coreutils/expand.c

   1 /* expand - convert tabs to spaces
   2  * unexpand - convert spaces to tabs
   3  *
   4  * Copyright (C) 89, 91, 1995-2006 Free Software Foundation, Inc.
   5  *
   6  * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
   7  *
   8  * David MacKenzie <djm@gnu.ai.mit.edu>
   9  *
  10  * Options for expand:
  11  * -t num  --tabs=NUM      Convert tabs to num spaces (default 8 spaces).
  12  * -i      --initial       Only convert initial tabs on each line to spaces.
  13  *
  14  * Options for unexpand:
  15  * -a      --all           Convert all blanks, instead of just initial blanks.
  16  * -f      --first-only    Convert only leading sequences of blanks (default).
  17  * -t num  --tabs=NUM      Have tabs num characters apart instead of 8.
  18  *
  19  *  Busybox version (C) 2007 by Tito Ragusa <farmatito@tiscali.it>
  20  *
  21  *  Caveat: this versions of expand and unexpand don't accept tab lists.
  22  */
  23 #include "libbb.h"
  24 #include "unicode.h"
  25
  26 enum {
  27         OPT_INITIAL     = 1 << 0,
  28         OPT_TABS        = 1 << 1,
  29         OPT_ALL         = 1 << 2,
  30 };
  31
  32 #if ENABLE_EXPAND
  33 static void expand(FILE *file, unsigned tab_size, unsigned opt)
  34 {
  35         char *line;
  36
  37         while ((line = xmalloc_fgets(file)) != NULL) {
  38                 unsigned char c;
  39                 char *ptr;
  40                 char *ptr_strbeg;
  41
  42                 ptr = ptr_strbeg = line;
  43                 while ((c = *ptr) != '\0') {
  44                         if ((opt & OPT_INITIAL) && !isblank(c)) {
  45                                 /* not space or tab */
  46                                 break;
  47                         }
  48                         if (c == '\t') {
  49                                 unsigned len;
  50                                 *ptr = '\0';
  51 # if ENABLE_FEATURE_ASSUME_UNICODE
  52                                 {
  53                                         uni_stat_t uni_stat;
  54                                         printable_string(&uni_stat, ptr_strbeg);
  55                                         len = uni_stat.unicode_width;
  56                                 }
  57 # else
  58                                 len = ptr - ptr_strbeg;
  59 # endif
  60                                 len = tab_size - (len % tab_size);
  61                                 /*while (ptr[1] == '\t') { ptr++; len += tab_size; } - can handle many tabs at once */
  62                                 printf("%s%*s", ptr_strbeg, len, "");
  63                                 ptr_strbeg = ptr + 1;
  64                         }
  65                         ptr++;
  66                 }
  67                 fputs(ptr_strbeg, stdout);
  68                 free(line);
  69         }
  70 }
  71 #endif
  72
  73 #if ENABLE_UNEXPAND
  74 static void unexpand(FILE *file, unsigned tab_size, unsigned opt)
  75 {
  76         char *line;
  77
  78         while ((line = xmalloc_fgets(file)) != NULL) {
  79                 char *ptr = line;
  80                 unsigned column = 0;
  81
  82                 while (*ptr) {
  83                         unsigned n;
  84                         unsigned len;
  85
  86                         while (*ptr == ' ') {
  87                                 column++;
  88                                 ptr++;
  89                         }
  90                         if (*ptr == '\t') {
  91                                 column += tab_size - (column % tab_size);
  92                                 ptr++;
  93                                 continue;
  94                         }
  95
  96                         n = column / tab_size;
  97                         column = column % tab_size;
  98                         while (n--)
  99                                 putchar('\t');
 100
 101                         if ((opt & OPT_INITIAL) && ptr != line) {
 102                                 printf("%*s%s", column, "", ptr);
 103                                 break;
 104                         }
 105                         n = strcspn(ptr, "\t ");
 106                         printf("%*s%.*s", column, "", n, ptr);
 107 # if ENABLE_FEATURE_ASSUME_UNICODE
 108                         {
 109                                 char c;
 110                                 uni_stat_t uni_stat;
 111                                 c = ptr[n];
 112                                 ptr[n] = '\0';
 113                                 printable_string(&uni_stat, ptr);
 114                                 len = uni_stat.unicode_width;
 115                                 ptr[n] = c;
 116                         }
 117 # else
 118                         len = n;
 119 # endif
 120                         ptr += n;
 121                         column = (column + len) % tab_size;
 122                 }
 123                 free(line);
 124         }
 125 }
 126 #endif
 127
 128 int expand_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 129 int expand_main(int argc UNUSED_PARAM, char **argv)
 130 {
 131         /* Default 8 spaces for 1 tab */
 132         const char *opt_t = "8";
 133         FILE *file;
 134         unsigned tab_size;
 135         unsigned opt;
 136         int exit_status = EXIT_SUCCESS;
 137
 138 #if ENABLE_FEATURE_EXPAND_LONG_OPTIONS
 139         static const char expand_longopts[] ALIGN1 =
 140                 /* name, has_arg, val */
 141                 "initial\0"          No_argument       "i"
 142                 "tabs\0"             Required_argument "t"
 143         ;
 144 #endif
 145 #if ENABLE_FEATURE_UNEXPAND_LONG_OPTIONS
 146         static const char unexpand_longopts[] ALIGN1 =
 147                 /* name, has_arg, val */
 148                 "first-only\0"       No_argument       "i"
 149                 "tabs\0"             Required_argument "t"
 150                 "all\0"              No_argument       "a"
 151         ;
 152 #endif
 153         init_unicode();
 154
 155         if (ENABLE_EXPAND && (!ENABLE_UNEXPAND || applet_name[0] == 'e')) {
 156                 IF_FEATURE_EXPAND_LONG_OPTIONS(applet_long_options = expand_longopts);
 157                 opt = getopt32(argv, "it:", &opt_t);
 158         } else {
 159                 IF_FEATURE_UNEXPAND_LONG_OPTIONS(applet_long_options = unexpand_longopts);
 160                 /* -t NUM sets also -a */
 161                 opt_complementary = "ta";
 162                 opt = getopt32(argv, "ft:a", &opt_t);
 163                 /* -f --first-only is the default */
 164                 if (!(opt & OPT_ALL)) opt |= OPT_INITIAL;
 165         }
 166         tab_size = xatou_range(opt_t, 1, UINT_MAX);
 167
 168         argv += optind;
 169
 170         if (!*argv) {
 171                 *--argv = (char*)bb_msg_standard_input;
 172         }
 173         do {
 174                 file = fopen_or_warn_stdin(*argv);
 175                 if (!file) {
 176                         exit_status = EXIT_FAILURE;
 177                         continue;
 178                 }
 179
 180                 if (ENABLE_EXPAND && (!ENABLE_UNEXPAND || applet_name[0] == 'e'))
 181                         IF_EXPAND(expand(file, tab_size, opt));
 182                 else
 183                         IF_UNEXPAND(unexpand(file, tab_size, opt));
 184
 185                 /* Check and close the file */
 186                 if (fclose_if_not_stdin(file)) {
 187                         bb_simple_perror_msg(*argv);
 188                         exit_status = EXIT_FAILURE;
 189                 }
 190                 /* If stdin also clear EOF */
 191                 if (file == stdin)
 192                         clearerr(file);
 193         } while (*++argv);
 194
 195         /* Now close stdin also */
 196         /* (if we didn't read from it, it's a no-op) */
 197         if (fclose(stdin))
 198                 bb_perror_msg_and_die(bb_msg_standard_input);
 199
 200         fflush_stdout_and_exit(exit_status);
 201 }