| 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 | |
| 8 | #define DISABLE_SIGN_COMPARE_WARNINGS |
| 9 | |
| 10 | #include "builtin.h" |
| 11 | #include "gettext.h" |
| 12 | #include "string-list.h" |
| 13 | #include "strbuf.h" |
| 14 | |
| 15 | static const char git_mailsplit_usage[] = |
| 16 | "git mailsplit [-d<prec>] [-f<n>] [-b] [--keep-cr] -o<directory> [(<mbox>|<Maildir>)...]"; |
| 17 | |
| 18 | static int is_from_line(const char *line, int len) |
| 19 | { |
| 20 | const char *colon; |
| 21 | |
| 22 | if (len < 20 || memcmp("From ", line, 5)) |
| 23 | return 0; |
| 24 | |
| 25 | colon = line + len - 2; |
| 26 | line += 5; |
| 27 | for (;;) { |
| 28 | if (colon < line) |
| 29 | return 0; |
| 30 | if (*--colon == ':') |
| 31 | break; |
| 32 | } |
| 33 | |
| 34 | if (!isdigit(colon[-4]) || |
| 35 | !isdigit(colon[-2]) || |
| 36 | !isdigit(colon[-1]) || |
| 37 | !isdigit(colon[ 1]) || |
| 38 | !isdigit(colon[ 2])) |
| 39 | return 0; |
| 40 | |
| 41 | /* year */ |
| 42 | if (strtol(colon+3, NULL, 10) <= 90) |
| 43 | return 0; |
| 44 | |
| 45 | /* Ok, close enough */ |
| 46 | return 1; |
| 47 | } |
| 48 | |
| 49 | static struct strbuf buf = STRBUF_INIT; |
| 50 | static int keep_cr; |
| 51 | static int mboxrd; |
| 52 | |
| 53 | static int is_gtfrom(const struct strbuf *buf) |
| 54 | { |
| 55 | size_t min = strlen(">From "); |
| 56 | size_t ngt; |
| 57 | |
| 58 | if (buf->len < min) |
| 59 | return 0; |
| 60 | |
| 61 | ngt = strspn(buf->buf, ">"); |
| 62 | return ngt && starts_with(buf->buf + ngt, "From "); |
| 63 | } |
| 64 | |
| 65 | /* Called with the first line (potentially partial) |
| 66 | * already in buf[] -- normally that should begin with |
| 67 | * the Unix "From " line. Write it into the specified |
| 68 | * file. |
| 69 | */ |
| 70 | static int split_one(FILE *mbox, const char *name, int allow_bare) |
| 71 | { |
| 72 | FILE *output; |
| 73 | int fd; |
| 74 | int status = 0; |
| 75 | int is_bare = !is_from_line(buf.buf, buf.len); |
| 76 | |
| 77 | if (is_bare && !allow_bare) { |
| 78 | fprintf(stderr, "corrupt mailbox\n"); |
| 79 | exit(1); |
| 80 | } |
| 81 | fd = xopen(name, O_WRONLY | O_CREAT | O_EXCL, 0666); |
| 82 | output = xfdopen(fd, "w"); |
| 83 | |
| 84 | /* Copy it out, while searching for a line that begins with |
| 85 | * "From " and having something that looks like a date format. |
| 86 | */ |
| 87 | for (;;) { |
| 88 | if (!keep_cr && buf.len > 1 && buf.buf[buf.len-1] == '\n' && |
| 89 | buf.buf[buf.len-2] == '\r') { |
| 90 | strbuf_setlen(&buf, buf.len-2); |
| 91 | strbuf_addch(&buf, '\n'); |
| 92 | } |
| 93 | |
| 94 | if (mboxrd && is_gtfrom(&buf)) |
| 95 | strbuf_remove(&buf, 0, 1); |
| 96 | |
| 97 | if (fwrite(buf.buf, 1, buf.len, output) != buf.len) |
| 98 | die_errno("cannot write output"); |
| 99 | |
| 100 | if (strbuf_getwholeline(&buf, mbox, '\n')) { |
| 101 | if (feof(mbox)) { |
| 102 | status = 1; |
| 103 | break; |
| 104 | } |
| 105 | die_errno("cannot read mbox"); |
| 106 | } |
| 107 | if (!is_bare && is_from_line(buf.buf, buf.len)) |
| 108 | break; /* done with one message */ |
| 109 | } |
| 110 | fclose(output); |
| 111 | return status; |
| 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 = NULL; |
| 119 | const char *subs[] = { "cur", "new", NULL }; |
| 120 | const char **sub; |
| 121 | int ret = -1; |
| 122 | |
| 123 | for (sub = subs; *sub; ++sub) { |
| 124 | free(name); |
| 125 | name = xstrfmt("%s/%s", path, *sub); |
| 126 | if (!(dir = opendir(name))) { |
| 127 | if (errno == ENOENT) |
| 128 | continue; |
| 129 | error_errno("cannot opendir %s", name); |
| 130 | goto out; |
| 131 | } |
| 132 | |
| 133 | while ((dent = readdir(dir)) != NULL) { |
| 134 | if (dent->d_name[0] == '.') |
| 135 | continue; |
| 136 | free(name); |
| 137 | name = xstrfmt("%s/%s", *sub, dent->d_name); |
| 138 | string_list_insert(list, name); |
| 139 | } |
| 140 | |
| 141 | closedir(dir); |
| 142 | } |
| 143 | |
| 144 | ret = 0; |
| 145 | |
| 146 | out: |
| 147 | free(name); |
| 148 | return ret; |
| 149 | } |
| 150 | |
| 151 | static int maildir_filename_cmp(const char *a, const char *b) |
| 152 | { |
| 153 | while (*a && *b) { |
| 154 | if (isdigit(*a) && isdigit(*b)) { |
| 155 | long int na, nb; |
| 156 | na = strtol(a, (char **)&a, 10); |
| 157 | nb = strtol(b, (char **)&b, 10); |
| 158 | if (na != nb) |
| 159 | return na - nb; |
| 160 | /* strtol advanced our pointers */ |
| 161 | } |
| 162 | else { |
| 163 | if (*a != *b) |
| 164 | return (unsigned char)*a - (unsigned char)*b; |
| 165 | a++; |
| 166 | b++; |
| 167 | } |
| 168 | } |
| 169 | return (unsigned char)*a - (unsigned char)*b; |
| 170 | } |
| 171 | |
| 172 | static int split_maildir(const char *maildir, const char *dir, |
| 173 | int nr_prec, int skip) |
| 174 | { |
| 175 | char *file = NULL; |
| 176 | FILE *f = NULL; |
| 177 | int ret = -1; |
| 178 | struct string_list list = STRING_LIST_INIT_DUP; |
| 179 | |
| 180 | list.cmp = maildir_filename_cmp; |
| 181 | |
| 182 | if (populate_maildir_list(&list, maildir) < 0) |
| 183 | goto out; |
| 184 | |
| 185 | for (size_t i = 0; i < list.nr; i++) { |
| 186 | char *name; |
| 187 | |
| 188 | free(file); |
| 189 | file = xstrfmt("%s/%s", maildir, list.items[i].string); |
| 190 | |
| 191 | f = fopen(file, "r"); |
| 192 | if (!f) { |
| 193 | error_errno("cannot open mail %s", file); |
| 194 | goto out; |
| 195 | } |
| 196 | |
| 197 | if (strbuf_getwholeline(&buf, f, '\n')) { |
| 198 | error_errno("cannot read mail %s", file); |
| 199 | goto out; |
| 200 | } |
| 201 | |
| 202 | name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip); |
| 203 | split_one(f, name, 1); |
| 204 | free(name); |
| 205 | |
| 206 | fclose(f); |
| 207 | f = NULL; |
| 208 | } |
| 209 | |
| 210 | ret = skip; |
| 211 | out: |
| 212 | if (f) |
| 213 | fclose(f); |
| 214 | free(file); |
| 215 | string_list_clear(&list, 1); |
| 216 | return ret; |
| 217 | } |
| 218 | |
| 219 | static int split_mbox(const char *file, const char *dir, int allow_bare, |
| 220 | int nr_prec, int skip) |
| 221 | { |
| 222 | int ret = -1; |
| 223 | int peek; |
| 224 | |
| 225 | FILE *f = !strcmp(file, "-") ? stdin : fopen(file, "r"); |
| 226 | int file_done = 0; |
| 227 | |
| 228 | if (isatty(fileno(f))) |
| 229 | warning(_("reading patches from stdin/tty...")); |
| 230 | |
| 231 | if (!f) { |
| 232 | error_errno("cannot open mbox %s", file); |
| 233 | goto out; |
| 234 | } |
| 235 | |
| 236 | do { |
| 237 | peek = fgetc(f); |
| 238 | if (peek == EOF) { |
| 239 | if (f == stdin) |
| 240 | /* empty stdin is OK */ |
| 241 | ret = skip; |
| 242 | else { |
| 243 | fclose(f); |
| 244 | error(_("empty mbox: '%s'"), file); |
| 245 | } |
| 246 | goto out; |
| 247 | } |
| 248 | } while (isspace(peek)); |
| 249 | ungetc(peek, f); |
| 250 | |
| 251 | if (strbuf_getwholeline(&buf, f, '\n')) { |
| 252 | /* empty stdin is OK */ |
| 253 | if (f != stdin) { |
| 254 | error("cannot read mbox %s", file); |
| 255 | goto out; |
| 256 | } |
| 257 | file_done = 1; |
| 258 | } |
| 259 | |
| 260 | while (!file_done) { |
| 261 | char *name = xstrfmt("%s/%0*d", dir, nr_prec, ++skip); |
| 262 | file_done = split_one(f, name, allow_bare); |
| 263 | free(name); |
| 264 | } |
| 265 | |
| 266 | if (f != stdin) |
| 267 | fclose(f); |
| 268 | |
| 269 | ret = skip; |
| 270 | out: |
| 271 | return ret; |
| 272 | } |
| 273 | |
| 274 | int cmd_mailsplit(int argc, |
| 275 | const char **argv, |
| 276 | const char *prefix, |
| 277 | struct repository *repo UNUSED) |
| 278 | { |
| 279 | int nr = 0, nr_prec = 4, num = 0; |
| 280 | int allow_bare = 0; |
| 281 | const char *dir = NULL; |
| 282 | const char **argp; |
| 283 | static const char *stdin_only[] = { "-", NULL }; |
| 284 | |
| 285 | BUG_ON_NON_EMPTY_PREFIX(prefix); |
| 286 | |
| 287 | show_usage_if_asked(argc, argv, git_mailsplit_usage); |
| 288 | |
| 289 | for (argp = argv+1; *argp; argp++) { |
| 290 | const char *arg = *argp; |
| 291 | |
| 292 | if (arg[0] != '-') |
| 293 | break; |
| 294 | /* do flags here */ |
| 295 | if ( arg[1] == 'd' ) { |
| 296 | nr_prec = strtol(arg+2, NULL, 10); |
| 297 | if (nr_prec < 3 || 10 <= nr_prec) |
| 298 | usage(git_mailsplit_usage); |
| 299 | continue; |
| 300 | } else if ( arg[1] == 'f' ) { |
| 301 | nr = strtol(arg+2, NULL, 10); |
| 302 | } else if ( arg[1] == 'b' && !arg[2] ) { |
| 303 | allow_bare = 1; |
| 304 | } else if (!strcmp(arg, "--keep-cr")) { |
| 305 | keep_cr = 1; |
| 306 | } else if ( arg[1] == 'o' && arg[2] ) { |
| 307 | dir = arg+2; |
| 308 | } else if (!strcmp(arg, "--mboxrd")) { |
| 309 | mboxrd = 1; |
| 310 | } else if ( arg[1] == '-' && !arg[2] ) { |
| 311 | argp++; /* -- marks end of options */ |
| 312 | break; |
| 313 | } else { |
| 314 | die("unknown option: %s", arg); |
| 315 | } |
| 316 | } |
| 317 | |
| 318 | if ( !dir ) { |
| 319 | /* Backwards compatibility: if no -o specified, accept |
| 320 | <mbox> <dir> or just <dir> */ |
| 321 | switch (argc - (argp-argv)) { |
| 322 | case 1: |
| 323 | dir = argp[0]; |
| 324 | argp = stdin_only; |
| 325 | break; |
| 326 | case 2: |
| 327 | stdin_only[0] = argp[0]; |
| 328 | dir = argp[1]; |
| 329 | argp = stdin_only; |
| 330 | break; |
| 331 | default: |
| 332 | usage(git_mailsplit_usage); |
| 333 | } |
| 334 | } else { |
| 335 | /* New usage: if no more argument, parse stdin */ |
| 336 | if ( !*argp ) |
| 337 | argp = stdin_only; |
| 338 | } |
| 339 | |
| 340 | while (*argp) { |
| 341 | const char *arg = *argp++; |
| 342 | struct stat argstat; |
| 343 | int ret = 0; |
| 344 | |
| 345 | if (arg[0] == '-' && arg[1] == 0) { |
| 346 | ret = split_mbox(arg, dir, allow_bare, nr_prec, nr); |
| 347 | if (ret < 0) { |
| 348 | error("cannot split patches from stdin"); |
| 349 | return 1; |
| 350 | } |
| 351 | num += (ret - nr); |
| 352 | nr = ret; |
| 353 | continue; |
| 354 | } |
| 355 | |
| 356 | if (stat(arg, &argstat) == -1) { |
| 357 | error_errno("cannot stat %s", arg); |
| 358 | return 1; |
| 359 | } |
| 360 | |
| 361 | if (S_ISDIR(argstat.st_mode)) |
| 362 | ret = split_maildir(arg, dir, nr_prec, nr); |
| 363 | else |
| 364 | ret = split_mbox(arg, dir, allow_bare, nr_prec, nr); |
| 365 | |
| 366 | if (ret < 0) { |
| 367 | error("cannot split patches from %s", arg); |
| 368 | return 1; |
| 369 | } |
| 370 | num += (ret - nr); |
| 371 | nr = ret; |
| 372 | } |
| 373 | |
| 374 | printf("%d\n", num); |
| 375 | |
| 376 | return 0; |
| 377 | } |