git/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 #include "path-list.h"
  10
  11 static const char git_mailsplit_usage[] =
  12 "git-mailsplit [-d<prec>] [-f<n>] [-b] -o<directory> <mbox>|<Maildir>...";
  13
  14 static int is_from_line(const char *line, int len)
  15 {
  16         const char *colon;
  17
  18         if (len < 20 || memcmp("From ", line, 5))
  19                 return 0;
  20
  21         colon = line + len - 2;
  22         line += 5;
  23         for (;;) {
  24                 if (colon < line)
  25                         return 0;
  26                 if (*--colon == ':')
  27                         break;
  28         }
  29
  30         if (!isdigit(colon[-4]) ||
  31             !isdigit(colon[-2]) ||
  32             !isdigit(colon[-1]) ||
  33             !isdigit(colon[ 1]) ||
  34             !isdigit(colon[ 2]))
  35                 return 0;
  36
  37         /* year */
  38         if (strtol(colon+3, NULL, 10) <= 90)
  39                 return 0;
  40
  41         /* Ok, close enough */
  42         return 1;
  43 }
  44
  45 /* Could be as small as 64, enough to hold a Unix "From " line. */
  46 static char buf[4096];
  47
  48 /* Called with the first line (potentially partial)
  49  * already in buf[] -- normally that should begin with
  50  * the Unix "From " line.  Write it into the specified
  51  * file.
  52  */
  53 static int split_one(FILE *mbox, const char *name, int allow_bare)
  54 {
  55         FILE *output = NULL;
  56         int len = strlen(buf);
  57         int fd;
  58         int status = 0;
  59         int is_bare = !is_from_line(buf, len);
  60
  61         if (is_bare && !allow_bare)
  62                 goto corrupt;
  63
  64         fd = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  65         if (fd < 0)
  66                 die("cannot open output file %s", name);
  67         output = fdopen(fd, "w");
  68
  69         /* Copy it out, while searching for a line that begins with
  70          * "From " and having something that looks like a date format.
  71          */
  72         for (;;) {
  73                 int is_partial = (buf[len-1] != '\n');
  74
  75                 if (fputs(buf, output) == EOF)
  76                         die("cannot write output");
  77
  78                 if (fgets(buf, sizeof(buf), mbox) == NULL) {
  79                         if (feof(mbox)) {
  80                                 status = 1;
  81                                 break;
  82                         }
  83                         die("cannot read mbox");
  84                 }
  85                 len = strlen(buf);
  86                 if (!is_partial && !is_bare && is_from_line(buf, len))
  87                         break; /* done with one message */
  88         }
  89         fclose(output);
  90         return status;
  91
  92  corrupt:
  93         if (output)
  94                 fclose(output);
  95         unlink(name);
  96         fprintf(stderr, "corrupt mailbox\n");
  97         exit(1);
  98 }
  99
 100 static int populate_maildir_list(struct path_list *list, const char *path)
 101 {
 102         DIR *dir;
 103         struct dirent *dent;
 104
 105         if ((dir = opendir(path)) == NULL) {
 106                 error("cannot opendir %s (%s)", path, strerror(errno));
 107                 return -1;
 108         }
 109
 110         while ((dent = readdir(dir)) != NULL) {
 111                 if (dent->d_name[0] == '.')
 112                         continue;
 113                 path_list_insert(dent->d_name, list);
 114         }
 115
 116         closedir(dir);
 117
 118         return 0;
 119 }
 120
 121 static int split_maildir(const char *maildir, const char *dir,
 122         int nr_prec, int skip)
 123 {
 124         char file[PATH_MAX];
 125         char curdir[PATH_MAX];
 126         char name[PATH_MAX];
 127         int ret = -1;
 128         int i;
 129         struct path_list list = {NULL, 0, 0, 1};
 130
 131         snprintf(curdir, sizeof(curdir), "%s/cur", maildir);
 132         if (populate_maildir_list(&list, curdir) < 0)
 133                 goto out;
 134
 135         for (i = 0; i < list.nr; i++) {
 136                 FILE *f;
 137                 snprintf(file, sizeof(file), "%s/%s", curdir, list.items[i].path);
 138                 f = fopen(file, "r");
 139                 if (!f) {
 140                         error("cannot open mail %s (%s)", file, strerror(errno));
 141                         goto out;
 142                 }
 143
 144                 if (fgets(buf, sizeof(buf), f) == NULL) {
 145                         error("cannot read mail %s (%s)", file, strerror(errno));
 146                         goto out;
 147                 }
 148
 149                 sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 150                 split_one(f, name, 1);
 151
 152                 fclose(f);
 153         }
 154
 155         path_list_clear(&list, 1);
 156
 157         ret = skip;
 158 out:
 159         return ret;
 160 }
 161
 162 static int split_mbox(const char *file, const char *dir, int allow_bare,
 163                       int nr_prec, int skip)
 164 {
 165         char name[PATH_MAX];
 166         int ret = -1;
 167
 168         FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 169         int file_done = 0;
 170
 171         if (!f) {
 172                 error("cannot open mbox %s", file);
 173                 goto out;
 174         }
 175
 176         if (fgets(buf, sizeof(buf), f) == NULL) {
 177                 /* empty stdin is OK */
 178                 if (f != stdin) {
 179                         error("cannot read mbox %s", file);
 180                         goto out;
 181                 }
 182                 file_done = 1;
 183         }
 184
 185         while (!file_done) {
 186                 sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 187                 file_done = split_one(f, name, allow_bare);
 188         }
 189
 190         if (f != stdin)
 191                 fclose(f);
 192
 193         ret = skip;
 194 out:
 195         return ret;
 196 }
 197
 198 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 199 {
 200         int nr = 0, nr_prec = 4, num = 0;
 201         int allow_bare = 0;
 202         const char *dir = NULL;
 203         const char **argp;
 204         static const char *stdin_only[] = { "-", NULL };
 205
 206         for (argp = argv+1; *argp; argp++) {
 207                 const char *arg = *argp;
 208
 209                 if (arg[0] != '-')
 210                         break;
 211                 /* do flags here */
 212                 if ( arg[1] == 'd' ) {
 213                         nr_prec = strtol(arg+2, NULL, 10);
 214                         if (nr_prec < 3 || 10 <= nr_prec)
 215                                 usage(git_mailsplit_usage);
 216                         continue;
 217                 } else if ( arg[1] == 'f' ) {
 218                         nr = strtol(arg+2, NULL, 10);
 219                 } else if ( arg[1] == 'b' && !arg[2] ) {
 220                         allow_bare = 1;
 221                 } else if ( arg[1] == 'o' && arg[2] ) {
 222                         dir = arg+2;
 223                 } else if ( arg[1] == '-' && !arg[2] ) {
 224                         argp++; /* -- marks end of options */
 225                         break;
 226                 } else {
 227                         die("unknown option: %s", arg);
 228                 }
 229         }
 230
 231         if ( !dir ) {
 232                 /* Backwards compatibility: if no -o specified, accept
 233                    <mbox> <dir> or just <dir> */
 234                 switch (argc - (argp-argv)) {
 235                 case 1:
 236                         dir = argp[0];
 237                         argp = stdin_only;
 238                         break;
 239                 case 2:
 240                         stdin_only[0] = argp[0];
 241                         dir = argp[1];
 242                         argp = stdin_only;
 243                         break;
 244                 default:
 245                         usage(git_mailsplit_usage);
 246                 }
 247         } else {
 248                 /* New usage: if no more argument, parse stdin */
 249                 if ( !*argp )
 250                         argp = stdin_only;
 251         }
 252
 253         while (*argp) {
 254                 const char *arg = *argp++;
 255                 struct stat argstat;
 256                 int ret = 0;
 257
 258                 if (arg[0] == '-' && arg[1] == 0) {
 259                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 260                         if (ret < 0) {
 261                                 error("cannot split patches from stdin");
 262                                 return 1;
 263                         }
 264                         num += (ret - nr);
 265                         nr = ret;
 266                         continue;
 267                 }
 268
 269                 if (stat(arg, &argstat) == -1) {
 270                         error("cannot stat %s (%s)", arg, strerror(errno));
 271                         return 1;
 272                 }
 273
 274                 if (S_ISDIR(argstat.st_mode))
 275                         ret = split_maildir(arg, dir, nr_prec, nr);
 276                 else
 277                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 278
 279                 if (ret < 0) {
 280                         error("cannot split patches from %s", arg);
 281                         return 1;
 282                 }
 283                 num += (ret - nr);
 284                 nr = ret;
 285         }
 286
 287         printf("%d\n", num);
 288
 289         return 0;
 290 }