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 <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 #include "builtin.h"
  16
  17 static const char git_mailsplit_usage[] =
  18 "git-mailsplit [-d<prec>] [-f<n>] [-b] -o<directory> <mbox>...";
  19
  20 static int is_from_line(const char *line, int len)
  21 {
  22         const char *colon;
  23
  24         if (len < 20 || memcmp("From ", line, 5))
  25                 return 0;
  26
  27         colon = line + len - 2;
  28         line += 5;
  29         for (;;) {
  30                 if (colon < line)
  31                         return 0;
  32                 if (*--colon == ':')
  33                         break;
  34         }
  35
  36         if (!isdigit(colon[-4]) ||
  37             !isdigit(colon[-2]) ||
  38             !isdigit(colon[-1]) ||
  39             !isdigit(colon[ 1]) ||
  40             !isdigit(colon[ 2]))
  41                 return 0;
  42
  43         /* year */
  44         if (strtol(colon+3, NULL, 10) <= 90)
  45                 return 0;
  46
  47         /* Ok, close enough */
  48         return 1;
  49 }
  50
  51 /* Could be as small as 64, enough to hold a Unix "From " line. */
  52 static char buf[4096];
  53
  54 /* Called with the first line (potentially partial)
  55  * already in buf[] -- normally that should begin with
  56  * the Unix "From " line.  Write it into the specified
  57  * file.
  58  */
  59 static int split_one(FILE *mbox, const char *name, int allow_bare)
  60 {
  61         FILE *output = NULL;
  62         int len = strlen(buf);
  63         int fd;
  64         int status = 0;
  65         int is_bare = !is_from_line(buf, len);
  66
  67         if (is_bare && !allow_bare)
  68                 goto corrupt;
  69
  70         fd = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  71         if (fd < 0)
  72                 die("cannot open output file %s", name);
  73         output = fdopen(fd, "w");
  74
  75         /* Copy it out, while searching for a line that begins with
  76          * "From " and having something that looks like a date format.
  77          */
  78         for (;;) {
  79                 int is_partial = (buf[len-1] != '\n');
  80
  81                 if (fputs(buf, output) == EOF)
  82                         die("cannot write output");
  83
  84                 if (fgets(buf, sizeof(buf), mbox) == NULL) {
  85                         if (feof(mbox)) {
  86                                 status = 1;
  87                                 break;
  88                         }
  89                         die("cannot read mbox");
  90                 }
  91                 len = strlen(buf);
  92                 if (!is_partial && !is_bare && is_from_line(buf, len))
  93                         break; /* done with one message */
  94         }
  95         fclose(output);
  96         return status;
  97
  98  corrupt:
  99         if (output)
 100                 fclose(output);
 101         unlink(name);
 102         fprintf(stderr, "corrupt mailbox\n");
 103         exit(1);
 104 }
 105
 106 int split_mbox(const char **mbox, const char *dir, int allow_bare, int nr_prec, int skip)
 107 {
 108         char *name = xmalloc(strlen(dir) + 2 + 3 * sizeof(skip));
 109         int ret = -1;
 110
 111         while (*mbox) {
 112                 const char *file = *mbox++;
 113                 FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 114                 int file_done = 0;
 115
 116                 if ( !f ) {
 117                         error("cannot open mbox %s", file);
 118                         goto out;
 119                 }
 120
 121                 if (fgets(buf, sizeof(buf), f) == NULL) {
 122                         if (f == stdin)
 123                                 break; /* empty stdin is OK */
 124                         error("cannot read mbox %s", file);
 125                         goto out;
 126                 }
 127
 128                 while (!file_done) {
 129                         sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 130                         file_done = split_one(f, name, allow_bare);
 131                 }
 132
 133                 if (f != stdin)
 134                         fclose(f);
 135         }
 136         ret = skip;
 137 out:
 138         free(name);
 139         return ret;
 140 }
 141 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 142 {
 143         int nr = 0, nr_prec = 4, ret;
 144         int allow_bare = 0;
 145         const char *dir = NULL;
 146         const char **argp;
 147         static const char *stdin_only[] = { "-", NULL };
 148
 149         for (argp = argv+1; *argp; argp++) {
 150                 const char *arg = *argp;
 151
 152                 if (arg[0] != '-')
 153                         break;
 154                 /* do flags here */
 155                 if ( arg[1] == 'd' ) {
 156                         nr_prec = strtol(arg+2, NULL, 10);
 157                         if (nr_prec < 3 || 10 <= nr_prec)
 158                                 usage(git_mailsplit_usage);
 159                         continue;
 160                 } else if ( arg[1] == 'f' ) {
 161                         nr = strtol(arg+2, NULL, 10);
 162                 } else if ( arg[1] == 'b' && !arg[2] ) {
 163                         allow_bare = 1;
 164                 } else if ( arg[1] == 'o' && arg[2] ) {
 165                         dir = arg+2;
 166                 } else if ( arg[1] == '-' && !arg[2] ) {
 167                         argp++; /* -- marks end of options */
 168                         break;
 169                 } else {
 170                         die("unknown option: %s", arg);
 171                 }
 172         }
 173
 174         if ( !dir ) {
 175                 /* Backwards compatibility: if no -o specified, accept
 176                    <mbox> <dir> or just <dir> */
 177                 switch (argc - (argp-argv)) {
 178                 case 1:
 179                         dir = argp[0];
 180                         argp = stdin_only;
 181                         break;
 182                 case 2:
 183                         stdin_only[0] = argp[0];
 184                         dir = argp[1];
 185                         argp = stdin_only;
 186                         break;
 187                 default:
 188                         usage(git_mailsplit_usage);
 189                 }
 190         } else {
 191                 /* New usage: if no more argument, parse stdin */
 192                 if ( !*argp )
 193                         argp = stdin_only;
 194         }
 195
 196         ret = split_mbox(argp, dir, allow_bare, nr_prec, nr);
 197         if (ret != -1)
 198                 printf("%d\n", ret);
 199
 200         return ret == -1;
 201 }