builtin-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 "cache.h"
   8 #include "builtin.h"
   9
  10 static const char git_mailsplit_usage[] =
  11 "git-mailsplit [-d<prec>] [-f<n>] [-b] -o<directory> <mbox>...";
  12
  13 static int is_from_line(const char *line, int len)
  14 {
  15         const char *colon;
  16
  17         if (len < 20 || memcmp("From ", line, 5))
  18                 return 0;
  19
  20         colon = line + len - 2;
  21         line += 5;
  22         for (;;) {
  23                 if (colon < line)
  24                         return 0;
  25                 if (*--colon == ':')
  26                         break;
  27         }
  28
  29         if (!isdigit(colon[-4]) ||
  30             !isdigit(colon[-2]) ||
  31             !isdigit(colon[-1]) ||
  32             !isdigit(colon[ 1]) ||
  33             !isdigit(colon[ 2]))
  34                 return 0;
  35
  36         /* year */
  37         if (strtol(colon+3, NULL, 10) <= 90)
  38                 return 0;
  39
  40         /* Ok, close enough */
  41         return 1;
  42 }
  43
  44 /* Could be as small as 64, enough to hold a Unix "From " line. */
  45 static char buf[4096];
  46
  47 /* Called with the first line (potentially partial)
  48  * already in buf[] -- normally that should begin with
  49  * the Unix "From " line.  Write it into the specified
  50  * file.
  51  */
  52 static int split_one(FILE *mbox, const char *name, int allow_bare)
  53 {
  54         FILE *output = NULL;
  55         int len = strlen(buf);
  56         int fd;
  57         int status = 0;
  58         int is_bare = !is_from_line(buf, len);
  59
  60         if (is_bare && !allow_bare)
  61                 goto corrupt;
  62
  63         fd = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  64         if (fd < 0)
  65                 die("cannot open output file %s", name);
  66         output = fdopen(fd, "w");
  67
  68         /* Copy it out, while searching for a line that begins with
  69          * "From " and having something that looks like a date format.
  70          */
  71         for (;;) {
  72                 int is_partial = (buf[len-1] != '\n');
  73
  74                 if (fputs(buf, output) == EOF)
  75                         die("cannot write output");
  76
  77                 if (fgets(buf, sizeof(buf), mbox) == NULL) {
  78                         if (feof(mbox)) {
  79                                 status = 1;
  80                                 break;
  81                         }
  82                         die("cannot read mbox");
  83                 }
  84                 len = strlen(buf);
  85                 if (!is_partial && !is_bare && is_from_line(buf, len))
  86                         break; /* done with one message */
  87         }
  88         fclose(output);
  89         return status;
  90
  91  corrupt:
  92         if (output)
  93                 fclose(output);
  94         unlink(name);
  95         fprintf(stderr, "corrupt mailbox\n");
  96         exit(1);
  97 }
  98
  99 int split_mbox(const char **mbox, const char *dir, int allow_bare, int nr_prec, int skip)
 100 {
 101         char *name = xmalloc(strlen(dir) + 2 + 3 * sizeof(skip));
 102         int ret = -1;
 103
 104         while (*mbox) {
 105                 const char *file = *mbox++;
 106                 FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 107                 int file_done = 0;
 108
 109                 if ( !f ) {
 110                         error("cannot open mbox %s", file);
 111                         goto out;
 112                 }
 113
 114                 if (fgets(buf, sizeof(buf), f) == NULL) {
 115                         if (f == stdin)
 116                                 break; /* empty stdin is OK */
 117                         error("cannot read mbox %s", file);
 118                         goto out;
 119                 }
 120
 121                 while (!file_done) {
 122                         sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 123                         file_done = split_one(f, name, allow_bare);
 124                 }
 125
 126                 if (f != stdin)
 127                         fclose(f);
 128         }
 129         ret = skip;
 130 out:
 131         free(name);
 132         return ret;
 133 }
 134 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 135 {
 136         int nr = 0, nr_prec = 4, ret;
 137         int allow_bare = 0;
 138         const char *dir = NULL;
 139         const char **argp;
 140         static const char *stdin_only[] = { "-", NULL };
 141
 142         for (argp = argv+1; *argp; argp++) {
 143                 const char *arg = *argp;
 144
 145                 if (arg[0] != '-')
 146                         break;
 147                 /* do flags here */
 148                 if ( arg[1] == 'd' ) {
 149                         nr_prec = strtol(arg+2, NULL, 10);
 150                         if (nr_prec < 3 || 10 <= nr_prec)
 151                                 usage(git_mailsplit_usage);
 152                         continue;
 153                 } else if ( arg[1] == 'f' ) {
 154                         nr = strtol(arg+2, NULL, 10);
 155                 } else if ( arg[1] == 'b' && !arg[2] ) {
 156                         allow_bare = 1;
 157                 } else if ( arg[1] == 'o' && arg[2] ) {
 158                         dir = arg+2;
 159                 } else if ( arg[1] == '-' && !arg[2] ) {
 160                         argp++; /* -- marks end of options */
 161                         break;
 162                 } else {
 163                         die("unknown option: %s", arg);
 164                 }
 165         }
 166
 167         if ( !dir ) {
 168                 /* Backwards compatibility: if no -o specified, accept
 169                    <mbox> <dir> or just <dir> */
 170                 switch (argc - (argp-argv)) {
 171                 case 1:
 172                         dir = argp[0];
 173                         argp = stdin_only;
 174                         break;
 175                 case 2:
 176                         stdin_only[0] = argp[0];
 177                         dir = argp[1];
 178                         argp = stdin_only;
 179                         break;
 180                 default:
 181                         usage(git_mailsplit_usage);
 182                 }
 183         } else {
 184                 /* New usage: if no more argument, parse stdin */
 185                 if ( !*argp )
 186                         argp = stdin_only;
 187         }
 188
 189         ret = split_mbox(argp, dir, allow_bare, nr_prec, nr);
 190         if (ret != -1)
 191                 printf("%d\n", ret);
 192
 193         return ret == -1;
 194 }