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