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 source tree.
   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 //config:config EXPAND
  24 //config:       bool "expand (5.3 kb)"
  25 //config:       default y
  26 //config:       help
  27 //config:       By default, convert all tabs to spaces.
  28 //config:
  29 //config:config UNEXPAND
  30 //config:       bool "unexpand (5.5 kb)"
  31 //config:       default y
  32 //config:       help
  33 //config:       By default, convert only leading sequences of blanks to tabs.
  34
  35 //applet:IF_EXPAND(APPLET(expand, BB_DIR_USR_BIN, BB_SUID_DROP))
  36 //                   APPLET_ODDNAME:name      main    location        suid_type     help
  37 //applet:IF_UNEXPAND(APPLET_ODDNAME(unexpand, expand, BB_DIR_USR_BIN, BB_SUID_DROP, unexpand))
  38
  39 //kbuild:lib-$(CONFIG_EXPAND) += expand.o
  40 //kbuild:lib-$(CONFIG_UNEXPAND) += expand.o
  41
  42 //usage:#define expand_trivial_usage
  43 //usage:       "[-i] [-t N] [FILE]..."
  44 //usage:#define expand_full_usage "\n\n"
  45 //usage:       "Convert tabs to spaces, writing to stdout\n"
  46 //usage:     "\n        -i      Don't convert tabs after non blanks"
  47 //usage:     "\n        -t      Tabstops every N chars"
  48
  49 //usage:#define unexpand_trivial_usage
  50 //usage:       "[-fa][-t N] [FILE]..."
  51 //usage:#define unexpand_full_usage "\n\n"
  52 //usage:       "Convert spaces to tabs, writing to stdout\n"
  53 //usage:     "\n        -a      Convert all blanks"
  54 //usage:     "\n        -f      Convert only leading blanks"
  55 //usage:     "\n        -t N    Tabstops every N chars"
  56
  57 #include "libbb.h"
  58 #include "unicode.h"
  59
  60 enum {
  61         OPT_INITIAL     = 1 << 0,
  62         OPT_TABS        = 1 << 1,
  63         OPT_ALL         = 1 << 2,
  64 };
  65
  66 //FIXME: does not work properly with input containing NULs
  67 //coreutils 8.30 preserves NULs but treats them as chars of width zero:
  68 //AB<nul><tab>C will expand <tab> to 6 spaces, not 5.
  69
  70 #if ENABLE_EXPAND
  71 static void expand(FILE *file, unsigned tab_size, unsigned opt)
  72 {
  73
  74         for (;;) {
  75                 char *line;
  76                 char *ptr;
  77                 char *ptr_strbeg;
  78 //commented-out code handles NULs, +90 bytes of code, not tested much
  79 //              size_t linelen;
  80 //              unsigned len = 0;
  81
  82 //              linelen = 1024 * 1024;
  83 //              line = xmalloc_fgets_str_len(file, "\n", &linelen);
  84                 line = xmalloc_fgets(file); //
  85                 if (!line)
  86                         break;
  87                 ptr = ptr_strbeg = line;
  88                 for (;;) {
  89                         unsigned char c = *ptr;
  90                         if (c == '\0') {
  91 //                              size_t rem = line + linelen - ptr;
  92 //                              if (rem > 0) {
  93 //# if ENABLE_UNICODE_SUPPORT
  94 //                                      len += unicode_strwidth(ptr_strbeg);
  95 //# else
  96 //                                      len += ptr - ptr_strbeg;
  97 //# endif
  98 //                                      printf("%s%c", ptr_strbeg, '\0');
  99 //                                      memmove(ptr, ptr + 1, rem + 1);
 100 //                                      ptr_strbeg = ptr;
 101 //                                      linelen--;
 102 //                                      continue;
 103 //                              }
 104                                 break;
 105                         }
 106                         if ((opt & OPT_INITIAL) && !isblank(c)) {
 107                                 /* not space or tab */
 108                                 break;
 109                         }
 110                         if (c == '\t') {
 111                                 unsigned len = 0; //
 112                                 *ptr = '\0';
 113 # if ENABLE_UNICODE_SUPPORT
 114                                 len += unicode_strwidth(ptr_strbeg);
 115 # else
 116                                 len += ptr - ptr_strbeg;
 117 # endif
 118                                 len = tab_size - (len % tab_size);
 119                                 /*while (ptr[1] == '\t') { ptr++; len += tab_size; } - can handle many tabs at once */
 120                                 printf("%s%*s", ptr_strbeg, len, "");
 121 //                              len = 0;
 122                                 ptr_strbeg = ptr + 1;
 123                         }
 124                         ptr++;
 125                 }
 126                 fputs_stdout(ptr_strbeg);
 127                 free(line);
 128         }
 129 }
 130 #endif
 131
 132 #if ENABLE_UNEXPAND
 133 static void unexpand(FILE *file, unsigned tab_size, unsigned opt)
 134 {
 135         char *line;
 136
 137         while ((line = xmalloc_fgets(file)) != NULL) {
 138                 char *ptr = line;
 139                 unsigned column = 0;
 140
 141                 while (*ptr) {
 142                         unsigned n;
 143                         unsigned len = 0;
 144
 145                         while (*ptr == ' ') {
 146                                 ptr++;
 147                                 len++;
 148                         }
 149                         column += len;
 150                         if (*ptr == '\t') {
 151                                 column += tab_size - (column % tab_size);
 152                                 ptr++;
 153                                 continue;
 154                         }
 155
 156                         n = column / tab_size;
 157                         if (n) {
 158                                 len = column = column % tab_size;
 159                                 while (n--)
 160                                         putchar('\t');
 161                         }
 162
 163                         if (!(opt & OPT_ALL) && ptr != line) {
 164                                 printf("%*s%s", len, "", ptr);
 165                                 break;
 166                         }
 167                         n = strcspn(ptr, "\t ");
 168                         printf("%*s%.*s", len, "", n, ptr);
 169 # if ENABLE_UNICODE_SUPPORT
 170                         {
 171                                 char c = ptr[n];
 172                                 ptr[n] = '\0';
 173                                 len = unicode_strwidth(ptr);
 174                                 ptr[n] = c;
 175                         }
 176 # else
 177                         len = n;
 178 # endif
 179                         ptr += n;
 180                         column = (column + len) % tab_size;
 181                 }
 182                 free(line);
 183         }
 184 }
 185 #endif
 186
 187 int expand_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 188 int expand_main(int argc UNUSED_PARAM, char **argv)
 189 {
 190         /* Default 8 spaces for 1 tab */
 191         const char *opt_t = "8";
 192         FILE *file;
 193         unsigned tab_size;
 194         unsigned opt;
 195         exitcode_t exit_status = EXIT_SUCCESS;
 196
 197         init_unicode();
 198
 199         if (ENABLE_EXPAND && (!ENABLE_UNEXPAND || applet_name[0] == 'e')) {
 200                 opt = getopt32long(argv, "it:",
 201                                 "initial\0"          No_argument       "i"
 202                                 "tabs\0"             Required_argument "t"
 203                                 , &opt_t
 204                 );
 205         } else {
 206                 opt = getopt32long(argv, "^"
 207                                 "ft:a"
 208                                 "\0"
 209                                 "ta" /* -t NUM sets -a */,
 210                                 "first-only\0"       No_argument       "f"
 211                                 "tabs\0"             Required_argument "t"
 212                                 "all\0"              No_argument       "a"
 213                                 , &opt_t
 214                 );
 215                 /* -t implies -a, but an explicit -f overrides */
 216                 if (opt & OPT_INITIAL) opt &= ~OPT_ALL;
 217         }
 218         tab_size = xatou_range(opt_t, 1, UINT_MAX);
 219
 220         argv += optind;
 221
 222         if (!*argv) {
 223                 *--argv = (char*)bb_msg_standard_input;
 224         }
 225         do {
 226                 file = fopen_or_warn_stdin(*argv);
 227                 if (!file) {
 228                         exit_status = EXIT_FAILURE;
 229                         continue;
 230                 }
 231
 232                 if (ENABLE_EXPAND && (!ENABLE_UNEXPAND || applet_name[0] == 'e'))
 233                         IF_EXPAND(expand(file, tab_size, opt));
 234                 else
 235                         IF_UNEXPAND(unexpand(file, tab_size, opt));
 236
 237                 /* Check and close the file */
 238                 if (fclose_if_not_stdin(file)) {
 239                         bb_simple_perror_msg(*argv);
 240                         exit_status = EXIT_FAILURE;
 241                 }
 242                 /* If stdin also clear EOF */
 243                 if (file == stdin)
 244                         clearerr(file);
 245         } while (*++argv);
 246
 247         /* Now close stdin also */
 248         /* (if we didn't read from it, it's a no-op) */
 249         if (fclose(stdin))
 250                 bb_simple_perror_msg_and_die(bb_msg_standard_input);
 251
 252         fflush_stdout_and_exit(exit_status);
 253 }