grep.c

   1 /*
   2  * Mini grep implementation for busybox using libc regex.
   3  *
   4  * Copyright (C) 1999,2000,2001 by Lineo, inc.
   5  * Written by Mark Whitley <markw@lineo.com>, <markw@codepoet.org>
   6  *
   7  * This program is free software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published by
   9  * the Free Software Foundation; either version 2 of the License, or
  10  * (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  15  * General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License
  18  * along with this program; if not, write to the Free Software
  19  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  20  *
  21  */
  22
  23 #include <stdio.h>
  24 #include <stdlib.h>
  25 #include <getopt.h>
  26 #include <regex.h>
  27 #include <string.h> /* for strerror() */
  28 #include <errno.h>
  29 #include "busybox.h"
  30
  31
  32 extern int optind; /* in unistd.h */
  33 extern void xregcomp(regex_t *preg, const char *regex, int cflags); /* in busybox.h */
  34
  35 /* options */
  36 static int reflags            = REG_NOSUB;
  37 static int print_filename     = 0;
  38 static int print_line_num     = 0;
  39 static int print_match_counts = 0;
  40 static int be_quiet           = 0;
  41 static int invert_search      = 0;
  42 static int suppress_err_msgs  = 0;
  43 static int print_files_with_matches  = 0;
  44
  45 #ifdef BB_FEATURE_GREP_CONTEXT
  46 extern char *optarg; /* in getopt.h */
  47 static int lines_before      = 0;
  48 static int lines_after       = 0;
  49 static char **before_buf     = NULL;
  50 static int last_line_printed = 0;
  51 #endif /* BB_FEATURE_GREP_CONTEXT */
  52
  53 /* globals used internally */
  54 static regex_t *regexes = NULL; /* growable array of compiled regular expressions */
  55 static int nregexes = 0; /* number of elements in above arrary */
  56 static int matched; /* keeps track of whether we ever matched */
  57 static char *cur_file = NULL; /* the current file we are reading */
  58
  59
  60 static void print_line(const char *line, int linenum, char decoration)
  61 {
  62 #ifdef BB_FEATURE_GREP_CONTEXT
  63         /* possibly print the little '--' seperator */
  64         if ((lines_before || lines_after) && last_line_printed &&
  65                         last_line_printed < linenum - 1) {
  66                 puts("--");
  67         }
  68         last_line_printed = linenum;
  69 #endif
  70         if (print_filename)
  71                 printf("%s%c", cur_file, decoration);
  72         if (print_line_num)
  73                 printf("%i%c", linenum, decoration);
  74         puts(line);
  75 }
  76
  77
  78 static void grep_file(FILE *file)
  79 {
  80         char *line = NULL;
  81         int ret;
  82         int linenum = 0;
  83         int nmatches = 0;
  84         int i;
  85 #ifdef BB_FEATURE_GREP_CONTEXT
  86         int print_n_lines_after = 0;
  87         int curpos = 0; /* track where we are in the circular 'before' buffer */
  88         int idx = 0; /* used for iteration through the circular buffer */
  89 #endif /* BB_FEATURE_GREP_CONTEXT */
  90
  91         while ((line = get_line_from_file(file)) != NULL) {
  92                 chomp(line);
  93                 linenum++;
  94
  95                 for (i = 0; i < nregexes; i++) {
  96                         /*
  97                          * test for a postitive-assertion match (regexec returns success (0)
  98                          * and the user did not specify invert search), or a negative-assertion
  99                          * match (regexec returns failure (REG_NOMATCH) and the user specified
 100                          * invert search)
 101                          */
 102                         ret = regexec(&regexes[i], line, 0, NULL, 0);
 103                         if ((ret == 0 && !invert_search) || (ret == REG_NOMATCH && invert_search)) {
 104
 105                                 /* if we found a match but were told to be quiet, stop here and
 106                                  * return success */
 107                                 if (be_quiet)
 108                                         exit(0);
 109
 110                                 /* keep track of matches */
 111                                 nmatches++;
 112
 113                                 /* if we're just printing filenames, we stop after the first match */
 114                                 if (print_files_with_matches)
 115                                         break;
 116
 117                                 /* print the matched line */
 118                                 if (print_match_counts == 0) {
 119 #ifdef BB_FEATURE_GREP_CONTEXT
 120                                         int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
 121
 122                                         /* if we were told to print 'before' lines and there is at least
 123                                          * one line in the circular buffer, print them */
 124                                         if (lines_before && before_buf[prevpos] != NULL) {
 125                                                 int first_buf_entry_line_num = linenum - lines_before;
 126
 127                                                 /* advance to the first entry in the circular buffer, and
 128                                                  * figure out the line number is of the first line in the
 129                                                  * buffer */
 130                                                 idx = curpos;
 131                                                 while (before_buf[idx] == NULL) {
 132                                                         idx = (idx + 1) % lines_before;
 133                                                         first_buf_entry_line_num++;
 134                                                 }
 135
 136                                                 /* now print each line in the buffer, clearing them as we go */
 137                                                 while (before_buf[idx] != NULL) {
 138                                                         print_line(before_buf[idx], first_buf_entry_line_num, '-');
 139                                                         free(before_buf[idx]);
 140                                                         before_buf[idx] = NULL;
 141                                                         idx = (idx + 1) % lines_before;
 142                                                         first_buf_entry_line_num++;
 143                                                 }
 144                                         }
 145
 146                                         /* make a note that we need to print 'after' lines */
 147                                         print_n_lines_after = lines_after;
 148 #endif /* BB_FEATURE_GREP_CONTEXT */
 149                                         print_line(line, linenum, ':');
 150                                 }
 151                         }
 152 #ifdef BB_FEATURE_GREP_CONTEXT
 153                         else { /* no match */
 154                                 /* Add the line to the circular 'before' buffer */
 155                                 if(lines_before) {
 156                                         if(before_buf[curpos])
 157                                                 free(before_buf[curpos]);
 158                                         before_buf[curpos] = strdup(line);
 159                                         curpos = (curpos + 1) % lines_before;
 160                                 }
 161                         }
 162
 163                         /* if we need to print some context lines after the last match, do so */
 164                         if (print_n_lines_after && (last_line_printed != linenum)) {
 165                                 print_line(line, linenum, '-');
 166                                 print_n_lines_after--;
 167                         }
 168 #endif /* BB_FEATURE_GREP_CONTEXT */
 169                 } /* for */
 170                 free(line);
 171         }
 172
 173
 174         /* special-case file post-processing for options where we don't print line
 175          * matches, just filenames and possibly match counts */
 176
 177         /* grep -c: print [filename:]count, even if count is zero */
 178         if (print_match_counts) {
 179                 if (print_filename)
 180                         printf("%s:", cur_file);
 181                 if (print_files_with_matches && nmatches > 0)
 182                         printf("1\n");
 183                 else
 184                     printf("%d\n", nmatches);
 185         }
 186
 187         /* grep -l: print just the filename, but only if we grepped the line in the file  */
 188         if (print_files_with_matches && nmatches > 0) {
 189                 puts(cur_file);
 190         }
 191
 192
 193         /* remember if we matched */
 194         if (nmatches != 0)
 195                 matched = 1;
 196 }
 197
 198
 199 static void add_regex(const char *restr)
 200 {
 201         regexes = xrealloc(regexes, sizeof(regex_t) * (++nregexes));
 202         xregcomp(&regexes[nregexes-1], restr, reflags);
 203 }
 204
 205
 206 static void     load_regexes_from_file(const char *filename)
 207 {
 208         char *line;
 209         FILE *f = xfopen(filename, "r");
 210         while ((line = get_line_from_file(f)) != NULL) {
 211                 chomp(line);
 212                 add_regex(line);
 213                 free(line);
 214         }
 215 }
 216
 217
 218 #ifdef BB_FEATURE_CLEAN_UP
 219 static void destroy_regexes()
 220 {
 221         if (regexes == NULL)
 222                 return;
 223
 224         /* destroy all the elments in the array */
 225         while (--nregexes >= 0) {
 226                 regfree(&regexes[nregexes]);
 227                 free(&regexes[nregexes]);
 228         }
 229 }
 230 #endif
 231
 232
 233 extern int grep_main(int argc, char **argv)
 234 {
 235         int opt;
 236 #ifdef BB_FEATURE_GREP_CONTEXT
 237         char *junk;
 238 #endif
 239
 240 #ifdef BB_FEATURE_CLEAN_UP
 241         /* destroy command strings on exit */
 242         if (atexit(destroy_regexes) == -1)
 243                 perror_msg_and_die("atexit");
 244 #endif
 245
 246         /* do normal option parsing */
 247         while ((opt = getopt(argc, argv, "iHhlnqvsce:f:"
 248 #ifdef BB_FEATURE_GREP_CONTEXT
 249 "A:B:C:"
 250 #endif
 251 )) > 0) {
 252                 switch (opt) {
 253                         case 'i':
 254                                 reflags |= REG_ICASE;
 255                                 break;
 256                         case 'l':
 257                                 print_files_with_matches++;
 258                                 break;
 259                         case 'H':
 260                                 print_filename++;
 261                                 break;
 262                         case 'h':
 263                                 print_filename--;
 264                                 break;
 265                         case 'n':
 266                                 print_line_num++;
 267                                 break;
 268                         case 'q':
 269                                 be_quiet++;
 270                                 break;
 271                         case 'v':
 272                                 invert_search++;
 273                                 break;
 274                         case 's':
 275                                 suppress_err_msgs++;
 276                                 break;
 277                         case 'c':
 278                                 print_match_counts++;
 279                                 break;
 280                         case 'e':
 281                                 add_regex(optarg);
 282                                 break;
 283                         case 'f':
 284                                 load_regexes_from_file(optarg);
 285                                 break;
 286 #ifdef BB_FEATURE_GREP_CONTEXT
 287                         case 'A':
 288                                 lines_after = strtoul(optarg, &junk, 10);
 289                                 if(*junk != '\0')
 290                                         error_msg_and_die("invalid context length argument");
 291                                 break;
 292                         case 'B':
 293                                 lines_before = strtoul(optarg, &junk, 10);
 294                                 if(*junk != '\0')
 295                                         error_msg_and_die("invalid context length argument");
 296                                 before_buf = (char **)calloc(lines_before, sizeof(char *));
 297                                 break;
 298                         case 'C':
 299                                 lines_after = lines_before = strtoul(optarg, &junk, 10);
 300                                 if(*junk != '\0')
 301                                         error_msg_and_die("invalid context length argument");
 302                                 before_buf = (char **)calloc(lines_before, sizeof(char *));
 303                                 break;
 304 #endif /* BB_FEATURE_GREP_CONTEXT */
 305                         default:
 306                                 show_usage();
 307                 }
 308         }
 309
 310         /* if we didn't get a pattern from a -e and no command file was specified,
 311          * argv[optind] should be the pattern. no pattern, no worky */
 312         if (nregexes == 0) {
 313                 if (argv[optind] == NULL)
 314                         show_usage();
 315                 else {
 316                         add_regex(argv[optind]);
 317                         optind++;
 318                 }
 319         }
 320
 321         /* sanity checks */
 322         if (print_match_counts || be_quiet || print_files_with_matches) {
 323                 print_line_num = 0;
 324 #ifdef BB_FEATURE_GREP_CONTEXT
 325                 lines_before = 0;
 326                 lines_after = 0;
 327 #endif
 328         }
 329
 330         /* argv[(optind)..(argc-1)] should be names of file to grep through. If
 331          * there is more than one file to grep, we will print the filenames */
 332         if ((argc-1) - (optind) > 0)
 333                 print_filename++;
 334
 335         /* If no files were specified, or '-' was specified, take input from
 336          * stdin. Otherwise, we grep through all the files specified. */
 337         if (argv[optind] == NULL || (strcmp(argv[optind], "-") == 0)) {
 338                 grep_file(stdin);
 339         }
 340         else {
 341                 int i;
 342                 FILE *file;
 343                 for (i = optind; i < argc; i++) {
 344                         cur_file = argv[i];
 345                         file = fopen(cur_file, "r");
 346                         if (file == NULL) {
 347                                 if (!suppress_err_msgs)
 348                                         perror_msg("%s", cur_file);
 349                         }
 350                         else {
 351                                 grep_file(file);
 352                                 fclose(file);
 353                         }
 354                 }
 355         }
 356
 357         return !matched; /* invert return value 0 = success, 1 = failed */
 358 }