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
  11 static const char git_mailsplit_usage[] =
  12 "git mailsplit [-d<prec>] [-f<n>] [-b] -o<directory> [<mbox>|<Maildir>...]";
  13
  14 static int is_from_line(const char *line, int len)
  15 {
  16         const char *colon;
  17
  18         if (len < 20 || memcmp("From ", line, 5))
  19                 return 0;
  20
  21         colon = line + len - 2;
  22         line += 5;
  23         for (;;) {
  24                 if (colon < line)
  25                         return 0;
  26                 if (*--colon == ':')
  27                         break;
  28         }
  29
  30         if (!isdigit(colon[-4]) ||
  31             !isdigit(colon[-2]) ||
  32             !isdigit(colon[-1]) ||
  33             !isdigit(colon[ 1]) ||
  34             !isdigit(colon[ 2]))
  35                 return 0;
  36
  37         /* year */
  38         if (strtol(colon+3, NULL, 10) <= 90)
  39                 return 0;
  40
  41         /* Ok, close enough */
  42         return 1;
  43 }
  44
  45 /* Could be as small as 64, enough to hold a Unix "From " line. */
  46 static char buf[4096];
  47
  48 /* We cannot use fgets() because our lines can contain NULs */
  49 int read_line_with_nul(char *buf, int size, FILE *in)
  50 {
  51         int len = 0, c;
  52
  53         for (;;) {
  54                 c = getc(in);
  55                 if (c == EOF)
  56                         break;
  57                 buf[len++] = c;
  58                 if (c == '\n' || len + 1 >= size)
  59                         break;
  60         }
  61         buf[len] = '\0';
  62
  63         return len;
  64 }
  65
  66 /* Called with the first line (potentially partial)
  67  * already in buf[] -- normally that should begin with
  68  * the Unix "From " line.  Write it into the specified
  69  * file.
  70  */
  71 static int split_one(FILE *mbox, const char *name, int allow_bare)
  72 {
  73         FILE *output = NULL;
  74         int len = strlen(buf);
  75         int fd;
  76         int status = 0;
  77         int is_bare = !is_from_line(buf, len);
  78
  79         if (is_bare && !allow_bare)
  80                 goto corrupt;
  81
  82         fd = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  83         if (fd < 0)
  84                 die_errno("cannot open output file '%s'", name);
  85         output = fdopen(fd, "w");
  86
  87         /* Copy it out, while searching for a line that begins with
  88          * "From " and having something that looks like a date format.
  89          */
  90         for (;;) {
  91                 int is_partial = len && buf[len-1] != '\n';
  92
  93                 if (fwrite(buf, 1, len, output) != len)
  94                         die_errno("cannot write output");
  95
  96                 len = read_line_with_nul(buf, sizeof(buf), mbox);
  97                 if (len == 0) {
  98                         if (feof(mbox)) {
  99                                 status = 1;
 100                                 break;
 101                         }
 102                         die_errno("cannot read mbox");
 103                 }
 104                 if (!is_partial && !is_bare && is_from_line(buf, len))
 105                         break; /* done with one message */
 106         }
 107         fclose(output);
 108         return status;
 109
 110  corrupt:
 111         if (output)
 112                 fclose(output);
 113         unlink(name);
 114         fprintf(stderr, "corrupt mailbox\n");
 115         exit(1);
 116 }
 117
 118 static int populate_maildir_list(struct string_list *list, const char *path)
 119 {
 120         DIR *dir;
 121         struct dirent *dent;
 122         char name[PATH_MAX];
 123         char *subs[] = { "cur", "new", NULL };
 124         char **sub;
 125
 126         for (sub = subs; *sub; ++sub) {
 127                 snprintf(name, sizeof(name), "%s/%s", path, *sub);
 128                 if ((dir = opendir(name)) == NULL) {
 129                         if (errno == ENOENT)
 130                                 continue;
 131                         error("cannot opendir %s (%s)", name, strerror(errno));
 132                         return -1;
 133                 }
 134
 135                 while ((dent = readdir(dir)) != NULL) {
 136                         if (dent->d_name[0] == '.')
 137                                 continue;
 138                         snprintf(name, sizeof(name), "%s/%s", *sub, dent->d_name);
 139                         string_list_insert(name, list);
 140                 }
 141
 142                 closedir(dir);
 143         }
 144
 145         return 0;
 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 = {NULL, 0, 0, 1};
 156
 157         if (populate_maildir_list(&list, maildir) < 0)
 158                 goto out;
 159
 160         for (i = 0; i < list.nr; i++) {
 161                 FILE *f;
 162                 snprintf(file, sizeof(file), "%s/%s", maildir, list.items[i].string);
 163                 f = fopen(file, "r");
 164                 if (!f) {
 165                         error("cannot open mail %s (%s)", file, strerror(errno));
 166                         goto out;
 167                 }
 168
 169                 if (fgets(buf, sizeof(buf), f) == NULL) {
 170                         error("cannot read mail %s (%s)", file, strerror(errno));
 171                         goto out;
 172                 }
 173
 174                 sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 175                 split_one(f, name, 1);
 176
 177                 fclose(f);
 178         }
 179
 180         ret = skip;
 181 out:
 182         string_list_clear(&list, 1);
 183         return ret;
 184 }
 185
 186 static int split_mbox(const char *file, const char *dir, int allow_bare,
 187                       int nr_prec, int skip)
 188 {
 189         char name[PATH_MAX];
 190         int ret = -1;
 191         int peek;
 192
 193         FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 194         int file_done = 0;
 195
 196         if (!f) {
 197                 error("cannot open mbox %s", file);
 198                 goto out;
 199         }
 200
 201         do {
 202                 peek = fgetc(f);
 203         } while (isspace(peek));
 204         ungetc(peek, f);
 205
 206         if (fgets(buf, sizeof(buf), f) == NULL) {
 207                 /* empty stdin is OK */
 208                 if (f != stdin) {
 209                         error("cannot read mbox %s", file);
 210                         goto out;
 211                 }
 212                 file_done = 1;
 213         }
 214
 215         while (!file_done) {
 216                 sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 217                 file_done = split_one(f, name, allow_bare);
 218         }
 219
 220         if (f != stdin)
 221                 fclose(f);
 222
 223         ret = skip;
 224 out:
 225         return ret;
 226 }
 227
 228 int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 229 {
 230         int nr = 0, nr_prec = 4, num = 0;
 231         int allow_bare = 0;
 232         const char *dir = NULL;
 233         const char **argp;
 234         static const char *stdin_only[] = { "-", NULL };
 235
 236         for (argp = argv+1; *argp; argp++) {
 237                 const char *arg = *argp;
 238
 239                 if (arg[0] != '-')
 240                         break;
 241                 /* do flags here */
 242                 if ( arg[1] == 'd' ) {
 243                         nr_prec = strtol(arg+2, NULL, 10);
 244                         if (nr_prec < 3 || 10 <= nr_prec)
 245                                 usage(git_mailsplit_usage);
 246                         continue;
 247                 } else if ( arg[1] == 'f' ) {
 248                         nr = strtol(arg+2, NULL, 10);
 249                 } else if ( arg[1] == 'b' && !arg[2] ) {
 250                         allow_bare = 1;
 251                 } else if ( arg[1] == 'o' && arg[2] ) {
 252                         dir = arg+2;
 253                 } else if ( arg[1] == '-' && !arg[2] ) {
 254                         argp++; /* -- marks end of options */
 255                         break;
 256                 } else {
 257                         die("unknown option: %s", arg);
 258                 }
 259         }
 260
 261         if ( !dir ) {
 262                 /* Backwards compatibility: if no -o specified, accept
 263                    <mbox> <dir> or just <dir> */
 264                 switch (argc - (argp-argv)) {
 265                 case 1:
 266                         dir = argp[0];
 267                         argp = stdin_only;
 268                         break;
 269                 case 2:
 270                         stdin_only[0] = argp[0];
 271                         dir = argp[1];
 272                         argp = stdin_only;
 273                         break;
 274                 default:
 275                         usage(git_mailsplit_usage);
 276                 }
 277         } else {
 278                 /* New usage: if no more argument, parse stdin */
 279                 if ( !*argp )
 280                         argp = stdin_only;
 281         }
 282
 283         while (*argp) {
 284                 const char *arg = *argp++;
 285                 struct stat argstat;
 286                 int ret = 0;
 287
 288                 if (arg[0] == '-' && arg[1] == 0) {
 289                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 290                         if (ret < 0) {
 291                                 error("cannot split patches from stdin");
 292                                 return 1;
 293                         }
 294                         num += (ret - nr);
 295                         nr = ret;
 296                         continue;
 297                 }
 298
 299                 if (stat(arg, &argstat) == -1) {
 300                         error("cannot stat %s (%s)", arg, strerror(errno));
 301                         return 1;
 302                 }
 303
 304                 if (S_ISDIR(argstat.st_mode))
 305                         ret = split_maildir(arg, dir, nr_prec, nr);
 306                 else
 307                         ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 308
 309                 if (ret < 0) {
 310                         error("cannot split patches from %s", arg);
 311                         return 1;
 312                 }
 313                 num += (ret - nr);
 314                 nr = ret;
 315         }
 316
 317         printf("%d\n", num);
 318
 319         return 0;
 320 }