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 static struct grep_expr *grep_true_expr(void)
 193 {
 194         struct grep_expr *z = xcalloc(1, sizeof(*z));
 195         z->node = GREP_NODE_TRUE;
 196         return z;
 197 }
 198
 199 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 200 {
 201         struct grep_expr *z = xcalloc(1, sizeof(*z));
 202         z->node = GREP_NODE_OR;
 203         z->u.binary.left = left;
 204         z->u.binary.right = right;
 205         return z;
 206 }
 207
 208 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 209 {
 210         struct grep_pat *p;
 211         struct grep_expr *header_expr;
 212         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 213         enum grep_header_field fld;
 214
 215         if (!opt->header_list)
 216                 return NULL;
 217         p = opt->header_list;
 218         for (p = opt->header_list; p; p = p->next) {
 219                 if (p->token != GREP_PATTERN_HEAD)
 220                         die("bug: a non-header pattern in grep header list.");
 221                 if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
 222                         die("bug: unknown header field %d", p->field);
 223                 compile_regexp(p, opt);
 224         }
 225
 226         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 227                 header_group[fld] = NULL;
 228
 229         for (p = opt->header_list; p; p = p->next) {
 230                 struct grep_expr *h;
 231                 struct grep_pat *pp = p;
 232
 233                 h = compile_pattern_atom(&pp);
 234                 if (!h || pp != p->next)
 235                         die("bug: malformed header expr");
 236                 if (!header_group[p->field]) {
 237                         header_group[p->field] = h;
 238                         continue;
 239                 }
 240                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 241         }
 242
 243         header_expr = NULL;
 244
 245         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 246                 if (!header_group[fld])
 247                         continue;
 248                 if (!header_expr)
 249                         header_expr = grep_true_expr();
 250                 header_expr = grep_or_expr(header_group[fld], header_expr);
 251         }
 252         return header_expr;
 253 }
 254
 255 void compile_grep_patterns(struct grep_opt *opt)
 256 {
 257         struct grep_pat *p;
 258         struct grep_expr *header_expr = prep_header_patterns(opt);
 259
 260         for (p = opt->pattern_list; p; p = p->next) {
 261                 switch (p->token) {
 262                 case GREP_PATTERN: /* atom */
 263                 case GREP_PATTERN_HEAD:
 264                 case GREP_PATTERN_BODY:
 265                         compile_regexp(p, opt);
 266                         break;
 267                 default:
 268                         opt->extended = 1;
 269                         break;
 270                 }
 271         }
 272
 273         if (opt->all_match || header_expr)
 274                 opt->extended = 1;
 275         else if (!opt->extended)
 276                 return;
 277
 278         p = opt->pattern_list;
 279         if (p)
 280                 opt->pattern_expression = compile_pattern_expr(&p);
 281         if (p)
 282                 die("incomplete pattern expression: %s", p->pattern);
 283
 284         if (!header_expr)
 285                 return;
 286
 287         if (!opt->pattern_expression)
 288                 opt->pattern_expression = header_expr;
 289         else
 290                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 291                                                        header_expr);
 292         opt->all_match = 1;
 293 }
 294
 295 static void free_pattern_expr(struct grep_expr *x)
 296 {
 297         switch (x->node) {
 298         case GREP_NODE_TRUE:
 299         case GREP_NODE_ATOM:
 300                 break;
 301         case GREP_NODE_NOT:
 302                 free_pattern_expr(x->u.unary);
 303                 break;
 304         case GREP_NODE_AND:
 305         case GREP_NODE_OR:
 306                 free_pattern_expr(x->u.binary.left);
 307                 free_pattern_expr(x->u.binary.right);
 308                 break;
 309         }
 310         free(x);
 311 }
 312
 313 void free_grep_patterns(struct grep_opt *opt)
 314 {
 315         struct grep_pat *p, *n;
 316
 317         for (p = opt->pattern_list; p; p = n) {
 318                 n = p->next;
 319                 switch (p->token) {
 320                 case GREP_PATTERN: /* atom */
 321                 case GREP_PATTERN_HEAD:
 322                 case GREP_PATTERN_BODY:
 323                         regfree(&p->regexp);
 324                         break;
 325                 default:
 326                         break;
 327                 }
 328                 free(p);
 329         }
 330
 331         if (!opt->extended)
 332                 return;
 333         free_pattern_expr(opt->pattern_expression);
 334 }
 335
 336 static char *end_of_line(char *cp, unsigned long *left)
 337 {
 338         unsigned long l = *left;
 339         while (l && *cp != '\n') {
 340                 l--;
 341                 cp++;
 342         }
 343         *left = l;
 344         return cp;
 345 }
 346
 347 static int word_char(char ch)
 348 {
 349         return isalnum(ch) || ch == '_';
 350 }
 351
 352 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 353                          const char *color)
 354 {
 355         if (opt->color && color && color[0]) {
 356                 opt->output(opt, color, strlen(color));
 357                 opt->output(opt, data, size);
 358                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 359         } else
 360                 opt->output(opt, data, size);
 361 }
 362
 363 static void output_sep(struct grep_opt *opt, char sign)
 364 {
 365         if (opt->null_following_name)
 366                 opt->output(opt, "\0", 1);
 367         else
 368                 output_color(opt, &sign, 1, opt->color_sep);
 369 }
 370
 371 static void show_name(struct grep_opt *opt, const char *name)
 372 {
 373         output_color(opt, name, strlen(name), opt->color_filename);
 374         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 375 }
 376
 377 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 378                     regmatch_t *match)
 379 {
 380         char *hit;
 381
 382         if (p->ignore_case) {
 383                 char *s = line;
 384                 do {
 385                         hit = strcasestr(s, p->pattern);
 386                         if (hit)
 387                                 break;
 388                         s += strlen(s) + 1;
 389                 } while (s < eol);
 390         } else
 391                 hit = memmem(line, eol - line, p->pattern, p->patternlen);
 392
 393         if (!hit) {
 394                 match->rm_so = match->rm_eo = -1;
 395                 return REG_NOMATCH;
 396         }
 397         else {
 398                 match->rm_so = hit - line;
 399                 match->rm_eo = match->rm_so + p->patternlen;
 400                 return 0;
 401         }
 402 }
 403
 404 static int regmatch(const regex_t *preg, char *line, char *eol,
 405                     regmatch_t *match, int eflags)
 406 {
 407 #ifdef REG_STARTEND
 408         match->rm_so = 0;
 409         match->rm_eo = eol - line;
 410         eflags |= REG_STARTEND;
 411 #endif
 412         return regexec(preg, line, 1, match, eflags);
 413 }
 414
 415 static int patmatch(struct grep_pat *p, char *line, char *eol,
 416                     regmatch_t *match, int eflags)
 417 {
 418         int hit;
 419
 420         if (p->fixed)
 421                 hit = !fixmatch(p, line, eol, match);
 422         else
 423                 hit = !regmatch(&p->regexp, line, eol, match, eflags);
 424
 425         return hit;
 426 }
 427
 428 static int strip_timestamp(char *bol, char **eol_p)
 429 {
 430         char *eol = *eol_p;
 431         int ch;
 432
 433         while (bol < --eol) {
 434                 if (*eol != '>')
 435                         continue;
 436                 *eol_p = ++eol;
 437                 ch = *eol;
 438                 *eol = '\0';
 439                 return ch;
 440         }
 441         return 0;
 442 }
 443
 444 static struct {
 445         const char *field;
 446         size_t len;
 447 } header_field[] = {
 448         { "author ", 7 },
 449         { "committer ", 10 },
 450 };
 451
 452 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 453                              enum grep_context ctx,
 454                              regmatch_t *pmatch, int eflags)
 455 {
 456         int hit = 0;
 457         int saved_ch = 0;
 458         const char *start = bol;
 459
 460         if ((p->token != GREP_PATTERN) &&
 461             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 462                 return 0;
 463
 464         if (p->token == GREP_PATTERN_HEAD) {
 465                 const char *field;
 466                 size_t len;
 467                 assert(p->field < ARRAY_SIZE(header_field));
 468                 field = header_field[p->field].field;
 469                 len = header_field[p->field].len;
 470                 if (strncmp(bol, field, len))
 471                         return 0;
 472                 bol += len;
 473                 saved_ch = strip_timestamp(bol, &eol);
 474         }
 475
 476  again:
 477         hit = patmatch(p, bol, eol, pmatch, eflags);
 478
 479         if (hit && p->word_regexp) {
 480                 if ((pmatch[0].rm_so < 0) ||
 481                     (eol - bol) < pmatch[0].rm_so ||
 482                     (pmatch[0].rm_eo < 0) ||
 483                     (eol - bol) < pmatch[0].rm_eo)
 484                         die("regexp returned nonsense");
 485
 486                 /* Match beginning must be either beginning of the
 487                  * line, or at word boundary (i.e. the last char must
 488                  * not be a word char).  Similarly, match end must be
 489                  * either end of the line, or at word boundary
 490                  * (i.e. the next char must not be a word char).
 491                  */
 492                 if ( ((pmatch[0].rm_so == 0) ||
 493                       !word_char(bol[pmatch[0].rm_so-1])) &&
 494                      ((pmatch[0].rm_eo == (eol-bol)) ||
 495                       !word_char(bol[pmatch[0].rm_eo])) )
 496                         ;
 497                 else
 498                         hit = 0;
 499
 500                 /* Words consist of at least one character. */
 501                 if (pmatch->rm_so == pmatch->rm_eo)
 502                         hit = 0;
 503
 504                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 505                         /* There could be more than one match on the
 506                          * line, and the first match might not be
 507                          * strict word match.  But later ones could be!
 508                          * Forward to the next possible start, i.e. the
 509                          * next position following a non-word char.
 510                          */
 511                         bol = pmatch[0].rm_so + bol + 1;
 512                         while (word_char(bol[-1]) && bol < eol)
 513                                 bol++;
 514                         eflags |= REG_NOTBOL;
 515                         if (bol < eol)
 516                                 goto again;
 517                 }
 518         }
 519         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 520                 *eol = saved_ch;
 521         if (hit) {
 522                 pmatch[0].rm_so += bol - start;
 523                 pmatch[0].rm_eo += bol - start;
 524         }
 525         return hit;
 526 }
 527
 528 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 529                            enum grep_context ctx, int collect_hits)
 530 {
 531         int h = 0;
 532         regmatch_t match;
 533
 534         if (!x)
 535                 die("Not a valid grep expression");
 536         switch (x->node) {
 537         case GREP_NODE_TRUE:
 538                 h = 1;
 539                 break;
 540         case GREP_NODE_ATOM:
 541                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 542                 break;
 543         case GREP_NODE_NOT:
 544                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 545                 break;
 546         case GREP_NODE_AND:
 547                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 548                         return 0;
 549                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 550                 break;
 551         case GREP_NODE_OR:
 552                 if (!collect_hits)
 553                         return (match_expr_eval(x->u.binary.left,
 554                                                 bol, eol, ctx, 0) ||
 555                                 match_expr_eval(x->u.binary.right,
 556                                                 bol, eol, ctx, 0));
 557                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 558                 x->u.binary.left->hit |= h;
 559                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 560                 break;
 561         default:
 562                 die("Unexpected node type (internal error) %d", x->node);
 563         }
 564         if (collect_hits)
 565                 x->hit |= h;
 566         return h;
 567 }
 568
 569 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 570                       enum grep_context ctx, int collect_hits)
 571 {
 572         struct grep_expr *x = opt->pattern_expression;
 573         return match_expr_eval(x, bol, eol, ctx, collect_hits);
 574 }
 575
 576 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 577                       enum grep_context ctx, int collect_hits)
 578 {
 579         struct grep_pat *p;
 580         regmatch_t match;
 581
 582         if (opt->extended)
 583                 return match_expr(opt, bol, eol, ctx, collect_hits);
 584
 585         /* we do not call with collect_hits without being extended */
 586         for (p = opt->pattern_list; p; p = p->next) {
 587                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 588                         return 1;
 589         }
 590         return 0;
 591 }
 592
 593 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 594                               enum grep_context ctx,
 595                               regmatch_t *pmatch, int eflags)
 596 {
 597         regmatch_t match;
 598
 599         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 600                 return 0;
 601         if (match.rm_so < 0 || match.rm_eo < 0)
 602                 return 0;
 603         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 604                 if (match.rm_so > pmatch->rm_so)
 605                         return 1;
 606                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 607                         return 1;
 608         }
 609         pmatch->rm_so = match.rm_so;
 610         pmatch->rm_eo = match.rm_eo;
 611         return 1;
 612 }
 613
 614 static int next_match(struct grep_opt *opt, char *bol, char *eol,
 615                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
 616 {
 617         struct grep_pat *p;
 618         int hit = 0;
 619
 620         pmatch->rm_so = pmatch->rm_eo = -1;
 621         if (bol < eol) {
 622                 for (p = opt->pattern_list; p; p = p->next) {
 623                         switch (p->token) {
 624                         case GREP_PATTERN: /* atom */
 625                         case GREP_PATTERN_HEAD:
 626                         case GREP_PATTERN_BODY:
 627                                 hit |= match_next_pattern(p, bol, eol, ctx,
 628                                                           pmatch, eflags);
 629                                 break;
 630                         default:
 631                                 break;
 632                         }
 633                 }
 634         }
 635         return hit;
 636 }
 637
 638 static void show_line(struct grep_opt *opt, char *bol, char *eol,
 639                       const char *name, unsigned lno, char sign)
 640 {
 641         int rest = eol - bol;
 642         char *line_color = NULL;
 643
 644         if (opt->pre_context || opt->post_context) {
 645                 if (opt->last_shown == 0) {
 646                         if (opt->show_hunk_mark) {
 647                                 output_color(opt, "--", 2, opt->color_sep);
 648                                 opt->output(opt, "\n", 1);
 649                         }
 650                 } else if (lno > opt->last_shown + 1) {
 651                         output_color(opt, "--", 2, opt->color_sep);
 652                         opt->output(opt, "\n", 1);
 653                 }
 654         }
 655         opt->last_shown = lno;
 656
 657         if (opt->pathname) {
 658                 output_color(opt, name, strlen(name), opt->color_filename);
 659                 output_sep(opt, sign);
 660         }
 661         if (opt->linenum) {
 662                 char buf[32];
 663                 snprintf(buf, sizeof(buf), "%d", lno);
 664                 output_color(opt, buf, strlen(buf), opt->color_lineno);
 665                 output_sep(opt, sign);
 666         }
 667         if (opt->color) {
 668                 regmatch_t match;
 669                 enum grep_context ctx = GREP_CONTEXT_BODY;
 670                 int ch = *eol;
 671                 int eflags = 0;
 672
 673                 if (sign == ':')
 674                         line_color = opt->color_selected;
 675                 else if (sign == '-')
 676                         line_color = opt->color_context;
 677                 else if (sign == '=')
 678                         line_color = opt->color_function;
 679                 *eol = '\0';
 680                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
 681                         if (match.rm_so == match.rm_eo)
 682                                 break;
 683
 684                         output_color(opt, bol, match.rm_so, line_color);
 685                         output_color(opt, bol + match.rm_so,
 686                                      match.rm_eo - match.rm_so,
 687                                      opt->color_match);
 688                         bol += match.rm_eo;
 689                         rest -= match.rm_eo;
 690                         eflags = REG_NOTBOL;
 691                 }
 692                 *eol = ch;
 693         }
 694         output_color(opt, bol, rest, line_color);
 695         opt->output(opt, "\n", 1);
 696 }
 697
 698 static int match_funcname(struct grep_opt *opt, char *bol, char *eol)
 699 {
 700         xdemitconf_t *xecfg = opt->priv;
 701         if (xecfg && xecfg->find_func) {
 702                 char buf[1];
 703                 return xecfg->find_func(bol, eol - bol, buf, 1,
 704                                         xecfg->find_func_priv) >= 0;
 705         }
 706
 707         if (bol == eol)
 708                 return 0;
 709         if (isalpha(*bol) || *bol == '_' || *bol == '$')
 710                 return 1;
 711         return 0;
 712 }
 713
 714 static void show_funcname_line(struct grep_opt *opt, const char *name,
 715                                char *buf, char *bol, unsigned lno)
 716 {
 717         while (bol > buf) {
 718                 char *eol = --bol;
 719
 720                 while (bol > buf && bol[-1] != '\n')
 721                         bol--;
 722                 lno--;
 723
 724                 if (lno <= opt->last_shown)
 725                         break;
 726
 727                 if (match_funcname(opt, bol, eol)) {
 728                         show_line(opt, bol, eol, name, lno, '=');
 729                         break;
 730                 }
 731         }
 732 }
 733
 734 static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
 735                              char *bol, unsigned lno)
 736 {
 737         unsigned cur = lno, from = 1, funcname_lno = 0;
 738         int funcname_needed = opt->funcname;
 739
 740         if (opt->pre_context < lno)
 741                 from = lno - opt->pre_context;
 742         if (from <= opt->last_shown)
 743                 from = opt->last_shown + 1;
 744
 745         /* Rewind. */
 746         while (bol > buf && cur > from) {
 747                 char *eol = --bol;
 748
 749                 while (bol > buf && bol[-1] != '\n')
 750                         bol--;
 751                 cur--;
 752                 if (funcname_needed && match_funcname(opt, bol, eol)) {
 753                         funcname_lno = cur;
 754                         funcname_needed = 0;
 755                 }
 756         }
 757
 758         /* We need to look even further back to find a function signature. */
 759         if (opt->funcname && funcname_needed)
 760                 show_funcname_line(opt, name, buf, bol, cur);
 761
 762         /* Back forward. */
 763         while (cur < lno) {
 764                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
 765
 766                 while (*eol != '\n')
 767                         eol++;
 768                 show_line(opt, bol, eol, name, cur, sign);
 769                 bol = eol + 1;
 770                 cur++;
 771         }
 772 }
 773
 774 static int should_lookahead(struct grep_opt *opt)
 775 {
 776         struct grep_pat *p;
 777
 778         if (opt->extended)
 779                 return 0; /* punt for too complex stuff */
 780         if (opt->invert)
 781                 return 0;
 782         for (p = opt->pattern_list; p; p = p->next) {
 783                 if (p->token != GREP_PATTERN)
 784                         return 0; /* punt for "header only" and stuff */
 785         }
 786         return 1;
 787 }
 788
 789 static int look_ahead(struct grep_opt *opt,
 790                       unsigned long *left_p,
 791                       unsigned *lno_p,
 792                       char **bol_p)
 793 {
 794         unsigned lno = *lno_p;
 795         char *bol = *bol_p;
 796         struct grep_pat *p;
 797         char *sp, *last_bol;
 798         regoff_t earliest = -1;
 799
 800         for (p = opt->pattern_list; p; p = p->next) {
 801                 int hit;
 802                 regmatch_t m;
 803
 804                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
 805                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
 806                         continue;
 807                 if (earliest < 0 || m.rm_so < earliest)
 808                         earliest = m.rm_so;
 809         }
 810
 811         if (earliest < 0) {
 812                 *bol_p = bol + *left_p;
 813                 *left_p = 0;
 814                 return 1;
 815         }
 816         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
 817                 ; /* find the beginning of the line */
 818         last_bol = sp;
 819
 820         for (sp = bol; sp < last_bol; sp++) {
 821                 if (*sp == '\n')
 822                         lno++;
 823         }
 824         *left_p -= last_bol - bol;
 825         *bol_p = last_bol;
 826         *lno_p = lno;
 827         return 0;
 828 }
 829
 830 int grep_threads_ok(const struct grep_opt *opt)
 831 {
 832         /* If this condition is true, then we may use the attribute
 833          * machinery in grep_buffer_1. The attribute code is not
 834          * thread safe, so we disable the use of threads.
 835          */
 836         if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
 837             !opt->name_only)
 838                 return 0;
 839
 840         return 1;
 841 }
 842
 843 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
 844 {
 845         fwrite(buf, size, 1, stdout);
 846 }
 847
 848 static int grep_buffer_1(struct grep_opt *opt, const char *name,
 849                          char *buf, unsigned long size, int collect_hits)
 850 {
 851         char *bol = buf;
 852         unsigned long left = size;
 853         unsigned lno = 1;
 854         unsigned last_hit = 0;
 855         int binary_match_only = 0;
 856         unsigned count = 0;
 857         int try_lookahead = 0;
 858         enum grep_context ctx = GREP_CONTEXT_HEAD;
 859         xdemitconf_t xecfg;
 860
 861         if (!opt->output)
 862                 opt->output = std_output;
 863
 864         if (opt->last_shown && (opt->pre_context || opt->post_context) &&
 865             opt->output == std_output)
 866                 opt->show_hunk_mark = 1;
 867         opt->last_shown = 0;
 868
 869         switch (opt->binary) {
 870         case GREP_BINARY_DEFAULT:
 871                 if (buffer_is_binary(buf, size))
 872                         binary_match_only = 1;
 873                 break;
 874         case GREP_BINARY_NOMATCH:
 875                 if (buffer_is_binary(buf, size))
 876                         return 0; /* Assume unmatch */
 877                 break;
 878         case GREP_BINARY_TEXT:
 879                 break;
 880         default:
 881                 die("bug: unknown binary handling mode");
 882         }
 883
 884         memset(&xecfg, 0, sizeof(xecfg));
 885         if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
 886             !opt->name_only && !binary_match_only && !collect_hits) {
 887                 struct userdiff_driver *drv = userdiff_find_by_path(name);
 888                 if (drv && drv->funcname.pattern) {
 889                         const struct userdiff_funcname *pe = &drv->funcname;
 890                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
 891                         opt->priv = &xecfg;
 892                 }
 893         }
 894         try_lookahead = should_lookahead(opt);
 895
 896         while (left) {
 897                 char *eol, ch;
 898                 int hit;
 899
 900                 /*
 901                  * look_ahead() skips quicly to the line that possibly
 902                  * has the next hit; don't call it if we need to do
 903                  * something more than just skipping the current line
 904                  * in response to an unmatch for the current line.  E.g.
 905                  * inside a post-context window, we will show the current
 906                  * line as a context around the previous hit when it
 907                  * doesn't hit.
 908                  */
 909                 if (try_lookahead
 910                     && !(last_hit
 911                          && lno <= last_hit + opt->post_context)
 912                     && look_ahead(opt, &left, &lno, &bol))
 913                         break;
 914                 eol = end_of_line(bol, &left);
 915                 ch = *eol;
 916                 *eol = 0;
 917
 918                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
 919                         ctx = GREP_CONTEXT_BODY;
 920
 921                 hit = match_line(opt, bol, eol, ctx, collect_hits);
 922                 *eol = ch;
 923
 924                 if (collect_hits)
 925                         goto next_line;
 926
 927                 /* "grep -v -e foo -e bla" should list lines
 928                  * that do not have either, so inversion should
 929                  * be done outside.
 930                  */
 931                 if (opt->invert)
 932                         hit = !hit;
 933                 if (opt->unmatch_name_only) {
 934                         if (hit)
 935                                 return 0;
 936                         goto next_line;
 937                 }
 938                 if (hit) {
 939                         count++;
 940                         if (opt->status_only)
 941                                 return 1;
 942                         if (opt->name_only) {
 943                                 show_name(opt, name);
 944                                 return 1;
 945                         }
 946                         if (opt->count)
 947                                 goto next_line;
 948                         if (binary_match_only) {
 949                                 opt->output(opt, "Binary file ", 12);
 950                                 output_color(opt, name, strlen(name),
 951                                              opt->color_filename);
 952                                 opt->output(opt, " matches\n", 9);
 953                                 return 1;
 954                         }
 955                         /* Hit at this line.  If we haven't shown the
 956                          * pre-context lines, we would need to show them.
 957                          */
 958                         if (opt->pre_context)
 959                                 show_pre_context(opt, name, buf, bol, lno);
 960                         else if (opt->funcname)
 961                                 show_funcname_line(opt, name, buf, bol, lno);
 962                         show_line(opt, bol, eol, name, lno, ':');
 963                         last_hit = lno;
 964                 }
 965                 else if (last_hit &&
 966                          lno <= last_hit + opt->post_context) {
 967                         /* If the last hit is within the post context,
 968                          * we need to show this line.
 969                          */
 970                         show_line(opt, bol, eol, name, lno, '-');
 971                 }
 972
 973         next_line:
 974                 bol = eol + 1;
 975                 if (!left)
 976                         break;
 977                 left--;
 978                 lno++;
 979         }
 980
 981         if (collect_hits)
 982                 return 0;
 983
 984         if (opt->status_only)
 985                 return 0;
 986         if (opt->unmatch_name_only) {
 987                 /* We did not see any hit, so we want to show this */
 988                 show_name(opt, name);
 989                 return 1;
 990         }
 991
 992         xdiff_clear_find_func(&xecfg);
 993         opt->priv = NULL;
 994
 995         /* NEEDSWORK:
 996          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 997          * which feels mostly useless but sometimes useful.  Maybe
 998          * make it another option?  For now suppress them.
 999          */
1000         if (opt->count && count) {
1001                 char buf[32];
1002                 output_color(opt, name, strlen(name), opt->color_filename);
1003                 output_sep(opt, ':');
1004                 snprintf(buf, sizeof(buf), "%u\n", count);
1005                 opt->output(opt, buf, strlen(buf));
1006                 return 1;
1007         }
1008         return !!last_hit;
1009 }
1010
1011 static void clr_hit_marker(struct grep_expr *x)
1012 {
1013         /* All-hit markers are meaningful only at the very top level
1014          * OR node.
1015          */
1016         while (1) {
1017                 x->hit = 0;
1018                 if (x->node != GREP_NODE_OR)
1019                         return;
1020                 x->u.binary.left->hit = 0;
1021                 x = x->u.binary.right;
1022         }
1023 }
1024
1025 static int chk_hit_marker(struct grep_expr *x)
1026 {
1027         /* Top level nodes have hit markers.  See if they all are hits */
1028         while (1) {
1029                 if (x->node != GREP_NODE_OR)
1030                         return x->hit;
1031                 if (!x->u.binary.left->hit)
1032                         return 0;
1033                 x = x->u.binary.right;
1034         }
1035 }
1036
1037 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
1038 {
1039         /*
1040          * we do not have to do the two-pass grep when we do not check
1041          * buffer-wide "all-match".
1042          */
1043         if (!opt->all_match)
1044                 return grep_buffer_1(opt, name, buf, size, 0);
1045
1046         /* Otherwise the toplevel "or" terms hit a bit differently.
1047          * We first clear hit markers from them.
1048          */
1049         clr_hit_marker(opt->pattern_expression);
1050         grep_buffer_1(opt, name, buf, size, 1);
1051
1052         if (!chk_hit_marker(opt->pattern_expression))
1053                 return 0;
1054
1055         return grep_buffer_1(opt, name, buf, size, 0);
1056 }