Andrew's git - gitweb.git/blob - 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
  12static const char git_mailsplit_usage[] =
  13"git mailsplit [-d<prec>] [-f<n>] [-b] -o<directory> [<mbox>|<Maildir>...]";
  14
  15static 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
  46static struct strbuf buf = STRBUF_INIT;
  47static 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 */
  54static int split_one(FILE *mbox, const char *name, int allow_bare)
  55{
  56        FILE *output = NULL;
  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                goto corrupt;
  63
  64        fd = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  65        if (fd < 0)
  66                die_errno("cannot open output file '%s'", name);
  67        output = xfdopen(fd, "w");
  68
  69        /* Copy it out, while searching for a line that begins with
  70         * "From " and having something that looks like a date format.
  71         */
  72        for (;;) {
  73                if (!keep_cr && buf.len > 1 && buf.buf[buf.len-1] == '\n' &&
  74                        buf.buf[buf.len-2] == '\r') {
  75                        strbuf_setlen(&buf, buf.len-2);
  76                        strbuf_addch(&buf, '\n');
  77                }
  78
  79                if (fwrite(buf.buf, 1, buf.len, output) != buf.len)
  80                        die_errno("cannot write output");
  81
  82                if (strbuf_getwholeline(&buf, mbox, '\n')) {
  83                        if (feof(mbox)) {
  84                                status = 1;
  85                                break;
  86                        }
  87                        die_errno("cannot read mbox");
  88                }
  89                if (!is_bare && is_from_line(buf.buf, buf.len))
  90                        break; /* done with one message */
  91        }
  92        fclose(output);
  93        return status;
  94
  95 corrupt:
  96        if (output)
  97                fclose(output);
  98        unlink(name);
  99        fprintf(stderr, "corrupt mailbox\n");
 100        exit(1);
 101}
 102
 103static int populate_maildir_list(struct string_list *list, const char *path)
 104{
 105        DIR *dir;
 106        struct dirent *dent;
 107        char name[PATH_MAX];
 108        char *subs[] = { "cur", "new", NULL };
 109        char **sub;
 110
 111        for (sub = subs; *sub; ++sub) {
 112                snprintf(name, sizeof(name), "%s/%s", path, *sub);
 113                if ((dir = opendir(name)) == NULL) {
 114                        if (errno == ENOENT)
 115                                continue;
 116                        error("cannot opendir %s (%s)", name, strerror(errno));
 117                        return -1;
 118                }
 119
 120                while ((dent = readdir(dir)) != NULL) {
 121                        if (dent->d_name[0] == '.')
 122                                continue;
 123                        snprintf(name, sizeof(name), "%s/%s", *sub, dent->d_name);
 124                        string_list_insert(name, list);
 125                }
 126
 127                closedir(dir);
 128        }
 129
 130        return 0;
 131}
 132
 133static int split_maildir(const char *maildir, const char *dir,
 134        int nr_prec, int skip)
 135{
 136        char file[PATH_MAX];
 137        char name[PATH_MAX];
 138        int ret = -1;
 139        int i;
 140        struct string_list list = {NULL, 0, 0, 1};
 141
 142        if (populate_maildir_list(&list, maildir) < 0)
 143                goto out;
 144
 145        for (i = 0; i < list.nr; i++) {
 146                FILE *f;
 147                snprintf(file, sizeof(file), "%s/%s", maildir, list.items[i].string);
 148                f = fopen(file, "r");
 149                if (!f) {
 150                        error("cannot open mail %s (%s)", file, strerror(errno));
 151                        goto out;
 152                }
 153
 154                if (strbuf_getwholeline(&buf, f, '\n')) {
 155                        error("cannot read mail %s (%s)", file, strerror(errno));
 156                        goto out;
 157                }
 158
 159                sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 160                split_one(f, name, 1);
 161
 162                fclose(f);
 163        }
 164
 165        ret = skip;
 166out:
 167        string_list_clear(&list, 1);
 168        return ret;
 169}
 170
 171static int split_mbox(const char *file, const char *dir, int allow_bare,
 172                      int nr_prec, int skip)
 173{
 174        char name[PATH_MAX];
 175        int ret = -1;
 176        int peek;
 177
 178        FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 179        int file_done = 0;
 180
 181        if (!f) {
 182                error("cannot open mbox %s", file);
 183                goto out;
 184        }
 185
 186        do {
 187                peek = fgetc(f);
 188        } while (isspace(peek));
 189        ungetc(peek, f);
 190
 191        if (strbuf_getwholeline(&buf, f, '\n')) {
 192                /* empty stdin is OK */
 193                if (f != stdin) {
 194                        error("cannot read mbox %s", file);
 195                        goto out;
 196                }
 197                file_done = 1;
 198        }
 199
 200        while (!file_done) {
 201                sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 202                file_done = split_one(f, name, allow_bare);
 203        }
 204
 205        if (f != stdin)
 206                fclose(f);
 207
 208        ret = skip;
 209out:
 210        return ret;
 211}
 212
 213int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 214{
 215        int nr = 0, nr_prec = 4, num = 0;
 216        int allow_bare = 0;
 217        const char *dir = NULL;
 218        const char **argp;
 219        static const char *stdin_only[] = { "-", NULL };
 220
 221        for (argp = argv+1; *argp; argp++) {
 222                const char *arg = *argp;
 223
 224                if (arg[0] != '-')
 225                        break;
 226                /* do flags here */
 227                if ( arg[1] == 'd' ) {
 228                        nr_prec = strtol(arg+2, NULL, 10);
 229                        if (nr_prec < 3 || 10 <= nr_prec)
 230                                usage(git_mailsplit_usage);
 231                        continue;
 232                } else if ( arg[1] == 'f' ) {
 233                        nr = strtol(arg+2, NULL, 10);
 234                } else if ( arg[1] == 'h' ) {
 235                        usage(git_mailsplit_usage);
 236                } else if ( arg[1] == 'b' && !arg[2] ) {
 237                        allow_bare = 1;
 238                } else if (!strcmp(arg, "--keep-cr")) {
 239                        keep_cr = 1;
 240                } else if ( arg[1] == 'o' && arg[2] ) {
 241                        dir = arg+2;
 242                } else if ( arg[1] == '-' && !arg[2] ) {
 243                        argp++; /* -- marks end of options */
 244                        break;
 245                } else {
 246                        die("unknown option: %s", arg);
 247                }
 248        }
 249
 250        if ( !dir ) {
 251                /* Backwards compatibility: if no -o specified, accept
 252                   <mbox> <dir> or just <dir> */
 253                switch (argc - (argp-argv)) {
 254                case 1:
 255                        dir = argp[0];
 256                        argp = stdin_only;
 257                        break;
 258                case 2:
 259                        stdin_only[0] = argp[0];
 260                        dir = argp[1];
 261                        argp = stdin_only;
 262                        break;
 263                default:
 264                        usage(git_mailsplit_usage);
 265                }
 266        } else {
 267                /* New usage: if no more argument, parse stdin */
 268                if ( !*argp )
 269                        argp = stdin_only;
 270        }
 271
 272        while (*argp) {
 273                const char *arg = *argp++;
 274                struct stat argstat;
 275                int ret = 0;
 276
 277                if (arg[0] == '-' && arg[1] == 0) {
 278                        ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 279                        if (ret < 0) {
 280                                error("cannot split patches from stdin");
 281                                return 1;
 282                        }
 283                        num += (ret - nr);
 284                        nr = ret;
 285                        continue;
 286                }
 287
 288                if (stat(arg, &argstat) == -1) {
 289                        error("cannot stat %s (%s)", arg, strerror(errno));
 290                        return 1;
 291                }
 292
 293                if (S_ISDIR(argstat.st_mode))
 294                        ret = split_maildir(arg, dir, nr_prec, nr);
 295                else
 296                        ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 297
 298                if (ret < 0) {
 299                        error("cannot split patches from %s", arg);
 300                        return 1;
 301                }
 302                num += (ret - nr);
 303                nr = ret;
 304        }
 305
 306        printf("%d\n", num);
 307
 308        return 0;
 309}