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 char *end_of_line(char *cp, unsigned long *left)
 171 {
 172         unsigned long l = *left;
 173         while (l && *cp != '\n') {
 174                 l--;
 175                 cp++;
 176         }
 177         *left = l;
 178         return cp;
 179 }
 180
 181 static int word_char(char ch)
 182 {
 183         return isalnum(ch) || ch == '_';
 184 }
 185
 186 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 187                       const char *name, unsigned lno, char sign)
 188 {
 189         if (opt->pathname)
 190                 printf("%s%c", name, sign);
 191         if (opt->linenum)
 192                 printf("%d%c", lno, sign);
 193         printf("%.*s\n", (int)(eol-bol), bol);
 194 }
 195
 196 /*
 197  * NEEDSWORK: share code with diff.c
 198  */
 199 #define FIRST_FEW_BYTES 8000
 200 static int buffer_is_binary(const char *ptr, unsigned long size)
 201 {
 202         if (FIRST_FEW_BYTES < size)
 203                 size = FIRST_FEW_BYTES;
 204         return !!memchr(ptr, 0, size);
 205 }
 206
 207 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 208 {
 209         char *hit = strstr(line, pattern);
 210         if (!hit) {
 211                 match->rm_so = match->rm_eo = -1;
 212                 return REG_NOMATCH;
 213         }
 214         else {
 215                 match->rm_so = hit - line;
 216                 match->rm_eo = match->rm_so + strlen(pattern);
 217                 return 0;
 218         }
 219 }
 220
 221 static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol, char *eol, enum grep_context ctx)
 222 {
 223         int hit = 0;
 224         int at_true_bol = 1;
 225         regmatch_t pmatch[10];
 226
 227         if ((p->token != GREP_PATTERN) &&
 228             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 229                 return 0;
 230
 231  again:
 232         if (!opt->fixed) {
 233                 regex_t *exp = &p->regexp;
 234                 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 235                                pmatch, 0);
 236         }
 237         else {
 238                 hit = !fixmatch(p->pattern, bol, pmatch);
 239         }
 240
 241         if (hit && opt->word_regexp) {
 242                 if ((pmatch[0].rm_so < 0) ||
 243                     (eol - bol) <= pmatch[0].rm_so ||
 244                     (pmatch[0].rm_eo < 0) ||
 245                     (eol - bol) < pmatch[0].rm_eo)
 246                         die("regexp returned nonsense");
 247
 248                 /* Match beginning must be either beginning of the
 249                  * line, or at word boundary (i.e. the last char must
 250                  * not be a word char).  Similarly, match end must be
 251                  * either end of the line, or at word boundary
 252                  * (i.e. the next char must not be a word char).
 253                  */
 254                 if ( ((pmatch[0].rm_so == 0 && at_true_bol) ||
 255                       !word_char(bol[pmatch[0].rm_so-1])) &&
 256                      ((pmatch[0].rm_eo == (eol-bol)) ||
 257                       !word_char(bol[pmatch[0].rm_eo])) )
 258                         ;
 259                 else
 260                         hit = 0;
 261
 262                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 263                         /* There could be more than one match on the
 264                          * line, and the first match might not be
 265                          * strict word match.  But later ones could be!
 266                          */
 267                         bol = pmatch[0].rm_so + bol + 1;
 268                         at_true_bol = 0;
 269                         goto again;
 270                 }
 271         }
 272         return hit;
 273 }
 274
 275 static int match_expr_eval(struct grep_opt *opt,
 276                            struct grep_expr *x,
 277                            char *bol, char *eol,
 278                            enum grep_context ctx)
 279 {
 280         switch (x->node) {
 281         case GREP_NODE_ATOM:
 282                 return match_one_pattern(opt, x->u.atom, bol, eol, ctx);
 283                 break;
 284         case GREP_NODE_NOT:
 285                 return !match_expr_eval(opt, x->u.unary, bol, eol, ctx);
 286         case GREP_NODE_AND:
 287                 return (match_expr_eval(opt, x->u.binary.left, bol, eol, ctx) &&
 288                         match_expr_eval(opt, x->u.binary.right, bol, eol, ctx));
 289         case GREP_NODE_OR:
 290                 return (match_expr_eval(opt, x->u.binary.left, bol, eol, ctx) ||
 291                         match_expr_eval(opt, x->u.binary.right, bol, eol, ctx));
 292         }
 293         die("Unexpected node type (internal error) %d\n", x->node);
 294 }
 295
 296 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 297                       enum grep_context ctx)
 298 {
 299         struct grep_expr *x = opt->pattern_expression;
 300         return match_expr_eval(opt, x, bol, eol, ctx);
 301 }
 302
 303 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 304                       enum grep_context ctx)
 305 {
 306         struct grep_pat *p;
 307         if (opt->extended)
 308                 return match_expr(opt, bol, eol, ctx);
 309         for (p = opt->pattern_list; p; p = p->next) {
 310                 if (match_one_pattern(opt, p, bol, eol, ctx))
 311                         return 1;
 312         }
 313         return 0;
 314 }
 315
 316 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
 317 {
 318         char *bol = buf;
 319         unsigned long left = size;
 320         unsigned lno = 1;
 321         struct pre_context_line {
 322                 char *bol;
 323                 char *eol;
 324         } *prev = NULL, *pcl;
 325         unsigned last_hit = 0;
 326         unsigned last_shown = 0;
 327         int binary_match_only = 0;
 328         const char *hunk_mark = "";
 329         unsigned count = 0;
 330         enum grep_context ctx = GREP_CONTEXT_HEAD;
 331
 332         if (buffer_is_binary(buf, size)) {
 333                 switch (opt->binary) {
 334                 case GREP_BINARY_DEFAULT:
 335                         binary_match_only = 1;
 336                         break;
 337                 case GREP_BINARY_NOMATCH:
 338                         return 0; /* Assume unmatch */
 339                         break;
 340                 default:
 341                         break;
 342                 }
 343         }
 344
 345         if (opt->pre_context)
 346                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 347         if (opt->pre_context || opt->post_context)
 348                 hunk_mark = "--\n";
 349
 350         while (left) {
 351                 char *eol, ch;
 352                 int hit = 0;
 353
 354                 eol = end_of_line(bol, &left);
 355                 ch = *eol;
 356                 *eol = 0;
 357
 358                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
 359                         ctx = GREP_CONTEXT_BODY;
 360
 361                 hit = match_line(opt, bol, eol, ctx);
 362                 *eol = ch;
 363
 364                 /* "grep -v -e foo -e bla" should list lines
 365                  * that do not have either, so inversion should
 366                  * be done outside.
 367                  */
 368                 if (opt->invert)
 369                         hit = !hit;
 370                 if (opt->unmatch_name_only) {
 371                         if (hit)
 372                                 return 0;
 373                         goto next_line;
 374                 }
 375                 if (hit) {
 376                         count++;
 377                         if (opt->status_only)
 378                                 return 1;
 379                         if (binary_match_only) {
 380                                 printf("Binary file %s matches\n", name);
 381                                 return 1;
 382                         }
 383                         if (opt->name_only) {
 384                                 printf("%s\n", name);
 385                                 return 1;
 386                         }
 387                         /* Hit at this line.  If we haven't shown the
 388                          * pre-context lines, we would need to show them.
 389                          * When asked to do "count", this still show
 390                          * the context which is nonsense, but the user
 391                          * deserves to get that ;-).
 392                          */
 393                         if (opt->pre_context) {
 394                                 unsigned from;
 395                                 if (opt->pre_context < lno)
 396                                         from = lno - opt->pre_context;
 397                                 else
 398                                         from = 1;
 399                                 if (from <= last_shown)
 400                                         from = last_shown + 1;
 401                                 if (last_shown && from != last_shown + 1)
 402                                         printf(hunk_mark);
 403                                 while (from < lno) {
 404                                         pcl = &prev[lno-from-1];
 405                                         show_line(opt, pcl->bol, pcl->eol,
 406                                                   name, from, '-');
 407                                         from++;
 408                                 }
 409                                 last_shown = lno-1;
 410                         }
 411                         if (last_shown && lno != last_shown + 1)
 412                                 printf(hunk_mark);
 413                         if (!opt->count)
 414                                 show_line(opt, bol, eol, name, lno, ':');
 415                         last_shown = last_hit = lno;
 416                 }
 417                 else if (last_hit &&
 418                          lno <= last_hit + opt->post_context) {
 419                         /* If the last hit is within the post context,
 420                          * we need to show this line.
 421                          */
 422                         if (last_shown && lno != last_shown + 1)
 423                                 printf(hunk_mark);
 424                         show_line(opt, bol, eol, name, lno, '-');
 425                         last_shown = lno;
 426                 }
 427                 if (opt->pre_context) {
 428                         memmove(prev+1, prev,
 429                                 (opt->pre_context-1) * sizeof(*prev));
 430                         prev->bol = bol;
 431                         prev->eol = eol;
 432                 }
 433
 434         next_line:
 435                 bol = eol + 1;
 436                 if (!left)
 437                         break;
 438                 left--;
 439                 lno++;
 440         }
 441
 442         if (opt->status_only)
 443                 return 0;
 444         if (opt->unmatch_name_only) {
 445                 /* We did not see any hit, so we want to show this */
 446                 printf("%s\n", name);
 447                 return 1;
 448         }
 449
 450         /* NEEDSWORK:
 451          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 452          * which feels mostly useless but sometimes useful.  Maybe
 453          * make it another option?  For now suppress them.
 454          */
 455         if (opt->count && count)
 456                 printf("%s:%u\n", name, count);
 457         return !!last_hit;
 458 }
 459