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