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