grep.c

   1 #include "cache.h"
   2 #include <regex.h>
   3 #include "grep.h"
   4
   5 void append_grep_pattern(struct grep_opt *opt, const char *pat,
   6                          const char *origin, int no, enum grep_pat_token t)
   7 {
   8         struct grep_pat *p = xcalloc(1, sizeof(*p));
   9         p->pattern = pat;
  10         p->origin = origin;
  11         p->no = no;
  12         p->token = t;
  13         *opt->pattern_tail = p;
  14         opt->pattern_tail = &p->next;
  15         p->next = NULL;
  16 }
  17
  18 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
  19 {
  20         int err = regcomp(&p->regexp, p->pattern, opt->regflags);
  21         if (err) {
  22                 char errbuf[1024];
  23                 char where[1024];
  24                 if (p->no)
  25                         sprintf(where, "In '%s' at %d, ",
  26                                 p->origin, p->no);
  27                 else if (p->origin)
  28                         sprintf(where, "%s, ", p->origin);
  29                 else
  30                         where[0] = 0;
  31                 regerror(err, &p->regexp, errbuf, 1024);
  32                 regfree(&p->regexp);
  33                 die("%s'%s': %s", where, p->pattern, errbuf);
  34         }
  35 }
  36
  37 static struct grep_expr *compile_pattern_expr(struct grep_pat **);
  38 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
  39 {
  40         struct grep_pat *p;
  41         struct grep_expr *x;
  42
  43         p = *list;
  44         switch (p->token) {
  45         case GREP_PATTERN: /* atom */
  46         case GREP_PATTERN_HEAD:
  47         case GREP_PATTERN_BODY:
  48                 x = xcalloc(1, sizeof (struct grep_expr));
  49                 x->node = GREP_NODE_ATOM;
  50                 x->u.atom = p;
  51                 *list = p->next;
  52                 return x;
  53         case GREP_OPEN_PAREN:
  54                 *list = p->next;
  55                 x = compile_pattern_expr(list);
  56                 if (!x)
  57                         return NULL;
  58                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
  59                         die("unmatched parenthesis");
  60                 *list = (*list)->next;
  61                 return x;
  62         default:
  63                 return NULL;
  64         }
  65 }
  66
  67 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
  68 {
  69         struct grep_pat *p;
  70         struct grep_expr *x;
  71
  72         p = *list;
  73         switch (p->token) {
  74         case GREP_NOT:
  75                 if (!p->next)
  76                         die("--not not followed by pattern expression");
  77                 *list = p->next;
  78                 x = xcalloc(1, sizeof (struct grep_expr));
  79                 x->node = GREP_NODE_NOT;
  80                 x->u.unary = compile_pattern_not(list);
  81                 if (!x->u.unary)
  82                         die("--not followed by non pattern expression");
  83                 return x;
  84         default:
  85                 return compile_pattern_atom(list);
  86         }
  87 }
  88
  89 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
  90 {
  91         struct grep_pat *p;
  92         struct grep_expr *x, *y, *z;
  93
  94         x = compile_pattern_not(list);
  95         p = *list;
  96         if (p && p->token == GREP_AND) {
  97                 if (!p->next)
  98                         die("--and not followed by pattern expression");
  99                 *list = p->next;
 100                 y = compile_pattern_and(list);
 101                 if (!y)
 102                         die("--and not followed by pattern expression");
 103                 z = xcalloc(1, sizeof (struct grep_expr));
 104                 z->node = GREP_NODE_AND;
 105                 z->u.binary.left = x;
 106                 z->u.binary.right = y;
 107                 return z;
 108         }
 109         return x;
 110 }
 111
 112 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 113 {
 114         struct grep_pat *p;
 115         struct grep_expr *x, *y, *z;
 116
 117         x = compile_pattern_and(list);
 118         p = *list;
 119         if (x && p && p->token != GREP_CLOSE_PAREN) {
 120                 y = compile_pattern_or(list);
 121                 if (!y)
 122                         die("not a pattern expression %s", p->pattern);
 123                 z = xcalloc(1, sizeof (struct grep_expr));
 124                 z->node = GREP_NODE_OR;
 125                 z->u.binary.left = x;
 126                 z->u.binary.right = y;
 127                 return z;
 128         }
 129         return x;
 130 }
 131
 132 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 133 {
 134         return compile_pattern_or(list);
 135 }
 136
 137 void compile_grep_patterns(struct grep_opt *opt)
 138 {
 139         struct grep_pat *p;
 140
 141         if (opt->fixed)
 142                 return;
 143
 144         /* First compile regexps */
 145         for (p = opt->pattern_list; p; p = p->next) {
 146                 switch (p->token) {
 147                 case GREP_PATTERN: /* atom */
 148                 case GREP_PATTERN_HEAD:
 149                 case GREP_PATTERN_BODY:
 150                         compile_regexp(p, opt);
 151                         break;
 152                 default:
 153                         opt->extended = 1;
 154                         break;
 155                 }
 156         }
 157
 158         if (!opt->extended)
 159                 return;
 160
 161         /* Then bundle them up in an expression.
 162          * A classic recursive descent parser would do.
 163          */
 164         p = opt->pattern_list;
 165         opt->pattern_expression = compile_pattern_expr(&p);
 166         if (p)
 167                 die("incomplete pattern expression: %s", p->pattern);
 168 }
 169
 170 static void free_pattern_expr(struct grep_expr *x)
 171 {
 172         switch (x->node) {
 173         case GREP_NODE_ATOM:
 174                 break;
 175         case GREP_NODE_NOT:
 176                 free_pattern_expr(x->u.unary);
 177                 break;
 178         case GREP_NODE_AND:
 179         case GREP_NODE_OR:
 180                 free_pattern_expr(x->u.binary.left);
 181                 free_pattern_expr(x->u.binary.right);
 182                 break;
 183         }
 184         free(x);
 185 }
 186
 187 void free_grep_patterns(struct grep_opt *opt)
 188 {
 189         struct grep_pat *p, *n;
 190
 191         for (p = opt->pattern_list; p; p = n) {
 192                 n = p->next;
 193                 switch (p->token) {
 194                 case GREP_PATTERN: /* atom */
 195                 case GREP_PATTERN_HEAD:
 196                 case GREP_PATTERN_BODY:
 197                         regfree(&p->regexp);
 198                         break;
 199                 default:
 200                         break;
 201                 }
 202                 free(p);
 203         }
 204
 205         if (!opt->extended)
 206                 return;
 207         free_pattern_expr(opt->pattern_expression);
 208 }
 209
 210 static char *end_of_line(char *cp, unsigned long *left)
 211 {
 212         unsigned long l = *left;
 213         while (l && *cp != '\n') {
 214                 l--;
 215                 cp++;
 216         }
 217         *left = l;
 218         return cp;
 219 }
 220
 221 static int word_char(char ch)
 222 {
 223         return isalnum(ch) || ch == '_';
 224 }
 225
 226 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 227                       const char *name, unsigned lno, char sign)
 228 {
 229         if (opt->pathname)
 230                 printf("%s%c", name, sign);
 231         if (opt->linenum)
 232                 printf("%d%c", lno, sign);
 233         printf("%.*s\n", (int)(eol-bol), bol);
 234 }
 235
 236 /*
 237  * NEEDSWORK: share code with diff.c
 238  */
 239 #define FIRST_FEW_BYTES 8000
 240 static int buffer_is_binary(const char *ptr, unsigned long size)
 241 {
 242         if (FIRST_FEW_BYTES < size)
 243                 size = FIRST_FEW_BYTES;
 244         return !!memchr(ptr, 0, size);
 245 }
 246
 247 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 248 {
 249         char *hit = strstr(line, pattern);
 250         if (!hit) {
 251                 match->rm_so = match->rm_eo = -1;
 252                 return REG_NOMATCH;
 253         }
 254         else {
 255                 match->rm_so = hit - line;
 256                 match->rm_eo = match->rm_so + strlen(pattern);
 257                 return 0;
 258         }
 259 }
 260
 261 static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol, char *eol, enum grep_context ctx)
 262 {
 263         int hit = 0;
 264         int at_true_bol = 1;
 265         regmatch_t pmatch[10];
 266
 267         if ((p->token != GREP_PATTERN) &&
 268             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 269                 return 0;
 270
 271  again:
 272         if (!opt->fixed) {
 273                 regex_t *exp = &p->regexp;
 274                 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 275                                pmatch, 0);
 276         }
 277         else {
 278                 hit = !fixmatch(p->pattern, bol, pmatch);
 279         }
 280
 281         if (hit && opt->word_regexp) {
 282                 if ((pmatch[0].rm_so < 0) ||
 283                     (eol - bol) <= pmatch[0].rm_so ||
 284                     (pmatch[0].rm_eo < 0) ||
 285                     (eol - bol) < pmatch[0].rm_eo)
 286                         die("regexp returned nonsense");
 287
 288                 /* Match beginning must be either beginning of the
 289                  * line, or at word boundary (i.e. the last char must
 290                  * not be a word char).  Similarly, match end must be
 291                  * either end of the line, or at word boundary
 292                  * (i.e. the next char must not be a word char).
 293                  */
 294                 if ( ((pmatch[0].rm_so == 0 && at_true_bol) ||
 295                       !word_char(bol[pmatch[0].rm_so-1])) &&
 296                      ((pmatch[0].rm_eo == (eol-bol)) ||
 297                       !word_char(bol[pmatch[0].rm_eo])) )
 298                         ;
 299                 else
 300                         hit = 0;
 301
 302                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 303                         /* There could be more than one match on the
 304                          * line, and the first match might not be
 305                          * strict word match.  But later ones could be!
 306                          */
 307                         bol = pmatch[0].rm_so + bol + 1;
 308                         at_true_bol = 0;
 309                         goto again;
 310                 }
 311         }
 312         return hit;
 313 }
 314
 315 static int match_expr_eval(struct grep_opt *opt,
 316                            struct grep_expr *x,
 317                            char *bol, char *eol,
 318                            enum grep_context ctx)
 319 {
 320         switch (x->node) {
 321         case GREP_NODE_ATOM:
 322                 return match_one_pattern(opt, x->u.atom, bol, eol, ctx);
 323                 break;
 324         case GREP_NODE_NOT:
 325                 return !match_expr_eval(opt, x->u.unary, bol, eol, ctx);
 326         case GREP_NODE_AND:
 327                 return (match_expr_eval(opt, x->u.binary.left, bol, eol, ctx) &&
 328                         match_expr_eval(opt, x->u.binary.right, bol, eol, ctx));
 329         case GREP_NODE_OR:
 330                 return (match_expr_eval(opt, x->u.binary.left, bol, eol, ctx) ||
 331                         match_expr_eval(opt, x->u.binary.right, bol, eol, ctx));
 332         }
 333         die("Unexpected node type (internal error) %d\n", x->node);
 334 }
 335
 336 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 337                       enum grep_context ctx)
 338 {
 339         struct grep_expr *x = opt->pattern_expression;
 340         return match_expr_eval(opt, x, bol, eol, ctx);
 341 }
 342
 343 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 344                       enum grep_context ctx)
 345 {
 346         struct grep_pat *p;
 347         if (opt->extended)
 348                 return match_expr(opt, bol, eol, ctx);
 349         for (p = opt->pattern_list; p; p = p->next) {
 350                 if (match_one_pattern(opt, p, bol, eol, ctx))
 351                         return 1;
 352         }
 353         return 0;
 354 }
 355
 356 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
 357 {
 358         char *bol = buf;
 359         unsigned long left = size;
 360         unsigned lno = 1;
 361         struct pre_context_line {
 362                 char *bol;
 363                 char *eol;
 364         } *prev = NULL, *pcl;
 365         unsigned last_hit = 0;
 366         unsigned last_shown = 0;
 367         int binary_match_only = 0;
 368         const char *hunk_mark = "";
 369         unsigned count = 0;
 370         enum grep_context ctx = GREP_CONTEXT_HEAD;
 371
 372         if (buffer_is_binary(buf, size)) {
 373                 switch (opt->binary) {
 374                 case GREP_BINARY_DEFAULT:
 375                         binary_match_only = 1;
 376                         break;
 377                 case GREP_BINARY_NOMATCH:
 378                         return 0; /* Assume unmatch */
 379                         break;
 380                 default:
 381                         break;
 382                 }
 383         }
 384
 385         if (opt->pre_context)
 386                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 387         if (opt->pre_context || opt->post_context)
 388                 hunk_mark = "--\n";
 389
 390         while (left) {
 391                 char *eol, ch;
 392                 int hit = 0;
 393
 394                 eol = end_of_line(bol, &left);
 395                 ch = *eol;
 396                 *eol = 0;
 397
 398                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
 399                         ctx = GREP_CONTEXT_BODY;
 400
 401                 hit = match_line(opt, bol, eol, ctx);
 402                 *eol = ch;
 403
 404                 /* "grep -v -e foo -e bla" should list lines
 405                  * that do not have either, so inversion should
 406                  * be done outside.
 407                  */
 408                 if (opt->invert)
 409                         hit = !hit;
 410                 if (opt->unmatch_name_only) {
 411                         if (hit)
 412                                 return 0;
 413                         goto next_line;
 414                 }
 415                 if (hit) {
 416                         count++;
 417                         if (opt->status_only)
 418                                 return 1;
 419                         if (binary_match_only) {
 420                                 printf("Binary file %s matches\n", name);
 421                                 return 1;
 422                         }
 423                         if (opt->name_only) {
 424                                 printf("%s\n", name);
 425                                 return 1;
 426                         }
 427                         /* Hit at this line.  If we haven't shown the
 428                          * pre-context lines, we would need to show them.
 429                          * When asked to do "count", this still show
 430                          * the context which is nonsense, but the user
 431                          * deserves to get that ;-).
 432                          */
 433                         if (opt->pre_context) {
 434                                 unsigned from;
 435                                 if (opt->pre_context < lno)
 436                                         from = lno - opt->pre_context;
 437                                 else
 438                                         from = 1;
 439                                 if (from <= last_shown)
 440                                         from = last_shown + 1;
 441                                 if (last_shown && from != last_shown + 1)
 442                                         printf(hunk_mark);
 443                                 while (from < lno) {
 444                                         pcl = &prev[lno-from-1];
 445                                         show_line(opt, pcl->bol, pcl->eol,
 446                                                   name, from, '-');
 447                                         from++;
 448                                 }
 449                                 last_shown = lno-1;
 450                         }
 451                         if (last_shown && lno != last_shown + 1)
 452                                 printf(hunk_mark);
 453                         if (!opt->count)
 454                                 show_line(opt, bol, eol, name, lno, ':');
 455                         last_shown = last_hit = lno;
 456                 }
 457                 else if (last_hit &&
 458                          lno <= last_hit + opt->post_context) {
 459                         /* If the last hit is within the post context,
 460                          * we need to show this line.
 461                          */
 462                         if (last_shown && lno != last_shown + 1)
 463                                 printf(hunk_mark);
 464                         show_line(opt, bol, eol, name, lno, '-');
 465                         last_shown = lno;
 466                 }
 467                 if (opt->pre_context) {
 468                         memmove(prev+1, prev,
 469                                 (opt->pre_context-1) * sizeof(*prev));
 470                         prev->bol = bol;
 471                         prev->eol = eol;
 472                 }
 473
 474         next_line:
 475                 bol = eol + 1;
 476                 if (!left)
 477                         break;
 478                 left--;
 479                 lno++;
 480         }
 481
 482         free(prev);
 483
 484         if (opt->status_only)
 485                 return 0;
 486         if (opt->unmatch_name_only) {
 487                 /* We did not see any hit, so we want to show this */
 488                 printf("%s\n", name);
 489                 return 1;
 490         }
 491
 492         /* NEEDSWORK:
 493          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 494          * which feels mostly useless but sometimes useful.  Maybe
 495          * make it another option?  For now suppress them.
 496          */
 497         if (opt->count && count)
 498                 printf("%s:%u\n", name, count);
 499         return !!last_hit;
 500 }
 501