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:       "Options:"
  17 //usage:     "\n        -b N[k|m]       Split by N (kilo|mega)bytes"
  18 //usage:     "\n        -l N            Split by N lines"
  19 //usage:     "\n        -a N            Use N letters as suffix"
  20 //usage:
  21 //usage:#define split_example_usage
  22 //usage:       "$ split TODO foo\n"
  23 //usage:       "$ cat TODO | split -a 2 -l 2 TODO_\n"
  24
  25 #include "libbb.h"
  26
  27 static const struct suffix_mult split_suffices[] = {
  28 #if ENABLE_FEATURE_SPLIT_FANCY
  29         { "b", 512 },
  30 #endif
  31         { "k", 1024 },
  32         { "m", 1024*1024 },
  33 #if ENABLE_FEATURE_SPLIT_FANCY
  34         { "g", 1024*1024*1024 },
  35 #endif
  36         { "", 0 }
  37 };
  38
  39 /* Increment the suffix part of the filename.
  40  * Returns NULL if we are out of filenames.
  41  */
  42 static char *next_file(char *old, unsigned suffix_len)
  43 {
  44         size_t end = strlen(old);
  45         unsigned i = 1;
  46         char *curr;
  47
  48         while (1) {
  49                 curr = old + end - i;
  50                 if (*curr < 'z') {
  51                         *curr += 1;
  52                         break;
  53                 }
  54                 i++;
  55                 if (i > suffix_len) {
  56                         return NULL;
  57                 }
  58                 *curr = 'a';
  59         }
  60
  61         return old;
  62 }
  63
  64 #define read_buffer bb_common_bufsiz1
  65 enum { READ_BUFFER_SIZE = COMMON_BUFSIZE - 1 };
  66
  67 #define SPLIT_OPT_l (1<<0)
  68 #define SPLIT_OPT_b (1<<1)
  69 #define SPLIT_OPT_a (1<<2)
  70
  71 int split_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  72 int split_main(int argc UNUSED_PARAM, char **argv)
  73 {
  74         unsigned suffix_len = 2;
  75         char *pfx;
  76         char *count_p;
  77         const char *sfx;
  78         off_t cnt = 1000;
  79         off_t remaining = 0;
  80         unsigned opt;
  81         ssize_t bytes_read, to_write;
  82         char *src;
  83
  84         opt_complementary = "?2:a+"; /* max 2 args; -a N */
  85         opt = getopt32(argv, "l:b:a:", &count_p, &count_p, &suffix_len);
  86
  87         if (opt & SPLIT_OPT_l)
  88                 cnt = XATOOFF(count_p);
  89         if (opt & SPLIT_OPT_b) // FIXME: also needs XATOOFF
  90                 cnt = xatoull_sfx(count_p, split_suffices);
  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 }