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