mailsplit.c

   1 /*
   2  * Totally braindamaged mbox splitter program.
   3  *
   4  * It just splits a mbox into a list of files: "0001" "0002" ..
   5  * so you can process them further from there.
   6  */
   7 #include <unistd.h>
   8 #include <stdlib.h>
   9 #include <fcntl.h>
  10 #include <sys/types.h>
  11 #include <sys/stat.h>
  12 #include <sys/mman.h>
  13 #include <string.h>
  14 #include <stdio.h>
  15 #include <ctype.h>
  16 #include <assert.h>
  17
  18 static int usage(void)
  19 {
  20         fprintf(stderr, "mailsplit <mbox> <directory>\n");
  21         exit(1);
  22 }
  23
  24 static int linelen(const char *map, unsigned long size)
  25 {
  26         int len = 0, c;
  27
  28         do {
  29                 c = *map;
  30                 map++;
  31                 size--;
  32                 len++;
  33         } while (size && c != '\n');
  34         return len;
  35 }
  36
  37 static int is_from_line(const char *line, int len)
  38 {
  39         const char *colon;
  40
  41         if (len < 20 || memcmp("From ", line, 5))
  42                 return 0;
  43
  44         colon = line + len - 2;
  45         line += 5;
  46         for (;;) {
  47                 if (colon < line)
  48                         return 0;
  49                 if (*--colon == ':')
  50                         break;
  51         }
  52
  53         if (!isdigit(colon[-4]) ||
  54             !isdigit(colon[-2]) ||
  55             !isdigit(colon[-1]) ||
  56             !isdigit(colon[ 1]) ||
  57             !isdigit(colon[ 2]))
  58                 return 0;
  59
  60         /* year */
  61         if (strtol(colon+3, NULL, 10) <= 90)
  62                 return 0;
  63
  64         /* Ok, close enough */
  65         return 1;
  66 }
  67
  68 static int parse_email(const void *map, unsigned long size)
  69 {
  70         unsigned long offset;
  71
  72         if (size < 6 || memcmp("From ", map, 5))
  73                 goto corrupt;
  74
  75         /* Make sure we don't trigger on this first line */
  76         map++; size--; offset=1;
  77
  78         /*
  79          * Search for a line beginning with "From ", and
  80          * having something that looks like a date format.
  81          */
  82         do {
  83                 int len = linelen(map, size);
  84                 if (is_from_line(map, len))
  85                         return offset;
  86                 map += len;
  87                 size -= len;
  88                 offset += len;
  89         } while (size);
  90         return offset;
  91
  92 corrupt:
  93         fprintf(stderr, "corrupt mailbox\n");
  94         exit(1);
  95 }
  96
  97 int main(int argc, char **argv)
  98 {
  99         int fd, nr;
 100         struct stat st;
 101         unsigned long size;
 102         void *map;
 103
 104         if (argc != 3)
 105                 usage();
 106         fd = open(argv[1], O_RDONLY);
 107         if (fd < 0) {
 108                 perror(argv[1]);
 109                 exit(1);
 110         }
 111         if (chdir(argv[2]) < 0)
 112                 usage();
 113         if (fstat(fd, &st) < 0) {
 114                 perror("stat");
 115                 exit(1);
 116         }
 117         size = st.st_size;
 118         map = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
 119         if (map == MAP_FAILED) {
 120                 perror("mmap");
 121                 close(fd);
 122                 exit(1);
 123         }
 124         close(fd);
 125         nr = 0;
 126         do {
 127                 char name[10];
 128                 unsigned long len = parse_email(map, size);
 129                 assert(len <= size);
 130                 sprintf(name, "%04d", ++nr);
 131                 fd = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
 132                 if (fd < 0) {
 133                         perror(name);
 134                         exit(1);
 135                 }
 136                 if (write(fd, map, len) != len) {
 137                         perror("write");
 138                         exit(1);
 139                 }
 140                 close(fd);
 141                 map += len;
 142                 size -= len;
 143         } while (size > 0);
 144         return 0;
 145 }