Andrew's git - gitweb.git/blob

d288fde2c2cef606a60507a7a56b81352ab5d06e
   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;
  47
  48/* Called with the first line (potentially partial)
  49 * already in buf[] -- normally that should begin with
  50 * the Unix "From " line.  Write it into the specified
  51 * file.
  52 */
  53static int split_one(FILE *mbox, const char *name, int allow_bare)
  54{
  55        FILE *output = NULL;
  56        int fd;
  57        int status = 0;
  58        int is_bare = !is_from_line(buf.buf, buf.len);
  59
  60        if (is_bare && !allow_bare)
  61                goto corrupt;
  62
  63        fd = open(name, O_WRONLY | O_CREAT | O_EXCL, 0666);
  64        if (fd < 0)
  65                die_errno("cannot open output file '%s'", name);
  66        output = fdopen(fd, "w");
  67
  68        /* Copy it out, while searching for a line that begins with
  69         * "From " and having something that looks like a date format.
  70         */
  71        for (;;) {
  72                if (fwrite(buf.buf, 1, buf.len, output) != buf.len)
  73                        die_errno("cannot write output");
  74
  75                if (strbuf_getwholeline(&buf, mbox, '\n')) {
  76                        if (feof(mbox)) {
  77                                status = 1;
  78                                break;
  79                        }
  80                        die_errno("cannot read mbox");
  81                }
  82                if (!is_bare && is_from_line(buf.buf, buf.len))
  83                        break; /* done with one message */
  84        }
  85        fclose(output);
  86        return status;
  87
  88 corrupt:
  89        if (output)
  90                fclose(output);
  91        unlink(name);
  92        fprintf(stderr, "corrupt mailbox\n");
  93        exit(1);
  94}
  95
  96static int populate_maildir_list(struct string_list *list, const char *path)
  97{
  98        DIR *dir;
  99        struct dirent *dent;
 100        char name[PATH_MAX];
 101        char *subs[] = { "cur", "new", NULL };
 102        char **sub;
 103
 104        for (sub = subs; *sub; ++sub) {
 105                snprintf(name, sizeof(name), "%s/%s", path, *sub);
 106                if ((dir = opendir(name)) == NULL) {
 107                        if (errno == ENOENT)
 108                                continue;
 109                        error("cannot opendir %s (%s)", name, strerror(errno));
 110                        return -1;
 111                }
 112
 113                while ((dent = readdir(dir)) != NULL) {
 114                        if (dent->d_name[0] == '.')
 115                                continue;
 116                        snprintf(name, sizeof(name), "%s/%s", *sub, dent->d_name);
 117                        string_list_insert(name, list);
 118                }
 119
 120                closedir(dir);
 121        }
 122
 123        return 0;
 124}
 125
 126static int split_maildir(const char *maildir, const char *dir,
 127        int nr_prec, int skip)
 128{
 129        char file[PATH_MAX];
 130        char name[PATH_MAX];
 131        int ret = -1;
 132        int i;
 133        struct string_list list = {NULL, 0, 0, 1};
 134
 135        if (populate_maildir_list(&list, maildir) < 0)
 136                goto out;
 137
 138        for (i = 0; i < list.nr; i++) {
 139                FILE *f;
 140                snprintf(file, sizeof(file), "%s/%s", maildir, list.items[i].string);
 141                f = fopen(file, "r");
 142                if (!f) {
 143                        error("cannot open mail %s (%s)", file, strerror(errno));
 144                        goto out;
 145                }
 146
 147                if (strbuf_getwholeline(&buf, f, '\n')) {
 148                        error("cannot read mail %s (%s)", file, strerror(errno));
 149                        goto out;
 150                }
 151
 152                sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 153                split_one(f, name, 1);
 154
 155                fclose(f);
 156        }
 157
 158        ret = skip;
 159out:
 160        string_list_clear(&list, 1);
 161        return ret;
 162}
 163
 164static int split_mbox(const char *file, const char *dir, int allow_bare,
 165                      int nr_prec, int skip)
 166{
 167        char name[PATH_MAX];
 168        int ret = -1;
 169        int peek;
 170
 171        FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r");
 172        int file_done = 0;
 173
 174        if (!f) {
 175                error("cannot open mbox %s", file);
 176                goto out;
 177        }
 178
 179        do {
 180                peek = fgetc(f);
 181        } while (isspace(peek));
 182        ungetc(peek, f);
 183
 184        if (strbuf_getwholeline(&buf, f, '\n')) {
 185                /* empty stdin is OK */
 186                if (f != stdin) {
 187                        error("cannot read mbox %s", file);
 188                        goto out;
 189                }
 190                file_done = 1;
 191        }
 192
 193        while (!file_done) {
 194                sprintf(name, "%s/%0*d", dir, nr_prec, ++skip);
 195                file_done = split_one(f, name, allow_bare);
 196        }
 197
 198        if (f != stdin)
 199                fclose(f);
 200
 201        ret = skip;
 202out:
 203        return ret;
 204}
 205
 206int cmd_mailsplit(int argc, const char **argv, const char *prefix)
 207{
 208        int nr = 0, nr_prec = 4, num = 0;
 209        int allow_bare = 0;
 210        const char *dir = NULL;
 211        const char **argp;
 212        static const char *stdin_only[] = { "-", NULL };
 213
 214        for (argp = argv+1; *argp; argp++) {
 215                const char *arg = *argp;
 216
 217                if (arg[0] != '-')
 218                        break;
 219                /* do flags here */
 220                if ( arg[1] == 'd' ) {
 221                        nr_prec = strtol(arg+2, NULL, 10);
 222                        if (nr_prec < 3 || 10 <= nr_prec)
 223                                usage(git_mailsplit_usage);
 224                        continue;
 225                } else if ( arg[1] == 'f' ) {
 226                        nr = strtol(arg+2, NULL, 10);
 227                } else if ( arg[1] == 'b' && !arg[2] ) {
 228                        allow_bare = 1;
 229                } else if ( arg[1] == 'o' && arg[2] ) {
 230                        dir = arg+2;
 231                } else if ( arg[1] == '-' && !arg[2] ) {
 232                        argp++; /* -- marks end of options */
 233                        break;
 234                } else {
 235                        die("unknown option: %s", arg);
 236                }
 237        }
 238
 239        if ( !dir ) {
 240                /* Backwards compatibility: if no -o specified, accept
 241                   <mbox> <dir> or just <dir> */
 242                switch (argc - (argp-argv)) {
 243                case 1:
 244                        dir = argp[0];
 245                        argp = stdin_only;
 246                        break;
 247                case 2:
 248                        stdin_only[0] = argp[0];
 249                        dir = argp[1];
 250                        argp = stdin_only;
 251                        break;
 252                default:
 253                        usage(git_mailsplit_usage);
 254                }
 255        } else {
 256                /* New usage: if no more argument, parse stdin */
 257                if ( !*argp )
 258                        argp = stdin_only;
 259        }
 260
 261        while (*argp) {
 262                const char *arg = *argp++;
 263                struct stat argstat;
 264                int ret = 0;
 265
 266                if (arg[0] == '-' && arg[1] == 0) {
 267                        ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 268                        if (ret < 0) {
 269                                error("cannot split patches from stdin");
 270                                return 1;
 271                        }
 272                        num += (ret - nr);
 273                        nr = ret;
 274                        continue;
 275                }
 276
 277                if (stat(arg, &argstat) == -1) {
 278                        error("cannot stat %s (%s)", arg, strerror(errno));
 279                        return 1;
 280                }
 281
 282                if (S_ISDIR(argstat.st_mode))
 283                        ret = split_maildir(arg, dir, nr_prec, nr);
 284                else
 285                        ret = split_mbox(arg, dir, allow_bare, nr_prec, nr);
 286
 287                if (ret < 0) {
 288                        error("cannot split patches from %s", arg);
 289                        return 1;
 290                }
 291                num += (ret - nr);
 292                nr = ret;
 293        }
 294
 295        printf("%d\n", num);
 296
 297        return 0;
 298}