grep.c

   1 #include "cache.h"
   2 #include "grep.h"
   3 #include "xdiff-interface.h"
   4
   5 void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat)
   6 {
   7         struct grep_pat *p = xcalloc(1, sizeof(*p));
   8         p->pattern = pat;
   9         p->origin = "header";
  10         p->no = 0;
  11         p->token = GREP_PATTERN_HEAD;
  12         p->field = field;
  13         *opt->pattern_tail = p;
  14         opt->pattern_tail = &p->next;
  15         p->next = NULL;
  16 }
  17
  18 void append_grep_pattern(struct grep_opt *opt, const char *pat,
  19                          const char *origin, int no, enum grep_pat_token t)
  20 {
  21         struct grep_pat *p = xcalloc(1, sizeof(*p));
  22         p->pattern = pat;
  23         p->origin = origin;
  24         p->no = no;
  25         p->token = t;
  26         *opt->pattern_tail = p;
  27         opt->pattern_tail = &p->next;
  28         p->next = NULL;
  29 }
  30
  31 static int is_fixed(const char *s)
  32 {
  33         while (*s && !is_regex_special(*s))
  34                 s++;
  35         return !*s;
  36 }
  37
  38 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
  39 {
  40         int err;
  41
  42         p->word_regexp = opt->word_regexp;
  43
  44         if (opt->fixed || is_fixed(p->pattern))
  45                 p->fixed = 1;
  46         if (opt->regflags & REG_ICASE)
  47                 p->fixed = 0;
  48         if (p->fixed)
  49                 return;
  50
  51         err = regcomp(&p->regexp, p->pattern, opt->regflags);
  52         if (err) {
  53                 char errbuf[1024];
  54                 char where[1024];
  55                 if (p->no)
  56                         sprintf(where, "In '%s' at %d, ",
  57                                 p->origin, p->no);
  58                 else if (p->origin)
  59                         sprintf(where, "%s, ", p->origin);
  60                 else
  61                         where[0] = 0;
  62                 regerror(err, &p->regexp, errbuf, 1024);
  63                 regfree(&p->regexp);
  64                 die("%s'%s': %s", where, p->pattern, errbuf);
  65         }
  66 }
  67
  68 static struct grep_expr *compile_pattern_or(struct grep_pat **);
  69 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
  70 {
  71         struct grep_pat *p;
  72         struct grep_expr *x;
  73
  74         p = *list;
  75         if (!p)
  76                 return NULL;
  77         switch (p->token) {
  78         case GREP_PATTERN: /* atom */
  79         case GREP_PATTERN_HEAD:
  80         case GREP_PATTERN_BODY:
  81                 x = xcalloc(1, sizeof (struct grep_expr));
  82                 x->node = GREP_NODE_ATOM;
  83                 x->u.atom = p;
  84                 *list = p->next;
  85                 return x;
  86         case GREP_OPEN_PAREN:
  87                 *list = p->next;
  88                 x = compile_pattern_or(list);
  89                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
  90                         die("unmatched parenthesis");
  91                 *list = (*list)->next;
  92                 return x;
  93         default:
  94                 return NULL;
  95         }
  96 }
  97
  98 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
  99 {
 100         struct grep_pat *p;
 101         struct grep_expr *x;
 102
 103         p = *list;
 104         if (!p)
 105                 return NULL;
 106         switch (p->token) {
 107         case GREP_NOT:
 108                 if (!p->next)
 109                         die("--not not followed by pattern expression");
 110                 *list = p->next;
 111                 x = xcalloc(1, sizeof (struct grep_expr));
 112                 x->node = GREP_NODE_NOT;
 113                 x->u.unary = compile_pattern_not(list);
 114                 if (!x->u.unary)
 115                         die("--not followed by non pattern expression");
 116                 return x;
 117         default:
 118                 return compile_pattern_atom(list);
 119         }
 120 }
 121
 122 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 123 {
 124         struct grep_pat *p;
 125         struct grep_expr *x, *y, *z;
 126
 127         x = compile_pattern_not(list);
 128         p = *list;
 129         if (p && p->token == GREP_AND) {
 130                 if (!p->next)
 131                         die("--and not followed by pattern expression");
 132                 *list = p->next;
 133                 y = compile_pattern_and(list);
 134                 if (!y)
 135                         die("--and not followed by pattern expression");
 136                 z = xcalloc(1, sizeof (struct grep_expr));
 137                 z->node = GREP_NODE_AND;
 138                 z->u.binary.left = x;
 139                 z->u.binary.right = y;
 140                 return z;
 141         }
 142         return x;
 143 }
 144
 145 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 146 {
 147         struct grep_pat *p;
 148         struct grep_expr *x, *y, *z;
 149
 150         x = compile_pattern_and(list);
 151         p = *list;
 152         if (x && p && p->token != GREP_CLOSE_PAREN) {
 153                 y = compile_pattern_or(list);
 154                 if (!y)
 155                         die("not a pattern expression %s", p->pattern);
 156                 z = xcalloc(1, sizeof (struct grep_expr));
 157                 z->node = GREP_NODE_OR;
 158                 z->u.binary.left = x;
 159                 z->u.binary.right = y;
 160                 return z;
 161         }
 162         return x;
 163 }
 164
 165 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 166 {
 167         return compile_pattern_or(list);
 168 }
 169
 170 void compile_grep_patterns(struct grep_opt *opt)
 171 {
 172         struct grep_pat *p;
 173
 174         if (opt->all_match)
 175                 opt->extended = 1;
 176
 177         for (p = opt->pattern_list; p; p = p->next) {
 178                 switch (p->token) {
 179                 case GREP_PATTERN: /* atom */
 180                 case GREP_PATTERN_HEAD:
 181                 case GREP_PATTERN_BODY:
 182                         compile_regexp(p, opt);
 183                         break;
 184                 default:
 185                         opt->extended = 1;
 186                         break;
 187                 }
 188         }
 189
 190         if (!opt->extended)
 191                 return;
 192
 193         /* Then bundle them up in an expression.
 194          * A classic recursive descent parser would do.
 195          */
 196         p = opt->pattern_list;
 197         if (p)
 198                 opt->pattern_expression = compile_pattern_expr(&p);
 199         if (p)
 200                 die("incomplete pattern expression: %s", p->pattern);
 201 }
 202
 203 static void free_pattern_expr(struct grep_expr *x)
 204 {
 205         switch (x->node) {
 206         case GREP_NODE_ATOM:
 207                 break;
 208         case GREP_NODE_NOT:
 209                 free_pattern_expr(x->u.unary);
 210                 break;
 211         case GREP_NODE_AND:
 212         case GREP_NODE_OR:
 213                 free_pattern_expr(x->u.binary.left);
 214                 free_pattern_expr(x->u.binary.right);
 215                 break;
 216         }
 217         free(x);
 218 }
 219
 220 void free_grep_patterns(struct grep_opt *opt)
 221 {
 222         struct grep_pat *p, *n;
 223
 224         for (p = opt->pattern_list; p; p = n) {
 225                 n = p->next;
 226                 switch (p->token) {
 227                 case GREP_PATTERN: /* atom */
 228                 case GREP_PATTERN_HEAD:
 229                 case GREP_PATTERN_BODY:
 230                         regfree(&p->regexp);
 231                         break;
 232                 default:
 233                         break;
 234                 }
 235                 free(p);
 236         }
 237
 238         if (!opt->extended)
 239                 return;
 240         free_pattern_expr(opt->pattern_expression);
 241 }
 242
 243 static char *end_of_line(char *cp, unsigned long *left)
 244 {
 245         unsigned long l = *left;
 246         while (l && *cp != '\n') {
 247                 l--;
 248                 cp++;
 249         }
 250         *left = l;
 251         return cp;
 252 }
 253
 254 static int word_char(char ch)
 255 {
 256         return isalnum(ch) || ch == '_';
 257 }
 258
 259 static void show_name(struct grep_opt *opt, const char *name)
 260 {
 261         printf("%s%c", name, opt->null_following_name ? '\0' : '\n');
 262 }
 263
 264 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 265 {
 266         char *hit = strstr(line, pattern);
 267         if (!hit) {
 268                 match->rm_so = match->rm_eo = -1;
 269                 return REG_NOMATCH;
 270         }
 271         else {
 272                 match->rm_so = hit - line;
 273                 match->rm_eo = match->rm_so + strlen(pattern);
 274                 return 0;
 275         }
 276 }
 277
 278 static int strip_timestamp(char *bol, char **eol_p)
 279 {
 280         char *eol = *eol_p;
 281         int ch;
 282
 283         while (bol < --eol) {
 284                 if (*eol != '>')
 285                         continue;
 286                 *eol_p = ++eol;
 287                 ch = *eol;
 288                 *eol = '\0';
 289                 return ch;
 290         }
 291         return 0;
 292 }
 293
 294 static struct {
 295         const char *field;
 296         size_t len;
 297 } header_field[] = {
 298         { "author ", 7 },
 299         { "committer ", 10 },
 300 };
 301
 302 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 303                              enum grep_context ctx,
 304                              regmatch_t *pmatch, int eflags)
 305 {
 306         int hit = 0;
 307         int saved_ch = 0;
 308         const char *start = bol;
 309
 310         if ((p->token != GREP_PATTERN) &&
 311             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 312                 return 0;
 313
 314         if (p->token == GREP_PATTERN_HEAD) {
 315                 const char *field;
 316                 size_t len;
 317                 assert(p->field < ARRAY_SIZE(header_field));
 318                 field = header_field[p->field].field;
 319                 len = header_field[p->field].len;
 320                 if (strncmp(bol, field, len))
 321                         return 0;
 322                 bol += len;
 323                 saved_ch = strip_timestamp(bol, &eol);
 324         }
 325
 326  again:
 327         if (p->fixed)
 328                 hit = !fixmatch(p->pattern, bol, pmatch);
 329         else
 330                 hit = !regexec(&p->regexp, bol, 1, pmatch, eflags);
 331
 332         if (hit && p->word_regexp) {
 333                 if ((pmatch[0].rm_so < 0) ||
 334                     (eol - bol) <= pmatch[0].rm_so ||
 335                     (pmatch[0].rm_eo < 0) ||
 336                     (eol - bol) < pmatch[0].rm_eo)
 337                         die("regexp returned nonsense");
 338
 339                 /* Match beginning must be either beginning of the
 340                  * line, or at word boundary (i.e. the last char must
 341                  * not be a word char).  Similarly, match end must be
 342                  * either end of the line, or at word boundary
 343                  * (i.e. the next char must not be a word char).
 344                  */
 345                 if ( ((pmatch[0].rm_so == 0) ||
 346                       !word_char(bol[pmatch[0].rm_so-1])) &&
 347                      ((pmatch[0].rm_eo == (eol-bol)) ||
 348                       !word_char(bol[pmatch[0].rm_eo])) )
 349                         ;
 350                 else
 351                         hit = 0;
 352
 353                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 354                         /* There could be more than one match on the
 355                          * line, and the first match might not be
 356                          * strict word match.  But later ones could be!
 357                          * Forward to the next possible start, i.e. the
 358                          * next position following a non-word char.
 359                          */
 360                         bol = pmatch[0].rm_so + bol + 1;
 361                         while (word_char(bol[-1]) && bol < eol)
 362                                 bol++;
 363                         eflags |= REG_NOTBOL;
 364                         if (bol < eol)
 365                                 goto again;
 366                 }
 367         }
 368         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 369                 *eol = saved_ch;
 370         if (hit) {
 371                 pmatch[0].rm_so += bol - start;
 372                 pmatch[0].rm_eo += bol - start;
 373         }
 374         return hit;
 375 }
 376
 377 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 378                            enum grep_context ctx, int collect_hits)
 379 {
 380         int h = 0;
 381         regmatch_t match;
 382
 383         if (!x)
 384                 die("Not a valid grep expression");
 385         switch (x->node) {
 386         case GREP_NODE_ATOM:
 387                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 388                 break;
 389         case GREP_NODE_NOT:
 390                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 391                 break;
 392         case GREP_NODE_AND:
 393                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 394                         return 0;
 395                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 396                 break;
 397         case GREP_NODE_OR:
 398                 if (!collect_hits)
 399                         return (match_expr_eval(x->u.binary.left,
 400                                                 bol, eol, ctx, 0) ||
 401                                 match_expr_eval(x->u.binary.right,
 402                                                 bol, eol, ctx, 0));
 403                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 404                 x->u.binary.left->hit |= h;
 405                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 406                 break;
 407         default:
 408                 die("Unexpected node type (internal error) %d", x->node);
 409         }
 410         if (collect_hits)
 411                 x->hit |= h;
 412         return h;
 413 }
 414
 415 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 416                       enum grep_context ctx, int collect_hits)
 417 {
 418         struct grep_expr *x = opt->pattern_expression;
 419         return match_expr_eval(x, bol, eol, ctx, collect_hits);
 420 }
 421
 422 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 423                       enum grep_context ctx, int collect_hits)
 424 {
 425         struct grep_pat *p;
 426         regmatch_t match;
 427
 428         if (opt->extended)
 429                 return match_expr(opt, bol, eol, ctx, collect_hits);
 430
 431         /* we do not call with collect_hits without being extended */
 432         for (p = opt->pattern_list; p; p = p->next) {
 433                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 434                         return 1;
 435         }
 436         return 0;
 437 }
 438
 439 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 440                               enum grep_context ctx,
 441                               regmatch_t *pmatch, int eflags)
 442 {
 443         regmatch_t match;
 444
 445         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 446                 return 0;
 447         if (match.rm_so < 0 || match.rm_eo < 0)
 448                 return 0;
 449         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 450                 if (match.rm_so > pmatch->rm_so)
 451                         return 1;
 452                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 453                         return 1;
 454         }
 455         pmatch->rm_so = match.rm_so;
 456         pmatch->rm_eo = match.rm_eo;
 457         return 1;
 458 }
 459
 460 static int next_match(struct grep_opt *opt, char *bol, char *eol,
 461                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
 462 {
 463         struct grep_pat *p;
 464         int hit = 0;
 465
 466         pmatch->rm_so = pmatch->rm_eo = -1;
 467         if (bol < eol) {
 468                 for (p = opt->pattern_list; p; p = p->next) {
 469                         switch (p->token) {
 470                         case GREP_PATTERN: /* atom */
 471                         case GREP_PATTERN_HEAD:
 472                         case GREP_PATTERN_BODY:
 473                                 hit |= match_next_pattern(p, bol, eol, ctx,
 474                                                           pmatch, eflags);
 475                                 break;
 476                         default:
 477                                 break;
 478                         }
 479                 }
 480         }
 481         return hit;
 482 }
 483
 484 static void show_line(struct grep_opt *opt, char *bol, char *eol,
 485                       const char *name, unsigned lno, char sign)
 486 {
 487         int rest = eol - bol;
 488
 489         if (opt->null_following_name)
 490                 sign = '\0';
 491         if (opt->pathname)
 492                 printf("%s%c", name, sign);
 493         if (opt->linenum)
 494                 printf("%d%c", lno, sign);
 495         if (opt->color) {
 496                 regmatch_t match;
 497                 enum grep_context ctx = GREP_CONTEXT_BODY;
 498                 int ch = *eol;
 499                 int eflags = 0;
 500
 501                 *eol = '\0';
 502                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
 503                         printf("%.*s%s%.*s%s",
 504                                (int)match.rm_so, bol,
 505                                opt->color_match,
 506                                (int)(match.rm_eo - match.rm_so), bol + match.rm_so,
 507                                GIT_COLOR_RESET);
 508                         bol += match.rm_eo;
 509                         rest -= match.rm_eo;
 510                         eflags = REG_NOTBOL;
 511                 }
 512                 *eol = ch;
 513         }
 514         printf("%.*s\n", rest, bol);
 515 }
 516
 517 static int grep_buffer_1(struct grep_opt *opt, const char *name,
 518                          char *buf, unsigned long size, int collect_hits)
 519 {
 520         char *bol = buf;
 521         unsigned long left = size;
 522         unsigned lno = 1;
 523         struct pre_context_line {
 524                 char *bol;
 525                 char *eol;
 526         } *prev = NULL, *pcl;
 527         unsigned last_hit = 0;
 528         unsigned last_shown = 0;
 529         int binary_match_only = 0;
 530         const char *hunk_mark = "";
 531         unsigned count = 0;
 532         enum grep_context ctx = GREP_CONTEXT_HEAD;
 533
 534         if (buffer_is_binary(buf, size)) {
 535                 switch (opt->binary) {
 536                 case GREP_BINARY_DEFAULT:
 537                         binary_match_only = 1;
 538                         break;
 539                 case GREP_BINARY_NOMATCH:
 540                         return 0; /* Assume unmatch */
 541                         break;
 542                 default:
 543                         break;
 544                 }
 545         }
 546
 547         if (opt->pre_context)
 548                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 549         if (opt->pre_context || opt->post_context)
 550                 hunk_mark = "--\n";
 551
 552         while (left) {
 553                 char *eol, ch;
 554                 int hit;
 555
 556                 eol = end_of_line(bol, &left);
 557                 ch = *eol;
 558                 *eol = 0;
 559
 560                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
 561                         ctx = GREP_CONTEXT_BODY;
 562
 563                 hit = match_line(opt, bol, eol, ctx, collect_hits);
 564                 *eol = ch;
 565
 566                 if (collect_hits)
 567                         goto next_line;
 568
 569                 /* "grep -v -e foo -e bla" should list lines
 570                  * that do not have either, so inversion should
 571                  * be done outside.
 572                  */
 573                 if (opt->invert)
 574                         hit = !hit;
 575                 if (opt->unmatch_name_only) {
 576                         if (hit)
 577                                 return 0;
 578                         goto next_line;
 579                 }
 580                 if (hit) {
 581                         count++;
 582                         if (opt->status_only)
 583                                 return 1;
 584                         if (binary_match_only) {
 585                                 printf("Binary file %s matches\n", name);
 586                                 return 1;
 587                         }
 588                         if (opt->name_only) {
 589                                 show_name(opt, name);
 590                                 return 1;
 591                         }
 592                         /* Hit at this line.  If we haven't shown the
 593                          * pre-context lines, we would need to show them.
 594                          * When asked to do "count", this still show
 595                          * the context which is nonsense, but the user
 596                          * deserves to get that ;-).
 597                          */
 598                         if (opt->pre_context) {
 599                                 unsigned from;
 600                                 if (opt->pre_context < lno)
 601                                         from = lno - opt->pre_context;
 602                                 else
 603                                         from = 1;
 604                                 if (from <= last_shown)
 605                                         from = last_shown + 1;
 606                                 if (last_shown && from != last_shown + 1)
 607                                         fputs(hunk_mark, stdout);
 608                                 while (from < lno) {
 609                                         pcl = &prev[lno-from-1];
 610                                         show_line(opt, pcl->bol, pcl->eol,
 611                                                   name, from, '-');
 612                                         from++;
 613                                 }
 614                                 last_shown = lno-1;
 615                         }
 616                         if (last_shown && lno != last_shown + 1)
 617                                 fputs(hunk_mark, stdout);
 618                         if (!opt->count)
 619                                 show_line(opt, bol, eol, name, lno, ':');
 620                         last_shown = last_hit = lno;
 621                 }
 622                 else if (last_hit &&
 623                          lno <= last_hit + opt->post_context) {
 624                         /* If the last hit is within the post context,
 625                          * we need to show this line.
 626                          */
 627                         if (last_shown && lno != last_shown + 1)
 628                                 fputs(hunk_mark, stdout);
 629                         show_line(opt, bol, eol, name, lno, '-');
 630                         last_shown = lno;
 631                 }
 632                 if (opt->pre_context) {
 633                         memmove(prev+1, prev,
 634                                 (opt->pre_context-1) * sizeof(*prev));
 635                         prev->bol = bol;
 636                         prev->eol = eol;
 637                 }
 638
 639         next_line:
 640                 bol = eol + 1;
 641                 if (!left)
 642                         break;
 643                 left--;
 644                 lno++;
 645         }
 646
 647         free(prev);
 648         if (collect_hits)
 649                 return 0;
 650
 651         if (opt->status_only)
 652                 return 0;
 653         if (opt->unmatch_name_only) {
 654                 /* We did not see any hit, so we want to show this */
 655                 show_name(opt, name);
 656                 return 1;
 657         }
 658
 659         /* NEEDSWORK:
 660          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 661          * which feels mostly useless but sometimes useful.  Maybe
 662          * make it another option?  For now suppress them.
 663          */
 664         if (opt->count && count)
 665                 printf("%s%c%u\n", name,
 666                        opt->null_following_name ? '\0' : ':', count);
 667         return !!last_hit;
 668 }
 669
 670 static void clr_hit_marker(struct grep_expr *x)
 671 {
 672         /* All-hit markers are meaningful only at the very top level
 673          * OR node.
 674          */
 675         while (1) {
 676                 x->hit = 0;
 677                 if (x->node != GREP_NODE_OR)
 678                         return;
 679                 x->u.binary.left->hit = 0;
 680                 x = x->u.binary.right;
 681         }
 682 }
 683
 684 static int chk_hit_marker(struct grep_expr *x)
 685 {
 686         /* Top level nodes have hit markers.  See if they all are hits */
 687         while (1) {
 688                 if (x->node != GREP_NODE_OR)
 689                         return x->hit;
 690                 if (!x->u.binary.left->hit)
 691                         return 0;
 692                 x = x->u.binary.right;
 693         }
 694 }
 695
 696 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
 697 {
 698         /*
 699          * we do not have to do the two-pass grep when we do not check
 700          * buffer-wide "all-match".
 701          */
 702         if (!opt->all_match)
 703                 return grep_buffer_1(opt, name, buf, size, 0);
 704
 705         /* Otherwise the toplevel "or" terms hit a bit differently.
 706          * We first clear hit markers from them.
 707          */
 708         clr_hit_marker(opt->pattern_expression);
 709         grep_buffer_1(opt, name, buf, size, 1);
 710
 711         if (!chk_hit_marker(opt->pattern_expression))
 712                 return 0;
 713
 714         return grep_buffer_1(opt, name, buf, size, 0);
 715 }