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 "gettext.h"
  10 #include "string-list.h"
  11 #include "strbuf.h"
  12
  13 static const char git_mailsplit_usage[] =
  14 "git mailsplit [-d<prec>] [-f<n>] [-b] [--keep-cr] -o<directory> [(<mbox>|<Maildir>)...]";
  15
  16 static int is_from_line(const char *line, int len)
  17 {
  18         const char *colon;
  19
  20         if (len < 20 || memcmp("From ", line, 5))
  21                 return 0;
  22
  23         colon = line + len - 2;
  24         line += 5;
  25         for (;;) {
  26                 if (colon < line)
  27                         return 0;
  28                 if (*--colon == ':')
  29                         break;
  30         }
  31
  32         if (!isdigit(colon[-4]) ||
  33             !isdigit(colon[-2]) ||
  34             !isdigit(colon[-1]) ||
  35             !isdigit(colon[ 1]) ||
  36             !isdigit(colon[ 2]))
  37                 return 0;
  38
  39         /* year */
  40         if (strtol(colon+3, NULL, 10) <= 90)
  41                 return 0;
  42
  43         /* Ok, close enough */
  44         return 1;
  45 }
  46
  47 static struct strbuf buf = STRBUF_INIT;
  48 static int keep_cr;
  49 static int mboxrd;
  50
  51 static int is_gtfrom(const struct strbuf *buf)
  52 {
  53         size_t min = strlen(">From ");
  54         size_t ngt;
  55
  56         if (buf->len < min)
  57                 return 0;
  58
  59         ngt = strspn(buf->buf, ">");
  60         return ngt && starts_with(buf->buf + ngt, "From ");
  61 }
  62
  63 /* Called with the first line (potentially partial)
  64  * already in buf[] -- normally that should begin with
  65  * the Unix "From " line.  Write it into the specified
  66  * file.
  67  */
  68 static int split_one(FILE *mbox, const char *name, int allow_bare)
  69 {
  70         FILE *output;
  71         int fd;
  72         int status = 0;
  73         int is_bare = !is_from_line(buf.buf, buf.len);
  74
  75         if (is_bare && !allow_bare) {
  76                 fprintf(stderr, "corrupt mailbox\n");
  77                 exit(1);
  78         }
  79         fd = xopen(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  80         output = xfdopen(fd, "w");
  81
  82         /* Copy it out, while searching for a line that begins with
  83          * "From " and having something that looks like a date format.
  84          */
  85         for (;;) {
  86                 if (!keep_cr && buf.len > 1 && buf.buf[buf.len-1] == '\n' &&
  87                         buf.buf[buf.len-2] == '\r') {
  88                         strbuf_setlen(&buf, buf.len-2);
  89                         strbuf_addch(&buf, '\n');
  90                 }
  91
  92                 if (mboxrd && is_gtfrom(&buf))
  93                         strbuf_remove(&buf, 0, 1);
  94
  95                 if (fwrite(buf.buf, 1, buf.len, output) != buf.len)
  96                         die_errno("cannot write output");
  97
  98                 if (strbuf_getwholeline(&buf, mbox, '\n')) {
  99                         if (feof(mbox)) {
 100                                 status = 1;
 101                                 break;
 102                         }
 103                         die_errno("cannot read mbox");
 104                 }
 105                 if (!is_bare && is_from_line(buf.buf, buf.len))
 106                         break; /* done with one message */
 107         }
 108         fclose(output);
 109         return status;
 110 }
 111
 112 static int populate_maildir_list(struct string_list *list, const char *path)
 113 {
 114         DIR *dir;
 115         struct dirent *dent;
 116         char *name = NULL;
 117         char *subs[] = { "cur", "new", NULL };
 118         char **sub;
 119         int ret = -1;
 120
 121         for (sub = subs; *sub; ++sub) {
 122                 free(name);
 123                 name = xstrfmt("%s/%s", path, *sub);
 124                 if (!(dir = opendir(name))) {
 125                         if (errno == ENOENT)
 126                                 continue;
 127                         error_errno("cannot opendir %s", name);
 128                         goto out;
 129                 }
 130
 131                 while ((dent = readdir(dir)) != NULL) {
 132                         if (dent->d_name[0] == '.')
 133                                 continue;
 134                         free(name);
 135                         name = xstrfmt("%s/%s", *sub, dent->d_name);
 136                         string_list_insert(list, name);
 137                 }
 138
 139                 closedir(dir);
 140         }
 141
 142         ret = 0;
 143
 144 out:
 145         free(name);
 146         return ret;
 147 }
 148
 149 static int maildir_filename_cmp(const char *a, const char *b)
 150 {
 151         while (*a && *b) {
 152                 if (isdigit(*a) && isdigit(*b)) {
 153                         long int na, nb;
 154                         na = strtol(a, (char **)&a, 10);
 155                         nb = strtol(b, (char **)&b, 10);
 156                         if (na != nb)
 157                                 return na - nb;
 158                         /* strtol advanced our pointers */
 159                 }
 160                 else {
 161                         if (*a != *b)
 162                                 return (unsigned char)*a - (unsigned char)*b;
 163                         a++;
 164                         b++;
 165                 }
 166         }
 167         return (unsigned char)*a - (unsigned char)*b;
 168 }
 169
 170 static int split_maildir(const char *maildir, const char *dir,
 171         int nr_prec, int skip)
 172 {
 173         char *file = NULL;
 174         FILE *f = NULL;
 175         int ret = -1;
 176         int i;
 177         struct string_list list = STRING_LIST_INIT_DUP;
 178
 179         list.cmp = maildir_filename_cmp;
 180
 181         if (populate_maildir_list(&list, maildir) < 0)
 182                 goto out;
 183
 184         for (i = 0; i < list.nr; i++) {
 185                 char *name;
 186
 187                 free(file);
 188                 file = xstrfmt("%s/%s", maildir, list.items[i].string);
 189
 190                 f = fopen(file, "r");
 191                 if (!f) {
 192                         error_errno("cannot open mail %s", file);
 193                         goto out;
 194                 }
 195
 196                 if (strbuf_getwholeline(&buf, f, '\n')) {
 197                         error_errno("cannot read mail %s", file);
 198                         goto out;
 199                 }
 200
 201                 name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip);
 202                 split_one(f, name, 1);
 203                 free(name);
 204
 205                 fclose(f);
 206                 f = NULL;
 207         }
 208
 209         ret = skip;
 210 out:
 211         if (f)
 212                 fclose(f);
 213         free(file);
 214         string_list_clear(&list, 1);
 215         return ret;
 216 }
 217
 218 static int split_mbox(const char *file, const char *dir, int allow_bare,
 219                       int nr_prec, int skip)
 220 {
 221         int ret = -1;
 222         int peek;
 223
 224         FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 225         int file_done = 0;
 226
 227         if (isatty(fileno(f)))
 228                 warning(_("reading patches from stdin/tty..."));
 229
 230         if (!f) {
 231                 error_errno("cannot open mbox %s", file);
 232                 goto out;
 233         }
 234
 235         do {
 236                 peek = fgetc(f);
 237                 if (peek == EOF) {
 238                         if (f == stdin)
 239                                 /* empty stdin is OK */
 240                                 ret = skip;
 241                         else {
 242                                 fclose(f);
 243                                 error(_("empty mbox: '%s'"), file);
 244                         }
 245                         goto out;
 246                 }
 247         } while (isspace(peek));
 248         ungetc(peek, f);
 249
 250         if (strbuf_getwholeline(&buf, f, '\n')) {
 251                 /* empty stdin is OK */
 252                 if (f != stdin) {
 253                         error("cannot read mbox %s", file);
 254                         goto out;
 255                 }
 256                 file_done = 1;
 257         }
 258
 259         while (!file_done) {
 260                 char *name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip);
 261                 file_done = split_one(f, name, allow_bare);
 262                 free(name);
 263         }
 264
 265         if (f != stdin)
 266                 fclose(f);
 267
 268         ret = skip;
 269 out:
 270         return ret;
 271 }
 272
 273 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 274 {
 275         int nr = 0, nr_prec = 4, num = 0;
 276         int allow_bare = 0;
 277         const char *dir = NULL;
 278         const char **argp;
 279         static const char *stdin_only[] = { "-", NULL };
 280
 281         BUG_ON_NON_EMPTY_PREFIX(prefix);
 282
 283         for (argp = argv+1; *argp; argp++) {
 284                 const char *arg = *argp;
 285
 286                 if (arg[0] != '-')
 287                         break;
 288                 /* do flags here */
 289                 if ( arg[1] == 'd' ) {
 290                         nr_prec = strtol(arg+2, NULL, 10);
 291                         if (nr_prec < 3 || 10 <= nr_prec)
 292                                 usage(git_mailsplit_usage);
 293                         continue;
 294                 } else if ( arg[1] == 'f' ) {
 295                         nr = strtol(arg+2, NULL, 10);
 296                 } else if ( arg[1] == 'h' ) {
 297                         usage(git_mailsplit_usage);
 298                 } else if ( arg[1] == 'b' && !arg[2] ) {
 299                         allow_bare = 1;
 300                 } else if (!strcmp(arg, "--keep-cr")) {
 301                         keep_cr = 1;
 302                 } else if ( arg[1] == 'o' && arg[2] ) {
 303                         dir = arg+2;
 304                 } else if (!strcmp(arg, "--mboxrd")) {
 305                         mboxrd = 1;
 306                 } else if ( arg[1] == '-' && !arg[2] ) {
 307                         argp++; /* -- marks end of options */
 308                         break;
 309                 } else {
 310                         die("unknown option: %s", arg);
 311                 }
 312         }
 313
 314         if ( !dir ) {
 315                 /* Backwards compatibility: if no -o specified, accept
 316                    <mbox> <dir> or just <dir> */
 317                 switch (argc - (argp-argv)) {
 318                 case 1:
 319                         dir = argp[0];
 320                         argp = stdin_only;
 321                         break;
 322                 case 2:
 323                         stdin_only[0] = argp[0];
 324                         dir = argp[1];
 325                         argp = stdin_only;
 326                         break;
 327                 default:
 328                         usage(git_mailsplit_usage);
 329                 }
 330         } else {
 331                 /* New usage: if no more argument, parse stdin */
 332                 if ( !*argp )
 333                         argp = stdin_only;
 334         }
 335
 336         while (*argp) {
 337                 const char *arg = *argp++;
 338                 struct stat argstat;
 339                 int ret = 0;
 340
 341                 if (arg[0] == '-' && arg[1] == 0) {
 342                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 343                         if (ret < 0) {
 344                                 error("cannot split patches from stdin");
 345                                 return 1;
 346                         }
 347                         num += (ret - nr);
 348                         nr = ret;
 349                         continue;
 350                 }
 351
 352                 if (stat(arg, &argstat) == -1) {
 353                         error_errno("cannot stat %s", arg);
 354                         return 1;
 355                 }
 356
 357                 if (S_ISDIR(argstat.st_mode))
 358                         ret = split_maildir(arg, dir, nr_prec, nr);
 359                 else
 360                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 361
 362                 if (ret < 0) {
 363                         error("cannot split patches from %s", arg);
 364                         return 1;
 365                 }
 366                 num += (ret - nr);
 367                 nr = ret;
 368         }
 369
 370         printf("%d\n", num);
 371
 372         return 0;
 373 }