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 static int mboxrd;
  49
  50 static int is_gtfrom(const struct strbuf *buf)
  51 {
  52         size_t min = strlen(">From ");
  53         size_t ngt;
  54
  55         if (buf->len < min)
  56                 return 0;
  57
  58         ngt = strspn(buf->buf, ">");
  59         return ngt && starts_with(buf->buf + ngt, "From ");
  60 }
  61
  62 /* Called with the first line (potentially partial)
  63  * already in buf[] -- normally that should begin with
  64  * the Unix "From " line.  Write it into the specified
  65  * file.
  66  */
  67 static int split_one(FILE *mbox, const char *name, int allow_bare)
  68 {
  69         FILE *output;
  70         int fd;
  71         int status = 0;
  72         int is_bare = !is_from_line(buf.buf, buf.len);
  73
  74         if (is_bare && !allow_bare) {
  75                 fprintf(stderr, "corrupt mailbox\n");
  76                 exit(1);
  77         }
  78         fd = xopen(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  79         output = xfdopen(fd, "w");
  80
  81         /* Copy it out, while searching for a line that begins with
  82          * "From " and having something that looks like a date format.
  83          */
  84         for (;;) {
  85                 if (!keep_cr && buf.len > 1 && buf.buf[buf.len-1] == '\n' &&
  86                         buf.buf[buf.len-2] == '\r') {
  87                         strbuf_setlen(&buf, buf.len-2);
  88                         strbuf_addch(&buf, '\n');
  89                 }
  90
  91                 if (mboxrd && is_gtfrom(&buf))
  92                         strbuf_remove(&buf, 0, 1);
  93
  94                 if (fwrite(buf.buf, 1, buf.len, output) != buf.len)
  95                         die_errno("cannot write output");
  96
  97                 if (strbuf_getwholeline(&buf, mbox, '\n')) {
  98                         if (feof(mbox)) {
  99                                 status = 1;
 100                                 break;
 101                         }
 102                         die_errno("cannot read mbox");
 103                 }
 104                 if (!is_bare && is_from_line(buf.buf, buf.len))
 105                         break; /* done with one message */
 106         }
 107         fclose(output);
 108         return status;
 109 }
 110
 111 static int populate_maildir_list(struct string_list *list, const char *path)
 112 {
 113         DIR *dir;
 114         struct dirent *dent;
 115         char *name = NULL;
 116         char *subs[] = { "cur", "new", NULL };
 117         char **sub;
 118         int ret = -1;
 119
 120         for (sub = subs; *sub; ++sub) {
 121                 free(name);
 122                 name = xstrfmt("%s/%s", path, *sub);
 123                 if ((dir = opendir(name)) == NULL) {
 124                         if (errno == ENOENT)
 125                                 continue;
 126                         error_errno("cannot opendir %s", name);
 127                         goto out;
 128                 }
 129
 130                 while ((dent = readdir(dir)) != NULL) {
 131                         if (dent->d_name[0] == '.')
 132                                 continue;
 133                         free(name);
 134                         name = xstrfmt("%s/%s", *sub, dent->d_name);
 135                         string_list_insert(list, name);
 136                 }
 137
 138                 closedir(dir);
 139         }
 140
 141         ret = 0;
 142
 143 out:
 144         free(name);
 145         return ret;
 146 }
 147
 148 static int maildir_filename_cmp(const char *a, const char *b)
 149 {
 150         while (*a && *b) {
 151                 if (isdigit(*a) && isdigit(*b)) {
 152                         long int na, nb;
 153                         na = strtol(a, (char **)&a, 10);
 154                         nb = strtol(b, (char **)&b, 10);
 155                         if (na != nb)
 156                                 return na - nb;
 157                         /* strtol advanced our pointers */
 158                 }
 159                 else {
 160                         if (*a != *b)
 161                                 return (unsigned char)*a - (unsigned char)*b;
 162                         a++;
 163                         b++;
 164                 }
 165         }
 166         return (unsigned char)*a - (unsigned char)*b;
 167 }
 168
 169 static int split_maildir(const char *maildir, const char *dir,
 170         int nr_prec, int skip)
 171 {
 172         char *file = NULL;
 173         FILE *f = NULL;
 174         int ret = -1;
 175         int i;
 176         struct string_list list = STRING_LIST_INIT_DUP;
 177
 178         list.cmp = maildir_filename_cmp;
 179
 180         if (populate_maildir_list(&list, maildir) < 0)
 181                 goto out;
 182
 183         for (i = 0; i < list.nr; i++) {
 184                 char *name;
 185
 186                 free(file);
 187                 file = xstrfmt("%s/%s", maildir, list.items[i].string);
 188
 189                 f = fopen(file, "r");
 190                 if (!f) {
 191                         error_errno("cannot open mail %s", file);
 192                         goto out;
 193                 }
 194
 195                 if (strbuf_getwholeline(&buf, f, '\n')) {
 196                         error_errno("cannot read mail %s", file);
 197                         goto out;
 198                 }
 199
 200                 name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip);
 201                 split_one(f, name, 1);
 202                 free(name);
 203
 204                 fclose(f);
 205                 f = NULL;
 206         }
 207
 208         ret = skip;
 209 out:
 210         if (f)
 211                 fclose(f);
 212         free(file);
 213         string_list_clear(&list, 1);
 214         return ret;
 215 }
 216
 217 static int split_mbox(const char *file, const char *dir, int allow_bare,
 218                       int nr_prec, int skip)
 219 {
 220         int ret = -1;
 221         int peek;
 222
 223         FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 224         int file_done = 0;
 225
 226         if (!f) {
 227                 error_errno("cannot open mbox %s", file);
 228                 goto out;
 229         }
 230
 231         do {
 232                 peek = fgetc(f);
 233                 if (peek == EOF) {
 234                         if (f == stdin)
 235                                 /* empty stdin is OK */
 236                                 ret = skip;
 237                         else {
 238                                 fclose(f);
 239                                 error(_("empty mbox: '%s'"), file);
 240                         }
 241                         goto out;
 242                 }
 243         } while (isspace(peek));
 244         ungetc(peek, f);
 245
 246         if (strbuf_getwholeline(&buf, f, '\n')) {
 247                 /* empty stdin is OK */
 248                 if (f != stdin) {
 249                         error("cannot read mbox %s", file);
 250                         goto out;
 251                 }
 252                 file_done = 1;
 253         }
 254
 255         while (!file_done) {
 256                 char *name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip);
 257                 file_done = split_one(f, name, allow_bare);
 258                 free(name);
 259         }
 260
 261         if (f != stdin)
 262                 fclose(f);
 263
 264         ret = skip;
 265 out:
 266         return ret;
 267 }
 268
 269 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 270 {
 271         int nr = 0, nr_prec = 4, num = 0;
 272         int allow_bare = 0;
 273         const char *dir = NULL;
 274         const char **argp;
 275         static const char *stdin_only[] = { "-", NULL };
 276
 277         for (argp = argv+1; *argp; argp++) {
 278                 const char *arg = *argp;
 279
 280                 if (arg[0] != '-')
 281                         break;
 282                 /* do flags here */
 283                 if ( arg[1] == 'd' ) {
 284                         nr_prec = strtol(arg+2, NULL, 10);
 285                         if (nr_prec < 3 || 10 <= nr_prec)
 286                                 usage(git_mailsplit_usage);
 287                         continue;
 288                 } else if ( arg[1] == 'f' ) {
 289                         nr = strtol(arg+2, NULL, 10);
 290                 } else if ( arg[1] == 'h' ) {
 291                         usage(git_mailsplit_usage);
 292                 } else if ( arg[1] == 'b' && !arg[2] ) {
 293                         allow_bare = 1;
 294                 } else if (!strcmp(arg, "--keep-cr")) {
 295                         keep_cr = 1;
 296                 } else if ( arg[1] == 'o' && arg[2] ) {
 297                         dir = arg+2;
 298                 } else if (!strcmp(arg, "--mboxrd")) {
 299                         mboxrd = 1;
 300                 } else if ( arg[1] == '-' && !arg[2] ) {
 301                         argp++; /* -- marks end of options */
 302                         break;
 303                 } else {
 304                         die("unknown option: %s", arg);
 305                 }
 306         }
 307
 308         if ( !dir ) {
 309                 /* Backwards compatibility: if no -o specified, accept
 310                    <mbox> <dir> or just <dir> */
 311                 switch (argc - (argp-argv)) {
 312                 case 1:
 313                         dir = argp[0];
 314                         argp = stdin_only;
 315                         break;
 316                 case 2:
 317                         stdin_only[0] = argp[0];
 318                         dir = argp[1];
 319                         argp = stdin_only;
 320                         break;
 321                 default:
 322                         usage(git_mailsplit_usage);
 323                 }
 324         } else {
 325                 /* New usage: if no more argument, parse stdin */
 326                 if ( !*argp )
 327                         argp = stdin_only;
 328         }
 329
 330         while (*argp) {
 331                 const char *arg = *argp++;
 332                 struct stat argstat;
 333                 int ret = 0;
 334
 335                 if (arg[0] == '-' && arg[1] == 0) {
 336                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 337                         if (ret < 0) {
 338                                 error("cannot split patches from stdin");
 339                                 return 1;
 340                         }
 341                         num += (ret - nr);
 342                         nr = ret;
 343                         continue;
 344                 }
 345
 346                 if (stat(arg, &argstat) == -1) {
 347                         error_errno("cannot stat %s", arg);
 348                         return 1;
 349                 }
 350
 351                 if (S_ISDIR(argstat.st_mode))
 352                         ret = split_maildir(arg, dir, nr_prec, nr);
 353                 else
 354                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 355
 356                 if (ret < 0) {
 357                         error("cannot split patches from %s", arg);
 358                         return 1;
 359                 }
 360                 num += (ret - nr);
 361                 nr = ret;
 362         }
 363
 364         printf("%d\n", num);
 365
 366         return 0;
 367 }