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