coreutils/split.c

   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 static const struct suffix_mult split_suffices[] = {
  27 #if ENABLE_FEATURE_SPLIT_FANCY
  28         { "b", 512 },
  29 #endif
  30         { "k", 1024 },
  31         { "m", 1024*1024 },
  32 #if ENABLE_FEATURE_SPLIT_FANCY
  33         { "g", 1024*1024*1024 },
  34 #endif
  35         { "", 0 }
  36 };
  37
  38 /* Increment the suffix part of the filename.
  39  * Returns NULL if we are out of filenames.
  40  */
  41 static char *next_file(char *old, unsigned suffix_len)
  42 {
  43         size_t end = strlen(old);
  44         unsigned i = 1;
  45         char *curr;
  46
  47         while (1) {
  48                 curr = old + end - i;
  49                 if (*curr < 'z') {
  50                         *curr += 1;
  51                         break;
  52                 }
  53                 i++;
  54                 if (i > suffix_len) {
  55                         return NULL;
  56                 }
  57                 *curr = 'a';
  58         }
  59
  60         return old;
  61 }
  62
  63 #define read_buffer bb_common_bufsiz1
  64 enum { READ_BUFFER_SIZE = COMMON_BUFSIZE - 1 };
  65
  66 #define SPLIT_OPT_l (1<<0)
  67 #define SPLIT_OPT_b (1<<1)
  68 #define SPLIT_OPT_a (1<<2)
  69
  70 int split_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  71 int split_main(int argc UNUSED_PARAM, char **argv)
  72 {
  73         unsigned suffix_len = 2;
  74         char *pfx;
  75         char *count_p;
  76         const char *sfx;
  77         off_t cnt = 1000;
  78         off_t remaining = 0;
  79         unsigned opt;
  80         ssize_t bytes_read, to_write;
  81         char *src;
  82
  83         opt_complementary = "?2:a+"; /* max 2 args; -a N */
  84         opt = getopt32(argv, "l:b:a:", &count_p, &count_p, &suffix_len);
  85
  86         if (opt & SPLIT_OPT_l)
  87                 cnt = XATOOFF(count_p);
  88         if (opt & SPLIT_OPT_b) // FIXME: also needs XATOOFF
  89                 cnt = xatoull_sfx(count_p, split_suffices);
  90         sfx = "x";
  91
  92         argv += optind;
  93         if (argv[0]) {
  94                 int fd;
  95                 if (argv[1])
  96                         sfx = argv[1];
  97                 fd = xopen_stdin(argv[0]);
  98                 xmove_fd(fd, STDIN_FILENO);
  99         } else {
 100                 argv[0] = (char *) bb_msg_standard_input;
 101         }
 102
 103         if (NAME_MAX < strlen(sfx) + suffix_len)
 104                 bb_error_msg_and_die("suffix too long");
 105
 106         {
 107                 char *char_p = xzalloc(suffix_len + 1);
 108                 memset(char_p, 'a', suffix_len);
 109                 pfx = xasprintf("%s%s", sfx, char_p);
 110                 if (ENABLE_FEATURE_CLEAN_UP)
 111                         free(char_p);
 112         }
 113
 114         while (1) {
 115                 bytes_read = safe_read(STDIN_FILENO, read_buffer, READ_BUFFER_SIZE);
 116                 if (!bytes_read)
 117                         break;
 118                 if (bytes_read < 0)
 119                         bb_simple_perror_msg_and_die(argv[0]);
 120                 src = read_buffer;
 121                 do {
 122                         if (!remaining) {
 123                                 if (!pfx)
 124                                         bb_error_msg_and_die("suffixes exhausted");
 125                                 xmove_fd(xopen(pfx, O_WRONLY | O_CREAT | O_TRUNC), 1);
 126                                 pfx = next_file(pfx, suffix_len);
 127                                 remaining = cnt;
 128                         }
 129
 130                         if (opt & SPLIT_OPT_b) {
 131                                 /* split by bytes */
 132                                 to_write = (bytes_read < remaining) ? bytes_read : remaining;
 133                                 remaining -= to_write;
 134                         } else {
 135                                 /* split by lines */
 136                                 /* can be sped up by using _memrchr_
 137                                  * and writing many lines at once... */
 138                                 char *end = memchr(src, '\n', bytes_read);
 139                                 if (end) {
 140                                         --remaining;
 141                                         to_write = end - src + 1;
 142                                 } else {
 143                                         to_write = bytes_read;
 144                                 }
 145                         }
 146
 147                         xwrite(STDOUT_FILENO, src, to_write);
 148                         bytes_read -= to_write;
 149                         src += to_write;
 150                 } while (bytes_read);
 151         }
 152         return EXIT_SUCCESS;
 153 }