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 "string-list.h"
  10 #include "strbuf.h"
  11
  12 static const char git_mailsplit_usage[] =
  13 "git mailsplit [-d<prec>] [-f<n>] [-b] [--keep-cr] -o<directory> [(<mbox>|<Maildir>)...]";
  14
  15 static int is_from_line(const char *line, int len)
  16 {
  17         const char *colon;
  18
  19         if (len < 20 || memcmp("From ", line, 5))
  20                 return 0;
  21
  22         colon = line + len - 2;
  23         line += 5;
  24         for (;;) {
  25                 if (colon < line)
  26                         return 0;
  27                 if (*--colon == ':')
  28                         break;
  29         }
  30
  31         if (!isdigit(colon[-4]) ||
  32             !isdigit(colon[-2]) ||
  33             !isdigit(colon[-1]) ||
  34             !isdigit(colon[ 1]) ||
  35             !isdigit(colon[ 2]))
  36                 return 0;
  37
  38         /* year */
  39         if (strtol(colon+3, NULL, 10) <= 90)
  40                 return 0;
  41
  42         /* Ok, close enough */
  43         return 1;
  44 }
  45
  46 static struct strbuf buf = STRBUF_INIT;
  47 static int keep_cr;
  48
  49 /* Called with the first line (potentially partial)
  50  * already in buf[] -- normally that should begin with
  51  * the Unix "From " line.  Write it into the specified
  52  * file.
  53  */
  54 static int split_one(FILE *mbox, const char *name, int allow_bare)
  55 {
  56         FILE *output = NULL;
  57         int fd;
  58         int status = 0;
  59         int is_bare = !is_from_line(buf.buf, 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_errno("cannot open output file '%s'", name);
  67         output = xfdopen(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                 if (!keep_cr && buf.len > 1 && buf.buf[buf.len-1] == '\n' &&
  74                         buf.buf[buf.len-2] == '\r') {
  75                         strbuf_setlen(&buf, buf.len-2);
  76                         strbuf_addch(&buf, '\n');
  77                 }
  78
  79                 if (fwrite(buf.buf, 1, buf.len, output) != buf.len)
  80                         die_errno("cannot write output");
  81
  82                 if (strbuf_getwholeline(&buf, mbox, '\n')) {
  83                         if (feof(mbox)) {
  84                                 status = 1;
  85                                 break;
  86                         }
  87                         die_errno("cannot read mbox");
  88                 }
  89                 if (!is_bare && is_from_line(buf.buf, buf.len))
  90                         break; /* done with one message */
  91         }
  92         fclose(output);
  93         return status;
  94
  95  corrupt:
  96         if (output)
  97                 fclose(output);
  98         unlink(name);
  99         fprintf(stderr, "corrupt mailbox\n");
 100         exit(1);
 101 }
 102
 103 static int populate_maildir_list(struct string_list *list, const char *path)
 104 {
 105         DIR *dir;
 106         struct dirent *dent;
 107         char name[PATH_MAX];
 108         char *subs[] = { "cur", "new", NULL };
 109         char **sub;
 110
 111         for (sub = subs; *sub; ++sub) {
 112                 snprintf(name, sizeof(name), "%s/%s", path, *sub);
 113                 if ((dir = opendir(name)) == NULL) {
 114                         if (errno == ENOENT)
 115                                 continue;
 116                         error("cannot opendir %s (%s)", name, strerror(errno));
 117                         return -1;
 118                 }
 119
 120                 while ((dent = readdir(dir)) != NULL) {
 121                         if (dent->d_name[0] == '.')
 122                                 continue;
 123                         snprintf(name, sizeof(name), "%s/%s", *sub, dent->d_name);
 124                         string_list_insert(list, name);
 125                 }
 126
 127                 closedir(dir);
 128         }
 129
 130         return 0;
 131 }
 132
 133 static int maildir_filename_cmp(const char *a, const char *b)
 134 {
 135         while (*a && *b) {
 136                 if (isdigit(*a) && isdigit(*b)) {
 137                         long int na, nb;
 138                         na = strtol(a, (char **)&a, 10);
 139                         nb = strtol(b, (char **)&b, 10);
 140                         if (na != nb)
 141                                 return na - nb;
 142                         /* strtol advanced our pointers */
 143                 }
 144                 else {
 145                         if (*a != *b)
 146                                 return (unsigned char)*a - (unsigned char)*b;
 147                         a++;
 148                         b++;
 149                 }
 150         }
 151         return (unsigned char)*a - (unsigned char)*b;
 152 }
 153
 154 static int split_maildir(const char *maildir, const char *dir,
 155         int nr_prec, int skip)
 156 {
 157         char file[PATH_MAX];
 158         char name[PATH_MAX];
 159         int ret = -1;
 160         int i;
 161         struct string_list list = STRING_LIST_INIT_DUP;
 162
 163         list.cmp = maildir_filename_cmp;
 164
 165         if (populate_maildir_list(&list, maildir) < 0)
 166                 goto out;
 167
 168         for (i = 0; i < list.nr; i++) {
 169                 FILE *f;
 170                 snprintf(file, sizeof(file), "%s/%s", maildir, list.items[i].string);
 171                 f = fopen(file, "r");
 172                 if (!f) {
 173                         error("cannot open mail %s (%s)", file, strerror(errno));
 174                         goto out;
 175                 }
 176
 177                 if (strbuf_getwholeline(&buf, f, '\n')) {
 178                         error("cannot read mail %s (%s)", file, strerror(errno));
 179                         goto out;
 180                 }
 181
 182                 sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 183                 split_one(f, name, 1);
 184
 185                 fclose(f);
 186         }
 187
 188         ret = skip;
 189 out:
 190         string_list_clear(&list, 1);
 191         return ret;
 192 }
 193
 194 static int split_mbox(const char *file, const char *dir, int allow_bare,
 195                       int nr_prec, int skip)
 196 {
 197         char name[PATH_MAX];
 198         int ret = -1;
 199         int peek;
 200
 201         FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 202         int file_done = 0;
 203
 204         if (!f) {
 205                 error("cannot open mbox %s", file);
 206                 goto out;
 207         }
 208
 209         do {
 210                 peek = fgetc(f);
 211         } while (isspace(peek));
 212         ungetc(peek, f);
 213
 214         if (strbuf_getwholeline(&buf, f, '\n')) {
 215                 /* empty stdin is OK */
 216                 if (f != stdin) {
 217                         error("cannot read mbox %s", file);
 218                         goto out;
 219                 }
 220                 file_done = 1;
 221         }
 222
 223         while (!file_done) {
 224                 sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 225                 file_done = split_one(f, name, allow_bare);
 226         }
 227
 228         if (f != stdin)
 229                 fclose(f);
 230
 231         ret = skip;
 232 out:
 233         return ret;
 234 }
 235
 236 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 237 {
 238         int nr = 0, nr_prec = 4, num = 0;
 239         int allow_bare = 0;
 240         const char *dir = NULL;
 241         const char **argp;
 242         static const char *stdin_only[] = { "-", NULL };
 243
 244         for (argp = argv+1; *argp; argp++) {
 245                 const char *arg = *argp;
 246
 247                 if (arg[0] != '-')
 248                         break;
 249                 /* do flags here */
 250                 if ( arg[1] == 'd' ) {
 251                         nr_prec = strtol(arg+2, NULL, 10);
 252                         if (nr_prec < 3 || 10 <= nr_prec)
 253                                 usage(git_mailsplit_usage);
 254                         continue;
 255                 } else if ( arg[1] == 'f' ) {
 256                         nr = strtol(arg+2, NULL, 10);
 257                 } else if ( arg[1] == 'h' ) {
 258                         usage(git_mailsplit_usage);
 259                 } else if ( arg[1] == 'b' && !arg[2] ) {
 260                         allow_bare = 1;
 261                 } else if (!strcmp(arg, "--keep-cr")) {
 262                         keep_cr = 1;
 263                 } else if ( arg[1] == 'o' && arg[2] ) {
 264                         dir = arg+2;
 265                 } else if ( arg[1] == '-' && !arg[2] ) {
 266                         argp++; /* -- marks end of options */
 267                         break;
 268                 } else {
 269                         die("unknown option: %s", arg);
 270                 }
 271         }
 272
 273         if ( !dir ) {
 274                 /* Backwards compatibility: if no -o specified, accept
 275                    <mbox> <dir> or just <dir> */
 276                 switch (argc - (argp-argv)) {
 277                 case 1:
 278                         dir = argp[0];
 279                         argp = stdin_only;
 280                         break;
 281                 case 2:
 282                         stdin_only[0] = argp[0];
 283                         dir = argp[1];
 284                         argp = stdin_only;
 285                         break;
 286                 default:
 287                         usage(git_mailsplit_usage);
 288                 }
 289         } else {
 290                 /* New usage: if no more argument, parse stdin */
 291                 if ( !*argp )
 292                         argp = stdin_only;
 293         }
 294
 295         while (*argp) {
 296                 const char *arg = *argp++;
 297                 struct stat argstat;
 298                 int ret = 0;
 299
 300                 if (arg[0] == '-' && arg[1] == 0) {
 301                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 302                         if (ret < 0) {
 303                                 error("cannot split patches from stdin");
 304                                 return 1;
 305                         }
 306                         num += (ret - nr);
 307                         nr = ret;
 308                         continue;
 309                 }
 310
 311                 if (stat(arg, &argstat) == -1) {
 312                         error("cannot stat %s (%s)", arg, strerror(errno));
 313                         return 1;
 314                 }
 315
 316                 if (S_ISDIR(argstat.st_mode))
 317                         ret = split_maildir(arg, dir, nr_prec, nr);
 318                 else
 319                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 320
 321                 if (ret < 0) {
 322                         error("cannot split patches from %s", arg);
 323                         return 1;
 324                 }
 325                 num += (ret - nr);
 326                 nr = ret;
 327         }
 328
 329         printf("%d\n", num);
 330
 331         return 0;
 332 }