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 #include "common_bufsiz.h"
  26
  27 #if ENABLE_FEATURE_SPLIT_FANCY
  28 static const struct suffix_mult split_suffixes[] = {
  29         { "b", 512 },
  30         { "k", 1024 },
  31         { "m", 1024*1024 },
  32         { "g", 1024*1024*1024 },
  33         { "", 0 }
  34 };
  35 #endif
  36
  37 /* Increment the suffix part of the filename.
  38  * Returns NULL if we are out of filenames.
  39  */
  40 static char *next_file(char *old, unsigned suffix_len)
  41 {
  42         size_t end = strlen(old);
  43         unsigned i = 1;
  44         char *curr;
  45
  46         while (1) {
  47                 curr = old + end - i;
  48                 if (*curr < 'z') {
  49                         *curr += 1;
  50                         break;
  51                 }
  52                 i++;
  53                 if (i > suffix_len) {
  54                         return NULL;
  55                 }
  56                 *curr = 'a';
  57         }
  58
  59         return old;
  60 }
  61
  62 #define read_buffer bb_common_bufsiz1
  63 enum { READ_BUFFER_SIZE = COMMON_BUFSIZE - 1 };
  64
  65 #define SPLIT_OPT_l (1<<0)
  66 #define SPLIT_OPT_b (1<<1)
  67 #define SPLIT_OPT_a (1<<2)
  68
  69 int split_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  70 int split_main(int argc UNUSED_PARAM, char **argv)
  71 {
  72         unsigned suffix_len = 2;
  73         char *pfx;
  74         char *count_p;
  75         const char *sfx;
  76         off_t cnt = 1000;
  77         off_t remaining = 0;
  78         unsigned opt;
  79         ssize_t bytes_read, to_write;
  80         char *src;
  81
  82         setup_common_bufsiz();
  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,
  91                                 IF_FEATURE_SPLIT_FANCY(split_suffixes)
  92                                 IF_NOT_FEATURE_SPLIT_FANCY(km_suffixes)
  93                 );
  94         sfx = "x";
  95
  96         argv += optind;
  97         if (argv[0]) {
  98                 int fd;
  99                 if (argv[1])
 100                         sfx = argv[1];
 101                 fd = xopen_stdin(argv[0]);
 102                 xmove_fd(fd, STDIN_FILENO);
 103         } else {
 104                 argv[0] = (char *) bb_msg_standard_input;
 105         }
 106
 107         if (NAME_MAX < strlen(sfx) + suffix_len)
 108                 bb_error_msg_and_die("suffix too long");
 109
 110         {
 111                 char *char_p = xzalloc(suffix_len + 1);
 112                 memset(char_p, 'a', suffix_len);
 113                 pfx = xasprintf("%s%s", sfx, char_p);
 114                 if (ENABLE_FEATURE_CLEAN_UP)
 115                         free(char_p);
 116         }
 117
 118         while (1) {
 119                 bytes_read = safe_read(STDIN_FILENO, read_buffer, READ_BUFFER_SIZE);
 120                 if (!bytes_read)
 121                         break;
 122                 if (bytes_read < 0)
 123                         bb_simple_perror_msg_and_die(argv[0]);
 124                 src = read_buffer;
 125                 do {
 126                         if (!remaining) {
 127                                 if (!pfx)
 128                                         bb_error_msg_and_die("suffixes exhausted");
 129                                 xmove_fd(xopen(pfx, O_WRONLY | O_CREAT | O_TRUNC), 1);
 130                                 pfx = next_file(pfx, suffix_len);
 131                                 remaining = cnt;
 132                         }
 133
 134                         if (opt & SPLIT_OPT_b) {
 135                                 /* split by bytes */
 136                                 to_write = (bytes_read < remaining) ? bytes_read : remaining;
 137                                 remaining -= to_write;
 138                         } else {
 139                                 /* split by lines */
 140                                 /* can be sped up by using _memrchr_
 141                                  * and writing many lines at once... */
 142                                 char *end = memchr(src, '\n', bytes_read);
 143                                 if (end) {
 144                                         --remaining;
 145                                         to_write = end - src + 1;
 146                                 } else {
 147                                         to_write = bytes_read;
 148                                 }
 149                         }
 150
 151                         xwrite(STDOUT_FILENO, src, to_write);
 152                         bytes_read -= to_write;
 153                         src += to_write;
 154                 } while (bytes_read);
 155         }
 156         return EXIT_SUCCESS;
 157 }