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_line(struct grep_opt *opt, const char *bol, const char *eol,
 257                       const char *name, unsigned lno, char sign)
 258 {
 259         if (opt->null_following_name)
 260                 sign = '\0';
 261         if (opt->pathname)
 262                 printf("%s%c", name, sign);
 263         if (opt->linenum)
 264                 printf("%d%c", lno, sign);
 265         printf("%.*s\n", (int)(eol-bol), bol);
 266 }
 267
 268 static void show_name(struct grep_opt *opt, const char *name)
 269 {
 270         printf("%s%c", name, opt->null_following_name ? '\0' : '\n');
 271 }
 272
 273 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 274 {
 275         char *hit = strstr(line, pattern);
 276         if (!hit) {
 277                 match->rm_so = match->rm_eo = -1;
 278                 return REG_NOMATCH;
 279         }
 280         else {
 281                 match->rm_so = hit - line;
 282                 match->rm_eo = match->rm_so + strlen(pattern);
 283                 return 0;
 284         }
 285 }
 286
 287 static int strip_timestamp(char *bol, char **eol_p)
 288 {
 289         char *eol = *eol_p;
 290         int ch;
 291
 292         while (bol < --eol) {
 293                 if (*eol != '>')
 294                         continue;
 295                 *eol_p = ++eol;
 296                 ch = *eol;
 297                 *eol = '\0';
 298                 return ch;
 299         }
 300         return 0;
 301 }
 302
 303 static struct {
 304         const char *field;
 305         size_t len;
 306 } header_field[] = {
 307         { "author ", 7 },
 308         { "committer ", 10 },
 309 };
 310
 311 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 312                              enum grep_context ctx)
 313 {
 314         int hit = 0;
 315         int saved_ch = 0;
 316         regmatch_t pmatch[10];
 317
 318         if ((p->token != GREP_PATTERN) &&
 319             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 320                 return 0;
 321
 322         if (p->token == GREP_PATTERN_HEAD) {
 323                 const char *field;
 324                 size_t len;
 325                 assert(p->field < ARRAY_SIZE(header_field));
 326                 field = header_field[p->field].field;
 327                 len = header_field[p->field].len;
 328                 if (strncmp(bol, field, len))
 329                         return 0;
 330                 bol += len;
 331                 saved_ch = strip_timestamp(bol, &eol);
 332         }
 333
 334  again:
 335         if (!p->fixed) {
 336                 regex_t *exp = &p->regexp;
 337                 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 338                                pmatch, 0);
 339         }
 340         else {
 341                 hit = !fixmatch(p->pattern, bol, pmatch);
 342         }
 343
 344         if (hit && p->word_regexp) {
 345                 if ((pmatch[0].rm_so < 0) ||
 346                     (eol - bol) <= pmatch[0].rm_so ||
 347                     (pmatch[0].rm_eo < 0) ||
 348                     (eol - bol) < pmatch[0].rm_eo)
 349                         die("regexp returned nonsense");
 350
 351                 /* Match beginning must be either beginning of the
 352                  * line, or at word boundary (i.e. the last char must
 353                  * not be a word char).  Similarly, match end must be
 354                  * either end of the line, or at word boundary
 355                  * (i.e. the next char must not be a word char).
 356                  */
 357                 if ( ((pmatch[0].rm_so == 0) ||
 358                       !word_char(bol[pmatch[0].rm_so-1])) &&
 359                      ((pmatch[0].rm_eo == (eol-bol)) ||
 360                       !word_char(bol[pmatch[0].rm_eo])) )
 361                         ;
 362                 else
 363                         hit = 0;
 364
 365                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 366                         /* There could be more than one match on the
 367                          * line, and the first match might not be
 368                          * strict word match.  But later ones could be!
 369                          * Forward to the next possible start, i.e. the
 370                          * next position following a non-word char.
 371                          */
 372                         bol = pmatch[0].rm_so + bol + 1;
 373                         while (word_char(bol[-1]) && bol < eol)
 374                                 bol++;
 375                         if (bol < eol)
 376                                 goto again;
 377                 }
 378         }
 379         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 380                 *eol = saved_ch;
 381         return hit;
 382 }
 383
 384 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 385                            enum grep_context ctx, int collect_hits)
 386 {
 387         int h = 0;
 388
 389         switch (x->node) {
 390         case GREP_NODE_ATOM:
 391                 h = match_one_pattern(x->u.atom, bol, eol, ctx);
 392                 break;
 393         case GREP_NODE_NOT:
 394                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 395                 break;
 396         case GREP_NODE_AND:
 397                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 398                         return 0;
 399                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 400                 break;
 401         case GREP_NODE_OR:
 402                 if (!collect_hits)
 403                         return (match_expr_eval(x->u.binary.left,
 404                                                 bol, eol, ctx, 0) ||
 405                                 match_expr_eval(x->u.binary.right,
 406                                                 bol, eol, ctx, 0));
 407                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 408                 x->u.binary.left->hit |= h;
 409                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 410                 break;
 411         default:
 412                 die("Unexpected node type (internal error) %d", x->node);
 413         }
 414         if (collect_hits)
 415                 x->hit |= h;
 416         return h;
 417 }
 418
 419 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 420                       enum grep_context ctx, int collect_hits)
 421 {
 422         struct grep_expr *x = opt->pattern_expression;
 423         return match_expr_eval(x, bol, eol, ctx, collect_hits);
 424 }
 425
 426 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 427                       enum grep_context ctx, int collect_hits)
 428 {
 429         struct grep_pat *p;
 430         if (opt->extended)
 431                 return match_expr(opt, bol, eol, ctx, collect_hits);
 432
 433         /* we do not call with collect_hits without being extended */
 434         for (p = opt->pattern_list; p; p = p->next) {
 435                 if (match_one_pattern(p, bol, eol, ctx))
 436                         return 1;
 437         }
 438         return 0;
 439 }
 440
 441 static int grep_buffer_1(struct grep_opt *opt, const char *name,
 442                          char *buf, unsigned long size, int collect_hits)
 443 {
 444         char *bol = buf;
 445         unsigned long left = size;
 446         unsigned lno = 1;
 447         struct pre_context_line {
 448                 char *bol;
 449                 char *eol;
 450         } *prev = NULL, *pcl;
 451         unsigned last_hit = 0;
 452         unsigned last_shown = 0;
 453         int binary_match_only = 0;
 454         const char *hunk_mark = "";
 455         unsigned count = 0;
 456         enum grep_context ctx = GREP_CONTEXT_HEAD;
 457
 458         if (buffer_is_binary(buf, size)) {
 459                 switch (opt->binary) {
 460                 case GREP_BINARY_DEFAULT:
 461                         binary_match_only = 1;
 462                         break;
 463                 case GREP_BINARY_NOMATCH:
 464                         return 0; /* Assume unmatch */
 465                         break;
 466                 default:
 467                         break;
 468                 }
 469         }
 470
 471         if (opt->pre_context)
 472                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 473         if (opt->pre_context || opt->post_context)
 474                 hunk_mark = "--\n";
 475
 476         while (left) {
 477                 char *eol, ch;
 478                 int hit;
 479
 480                 eol = end_of_line(bol, &left);
 481                 ch = *eol;
 482                 *eol = 0;
 483
 484                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
 485                         ctx = GREP_CONTEXT_BODY;
 486
 487                 hit = match_line(opt, bol, eol, ctx, collect_hits);
 488                 *eol = ch;
 489
 490                 if (collect_hits)
 491                         goto next_line;
 492
 493                 /* "grep -v -e foo -e bla" should list lines
 494                  * that do not have either, so inversion should
 495                  * be done outside.
 496                  */
 497                 if (opt->invert)
 498                         hit = !hit;
 499                 if (opt->unmatch_name_only) {
 500                         if (hit)
 501                                 return 0;
 502                         goto next_line;
 503                 }
 504                 if (hit) {
 505                         count++;
 506                         if (opt->status_only)
 507                                 return 1;
 508                         if (binary_match_only) {
 509                                 printf("Binary file %s matches\n", name);
 510                                 return 1;
 511                         }
 512                         if (opt->name_only) {
 513                                 show_name(opt, name);
 514                                 return 1;
 515                         }
 516                         /* Hit at this line.  If we haven't shown the
 517                          * pre-context lines, we would need to show them.
 518                          * When asked to do "count", this still show
 519                          * the context which is nonsense, but the user
 520                          * deserves to get that ;-).
 521                          */
 522                         if (opt->pre_context) {
 523                                 unsigned from;
 524                                 if (opt->pre_context < lno)
 525                                         from = lno - opt->pre_context;
 526                                 else
 527                                         from = 1;
 528                                 if (from <= last_shown)
 529                                         from = last_shown + 1;
 530                                 if (last_shown && from != last_shown + 1)
 531                                         fputs(hunk_mark, stdout);
 532                                 while (from < lno) {
 533                                         pcl = &prev[lno-from-1];
 534                                         show_line(opt, pcl->bol, pcl->eol,
 535                                                   name, from, '-');
 536                                         from++;
 537                                 }
 538                                 last_shown = lno-1;
 539                         }
 540                         if (last_shown && lno != last_shown + 1)
 541                                 fputs(hunk_mark, stdout);
 542                         if (!opt->count)
 543                                 show_line(opt, bol, eol, name, lno, ':');
 544                         last_shown = last_hit = lno;
 545                 }
 546                 else if (last_hit &&
 547                          lno <= last_hit + opt->post_context) {
 548                         /* If the last hit is within the post context,
 549                          * we need to show this line.
 550                          */
 551                         if (last_shown && lno != last_shown + 1)
 552                                 fputs(hunk_mark, stdout);
 553                         show_line(opt, bol, eol, name, lno, '-');
 554                         last_shown = lno;
 555                 }
 556                 if (opt->pre_context) {
 557                         memmove(prev+1, prev,
 558                                 (opt->pre_context-1) * sizeof(*prev));
 559                         prev->bol = bol;
 560                         prev->eol = eol;
 561                 }
 562
 563         next_line:
 564                 bol = eol + 1;
 565                 if (!left)
 566                         break;
 567                 left--;
 568                 lno++;
 569         }
 570
 571         free(prev);
 572         if (collect_hits)
 573                 return 0;
 574
 575         if (opt->status_only)
 576                 return 0;
 577         if (opt->unmatch_name_only) {
 578                 /* We did not see any hit, so we want to show this */
 579                 show_name(opt, name);
 580                 return 1;
 581         }
 582
 583         /* NEEDSWORK:
 584          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 585          * which feels mostly useless but sometimes useful.  Maybe
 586          * make it another option?  For now suppress them.
 587          */
 588         if (opt->count && count)
 589                 printf("%s%c%u\n", name,
 590                        opt->null_following_name ? '\0' : ':', count);
 591         return !!last_hit;
 592 }
 593
 594 static void clr_hit_marker(struct grep_expr *x)
 595 {
 596         /* All-hit markers are meaningful only at the very top level
 597          * OR node.
 598          */
 599         while (1) {
 600                 x->hit = 0;
 601                 if (x->node != GREP_NODE_OR)
 602                         return;
 603                 x->u.binary.left->hit = 0;
 604                 x = x->u.binary.right;
 605         }
 606 }
 607
 608 static int chk_hit_marker(struct grep_expr *x)
 609 {
 610         /* Top level nodes have hit markers.  See if they all are hits */
 611         while (1) {
 612                 if (x->node != GREP_NODE_OR)
 613                         return x->hit;
 614                 if (!x->u.binary.left->hit)
 615                         return 0;
 616                 x = x->u.binary.right;
 617         }
 618 }
 619
 620 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
 621 {
 622         /*
 623          * we do not have to do the two-pass grep when we do not check
 624          * buffer-wide "all-match".
 625          */
 626         if (!opt->all_match)
 627                 return grep_buffer_1(opt, name, buf, size, 0);
 628
 629         /* Otherwise the toplevel "or" terms hit a bit differently.
 630          * We first clear hit markers from them.
 631          */
 632         clr_hit_marker(opt->pattern_expression);
 633         grep_buffer_1(opt, name, buf, size, 1);
 634
 635         if (!chk_hit_marker(opt->pattern_expression))
 636                 return 0;
 637
 638         return grep_buffer_1(opt, name, buf, size, 0);
 639 }