coreutils/fold.c

   1 /* vi: set sw=4 ts=4: */
   2 /* fold -- wrap each input line to fit in specified width.
   3
   4    Written by David MacKenzie, djm@gnu.ai.mit.edu.
   5    Copyright (C) 91, 1995-2002 Free Software Foundation, Inc.
   6
   7    Modified for busybox based on coreutils v 5.0
   8    Copyright (C) 2003 Glenn McGrath
   9
  10    Licensed under GPLv2 or later, see file LICENSE in this source tree.
  11 */
  12
  13 //usage:#define fold_trivial_usage
  14 //usage:       "[-bs] [-w WIDTH] [FILE]..."
  15 //usage:#define fold_full_usage "\n\n"
  16 //usage:       "Wrap input lines in each FILE (or stdin), writing to stdout\n"
  17 //usage:     "\n        -b      Count bytes rather than columns"
  18 //usage:     "\n        -s      Break at spaces"
  19 //usage:     "\n        -w      Use WIDTH columns instead of 80"
  20
  21 #include "libbb.h"
  22 #include "unicode.h"
  23
  24 /* This is a NOEXEC applet. Be very careful! */
  25
  26 /* Must match getopt32 call */
  27 #define FLAG_COUNT_BYTES        1
  28 #define FLAG_BREAK_SPACES       2
  29 #define FLAG_WIDTH              4
  30
  31 /* Assuming the current column is COLUMN, return the column that
  32    printing C will move the cursor to.
  33    The first column is 0. */
  34 static int adjust_column(unsigned column, char c)
  35 {
  36         if (option_mask32 & FLAG_COUNT_BYTES)
  37                 return ++column;
  38
  39         if (c == '\t')
  40                 return column + 8 - column % 8;
  41
  42         if (c == '\b') {
  43                 if ((int)--column < 0)
  44                         column = 0;
  45         }
  46         else if (c == '\r')
  47                 column = 0;
  48         else { /* just a printable char */
  49                 if (unicode_status != UNICODE_ON /* every byte is a new char */
  50                  || (c & 0xc0) != 0x80 /* it isn't a 2nd+ byte of a Unicode char */
  51                 ) {
  52                         column++;
  53                 }
  54         }
  55         return column;
  56 }
  57
  58 /* Note that this function can write NULs, unlike fputs etc. */
  59 static void write2stdout(const void *buf, unsigned size)
  60 {
  61         fwrite(buf, 1, size, stdout);
  62 }
  63
  64 int fold_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  65 int fold_main(int argc UNUSED_PARAM, char **argv)
  66 {
  67         char *line_out = NULL;
  68         const char *w_opt = "80";
  69         unsigned width;
  70         smallint exitcode = EXIT_SUCCESS;
  71
  72         init_unicode();
  73
  74         if (ENABLE_INCLUDE_SUSv2) {
  75                 /* Turn any numeric options into -w options.  */
  76                 int i;
  77                 for (i = 1; argv[i]; i++) {
  78                         const char *a = argv[i];
  79                         if (*a == '-') {
  80                                 a++;
  81                                 if (*a == '-' && !a[1]) /* "--" */
  82                                         break;
  83                                 if (isdigit(*a))
  84                                         argv[i] = xasprintf("-w%s", a);
  85                         }
  86                 }
  87         }
  88
  89         getopt32(argv, "bsw:", &w_opt);
  90         width = xatou_range(w_opt, 1, 10000);
  91
  92         argv += optind;
  93         if (!*argv)
  94                 *--argv = (char*)"-";
  95
  96         do {
  97                 FILE *istream = fopen_or_warn_stdin(*argv);
  98                 int c;
  99                 unsigned column = 0;     /* Screen column where next char will go */
 100                 unsigned offset_out = 0; /* Index in 'line_out' for next char */
 101
 102                 if (istream == NULL) {
 103                         exitcode = EXIT_FAILURE;
 104                         continue;
 105                 }
 106
 107                 while ((c = getc(istream)) != EOF) {
 108                         /* We grow line_out in chunks of 0x1000 bytes */
 109                         if ((offset_out & 0xfff) == 0) {
 110                                 line_out = xrealloc(line_out, offset_out + 0x1000);
 111                         }
 112  rescan:
 113                         line_out[offset_out] = c;
 114                         if (c == '\n') {
 115                                 write2stdout(line_out, offset_out + 1);
 116                                 column = offset_out = 0;
 117                                 continue;
 118                         }
 119                         column = adjust_column(column, c);
 120                         if (column <= width || offset_out == 0) {
 121                                 /* offset_out == 0 case happens
 122                                  * with small width (say, 1) and tabs.
 123                                  * The very first tab already goes to column 8,
 124                                  * but we must not wrap it */
 125                                 offset_out++;
 126                                 continue;
 127                         }
 128
 129                         /* This character would make the line too long.
 130                          * Print the line plus a newline, and make this character
 131                          * start the next line */
 132                         if (option_mask32 & FLAG_BREAK_SPACES) {
 133                                 unsigned i;
 134                                 unsigned logical_end;
 135
 136                                 /* Look for the last blank. */
 137                                 for (logical_end = offset_out - 1; (int)logical_end >= 0; logical_end--) {
 138                                         if (!isblank(line_out[logical_end]))
 139                                                 continue;
 140
 141                                         /* Found a space or tab.
 142                                          * Output up to and including it, and start a new line */
 143                                         logical_end++;
 144                                         /*line_out[logical_end] = '\n'; - NO! this nukes one buffered character */
 145                                         write2stdout(line_out, logical_end);
 146                                         putchar('\n');
 147                                         /* Move the remainder to the beginning of the next line.
 148                                          * The areas being copied here might overlap. */
 149                                         memmove(line_out, line_out + logical_end, offset_out - logical_end);
 150                                         offset_out -= logical_end;
 151                                         for (column = i = 0; i < offset_out; i++) {
 152                                                 column = adjust_column(column, line_out[i]);
 153                                         }
 154                                         goto rescan;
 155                                 }
 156                                 /* No blank found, wrap will split the overlong word */
 157                         }
 158                         /* Output what we accumulated up to now, and start a new line */
 159                         line_out[offset_out] = '\n';
 160                         write2stdout(line_out, offset_out + 1);
 161                         column = offset_out = 0;
 162                         goto rescan;
 163                 } /* while (not EOF) */
 164
 165                 if (offset_out) {
 166                         write2stdout(line_out, offset_out);
 167                 }
 168
 169                 if (fclose_if_not_stdin(istream)) {
 170                         bb_simple_perror_msg(*argv);
 171                         exitcode = EXIT_FAILURE;
 172                 }
 173         } while (*++argv);
 174
 175         fflush_stdout_and_exit(exitcode);
 176 }