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