->node = GREP_NODE_ATOM;
 105                 x->u.atom = p;
 106                 *list = p->next;
 107                 return x;
 108         case GREP_OPEN_PAREN:
 109                 *list = p->next;
 110                 x = compile_pattern_or(list);
 111                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 112                         die("unmatched parenthesis");
 113                 *list = (*list)->next;
 114                 return x;
 115         default:
 116                 return NULL;
 117         }
 118 }
 119
 120 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 121 {
 122         struct grep_pat *p;
 123         struct grep_expr *x;
 124
 125         p = *list;
 126         if (!p)
 127                 return NULL;
 128         switch (p->token) {
 129         case GREP_NOT:
 130                 if (!p->next)
 131                         die("--not not followed by pattern expression");
 132                 *list = p->next;
 133                 x = xcalloc(1, sizeof (struct grep_expr));
 134                 x->node = GREP_NODE_NOT;
 135                 x->u.unary = compile_pattern_not(list);
 136                 if (!x->u.unary)
 137                         die("--not followed by non pattern expression");
 138                 return x;
 139         default:
 140                 return compile_pattern_atom(list);
 141         }
 142 }
 143
 144 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 145 {
 146         struct grep_pat *p;
 147         struct grep_expr *x, *y, *z;
 148
 149         x = compile_pattern_not(list);
 150         p = *list;
 151         if (p && p->token == GREP_AND) {
 152                 if (!p->next)
 153                         die("--and not followed by pattern expression");
 154                 *list = p->next;
 155                 y = compile_pattern_and(list);
 156                 if (!y)
 157                         die("--and not followed by pattern expression");
 158                 z = xcalloc(1, sizeof (struct grep_expr));
 159                 z->node = GREP_NODE_AND;
 160                 z->u.binary.left = x;
 161                 z->u.binary.right = y;
 162                 return z;
 163         }
 164         return x;
 165 }
 166
 167 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 168 {
 169         struct grep_pat *p;
 170         struct grep_expr *x, *y, *z;
 171
 172         x = compile_pattern_and(list);
 173         p = *list;
 174         if (x && p && p->token != GREP_CLOSE_PAREN) {
 175                 y = compile_pattern_or(list);
 176                 if (!y)
 177                         die("not a pattern expression %s", p->pattern);
 178                 z = xcalloc(1, sizeof (struct grep_expr));
 179                 z->node = GREP_NODE_OR;
 180                 z->u.binary.left = x;
 181                 z->u.binary.right = y;
 182                 return z;
 183         }
 184         return x;
 185 }
 186
 187 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 188 {
 189         return compile_pattern_or(list);
 190 }
 191
 192 void compile_grep_patterns(struct grep_opt *opt)
 193 {
 194         struct grep_pat *p;
 195         struct grep_expr *header_expr = NULL;
 196
 197         if (opt->header_list) {
 198                 p = opt->header_list;
 199                 header_expr = compile_pattern_expr(&p);
 200                 if (p)
 201                         die("incomplete pattern expression: %s", p->pattern);
 202                 for (p = opt->header_list; p; p = p->next) {
 203                         switch (p->token) {
 204                         case GREP_PATTERN: /* atom */
 205                         case GREP_PATTERN_HEAD:
 206                         case GREP_PATTERN_BODY:
 207                                 compile_regexp(p, opt);
 208                                 break;
 209                         default:
 210                                 opt->extended = 1;
 211                                 break;
 212                         }
 213                 }
 214         }
 215
 216         for (p = opt->pattern_list; p; p = p->next) {
 217                 switch (p->token) {
 218                 case GREP_PATTERN: /* atom */
 219                 case GREP_PATTERN_HEAD:
 220                 case GREP_PATTERN_BODY:
 221                         compile_regexp(p, opt);
 222                         break;
 223                 default:
 224                         opt->extended = 1;
 225                         break;
 226                 }
 227         }
 228
 229         if (opt->all_match || header_expr)
 230                 opt->extended = 1;
 231         else if (!opt->extended)
 232                 return;
 233
 234         /* Then bundle them up in an expression.
 235          * A classic recursive descent parser would do.
 236          */
 237         p = opt->pattern_list;
 238         if (p)
 239                 opt->pattern_expression = compile_pattern_expr(&p);
 240         if (p)
 241                 die("incomplete pattern expression: %s", p->pattern);
 242
 243         if (!header_expr)
 244                 return;
 245
 246         if (opt->pattern_expression) {
 247                 struct grep_expr *z;
 248                 z = xcalloc(1, sizeof(*z));
 249                 z->node = GREP_NODE_OR;
 250                 z->u.binary.left = opt->pattern_expression;
 251                 z->u.binary.right = header_expr;
 252                 opt->pattern_expression = z;
 253         } else {
 254                 opt->pattern_expression = header_expr;
 255         }
 256         opt->all_match = 1;
 257 }
 258
 259 static void free_pattern_expr(struct grep_expr *x)
 260 {
 261         switch (x->node) {
 262         case GREP_NODE_ATOM:
 263                 break;
 264         case GREP_NODE_NOT:
 265                 free_pattern_expr(x->u.unary);
 266                 break;
 267         case GREP_NODE_AND:
 268         case GREP_NODE_OR:
 269                 free_pattern_expr(x->u.binary.left);
 270                 free_pattern_expr(x->u.binary.right);
 271                 break;
 272         }
 273         free(x);
 274 }
 275
 276 void free_grep_patterns(struct grep_opt *opt)
 277 {
 278         struct grep_pat *p, *n;
 279
 280         for (p = opt->pattern_list; p; p = n) {
 281                 n = p->next;
 282                 switch (p->token) {
 283                 case GREP_PATTERN: /* atom */
 284                 case GREP_PATTERN_HEAD:
 285                 case GREP_PATTERN_BODY:
 286                         regfree(&p->regexp);
 287                         break;
 288                 default:
 289                         break;
 290                 }
 291                 free(p);
 292         }
 293
 294         if (!opt->extended)
 295                 return;
 296         free_pattern_expr(opt->pattern_expression);
 297 }
 298
 299 static char *end_of_line(char *cp, unsigned long *left)
 300 {
 301         unsigned long l = *left;
 302         while (l && *cp != '\n') {
 303                 l--;
 304                 cp++;
 305         }
 306         *left = l;
 307         return cp;
 308 }
 309
 310 static int word_char(char ch)
 311 {
 312         return isalnum(ch) || ch == '_';
 313 }
 314
 315 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 316                          const char *color)
 317 {
 318         if (opt->color && color && color[0]) {
 319                 opt->output(opt, color, strlen(color));
 320                 opt->output(opt, data, size);
 321                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 322         } else
 323                 opt->output(opt, data, size);
 324 }
 325
 326 static void output_sep(struct grep_opt *opt, char sign)
 327 {
 328         if (opt->null_following_name)
 329                 opt->output(opt, "\0", 1);
 330         else
 331                 output_color(opt, &sign, 1, opt->color_sep);
 332 }
 333
 334 static void show_name(struct grep_opt *opt, const char *name)
 335 {
 336         output_color(opt, name, strlen(name), opt->color_filename);
 337         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 338 }
 339
 340 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 341                     regmatch_t *match)
 342 {
 343         char *hit;
 344
 345         if (p->ignore_case) {
 346                 char *s = line;
 347                 do {
 348                         hit = strcasestr(s, p->pattern);
 349                         if (hit)
 350                                 break;
 351                         s += strlen(s) + 1;
 352                 } while (s < eol);
 353         } else
 354                 hit = memmem(line, eol - line, p->pattern, p->patternlen);
 355
 356         if (!hit) {
 357                 match->rm_so = match->rm_eo = -1;
 358                 return REG_NOMATCH;
 359         }
 360         else {
 361                 match->rm_so = hit - line;
 362                 match->rm_eo = match->rm_so + p->patternlen;
 363                 return 0;
 364         }
 365 }
 366
 367 static int regmatch(const regex_t *preg, char *line, char *eol,
 368                     regmatch_t *match, int eflags)
 369 {
 370 #ifdef REG_STARTEND
 371         match->rm_so = 0;
 372         match->rm_eo = eol - line;
 373         eflags |= REG_STARTEND;
 374 #endif
 375         return regexec(preg, line, 1, match, eflags);
 376 }
 377
 378 static int strip_timestamp(char *bol, char **eol_p)
 379 {
 380         char *eol = *eol_p;
 381         int ch;
 382
 383         while (bol < --eol) {
 384                 if (*eol != '>')
 385                         continue;
 386                 *eol_p = ++eol;
 387                 ch = *eol;
 388                 *eol = '\0';
 389                 return ch;
 390         }
 391         return 0;
 392 }
 393
 394 static struct {
 395         const char *field;
 396         size_t len;
 397 } header_field[] = {
 398         { "author ", 7 },
 399         { "committer ", 10 },
 400 };
 401
 402 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 403                              enum grep_context ctx,
 404                              regmatch_t *pmatch, int eflags)
 405 {
 406         int hit = 0;
 407         int saved_ch = 0;
 408         const char *start = bol;
 409
 410         if ((p->token != GREP_PATTERN) &&
 411             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 412                 return 0;
 413
 414         if (p->token == GREP_PATTERN_HEAD) {
 415                 const char *field;
 416                 size_t len;
 417                 assert(p->field < ARRAY_SIZE(header_field));
 418                 field = header_field[p->field].field;
 419                 len = header_field[p->field].len;
 420                 if (strncmp(bol, field, len))
 421                         return 0;
 422                 bol += len;
 423                 saved_ch = strip_timestamp(bol, &eol);
 424         }
 425
 426  again:
 427         if (p->fixed)
 428                 hit = !fixmatch(p, bol, eol, pmatch);
 429         else
 430                 hit = !regmatch(&p->regexp, bol, eol, pmatch, eflags);
 431
 432         if (hit && p->word_regexp) {
 433                 if ((pmatch[0].rm_so < 0) ||
 434                     (eol - bol) < pmatch[0].rm_so ||
 435                     (pmatch[0].rm_eo < 0) ||
 436                     (eol - bol) < pmatch[0].rm_eo)
 437                         die("regexp returned nonsense");
 438
 439                 /* Match beginning must be either beginning of the
 440                  * line, or at word boundary (i.e. the last char must
 441                  * not be a word char).  Similarly, match end must be
 442                  * either end of the line, or at word boundary
 443                  * (i.e. the next char must not be a word char).
 444                  */
 445                 if ( ((pmatch[0].rm_so == 0) ||
 446                       !word_char(bol[pmatch[0].rm_so-1])) &&
 447                      ((pmatch[0].rm_eo == (eol-bol)) ||
 448                       !word_char(bol[pmatch[0].rm_eo])) )
 449                         ;
 450                 else
 451                         hit = 0;
 452
 453                 /* Words consist of at least one character. */
 454                 if (pmatch->rm_so == pmatch->rm_eo)
 455                         hit = 0;
 456
 457                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 458                         /* There could be more than one match on the
 459                          * line, and the first match might not be
 460                          * strict word match.  But later ones could be!
 461                          * Forward to the next possible start, i.e. the
 462                          * next position following a non-word char.
 463                          */
 464                         bol = pmatch[0].rm_so + bol + 1;
 465                         while (word_char(bol[-1]) && bol < eol)
 466                                 bol++;
 467                         eflags |= REG_NOTBOL;
 468                         if (bol < eol)
 469                                 goto again;
 470                 }
 471         }
 472         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 473                 *eol = saved_ch;
 474         if (hit) {
 475                 pmatch[0].rm_so += bol - start;
 476                 pmatch[0].rm_eo += bol - start;
 477         }
 478         return hit;
 479 }
 480
 481 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 482                            enum grep_context ctx, int collect_hits)
 483 {
 484         int h = 0;
 485         regmatch_t match;
 486
 487         if (!x)
 488                 die("Not a valid grep expression");
 489         switch (x->node) {
 490         case GREP_NODE_ATOM:
 491                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 492                 break;
 493         case GREP_NODE_NOT:
 494                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 495                 break;
 496         case GREP_NODE_AND:
 497                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 498                         return 0;
 499                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 500                 break;
 501         case GREP_NODE_OR:
 502                 if (!collect_hits)
 503                         return (match_expr_eval(x->u.binary.left,
 504                                                 bol, eol, ctx, 0) ||
 505                                 match_expr_eval(x->u.binary.right,
 506                                                 bol, eol, ctx, 0));
 507                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 508                 x->u.binary.left->hit |= h;
 509                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 510                 break;
 511         default:
 512                 die("Unexpected node type (internal error) %d", x->node);
 513         }
 514         if (collect_hits)
 515                 x->hit |= h;
 516         return h;
 517 }
 518
 519 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 520                       enum grep_context ctx, int collect_hits)
 521 {
 522         struct grep_expr *x = opt->pattern_expression;
 523         return match_expr_eval(x, bol, eol, ctx, collect_hits);
 524 }
 525
 526 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 527                       enum grep_context ctx, int collect_hits)
 528 {
 529         struct grep_pat *p;
 530         regmatch_t match;
 531
 532         if (opt->extended)
 533                 return match_expr(opt, bol, eol, ctx, collect_hits);
 534
 535         /* we do not call with collect_hits without being extended */
 536         for (p = opt->pattern_list; p; p = p->next) {
 537                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 538                         return 1;
 539         }
 540         return 0;
 541 }
 542
 543 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 544                               enum grep_context ctx,
 545                               regmatch_t *pmatch, int eflags)
 546 {
 547         regmatch_t match;
 548
 549         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 550                 return 0;
 551         if (match.rm_so < 0 || match.rm_eo < 0)
 552                 return 0;
 553         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 554                 if (match.rm_so > pmatch->rm_so)
 555                         return 1;
 556                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 557                         return 1;
 558         }
 559         pmatch->rm_so = match.rm_so;
 560         pmatch->rm_eo = match.rm_eo;
 561         return 1;
 562 }
 563
 564 static int next_match(struct grep_opt *opt, char *bol, char *eol,
 565                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
 566 {
 567         struct grep_pat *p;
 568         int hit = 0;
 569
 570         pmatch->rm_so = pmatch->rm_eo = -1;
 571         if (bol < eol) {
 572                 for (p = opt->pattern_list; p; p = p->next) {
 573                         switch (p->token) {
 574                         case GREP_PATTERN: /* atom */
 575                         case GREP_PATTERN_HEAD:
 576                         case GREP_PATTERN_BODY:
 577                                 hit |= match_next_pattern(p, bol, eol, ctx,
 578                                                           pmatch, eflags);
 579                                 break;
 580                         default:
 581                                 break;
 582                         }
 583                 }
 584         }
 585         return hit;
 586 }
 587
 588 static void show_line(struct grep_opt *opt, char *bol, char *eol,
 589                       const char *name, unsigned lno, char sign)
 590 {
 591         int rest = eol - bol;
 592         char *line_color = NULL;
 593
 594         if (opt->pre_context || opt->post_context) {
 595                 if (opt->last_shown == 0) {
 596                         if (opt->show_hunk_mark) {
 597                                 output_color(opt, "--", 2, opt->color_sep);
 598                                 opt->output(opt, "\n", 1);
 599                         }
 600                 } else if (lno > opt->last_shown + 1) {
 601                         output_color(opt, "--", 2, opt->color_sep);
 602                         opt->output(opt, "\n", 1);
 603                 }
 604         }
 605         opt->last_shown = lno;
 606
 607         if (opt->pathname) {
 608                 output_color(opt, name, strlen(name), opt->color_filename);
 609                 output_sep(opt, sign);
 610         }
 611         if (opt->linenum) {
 612                 char buf[32];
 613                 snprintf(buf, sizeof(buf), "%d", lno);
 614                 output_color(opt, buf, strlen(buf), opt->color_lineno);
 615                 output_sep(opt, sign);
 616         }
 617         if (opt->color) {
 618                 regmatch_t match;
 619                 enum grep_context ctx = GREP_CONTEXT_BODY;
 620                 int ch = *eol;
 621                 int eflags = 0;
 622
 623                 if (sign == ':')
 624                         line_color = opt->color_selected;
 625                 else if (sign == '-')
 626                         line_color = opt->color_context;
 627                 else if (sign == '=')
 628                         line_color = opt->color_function;
 629                 *eol = '\0';
 630                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
 631                         if (match.rm_so == match.rm_eo)
 632                                 break;
 633
 634                         output_color(opt, bol, match.rm_so, line_color);
 635                         output_color(opt, bol + match.rm_so,
 636                                      match.rm_eo - match.rm_so,
 637                                      opt->color_match);
 638                         bol += match.rm_eo;
 639                         rest -= match.rm_eo;
 640                         eflags = REG_NOTBOL;
 641                 }
 642                 *eol = ch;
 643         }
 644         output_color(opt, bol, rest, line_color);
 645         opt->output(opt, "\n", 1);
 646 }
 647
 648 static int match_funcname(struct grep_opt *opt, char *bol, char *eol)
 649 {
 650         xdemitconf_t *xecfg = opt->priv;
 651         if (xecfg && xecfg->find_func) {
 652                 char buf[1];
 653                 return xecfg->find_func(bol, eol - bol, buf, 1,
 654                                         xecfg->find_func_priv) >= 0;
 655         }
 656
 657         if (bol == eol)
 658                 return 0;
 659         if (isalpha(*bol) || *bol == '_' || *bol == '$')
 660                 return 1;
 661         return 0;
 662 }
 663
 664 static void show_funcname_line(struct grep_opt *opt, const char *name,
 665                                char *buf, char *bol, unsigned lno)
 666 {
 667         while (bol > buf) {
 668                 char *eol = --bol;
 669
 670                 while (bol > buf && bol[-1] != '\n')
 671                         bol--;
 672                 lno--;
 673
 674                 if (lno <= opt->last_shown)
 675                         break;
 676
 677                 if (match_funcname(opt, bol, eol)) {
 678                         show_line(opt, bol, eol, name, lno, '=');
 679                         break;
 680                 }
 681         }
 682 }
 683
 684 static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
 685                              char *bol, unsigned lno)
 686 {
 687         unsigned cur = lno, from = 1, funcname_lno = 0;
 688         int funcname_needed = opt->funcname;
 689
 690         if (opt->pre_context < lno)
 691                 from = lno - opt->pre_context;
 692         if (from <= opt->last_shown)
 693                 from = opt->last_shown + 1;
 694
 695         /* Rewind. */
 696         while (bol > buf && cur > from) {
 697                 char *eol = --bol;
 698
 699                 while (bol > buf && bol[-1] != '\n')
 700                         bol--;
 701                 cur--;
 702                 if (funcname_needed && match_funcname(opt, bol, eol)) {
 703                         funcname_lno = cur;
 704                         funcname_needed = 0;
 705                 }
 706         }
 707
 708         /* We need to look even further back to find a function signature. */
 709         if (opt->funcname && funcname_needed)
 710                 show_funcname_line(opt, name, buf, bol, cur);
 711
 712         /* Back forward. */
 713         while (cur < lno) {
 714                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
 715
 716                 while (*eol != '\n')
 717                         eol++;
 718                 show_line(opt, bol, eol, name, cur, sign);
 719                 bol = eol + 1;
 720                 cur++;
 721         }
 722 }
 723
 724 static int should_lookahead(struct grep_opt *opt)
 725 {
 726         struct grep_pat *p;
 727
 728         if (opt->extended)
 729                 return 0; /* punt for too complex stuff */
 730         if (opt->invert)
 731                 return 0;
 732         for (p = opt->pattern_list; p; p = p->next) {
 733                 if (p->token != GREP_PATTERN)
 734                         return 0; /* punt for "header only" and stuff */
 735         }
 736         return 1;
 737 }
 738
 739 static int look_ahead(struct grep_opt *opt,
 740                       unsigned long *left_p,
 741                       unsigned *lno_p,
 742                       char **bol_p)
 743 {
 744         unsigned lno = *lno_p;
 745         char *bol = *bol_p;
 746         struct grep_pat *p;
 747         char *sp, *last_bol;
 748         regoff_t earliest = -1;
 749
 750         for (p = opt->pattern_list; p; p = p->next) {
 751                 int hit;
 752                 regmatch_t m;
 753
 754                 if (p->fixed)
 755                         hit = !fixmatch(p, bol, bol + *left_p, &m);
 756                 else
 757                         hit = !regmatch(&p->regexp, bol, bol + *left_p, &m, 0);
 758                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
 759                         continue;
 760                 if (earliest < 0 || m.rm_so < earliest)
 761                         earliest = m.rm_so;
 762         }
 763
 764         if (earliest < 0) {
 765                 *bol_p = bol + *left_p;
 766                 *left_p = 0;
 767                 return 1;
 768         }
 769         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
 770                 ; /* find the beginning of the line */
 771         last_bol = sp;
 772
 773         for (sp = bol; sp < last_bol; sp++) {
 774                 if (*sp == '\n')
 775                         lno++;
 776         }
 777         *left_p -= last_bol - bol;
 778         *bol_p = last_bol;
 779         *lno_p = lno;
 780         return 0;
 781 }
 782
 783 int grep_threads_ok(const struct grep_opt *opt)
 784 {
 785         /* If this condition is true, then we may use the attribute
 786          * machinery in grep_buffer_1. The attribute code is not
 787          * thread safe, so we disable the use of threads.
 788          */
 789         if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
 790             !opt->name_only)
 791                 return 0;
 792
 793         return 1;
 794 }
 795
 796 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
 797 {
 798         fwrite(buf, size, 1, stdout);
 799 }
 800
 801 static int grep_buffer_1(struct grep_opt *opt, const char *name,
 802                          char *buf, unsigned long size, int collect_hits)
 803 {
 804         char *bol = buf;
 805         unsigned long left = size;
 806         unsigned lno = 1;
 807         unsigned last_hit = 0;
 808         int binary_match_only = 0;
 809         unsigned count = 0;
 810         int try_lookahead = 0;
 811         enum grep_context ctx = GREP_CONTEXT_HEAD;
 812         xdemitconf_t xecfg;
 813
 814         if (!opt->output)
 815                 opt->output = std_output;
 816
 817         if (opt->last_shown && (opt->pre_context || opt->post_context) &&
 818             opt->output == std_output)
 819                 opt->show_hunk_mark = 1;
 820         opt->last_shown = 0;
 821
 822         switch (opt->binary) {
 823         case GREP_BINARY_DEFAULT:
 824                 if (buffer_is_binary(buf, size))
 825                         binary_match_only = 1;
 826                 break;
 827         case GREP_BINARY_NOMATCH:
 828                 if (buffer_is_binary(buf, size))
 829                         return 0; /* Assume unmatch */
 830                 break;
 831         case GREP_BINARY_TEXT:
 832                 break;
 833         default:
 834                 die("bug: unknown binary handling mode");
 835         }
 836
 837         memset(&xecfg, 0, sizeof(xecfg));
 838         if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
 839             !opt->name_only && !binary_match_only && !collect_hits) {
 840                 struct userdiff_driver *drv = userdiff_find_by_path(name);
 841                 if (drv && drv->funcname.pattern) {
 842                         const struct userdiff_funcname *pe = &drv->funcname;
 843                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
 844                         opt->priv = &xecfg;
 845                 }
 846         }
 847         try_lookahead = should_lookahead(opt);
 848
 849         while (left) {
 850                 char *eol, ch;
 851                 int hit;
 852
 853                 /*
 854                  * look_ahead() skips quicly to the line that possibly
 855                  * has the next hit; don't call it if we need to do
 856                  * something more than just skipping the current line
 857                  * in response to an unmatch for the current line.  E.g.
 858                  * inside a post-context window, we will show the current
 859                  * line as a context around the previous hit when it
 860                  * doesn't hit.
 861                  */
 862                 if (try_lookahead
 863                     && !(last_hit
 864                          && lno <= last_hit + opt->post_context)
 865                     && look_ahead(opt, &left, &lno, &bol))
 866                         break;
 867                 eol = end_of_line(bol, &left);
 868                 ch = *eol;
 869                 *eol = 0;
 870
 871                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
 872                         ctx = GREP_CONTEXT_BODY;
 873
 874                 hit = match_line(opt, bol, eol, ctx, collect_hits);
 875                 *eol = ch;
 876
 877                 if (collect_hits)
 878                         goto next_line;
 879
 880                 /* "grep -v -e foo -e bla" should list lines
 881                  * that do not have either, so inversion should
 882                  * be done outside.
 883                  */
 884                 if (opt->invert)
 885                         hit = !hit;
 886                 if (opt->unmatch_name_only) {
 887                         if (hit)
 888                                 return 0;
 889                         goto next_line;
 890                 }
 891                 if (hit) {
 892                         count++;
 893                         if (opt->status_only)
 894                                 return 1;
 895                         if (opt->name_only) {
 896                                 show_name(opt, name);
 897                                 return 1;
 898                         }
 899                         if (opt->count)
 900                                 goto next_line;
 901                         if (binary_match_only) {
 902                                 opt->output(opt, "Binary file ", 12);
 903                                 output_color(opt, name, strlen(name),
 904                                              opt->color_filename);
 905                                 opt->output(opt, " matches\n", 9);
 906                                 return 1;
 907                         }
 908                         /* Hit at this line.  If we haven't shown the
 909                          * pre-context lines, we would need to show them.
 910                          */
 911                         if (opt->pre_context)
 912                                 show_pre_context(opt, name, buf, bol, lno);
 913                         else if (opt->funcname)
 914                                 show_funcname_line(opt, name, buf, bol, lno);
 915                         show_line(opt, bol, eol, name, lno, ':');
 916                         last_hit = lno;
 917                 }
 918                 else if (last_hit &&
 919                          lno <= last_hit + opt->post_context) {
 920                         /* If the last hit is within the post context,
 921                          * we need to show this line.
 922                          */
 923                         show_line(opt, bol, eol, name, lno, '-');
 924                 }
 925
 926         next_line:
 927                 bol = eol + 1;
 928                 if (!left)
 929                         break;
 930                 left--;
 931                 lno++;
 932         }
 933
 934         if (collect_hits)
 935                 return 0;
 936
 937         if (opt->status_only)
 938                 return 0;
 939         if (opt->unmatch_name_only) {
 940                 /* We did not see any hit, so we want to show this */
 941                 show_name(opt, name);
 942                 return 1;
 943         }
 944
 945         xdiff_clear_find_func(&xecfg);
 946         opt->priv = NULL;
 947
 948         /* NEEDSWORK:
 949          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 950          * which feels mostly useless but sometimes useful.  Maybe
 951          * make it another option?  For now suppress them.
 952          */
 953         if (opt->count && count) {
 954                 char buf[32];
 955                 output_color(opt, name, strlen(name), opt->color_filename);
 956                 output_sep(opt, ':');
 957                 snprintf(buf, sizeof(buf), "%u\n", count);
 958                 opt->output(opt, buf, strlen(buf));
 959                 return 1;
 960         }
 961         return !!last_hit;
 962 }
 963
 964 static void clr_hit_marker(struct grep_expr *x)
 965 {
 966         /* All-hit markers are meaningful only at the very top level
 967          * OR node.
 968          */
 969         while (1) {
 970                 x->hit = 0;
 971                 if (x->node != GREP_NODE_OR)
 972                         return;
 973                 x->u.binary.left->hit = 0;
 974                 x = x->u.binary.right;
 975         }
 976 }
 977
 978 static int chk_hit_marker(struct grep_expr *x)
 979 {
 980         /* Top level nodes have hit markers.  See if they all are hits */
 981         while (1) {
 982                 if (x->node != GREP_NODE_OR)
 983                         return x->hit;
 984                 if (!x->u.binary.left->hit)
 985                         return 0;
 986                 x = x->u.binary.right;
 987         }
 988 }
 989
 990 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
 991 {
 992         /*
 993          * we do not have to do the two-pass grep when we do not check
 994          * buffer-wide "all-match".
 995          */
 996         if (!opt->all_match)
 997                 return grep_buffer_1(opt, name, buf, size, 0);
 998
 999         /* Otherwise the toplevel "or" terms hit a bit differently.
1000          * We first clear hit markers from them.
1001          */
1002         clr_hit_marker(opt->pattern_expression);
1003         grep_buffer_1(opt, name, buf, size, 1);
1004
1005         if (!chk_hit_marker(opt->pattern_expression))
1006                 return 0;
1007
1008         return grep_buffer_1(opt, name, buf, size, 0);
1009 }