Re: Split -b Question

[Paul Eggert]

Luke Hassell <address@hidden> writes:

> I must have missed something obvious.

No, you haven't.  split doesn't support large files in coreutils 5.0;
this has been a FIXME for a while, I think.  Here's a patch.

2003-04-08  Paul Eggert  <address@hidden>

        * src/split.c: Add support for large files.
        Include "inttostr.h".
        (bytes_split, lines_split, line_bytes_split, main):
        Use uintmax_t, not size_t, for file sizes.
        Use umaxtostr to print file sizes.

--- split.c     2003/03/11 20:44:40     5.0
+++ split.c     2003/04/08 20:46:46
@@ -33,6 +33,7 @@
 #include "error.h"
 #include "full-read.h"
 #include "full-write.h"
+#include "inttostr.h"
 #include "posixver.h"
 #include "safe-read.h"
 #include "xstrtol.h"
@@ -200,12 +201,12 @@ cwrite (int new_file_flag, const char *b
    Use buffer BUF, whose size is BUFSIZE.  */
 
 static void
-bytes_split (size_t nchars, char *buf, size_t bufsize)
+bytes_split (uintmax_t nchars, char *buf, size_t bufsize)
 {
   size_t n_read;
   int new_file_flag = 1;
   size_t to_read;
-  size_t to_write = nchars;
+  uintmax_t to_write = nchars;
   char *bp_out;
 
   do
@@ -227,12 +228,15 @@ bytes_split (size_t nchars, char *buf, s
                }
              break;
            }
-
-         cwrite (new_file_flag, bp_out, to_write);
-         bp_out += to_write;
-         to_read -= to_write;
-         new_file_flag = 1;
-         to_write = nchars;
+         else
+           {
+             size_t w = to_write;
+             cwrite (new_file_flag, bp_out, w);
+             bp_out += w;
+             to_read -= w;
+             new_file_flag = 1;
+             to_write = nchars;
+           }
        }
     }
   while (n_read == bufsize);
@@ -242,12 +246,12 @@ bytes_split (size_t nchars, char *buf, s
    Use buffer BUF, whose size is BUFSIZE.  */
 
 static void
-lines_split (size_t nlines, char *buf, size_t bufsize)
+lines_split (uintmax_t nlines, char *buf, size_t bufsize)
 {
   size_t n_read;
   char *bp, *bp_out, *eob;
   int new_file_flag = 1;
-  size_t n = 0;
+  uintmax_t n = 0;
 
   do
     {
@@ -289,13 +293,14 @@ lines_split (size_t nlines, char *buf, s
    where lines longer than NCHARS bytes occur. */
 
 static void
-line_bytes_split (size_t nchars)
+line_bytes_split (uintmax_t nchars)
 {
   size_t n_read;
   char *bp;
   int eof = 0;
   size_t n_buffered = 0;
-  char *buf = (char *) xmalloc (nchars);
+  size_t allocated = nchars;
+  char *buf = (char *) xmalloc (allocated);
 
   do
     {
@@ -347,14 +352,14 @@ int
 main (int argc, char **argv)
 {
   struct stat stat_buf;
-  size_t num;                  /* numeric argument from command line */
+  uintmax_t num;               /* numeric argument from command line */
   enum
     {
       type_undef, type_bytes, type_byteslines, type_lines, type_digits
     } split_type = type_undef;
   size_t in_blk_size;          /* optimal block size of input file device */
   char *buf;                   /* file i/o buffer */
-  size_t accum = 0;
+  uintmax_t accum = 0;
   int c;
   int digits_optind = 0;
 
@@ -374,7 +379,6 @@ main (int argc, char **argv)
     {
       /* This is the argv-index of the option we will read next.  */
       int this_optind = optind ? optind : 1;
-      long int tmp_long;
 
       c = getopt_long (argc, argv, "0123456789C:a:b:l:", longopts, NULL);
       if (c == -1)
@@ -402,39 +406,34 @@ main (int argc, char **argv)
          if (split_type != type_undef)
            FAIL_ONLY_ONE_WAY ();
          split_type = type_bytes;
-         if (xstrtol (optarg, NULL, 10, &tmp_long, "bkm") != LONGINT_OK
-             || tmp_long < 0 || tmp_long > INT_MAX)
+         if (xstrtoumax (optarg, NULL, 10, &accum, "bkm") != LONGINT_OK)
            {
              error (0, 0, _("%s: invalid number of bytes"), optarg);
              usage (EXIT_FAILURE);
            }
-         accum = /* FIXME: */ (int) tmp_long;
          break;
 
        case 'l':
          if (split_type != type_undef)
            FAIL_ONLY_ONE_WAY ();
          split_type = type_lines;
-         if (xstrtol (optarg, NULL, 10, &tmp_long, "") != LONGINT_OK
-             || tmp_long < 0 || tmp_long > INT_MAX)
+         if (xstrtoumax (optarg, NULL, 10, &accum, "") != LONGINT_OK)
            {
              error (0, 0, _("%s: invalid number of lines"), optarg);
              usage (EXIT_FAILURE);
            }
-         accum = /* FIXME */ (int) tmp_long;
          break;
 
        case 'C':
          if (split_type != type_undef)
            FAIL_ONLY_ONE_WAY ();
          split_type = type_byteslines;
-         if (xstrtol (optarg, NULL, 10, &tmp_long, "bkm") != LONGINT_OK
-             || tmp_long < 0 ||  tmp_long > INT_MAX)
+         if (xstrtoumax (optarg, NULL, 10, &accum, "bkm") != LONGINT_OK
+             || SIZE_MAX < accum)
            {
              error (0, 0, _("%s: invalid number of bytes"), optarg);
              usage (EXIT_FAILURE);
            }
-         accum = /* FIXME */ (int) tmp_long;
          break;
 
        case '0':
@@ -467,7 +466,9 @@ main (int argc, char **argv)
 
   if (digits_optind && 200112 <= posix2_version ())
     {
-      error (0, 0, _("`-%d' option is obsolete; use `-l %d'"), accum, accum);
+      char buf[INT_BUFSIZE_BOUND (uintmax_t)];
+      char const *a = umaxtostr (accum, buf);
+      error (0, 0, _("`-%s' option is obsolete; use `-l %s'"), a, a);
       usage (EXIT_FAILURE);
     }