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;
  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                 fprintf(stderr, "corrupt mailbox\n");
  63                 exit(1);
  64         }
  65         fd = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  66         if (fd < 0)
  67                 die_errno("cannot open output file '%s'", name);
  68         output = xfdopen(fd, "w");
  69
  70         /* Copy it out, while searching for a line that begins with
  71          * "From " and having something that looks like a date format.
  72          */
  73         for (;;) {
  74                 if (!keep_cr && buf.len > 1 && buf.buf[buf.len-1] == '\n' &&
  75                         buf.buf[buf.len-2] == '\r') {
  76                         strbuf_setlen(&buf, buf.len-2);
  77                         strbuf_addch(&buf, '\n');
  78                 }
  79
  80                 if (fwrite(buf.buf, 1, buf.len, output) != buf.len)
  81                         die_errno("cannot write output");
  82
  83                 if (strbuf_getwholeline(&buf, mbox, '\n')) {
  84                         if (feof(mbox)) {
  85                                 status = 1;
  86                                 break;
  87                         }
  88                         die_errno("cannot read mbox");
  89                 }
  90                 if (!is_bare && is_from_line(buf.buf, buf.len))
  91                         break; /* done with one message */
  92         }
  93         fclose(output);
  94         return status;
  95 }
  96
  97 static int populate_maildir_list(struct string_list *list, const char *path)
  98 {
  99         DIR *dir;
 100         struct dirent *dent;
 101         char *name = NULL;
 102         char *subs[] = { "cur", "new", NULL };
 103         char **sub;
 104         int ret = -1;
 105
 106         for (sub = subs; *sub; ++sub) {
 107                 free(name);
 108                 name = xstrfmt("%s/%s", path, *sub);
 109                 if ((dir = opendir(name)) == NULL) {
 110                         if (errno == ENOENT)
 111                                 continue;
 112                         error_errno("cannot opendir %s", name);
 113                         goto out;
 114                 }
 115
 116                 while ((dent = readdir(dir)) != NULL) {
 117                         if (dent->d_name[0] == '.')
 118                                 continue;
 119                         free(name);
 120                         name = xstrfmt("%s/%s", *sub, dent->d_name);
 121                         string_list_insert(list, name);
 122                 }
 123
 124                 closedir(dir);
 125         }
 126
 127         ret = 0;
 128
 129 out:
 130         free(name);
 131         return ret;
 132 }
 133
 134 static int maildir_filename_cmp(const char *a, const char *b)
 135 {
 136         while (*a && *b) {
 137                 if (isdigit(*a) && isdigit(*b)) {
 138                         long int na, nb;
 139                         na = strtol(a, (char **)&a, 10);
 140                         nb = strtol(b, (char **)&b, 10);
 141                         if (na != nb)
 142                                 return na - nb;
 143                         /* strtol advanced our pointers */
 144                 }
 145                 else {
 146                         if (*a != *b)
 147                                 return (unsigned char)*a - (unsigned char)*b;
 148                         a++;
 149                         b++;
 150                 }
 151         }
 152         return (unsigned char)*a - (unsigned char)*b;
 153 }
 154
 155 static int split_maildir(const char *maildir, const char *dir,
 156         int nr_prec, int skip)
 157 {
 158         char *file = NULL;
 159         FILE *f = NULL;
 160         int ret = -1;
 161         int i;
 162         struct string_list list = STRING_LIST_INIT_DUP;
 163
 164         list.cmp = maildir_filename_cmp;
 165
 166         if (populate_maildir_list(&list, maildir) < 0)
 167                 goto out;
 168
 169         for (i = 0; i < list.nr; i++) {
 170                 char *name;
 171
 172                 free(file);
 173                 file = xstrfmt("%s/%s", maildir, list.items[i].string);
 174
 175                 f = fopen(file, "r");
 176                 if (!f) {
 177                         error_errno("cannot open mail %s", file);
 178                         goto out;
 179                 }
 180
 181                 if (strbuf_getwholeline(&buf, f, '\n')) {
 182                         error_errno("cannot read mail %s", file);
 183                         goto out;
 184                 }
 185
 186                 name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip);
 187                 split_one(f, name, 1);
 188                 free(name);
 189
 190                 fclose(f);
 191                 f = NULL;
 192         }
 193
 194         ret = skip;
 195 out:
 196         if (f)
 197                 fclose(f);
 198         free(file);
 199         string_list_clear(&list, 1);
 200         return ret;
 201 }
 202
 203 static int split_mbox(const char *file, const char *dir, int allow_bare,
 204                       int nr_prec, int skip)
 205 {
 206         int ret = -1;
 207         int peek;
 208
 209         FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 210         int file_done = 0;
 211
 212         if (!f) {
 213                 error_errno("cannot open mbox %s", file);
 214                 goto out;
 215         }
 216
 217         do {
 218                 peek = fgetc(f);
 219         } while (isspace(peek));
 220         ungetc(peek, f);
 221
 222         if (strbuf_getwholeline(&buf, f, '\n')) {
 223                 /* empty stdin is OK */
 224                 if (f != stdin) {
 225                         error("cannot read mbox %s", file);
 226                         goto out;
 227                 }
 228                 file_done = 1;
 229         }
 230
 231         while (!file_done) {
 232                 char *name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip);
 233                 file_done = split_one(f, name, allow_bare);
 234                 free(name);
 235         }
 236
 237         if (f != stdin)
 238                 fclose(f);
 239
 240         ret = skip;
 241 out:
 242         return ret;
 243 }
 244
 245 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 246 {
 247         int nr = 0, nr_prec = 4, num = 0;
 248         int allow_bare = 0;
 249         const char *dir = NULL;
 250         const char **argp;
 251         static const char *stdin_only[] = { "-", NULL };
 252
 253         for (argp = argv+1; *argp; argp++) {
 254                 const char *arg = *argp;
 255
 256                 if (arg[0] != '-')
 257                         break;
 258                 /* do flags here */
 259                 if ( arg[1] == 'd' ) {
 260                         nr_prec = strtol(arg+2, NULL, 10);
 261                         if (nr_prec < 3 || 10 <= nr_prec)
 262                                 usage(git_mailsplit_usage);
 263                         continue;
 264                 } else if ( arg[1] == 'f' ) {
 265                         nr = strtol(arg+2, NULL, 10);
 266                 } else if ( arg[1] == 'h' ) {
 267                         usage(git_mailsplit_usage);
 268                 } else if ( arg[1] == 'b' && !arg[2] ) {
 269                         allow_bare = 1;
 270                 } else if (!strcmp(arg, "--keep-cr")) {
 271                         keep_cr = 1;
 272                 } else if ( arg[1] == 'o' && arg[2] ) {
 273                         dir = arg+2;
 274                 } else if ( arg[1] == '-' && !arg[2] ) {
 275                         argp++; /* -- marks end of options */
 276                         break;
 277                 } else {
 278                         die("unknown option: %s", arg);
 279                 }
 280         }
 281
 282         if ( !dir ) {
 283                 /* Backwards compatibility: if no -o specified, accept
 284                    <mbox> <dir> or just <dir> */
 285                 switch (argc - (argp-argv)) {
 286                 case 1:
 287                         dir = argp[0];
 288                         argp = stdin_only;
 289                         break;
 290                 case 2:
 291                         stdin_only[0] = argp[0];
 292                         dir = argp[1];
 293                         argp = stdin_only;
 294                         break;
 295                 default:
 296                         usage(git_mailsplit_usage);
 297                 }
 298         } else {
 299                 /* New usage: if no more argument, parse stdin */
 300                 if ( !*argp )
 301                         argp = stdin_only;
 302         }
 303
 304         while (*argp) {
 305                 const char *arg = *argp++;
 306                 struct stat argstat;
 307                 int ret = 0;
 308
 309                 if (arg[0] == '-' && arg[1] == 0) {
 310                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 311                         if (ret < 0) {
 312                                 error("cannot split patches from stdin");
 313                                 return 1;
 314                         }
 315                         num += (ret - nr);
 316                         nr = ret;
 317                         continue;
 318                 }
 319
 320                 if (stat(arg, &argstat) == -1) {
 321                         error_errno("cannot stat %s", arg);
 322                         return 1;
 323                 }
 324
 325                 if (S_ISDIR(argstat.st_mode))
 326                         ret = split_maildir(arg, dir, nr_prec, nr);
 327                 else
 328                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 329
 330                 if (ret < 0) {
 331                         error("cannot split patches from %s", arg);
 332                         return 1;
 333                 }
 334                 num += (ret - nr);
 335                 nr = ret;
 336         }
 337
 338         printf("%d\n", num);
 339
 340         return 0;
 341 }