Home | History | Annotate | Download | only in posix 
      1 /* split.c - split a file into smaller files
      2  *
      3  * Copyright 2013 Rob Landley <rob (at) landley.net>
      4  *
      5  * See http://pubs.opengroup.org/onlinepubs/9699919799/utilities/split.html
      6  *
      7  * Standard does not cover:
      8  * - should splitting an empty file produce an empty outfile? (Went with "no".)
      9  * - permissions on output file
     10 
     11 USE_SPLIT(NEWTOY(split, ">2a#<1=2>9b#<1l#<1[!bl]", TOYFLAG_USR|TOYFLAG_BIN))
     12 
     13 config SPLIT
     14   bool "split"
     15   default y
     16   help
     17     usage: split [-a SUFFIX_LEN] [-b BYTES] [-l LINES] [INPUT [OUTPUT]]
     18 
     19     Copy INPUT (or stdin) data to a series of OUTPUT (or "x") files with
     20     alphabetically increasing suffix (aa, ab, ac... az, ba, bb...).
     21 
     22     -a	Suffix length (default 2)
     23     -b	BYTES/file (10, 10k, 10m, 10g...)
     24     -l	LINES/file (default 1000)
     25 */
     26 
     27 #define FOR_split
     28 #include "toys.h"
     29 
     30 GLOBALS(
     31   long l, b, a;
     32 
     33   char *outfile;
     34 )
     35 
     36 static void do_split(int infd, char *in)
     37 {
     38   unsigned long bytesleft, linesleft, filenum, len, pos;
     39   int outfd = -1;
     40   struct stat st;
     41 
     42   // posix doesn't cover permissions on output file, so copy input (or 0777)
     43   st.st_mode = 0777;
     44   fstat(infd, &st);
     45 
     46   len = pos = filenum = bytesleft = linesleft = 0;
     47   for (;;) {
     48     int i, j;
     49 
     50     // Refill toybuf?
     51     if (len == pos) {
     52       if (!(len = xread(infd, toybuf, sizeof(toybuf)))) break;
     53       pos = 0;
     54     }
     55 
     56     // Start new output file?
     57     if ((TT.b && !bytesleft) || (TT.l && !linesleft)) {
     58       char *s = TT.outfile + strlen(TT.outfile);
     59 
     60       j = filenum++;
     61       for (i = 0; i<TT.a; i++) {
     62         *(--s) = 'a'+(j%26);
     63         j /= 26;
     64       }
     65       if (j) error_exit("bad suffix");
     66       bytesleft = TT.b;
     67       linesleft = TT.l;
     68       if (outfd != -1) close(outfd);
     69       outfd = xcreate(TT.outfile, O_RDWR|O_CREAT|O_TRUNC, st.st_mode & 0777);
     70     }
     71 
     72     // Write next chunk of output.
     73     if (TT.l) {
     74       for (i = pos; i < len; ) {
     75         if (toybuf[i++] == '\n' && !--linesleft) break;
     76         if (!--bytesleft) break;
     77       }
     78       j = i - pos;
     79     } else {
     80       j = len - pos;
     81       if (j > bytesleft) j = bytesleft;
     82       bytesleft -= j;
     83     }
     84     xwrite(outfd, toybuf+pos, j);
     85     pos += j;
     86   }
     87 
     88   if (CFG_TOYBOX_FREE) {
     89     if (outfd != -1) close(outfd);
     90     if (infd) close(infd);
     91     free(TT.outfile);
     92   }
     93   xexit();
     94 }
     95 
     96 void split_main(void)
     97 {
     98   if (!TT.b && !TT.l) TT.l = 1000;
     99 
    100   // Allocate template for output filenames
    101   TT.outfile = xmprintf("%s%*c", (toys.optc == 2) ? toys.optargs[1] : "x",
    102     (int)TT.a, ' ');
    103 
    104   // We only ever use one input, but this handles '-' or no input for us.
    105   loopfiles(toys.optargs, do_split);
    106 }
    107