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[PATH_MAX];
 102         char *subs[] = { "cur", "new", NULL };
 103         char **sub;
 104
 105         for (sub = subs; *sub; ++sub) {
 106                 snprintf(name, sizeof(name), "%s/%s", path, *sub);
 107                 if ((dir = opendir(name)) == NULL) {
 108                         if (errno == ENOENT)
 109                                 continue;
 110                         error("cannot opendir %s (%s)", name, strerror(errno));
 111                         return -1;
 112                 }
 113
 114                 while ((dent = readdir(dir)) != NULL) {
 115                         if (dent->d_name[0] == '.')
 116                                 continue;
 117                         snprintf(name, sizeof(name), "%s/%s", *sub, dent->d_name);
 118                         string_list_insert(list, name);
 119                 }
 120
 121                 closedir(dir);
 122         }
 123
 124         return 0;
 125 }
 126
 127 static int maildir_filename_cmp(const char *a, const char *b)
 128 {
 129         while (*a && *b) {
 130                 if (isdigit(*a) && isdigit(*b)) {
 131                         long int na, nb;
 132                         na = strtol(a, (char **)&a, 10);
 133                         nb = strtol(b, (char **)&b, 10);
 134                         if (na != nb)
 135                                 return na - nb;
 136                         /* strtol advanced our pointers */
 137                 }
 138                 else {
 139                         if (*a != *b)
 140                                 return (unsigned char)*a - (unsigned char)*b;
 141                         a++;
 142                         b++;
 143                 }
 144         }
 145         return (unsigned char)*a - (unsigned char)*b;
 146 }
 147
 148 static int split_maildir(const char *maildir, const char *dir,
 149         int nr_prec, int skip)
 150 {
 151         char file[PATH_MAX];
 152         char name[PATH_MAX];
 153         int ret = -1;
 154         int i;
 155         struct string_list list = STRING_LIST_INIT_DUP;
 156
 157         list.cmp = maildir_filename_cmp;
 158
 159         if (populate_maildir_list(&list, maildir) < 0)
 160                 goto out;
 161
 162         for (i = 0; i < list.nr; i++) {
 163                 FILE *f;
 164                 snprintf(file, sizeof(file), "%s/%s", maildir, list.items[i].string);
 165                 f = fopen(file, "r");
 166                 if (!f) {
 167                         error("cannot open mail %s (%s)", file, strerror(errno));
 168                         goto out;
 169                 }
 170
 171                 if (strbuf_getwholeline(&buf, f, '\n')) {
 172                         error("cannot read mail %s (%s)", file, strerror(errno));
 173                         goto out;
 174                 }
 175
 176                 sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 177                 split_one(f, name, 1);
 178
 179                 fclose(f);
 180         }
 181
 182         ret = skip;
 183 out:
 184         string_list_clear(&list, 1);
 185         return ret;
 186 }
 187
 188 static int split_mbox(const char *file, const char *dir, int allow_bare,
 189                       int nr_prec, int skip)
 190 {
 191         char name[PATH_MAX];
 192         int ret = -1;
 193         int peek;
 194
 195         FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 196         int file_done = 0;
 197
 198         if (!f) {
 199                 error("cannot open mbox %s", file);
 200                 goto out;
 201         }
 202
 203         do {
 204                 peek = fgetc(f);
 205         } while (isspace(peek));
 206         ungetc(peek, f);
 207
 208         if (strbuf_getwholeline(&buf, f, '\n')) {
 209                 /* empty stdin is OK */
 210                 if (f != stdin) {
 211                         error("cannot read mbox %s", file);
 212                         goto out;
 213                 }
 214                 file_done = 1;
 215         }
 216
 217         while (!file_done) {
 218                 sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 219                 file_done = split_one(f, name, allow_bare);
 220         }
 221
 222         if (f != stdin)
 223                 fclose(f);
 224
 225         ret = skip;
 226 out:
 227         return ret;
 228 }
 229
 230 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 231 {
 232         int nr = 0, nr_prec = 4, num = 0;
 233         int allow_bare = 0;
 234         const char *dir = NULL;
 235         const char **argp;
 236         static const char *stdin_only[] = { "-", NULL };
 237
 238         for (argp = argv+1; *argp; argp++) {
 239                 const char *arg = *argp;
 240
 241                 if (arg[0] != '-')
 242                         break;
 243                 /* do flags here */
 244                 if ( arg[1] == 'd' ) {
 245                         nr_prec = strtol(arg+2, NULL, 10);
 246                         if (nr_prec < 3 || 10 <= nr_prec)
 247                                 usage(git_mailsplit_usage);
 248                         continue;
 249                 } else if ( arg[1] == 'f' ) {
 250                         nr = strtol(arg+2, NULL, 10);
 251                 } else if ( arg[1] == 'h' ) {
 252                         usage(git_mailsplit_usage);
 253                 } else if ( arg[1] == 'b' && !arg[2] ) {
 254                         allow_bare = 1;
 255                 } else if (!strcmp(arg, "--keep-cr")) {
 256                         keep_cr = 1;
 257                 } else if ( arg[1] == 'o' && arg[2] ) {
 258                         dir = arg+2;
 259                 } else if ( arg[1] == '-' && !arg[2] ) {
 260                         argp++; /* -- marks end of options */
 261                         break;
 262                 } else {
 263                         die("unknown option: %s", arg);
 264                 }
 265         }
 266
 267         if ( !dir ) {
 268                 /* Backwards compatibility: if no -o specified, accept
 269                    <mbox> <dir> or just <dir> */
 270                 switch (argc - (argp-argv)) {
 271                 case 1:
 272                         dir = argp[0];
 273                         argp = stdin_only;
 274                         break;
 275                 case 2:
 276                         stdin_only[0] = argp[0];
 277                         dir = argp[1];
 278                         argp = stdin_only;
 279                         break;
 280                 default:
 281                         usage(git_mailsplit_usage);
 282                 }
 283         } else {
 284                 /* New usage: if no more argument, parse stdin */
 285                 if ( !*argp )
 286                         argp = stdin_only;
 287         }
 288
 289         while (*argp) {
 290                 const char *arg = *argp++;
 291                 struct stat argstat;
 292                 int ret = 0;
 293
 294                 if (arg[0] == '-' && arg[1] == 0) {
 295                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 296                         if (ret < 0) {
 297                                 error("cannot split patches from stdin");
 298                                 return 1;
 299                         }
 300                         num += (ret - nr);
 301                         nr = ret;
 302                         continue;
 303                 }
 304
 305                 if (stat(arg, &argstat) == -1) {
 306                         error("cannot stat %s (%s)", arg, strerror(errno));
 307                         return 1;
 308                 }
 309
 310                 if (S_ISDIR(argstat.st_mode))
 311                         ret = split_maildir(arg, dir, nr_prec, nr);
 312                 else
 313                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 314
 315                 if (ret < 0) {
 316                         error("cannot split patches from %s", arg);
 317                         return 1;
 318                 }
 319                 num += (ret - nr);
 320                 nr = ret;
 321         }
 322
 323         printf("%d\n", num);
 324
 325         return 0;
 326 }