924
|
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
|
|
11 USE_SPLIT(NEWTOY(split, ">2a#<1=2>9b#<1l#<1", 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 lines;
|
|
32 long bytes;
|
|
33 long suflen;
|
|
34
|
|
35 char *outfile;
|
|
36 )
|
|
37
|
|
38 void do_split(int infd, char *in)
|
|
39 {
|
|
40 unsigned long bytesleft, linesleft, filenum, len, pos;
|
|
41 int outfd = -1;
|
|
42 struct stat st;
|
|
43
|
|
44 // posix doesn't cover permissions on output file, so copy input (or 0777)
|
|
45 st.st_mode = 0777;
|
|
46 fstat(infd, &st);
|
|
47
|
|
48 len = pos = filenum = bytesleft = linesleft = 0;
|
|
49 for (;;) {
|
|
50 int i, j;
|
|
51
|
|
52 // Refill toybuf?
|
|
53 if (len == pos) {
|
|
54 if (!(len = xread(infd, toybuf, sizeof(toybuf)))) break;
|
|
55 pos = 0;
|
|
56 }
|
|
57
|
|
58 // Start new output file?
|
|
59 if ((TT.bytes && !bytesleft) || (TT.lines && !linesleft)) {
|
|
60 char *s = TT.outfile + strlen(TT.outfile);
|
|
61
|
|
62 j = filenum++;
|
|
63 for (i = 0; i<TT.suflen; i++) {
|
|
64 *(--s) = 'a'+(j%26);
|
|
65 j /= 26;
|
|
66 }
|
|
67 if (j) error_exit("bad suffix");
|
|
68 bytesleft = TT.bytes;
|
|
69 linesleft = TT.lines;
|
|
70 if (outfd != -1) close(outfd);
|
|
71 outfd = xcreate(TT.outfile, O_RDWR|O_CREAT|O_TRUNC, st.st_mode & 0777);
|
|
72 }
|
|
73
|
|
74 // Write next chunk of output.
|
|
75 if (TT.lines) {
|
|
76 for (i = pos; i < len; ) {
|
|
77 if (toybuf[i++] == '\n' && !--linesleft) break;
|
|
78 if (!--bytesleft) break;
|
|
79 }
|
|
80 j = i - pos;
|
|
81 } else {
|
|
82 j = len - pos;
|
|
83 if (j > bytesleft) j = bytesleft;
|
|
84 bytesleft -= j;
|
|
85 }
|
|
86 xwrite(outfd, toybuf+pos, j);
|
|
87 pos += j;
|
|
88 }
|
|
89
|
|
90 if (CFG_TOYBOX_FREE) {
|
|
91 if (outfd != -1) close(outfd);
|
|
92 if (infd) close(infd);
|
|
93 free(TT.outfile);
|
|
94 }
|
|
95 xexit();
|
|
96 }
|
|
97
|
|
98 void split_main(void)
|
|
99 {
|
|
100 if (!TT.bytes && !TT.lines) TT.lines = 1000;
|
|
101
|
|
102 // Allocate template for output filenames
|
|
103 TT.outfile = xmsprintf("%s% *c", (toys.optc == 2) ? toys.optargs[1] : "x",
|
|
104 (int)TT.suflen, ' ');
|
|
105
|
|
106 // We only ever use one input, but this handles '-' or no input for us.
|
|
107 loopfiles(toys.optargs, do_split);
|
|
108 }
|