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