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 = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  79         if (fd < 0)
  80                 die_errno("cannot open output file '%s'", name);
  81         output = xfdopen(fd, "w");
  82
  83         /* Copy it out, while searching for a line that begins with
  84          * "From " and having something that looks like a date format.
  85          */
  86         for (;;) {
  87                 if (!keep_cr && buf.len > 1 && buf.buf[buf.len-1] == '\n' &&
  88                         buf.buf[buf.len-2] == '\r') {
  89                         strbuf_setlen(&buf, buf.len-2);
  90                         strbuf_addch(&buf, '\n');
  91                 }
  92
  93                 if (mboxrd && is_gtfrom(&buf))
  94                         strbuf_remove(&buf, 0, 1);
  95
  96                 if (fwrite(buf.buf, 1, buf.len, output) != buf.len)
  97                         die_errno("cannot write output");
  98
  99                 if (strbuf_getwholeline(&buf, mbox, '\n')) {
 100                         if (feof(mbox)) {
 101                                 status = 1;
 102                                 break;
 103                         }
 104                         die_errno("cannot read mbox");
 105                 }
 106                 if (!is_bare && is_from_line(buf.buf, buf.len))
 107                         break; /* done with one message */
 108         }
 109         fclose(output);
 110         return status;
 111 }
 112
 113 static int populate_maildir_list(struct string_list *list, const char *path)
 114 {
 115         DIR *dir;
 116         struct dirent *dent;
 117         char *name = NULL;
 118         char *subs[] = { "cur", "new", NULL };
 119         char **sub;
 120         int ret = -1;
 121
 122         for (sub = subs; *sub; ++sub) {
 123                 free(name);
 124                 name = xstrfmt("%s/%s", path, *sub);
 125                 if ((dir = opendir(name)) == NULL) {
 126                         if (errno == ENOENT)
 127                                 continue;
 128                         error_errno("cannot opendir %s", name);
 129                         goto out;
 130                 }
 131
 132                 while ((dent = readdir(dir)) != NULL) {
 133                         if (dent->d_name[0] == '.')
 134                                 continue;
 135                         free(name);
 136                         name = xstrfmt("%s/%s", *sub, dent->d_name);
 137                         string_list_insert(list, name);
 138                 }
 139
 140                 closedir(dir);
 141         }
 142
 143         ret = 0;
 144
 145 out:
 146         free(name);
 147         return ret;
 148 }
 149
 150 static int maildir_filename_cmp(const char *a, const char *b)
 151 {
 152         while (*a && *b) {
 153                 if (isdigit(*a) && isdigit(*b)) {
 154                         long int na, nb;
 155                         na = strtol(a, (char **)&a, 10);
 156                         nb = strtol(b, (char **)&b, 10);
 157                         if (na != nb)
 158                                 return na - nb;
 159                         /* strtol advanced our pointers */
 160                 }
 161                 else {
 162                         if (*a != *b)
 163                                 return (unsigned char)*a - (unsigned char)*b;
 164                         a++;
 165                         b++;
 166                 }
 167         }
 168         return (unsigned char)*a - (unsigned char)*b;
 169 }
 170
 171 static int split_maildir(const char *maildir, const char *dir,
 172         int nr_prec, int skip)
 173 {
 174         char *file = NULL;
 175         FILE *f = NULL;
 176         int ret = -1;
 177         int i;
 178         struct string_list list = STRING_LIST_INIT_DUP;
 179
 180         list.cmp = maildir_filename_cmp;
 181
 182         if (populate_maildir_list(&list, maildir) < 0)
 183                 goto out;
 184
 185         for (i = 0; i < list.nr; i++) {
 186                 char *name;
 187
 188                 free(file);
 189                 file = xstrfmt("%s/%s", maildir, list.items[i].string);
 190
 191                 f = fopen(file, "r");
 192                 if (!f) {
 193                         error_errno("cannot open mail %s", file);
 194                         goto out;
 195                 }
 196
 197                 if (strbuf_getwholeline(&buf, f, '\n')) {
 198                         error_errno("cannot read mail %s", file);
 199                         goto out;
 200                 }
 201
 202                 name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip);
 203                 split_one(f, name, 1);
 204                 free(name);
 205
 206                 fclose(f);
 207                 f = NULL;
 208         }
 209
 210         ret = skip;
 211 out:
 212         if (f)
 213                 fclose(f);
 214         free(file);
 215         string_list_clear(&list, 1);
 216         return ret;
 217 }
 218
 219 static int split_mbox(const char *file, const char *dir, int allow_bare,
 220                       int nr_prec, int skip)
 221 {
 222         int ret = -1;
 223         int peek;
 224
 225         FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 226         int file_done = 0;
 227
 228         if (!f) {
 229                 error_errno("cannot open mbox %s", file);
 230                 goto out;
 231         }
 232
 233         do {
 234                 peek = fgetc(f);
 235                 if (peek == EOF) {
 236                         if (f == stdin)
 237                                 /* empty stdin is OK */
 238                                 ret = skip;
 239                         else {
 240                                 fclose(f);
 241                                 error(_("empty mbox: '%s'"), file);
 242                         }
 243                         goto out;
 244                 }
 245         } while (isspace(peek));
 246         ungetc(peek, f);
 247
 248         if (strbuf_getwholeline(&buf, f, '\n')) {
 249                 /* empty stdin is OK */
 250                 if (f != stdin) {
 251                         error("cannot read mbox %s", file);
 252                         goto out;
 253                 }
 254                 file_done = 1;
 255         }
 256
 257         while (!file_done) {
 258                 char *name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip);
 259                 file_done = split_one(f, name, allow_bare);
 260                 free(name);
 261         }
 262
 263         if (f != stdin)
 264                 fclose(f);
 265
 266         ret = skip;
 267 out:
 268         return ret;
 269 }
 270
 271 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 272 {
 273         int nr = 0, nr_prec = 4, num = 0;
 274         int allow_bare = 0;
 275         const char *dir = NULL;
 276         const char **argp;
 277         static const char *stdin_only[] = { "-", NULL };
 278
 279         for (argp = argv+1; *argp; argp++) {
 280                 const char *arg = *argp;
 281
 282                 if (arg[0] != '-')
 283                         break;
 284                 /* do flags here */
 285                 if ( arg[1] == 'd' ) {
 286                         nr_prec = strtol(arg+2, NULL, 10);
 287                         if (nr_prec < 3 || 10 <= nr_prec)
 288                                 usage(git_mailsplit_usage);
 289                         continue;
 290                 } else if ( arg[1] == 'f' ) {
 291                         nr = strtol(arg+2, NULL, 10);
 292                 } else if ( arg[1] == 'h' ) {
 293                         usage(git_mailsplit_usage);
 294                 } else if ( arg[1] == 'b' && !arg[2] ) {
 295                         allow_bare = 1;
 296                 } else if (!strcmp(arg, "--keep-cr")) {
 297                         keep_cr = 1;
 298                 } else if ( arg[1] == 'o' && arg[2] ) {
 299                         dir = arg+2;
 300                 } else if (!strcmp(arg, "--mboxrd")) {
 301                         mboxrd = 1;
 302                 } else if ( arg[1] == '-' && !arg[2] ) {
 303                         argp++; /* -- marks end of options */
 304                         break;
 305                 } else {
 306                         die("unknown option: %s", arg);
 307                 }
 308         }
 309
 310         if ( !dir ) {
 311                 /* Backwards compatibility: if no -o specified, accept
 312                    <mbox> <dir> or just <dir> */
 313                 switch (argc - (argp-argv)) {
 314                 case 1:
 315                         dir = argp[0];
 316                         argp = stdin_only;
 317                         break;
 318                 case 2:
 319                         stdin_only[0] = argp[0];
 320                         dir = argp[1];
 321                         argp = stdin_only;
 322                         break;
 323                 default:
 324                         usage(git_mailsplit_usage);
 325                 }
 326         } else {
 327                 /* New usage: if no more argument, parse stdin */
 328                 if ( !*argp )
 329                         argp = stdin_only;
 330         }
 331
 332         while (*argp) {
 333                 const char *arg = *argp++;
 334                 struct stat argstat;
 335                 int ret = 0;
 336
 337                 if (arg[0] == '-' && arg[1] == 0) {
 338                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 339                         if (ret < 0) {
 340                                 error("cannot split patches from stdin");
 341                                 return 1;
 342                         }
 343                         num += (ret - nr);
 344                         nr = ret;
 345                         continue;
 346                 }
 347
 348                 if (stat(arg, &argstat) == -1) {
 349                         error_errno("cannot stat %s", arg);
 350                         return 1;
 351                 }
 352
 353                 if (S_ISDIR(argstat.st_mode))
 354                         ret = split_maildir(arg, dir, nr_prec, nr);
 355                 else
 356                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 357
 358                 if (ret < 0) {
 359                         error("cannot split patches from %s", arg);
 360                         return 1;
 361                 }
 362                 num += (ret - nr);
 363                 nr = ret;
 364         }
 365
 366         printf("%d\n", num);
 367
 368         return 0;
 369 }