path: root/coreutils/split.c (plain)
blob: 1e1673efbe2c4b9fa51611e13c25752488ad90c4

1	/* vi: set sw=4 ts=4: */
2	/*
3	* split - split a file into pieces
4	* Copyright (c) 2007 Bernhard Reutner-Fischer
5	*
6	* Licensed under GPLv2 or later, see file LICENSE in this source tree.
7	*/
8	/* BB_AUDIT: SUSv3 compliant
9	* SUSv3 requirements:
10	* http://www.opengroup.org/onlinepubs/009695399/utilities/split.html
11	*/
12
13	//usage:#define split_trivial_usage
14	//usage: "[OPTIONS] [INPUT [PREFIX]]"
15	//usage:#define split_full_usage "\n\n"
16	//usage: " -b N[k|m] Split by N (kilo|mega)bytes"
17	//usage: "\n -l N Split by N lines"
18	//usage: "\n -a N Use N letters as suffix"
19	//usage:
20	//usage:#define split_example_usage
21	//usage: "$ split TODO foo\n"
22	//usage: "$ cat TODO | split -a 2 -l 2 TODO_\n"
23
24	#include "libbb.h"
25
26	#if ENABLE_FEATURE_SPLIT_FANCY
27	static const struct suffix_mult split_suffixes[] = {
28	{ "b", 512 },
29	{ "k", 1024 },
30	{ "m", 1024*1024 },
31	{ "g", 1024*1024*1024 },
32	{ "", 0 }
33	};
34	#endif
35
36	/* Increment the suffix part of the filename.
37	* Returns NULL if we are out of filenames.
38	*/
39	static char *next_file(char *old, unsigned suffix_len)
40	{
41	size_t end = strlen(old);
42	unsigned i = 1;
43	char *curr;
44
45	while (1) {
46	curr = old + end - i;
47	if (*curr < 'z') {
48	*curr += 1;
49	break;
50	}
51	i++;
52	if (i > suffix_len) {
53	return NULL;
54	}
55	*curr = 'a';
56	}
57
58	return old;
59	}
60
61	#define read_buffer bb_common_bufsiz1
62	enum { READ_BUFFER_SIZE = COMMON_BUFSIZE - 1 };
63
64	#define SPLIT_OPT_l (1<<0)
65	#define SPLIT_OPT_b (1<<1)
66	#define SPLIT_OPT_a (1<<2)
67
68	int split_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
69	int split_main(int argc UNUSED_PARAM, char **argv)
70	{
71	unsigned suffix_len = 2;
72	char *pfx;
73	char *count_p;
74	const char *sfx;
75	off_t cnt = 1000;
76	off_t remaining = 0;
77	unsigned opt;
78	ssize_t bytes_read, to_write;
79	char *src;
80
81	opt_complementary = "?2:a+"; /* max 2 args; -a N */
82	opt = getopt32(argv, "l:b:a:", &count_p, &count_p, &suffix_len);
83
84	if (opt & SPLIT_OPT_l)
85	cnt = XATOOFF(count_p);
86	if (opt & SPLIT_OPT_b) // FIXME: also needs XATOOFF
87	cnt = xatoull_sfx(count_p,
88	IF_FEATURE_SPLIT_FANCY(split_suffixes)
89	IF_NOT_FEATURE_SPLIT_FANCY(km_suffixes)
90	);
91	sfx = "x";
92
93	argv += optind;
94	if (argv[0]) {
95	int fd;
96	if (argv[1])
97	sfx = argv[1];
98	fd = xopen_stdin(argv[0]);
99	xmove_fd(fd, STDIN_FILENO);
100	} else {
101	argv[0] = (char *) bb_msg_standard_input;
102	}
103
104	if (NAME_MAX < strlen(sfx) + suffix_len)
105	bb_error_msg_and_die("suffix too long");
106
107	{
108	char *char_p = xzalloc(suffix_len + 1);
109	memset(char_p, 'a', suffix_len);
110	pfx = xasprintf("%s%s", sfx, char_p);
111	if (ENABLE_FEATURE_CLEAN_UP)
112	free(char_p);
113	}
114
115	while (1) {
116	bytes_read = safe_read(STDIN_FILENO, read_buffer, READ_BUFFER_SIZE);
117	if (!bytes_read)
118	break;
119	if (bytes_read < 0)
120	bb_simple_perror_msg_and_die(argv[0]);
121	src = read_buffer;
122	do {
123	if (!remaining) {
124	if (!pfx)
125	bb_error_msg_and_die("suffixes exhausted");
126	xmove_fd(xopen(pfx, O_WRONLY | O_CREAT | O_TRUNC), 1);
127	pfx = next_file(pfx, suffix_len);
128	remaining = cnt;
129	}
130
131	if (opt & SPLIT_OPT_b) {
132	/* split by bytes */
133	to_write = (bytes_read < remaining) ? bytes_read : remaining;
134	remaining -= to_write;
135	} else {
136	/* split by lines */
137	/* can be sped up by using _memrchr_
138	* and writing many lines at once... */
139	char *end = memchr(src, '\n', bytes_read);
140	if (end) {
141	--remaining;
142	to_write = end - src + 1;
143	} else {
144	to_write = bytes_read;
145	}
146	}
147
148	xwrite(STDOUT_FILENO, src, to_write);
149	bytes_read -= to_write;
150	src += to_write;
151	} while (bytes_read);
152	}
153	return EXIT_SUCCESS;
154	}
155