LXR toybox/toys/posix/split.c

   1/* split.c - split a file into smaller files
   2 *
   3 * Copyright 2013 Rob Landley <rob@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
  11USE_SPLIT(NEWTOY(split, ">2a#<1=2>9b#<1l#<1n#<1[!bl][!bn][!ln]", TOYFLAG_USR|TOYFLAG_BIN))
  12
  13config SPLIT
  14  bool "split"
  15  default y
  16  help
  17    usage: split [-a SUFFIX_LEN] [-b BYTES] [-l LINES] [-n PARTS] [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    -n  PARTS many equal length files
  26*/
  27
  28#define FOR_split
  29#include "toys.h"
  30
  31GLOBALS(
  32  long n, l, b, a;
  33
  34  char *outfile;
  35)
  36
  37static void do_split(int infd, char *in)
  38{
  39  unsigned long bytesleft, linesleft, filenum, len, pos;
  40  int outfd = -1;
  41  struct stat st;
  42
  43  // posix doesn't cover permissions on output file, so copy input (or 0777)
  44  st.st_mode = 0777;
  45  st.st_size = 0;
  46  fstat(infd, &st);
  47
  48  if (TT.n && (TT.b = st.st_size/TT.n)<1) return error_msg("%s: no size", in);
  49  len = pos = filenum = bytesleft = linesleft = 0;
  50  for (;;) {
  51    int i, j;
  52
  53    // Refill toybuf?
  54    if (len == pos) {
  55      if (!(len = xread(infd, toybuf, sizeof(toybuf)))) break;
  56      pos = 0;
  57    }
  58
  59    // Start new output file?
  60    if ((TT.b && !bytesleft) || (TT.l && !linesleft)) {
  61      char *s = TT.outfile + strlen(TT.outfile);
  62
  63      j = filenum++;
  64      for (i = 0; i<TT.a; i++) {
  65        *(--s) = 'a'+(j%26);
  66        j /= 26;
  67      }
  68      if (j) error_exit("bad suffix");
  69      bytesleft = TT.b + ((filenum == TT.n) ? st.st_size%TT.n : 0);
  70      linesleft = TT.l;
  71      xclose(outfd);
  72      outfd = xcreate(TT.outfile, O_RDWR|O_CREAT|O_TRUNC, st.st_mode & 0777);
  73    }
  74
  75    // Write next chunk of output.
  76    if (TT.l) {
  77      for (i = pos; i < len; ) {
  78        if (toybuf[i++] == '\n' && !--linesleft) break;
  79        if (!--bytesleft) break;
  80      }
  81      j = i - pos;
  82    } else {
  83      j = len - pos;
  84      if (j > bytesleft) j = bytesleft;
  85      bytesleft -= j;
  86    }
  87    xwrite(outfd, toybuf+pos, j);
  88    pos += j;
  89  }
  90
  91  if (CFG_TOYBOX_FREE) {
  92    xclose(outfd);
  93    if (infd) close(infd);
  94    free(TT.outfile);
  95  }
  96  xexit();
  97}
  98
  99void split_main(void)
 100{
 101  if (!TT.b && !TT.l && !TT.n) TT.l = 1000;
 102
 103  // Allocate template for output filenames
 104  TT.outfile = xmprintf("%s%*c", (toys.optc == 2) ? toys.optargs[1] : "x",
 105    (int)TT.a, ' ');
 106
 107  // We only ever use one input, but this handles '-' or no input for us.
 108  loopfiles(toys.optargs, do_split);
 109}
 110