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 <string.h>
  13 #include <stdio.h>
  14 #include "cache.h"
  15
  16 static const char git_mailsplit_usage[] =
  17 "git-mailsplit [-d<prec>] [-f<n>] [-b] -o<directory> <mbox>...";
  18
  19 static int is_from_line(const char *line, int len)
  20 {
  21         const char *colon;
  22
  23         if (len < 20 || memcmp("From ", line, 5))
  24                 return 0;
  25
  26         colon = line + len - 2;
  27         line += 5;
  28         for (;;) {
  29                 if (colon < line)
  30                         return 0;
  31                 if (*--colon == ':')
  32                         break;
  33         }
  34
  35         if (!isdigit(colon[-4]) ||
  36             !isdigit(colon[-2]) ||
  37             !isdigit(colon[-1]) ||
  38             !isdigit(colon[ 1]) ||
  39             !isdigit(colon[ 2]))
  40                 return 0;
  41
  42         /* year */
  43         if (strtol(colon+3, NULL, 10) <= 90)
  44                 return 0;
  45
  46         /* Ok, close enough */
  47         return 1;
  48 }
  49
  50 /* Could be as small as 64, enough to hold a Unix "From " line. */
  51 static char buf[4096];
  52
  53 /* Called with the first line (potentially partial)
  54  * already in buf[] -- normally that should begin with
  55  * the Unix "From " line.  Write it into the specified
  56  * file.
  57  */
  58 static int split_one(FILE *mbox, const char *name, int allow_bare)
  59 {
  60         FILE *output = NULL;
  61         int len = strlen(buf);
  62         int fd;
  63         int status = 0;
  64         int is_bare = !is_from_line(buf, len);
  65
  66         if (is_bare && !allow_bare)
  67                 goto corrupt;
  68
  69         fd = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  70         if (fd < 0)
  71                 die("cannot open output file %s", name);
  72         output = fdopen(fd, "w");
  73
  74         /* Copy it out, while searching for a line that begins with
  75          * "From " and having something that looks like a date format.
  76          */
  77         for (;;) {
  78                 int is_partial = (buf[len-1] != '\n');
  79
  80                 if (fputs(buf, output) == EOF)
  81                         die("cannot write output");
  82
  83                 if (fgets(buf, sizeof(buf), mbox) == NULL) {
  84                         if (feof(mbox)) {
  85                                 status = 1;
  86                                 break;
  87                         }
  88                         die("cannot read mbox");
  89                 }
  90                 len = strlen(buf);
  91                 if (!is_partial && !is_bare && is_from_line(buf, len))
  92                         break; /* done with one message */
  93         }
  94         fclose(output);
  95         return status;
  96
  97  corrupt:
  98         if (output)
  99                 fclose(output);
 100         unlink(name);
 101         fprintf(stderr, "corrupt mailbox\n");
 102         exit(1);
 103 }
 104
 105 int main(int argc, const char **argv)
 106 {
 107         int nr = 0, nr_prec = 4;
 108         int allow_bare = 0;
 109         const char *dir = NULL;
 110         const char **argp;
 111         static const char *stdin_only[] = { "-", NULL };
 112         char *name;
 113
 114         for (argp = argv+1; *argp; argp++) {
 115                 const char *arg = *argp;
 116
 117                 if (arg[0] != '-')
 118                         break;
 119                 /* do flags here */
 120                 if ( arg[1] == 'd' ) {
 121                         nr_prec = strtol(arg+2, NULL, 10);
 122                         if (nr_prec < 3 || 10 <= nr_prec)
 123                                 usage(git_mailsplit_usage);
 124                         continue;
 125                 } else if ( arg[1] == 'f' ) {
 126                         nr = strtol(arg+2, NULL, 10);
 127                 } else if ( arg[1] == 'b' && !arg[2] ) {
 128                         allow_bare = 1;
 129                 } else if ( arg[1] == 'o' && arg[2] ) {
 130                         dir = arg+2;
 131                 } else if ( arg[1] == '-' && !arg[2] ) {
 132                         argp++; /* -- marks end of options */
 133                         break;
 134                 } else {
 135                         die("unknown option: %s", arg);
 136                 }
 137         }
 138
 139         if ( !dir ) {
 140                 /* Backwards compatibility: if no -o specified, accept
 141                    <mbox> <dir> or just <dir> */
 142                 switch (argc - (argp-argv)) {
 143                 case 1:
 144                         dir = argp[0];
 145                         argp = stdin_only;
 146                         break;
 147                 case 2:
 148                         stdin_only[0] = argp[0];
 149                         dir = argp[1];
 150                         argp = stdin_only;
 151                         break;
 152                 default:
 153                         usage(git_mailsplit_usage);
 154                 }
 155         } else {
 156                 /* New usage: if no more argument, parse stdin */
 157                 if ( !*argp )
 158                         argp = stdin_only;
 159         }
 160
 161         name = xmalloc(strlen(dir) + 2 + 3 * sizeof(nr));
 162
 163         while (*argp) {
 164                 const char *file = *argp++;
 165                 FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "rt");
 166                 int file_done = 0;
 167
 168                 if ( !f )
 169                         die ("cannot open mbox %s", file);
 170
 171                 if (fgets(buf, sizeof(buf), f) == NULL) {
 172                         if (f == stdin)
 173                                 break; /* empty stdin is OK */
 174                         die("cannot read mbox %s", file);
 175                 }
 176
 177                 while (!file_done) {
 178                         sprintf(name, "%s/%0*d", dir, nr_prec, ++nr);
 179                         file_done = split_one(f, name, allow_bare);
 180                 }
 181
 182                 if (f != stdin)
 183                         fclose(f);
 184         }
 185
 186         printf("%d\n", nr);
 187         return 0;
 188 }