builtin-mailinfo.c

   1 /*
   2  * Another stupid program, this one parsing the headers of an
   3  * email to figure out authorship and subject
   4  */
   5 #include "cache.h"
   6 #include "builtin.h"
   7 #include "utf8.h"
   8
   9 static FILE *cmitmsg, *patchfile, *fin, *fout;
  10
  11 static int keep_subject;
  12 static const char *metainfo_charset;
  13 static char line[1000];
  14 static char date[1000];
  15 static char name[1000];
  16 static char email[1000];
  17 static char subject[1000];
  18
  19 static enum  {
  20         TE_DONTCARE, TE_QP, TE_BASE64,
  21 } transfer_encoding;
  22 static char charset[256];
  23
  24 static char multipart_boundary[1000];
  25 static int multipart_boundary_len;
  26 static int patch_lines;
  27
  28 static char *sanity_check(char *name, char *email)
  29 {
  30         int len = strlen(name);
  31         if (len < 3 || len > 60)
  32                 return email;
  33         if (strchr(name, '@') || strchr(name, '<') || strchr(name, '>'))
  34                 return email;
  35         return name;
  36 }
  37
  38 static int bogus_from(char *line)
  39 {
  40         /* John Doe <johndoe> */
  41         char *bra, *ket, *dst, *cp;
  42
  43         /* This is fallback, so do not bother if we already have an
  44          * e-mail address.
  45          */
  46         if (*email)
  47                 return 0;
  48
  49         bra = strchr(line, '<');
  50         if (!bra)
  51                 return 0;
  52         ket = strchr(bra, '>');
  53         if (!ket)
  54                 return 0;
  55
  56         for (dst = email, cp = bra+1; cp < ket; )
  57                 *dst++ = *cp++;
  58         *dst = 0;
  59         for (cp = line; isspace(*cp); cp++)
  60                 ;
  61         for (bra--; isspace(*bra); bra--)
  62                 *bra = 0;
  63         cp = sanity_check(cp, email);
  64         strcpy(name, cp);
  65         return 1;
  66 }
  67
  68 static int handle_from(char *in_line)
  69 {
  70         char line[1000];
  71         char *at;
  72         char *dst;
  73
  74         strcpy(line, in_line);
  75         at = strchr(line, '@');
  76         if (!at)
  77                 return bogus_from(line);
  78
  79         /*
  80          * If we already have one email, don't take any confusing lines
  81          */
  82         if (*email && strchr(at+1, '@'))
  83                 return 0;
  84
  85         /* Pick up the string around '@', possibly delimited with <>
  86          * pair; that is the email part.  White them out while copying.
  87          */
  88         while (at > line) {
  89                 char c = at[-1];
  90                 if (isspace(c))
  91                         break;
  92                 if (c == '<') {
  93                         at[-1] = ' ';
  94                         break;
  95                 }
  96                 at--;
  97         }
  98         dst = email;
  99         for (;;) {
 100                 unsigned char c = *at;
 101                 if (!c || c == '>' || isspace(c)) {
 102                         if (c == '>')
 103                                 *at = ' ';
 104                         break;
 105                 }
 106                 *at++ = ' ';
 107                 *dst++ = c;
 108         }
 109         *dst++ = 0;
 110
 111         /* The remainder is name.  It could be "John Doe <john.doe@xz>"
 112          * or "john.doe@xz (John Doe)", but we have whited out the
 113          * email part, so trim from both ends, possibly removing
 114          * the () pair at the end.
 115          */
 116         at = line + strlen(line);
 117         while (at > line) {
 118                 unsigned char c = *--at;
 119                 if (!isspace(c)) {
 120                         at[(c == ')') ? 0 : 1] = 0;
 121                         break;
 122                 }
 123         }
 124
 125         at = line;
 126         for (;;) {
 127                 unsigned char c = *at;
 128                 if (!c || !isspace(c)) {
 129                         if (c == '(')
 130                                 at++;
 131                         break;
 132                 }
 133                 at++;
 134         }
 135         at = sanity_check(at, email);
 136         strcpy(name, at);
 137         return 1;
 138 }
 139
 140 static int handle_date(char *line)
 141 {
 142         strcpy(date, line);
 143         return 0;
 144 }
 145
 146 static int handle_subject(char *line)
 147 {
 148         strcpy(subject, line);
 149         return 0;
 150 }
 151
 152 /* NOTE NOTE NOTE.  We do not claim we do full MIME.  We just attempt
 153  * to have enough heuristics to grok MIME encoded patches often found
 154  * on our mailing lists.  For example, we do not even treat header lines
 155  * case insensitively.
 156  */
 157
 158 static int slurp_attr(const char *line, const char *name, char *attr)
 159 {
 160         const char *ends, *ap = strcasestr(line, name);
 161         size_t sz;
 162
 163         if (!ap) {
 164                 *attr = 0;
 165                 return 0;
 166         }
 167         ap += strlen(name);
 168         if (*ap == '"') {
 169                 ap++;
 170                 ends = "\"";
 171         }
 172         else
 173                 ends = "; \t";
 174         sz = strcspn(ap, ends);
 175         memcpy(attr, ap, sz);
 176         attr[sz] = 0;
 177         return 1;
 178 }
 179
 180 static int handle_subcontent_type(char *line)
 181 {
 182         /* We do not want to mess with boundary.  Note that we do not
 183          * handle nested multipart.
 184          */
 185         if (strcasestr(line, "boundary=")) {
 186                 fprintf(stderr, "Not handling nested multipart message.\n");
 187                 exit(1);
 188         }
 189         slurp_attr(line, "charset=", charset);
 190         if (*charset) {
 191                 int i, c;
 192                 for (i = 0; (c = charset[i]) != 0; i++)
 193                         charset[i] = tolower(c);
 194         }
 195         return 0;
 196 }
 197
 198 static int handle_content_type(char *line)
 199 {
 200         *multipart_boundary = 0;
 201         if (slurp_attr(line, "boundary=", multipart_boundary + 2)) {
 202                 memcpy(multipart_boundary, "--", 2);
 203                 multipart_boundary_len = strlen(multipart_boundary);
 204         }
 205         slurp_attr(line, "charset=", charset);
 206         return 0;
 207 }
 208
 209 static int handle_content_transfer_encoding(char *line)
 210 {
 211         if (strcasestr(line, "base64"))
 212                 transfer_encoding = TE_BASE64;
 213         else if (strcasestr(line, "quoted-printable"))
 214                 transfer_encoding = TE_QP;
 215         else
 216                 transfer_encoding = TE_DONTCARE;
 217         return 0;
 218 }
 219
 220 static int is_multipart_boundary(const char *line)
 221 {
 222         return (!memcmp(line, multipart_boundary, multipart_boundary_len));
 223 }
 224
 225 static int eatspace(char *line)
 226 {
 227         int len = strlen(line);
 228         while (len > 0 && isspace(line[len-1]))
 229                 line[--len] = 0;
 230         return len;
 231 }
 232
 233 #define SEEN_FROM 01
 234 #define SEEN_DATE 02
 235 #define SEEN_SUBJECT 04
 236 #define SEEN_BOGUS_UNIX_FROM 010
 237 #define SEEN_PREFIX  020
 238
 239 /* First lines of body can have From:, Date:, and Subject: or empty */
 240 static void handle_inbody_header(int *seen, char *line)
 241 {
 242         if (*seen & SEEN_PREFIX)
 243                 return;
 244         if (isspace(*line)) {
 245                 char *cp;
 246                 for (cp = line + 1; *cp; cp++) {
 247                         if (!isspace(*cp))
 248                                 break;
 249                 }
 250                 if (!*cp)
 251                         return;
 252         }
 253         if (!memcmp(">From", line, 5) && isspace(line[5])) {
 254                 if (!(*seen & SEEN_BOGUS_UNIX_FROM)) {
 255                         *seen |= SEEN_BOGUS_UNIX_FROM;
 256                         return;
 257                 }
 258         }
 259         if (!memcmp("From:", line, 5) && isspace(line[5])) {
 260                 if (!(*seen & SEEN_FROM) && handle_from(line+6)) {
 261                         *seen |= SEEN_FROM;
 262                         return;
 263                 }
 264         }
 265         if (!memcmp("Date:", line, 5) && isspace(line[5])) {
 266                 if (!(*seen & SEEN_DATE)) {
 267                         handle_date(line+6);
 268                         *seen |= SEEN_DATE;
 269                         return;
 270                 }
 271         }
 272         if (!memcmp("Subject:", line, 8) && isspace(line[8])) {
 273                 if (!(*seen & SEEN_SUBJECT)) {
 274                         handle_subject(line+9);
 275                         *seen |= SEEN_SUBJECT;
 276                         return;
 277                 }
 278         }
 279         if (!memcmp("[PATCH]", line, 7) && isspace(line[7])) {
 280                 if (!(*seen & SEEN_SUBJECT)) {
 281                         handle_subject(line);
 282                         *seen |= SEEN_SUBJECT;
 283                         return;
 284                 }
 285         }
 286         *seen |= SEEN_PREFIX;
 287 }
 288
 289 static char *cleanup_subject(char *subject)
 290 {
 291         if (keep_subject)
 292                 return subject;
 293         for (;;) {
 294                 char *p;
 295                 int len, remove;
 296                 switch (*subject) {
 297                 case 'r': case 'R':
 298                         if (!memcmp("e:", subject+1, 2)) {
 299                                 subject +=3;
 300                                 continue;
 301                         }
 302                         break;
 303                 case ' ': case '\t': case ':':
 304                         subject++;
 305                         continue;
 306
 307                 case '[':
 308                         p = strchr(subject, ']');
 309                         if (!p) {
 310                                 subject++;
 311                                 continue;
 312                         }
 313                         len = strlen(p);
 314                         remove = p - subject;
 315                         if (remove <= len *2) {
 316                                 subject = p+1;
 317                                 continue;
 318                         }
 319                         break;
 320                 }
 321                 eatspace(subject);
 322                 return subject;
 323         }
 324 }
 325
 326 static void cleanup_space(char *buf)
 327 {
 328         unsigned char c;
 329         while ((c = *buf) != 0) {
 330                 buf++;
 331                 if (isspace(c)) {
 332                         buf[-1] = ' ';
 333                         c = *buf;
 334                         while (isspace(c)) {
 335                                 int len = strlen(buf);
 336                                 memmove(buf, buf+1, len);
 337                                 c = *buf;
 338                         }
 339                 }
 340         }
 341 }
 342
 343 static void decode_header(char *it);
 344 typedef int (*header_fn_t)(char *);
 345 struct header_def {
 346         const char *name;
 347         header_fn_t func;
 348         int namelen;
 349 };
 350
 351 static void check_header(char *line, struct header_def *header)
 352 {
 353         int i;
 354
 355         if (header[0].namelen <= 0) {
 356                 for (i = 0; header[i].name; i++)
 357                         header[i].namelen = strlen(header[i].name);
 358         }
 359         for (i = 0; header[i].name; i++) {
 360                 int len = header[i].namelen;
 361                 if (!strncasecmp(line, header[i].name, len) &&
 362                     line[len] == ':' && isspace(line[len + 1])) {
 363                         /* Unwrap inline B and Q encoding, and optionally
 364                          * normalize the meta information to utf8.
 365                          */
 366                         decode_header(line + len + 2);
 367                         header[i].func(line + len + 2);
 368                         break;
 369                 }
 370         }
 371 }
 372
 373 static void check_subheader_line(char *line)
 374 {
 375         static struct header_def header[] = {
 376                 { "Content-Type", handle_subcontent_type },
 377                 { "Content-Transfer-Encoding",
 378                   handle_content_transfer_encoding },
 379                 { NULL },
 380         };
 381         check_header(line, header);
 382 }
 383 static void check_header_line(char *line)
 384 {
 385         static struct header_def header[] = {
 386                 { "From", handle_from },
 387                 { "Date", handle_date },
 388                 { "Subject", handle_subject },
 389                 { "Content-Type", handle_content_type },
 390                 { "Content-Transfer-Encoding",
 391                   handle_content_transfer_encoding },
 392                 { NULL },
 393         };
 394         check_header(line, header);
 395 }
 396
 397 static int is_rfc2822_header(char *line)
 398 {
 399         /*
 400          * The section that defines the loosest possible
 401          * field name is "3.6.8 Optional fields".
 402          *
 403          * optional-field = field-name ":" unstructured CRLF
 404          * field-name = 1*ftext
 405          * ftext = %d33-57 / %59-126
 406          */
 407         int ch;
 408         char *cp = line;
 409         while ((ch = *cp++)) {
 410                 if (ch == ':')
 411                         return cp != line;
 412                 if ((33 <= ch && ch <= 57) ||
 413                     (59 <= ch && ch <= 126))
 414                         continue;
 415                 break;
 416         }
 417         return 0;
 418 }
 419
 420 static int read_one_header_line(char *line, int sz, FILE *in)
 421 {
 422         int ofs = 0;
 423         while (ofs < sz) {
 424                 int peek, len;
 425                 if (fgets(line + ofs, sz - ofs, in) == NULL)
 426                         break;
 427                 len = eatspace(line + ofs);
 428                 if ((len == 0) || !is_rfc2822_header(line)) {
 429                         /* Re-add the newline */
 430                         line[ofs + len] = '\n';
 431                         line[ofs + len + 1] = '\0';
 432                         break;
 433                 }
 434                 ofs += len;
 435                 /* Yuck, 2822 header "folding" */
 436                 peek = fgetc(in); ungetc(peek, in);
 437                 if (peek != ' ' && peek != '\t')
 438                         break;
 439         }
 440         /* Count mbox From headers as headers */
 441         if (!ofs && (!memcmp(line, "From ", 5) || !memcmp(line, ">From ", 6)))
 442                 ofs = 1;
 443         return ofs;
 444 }
 445
 446 static int decode_q_segment(char *in, char *ot, char *ep, int rfc2047)
 447 {
 448         int c;
 449         while ((c = *in++) != 0 && (in <= ep)) {
 450                 if (c == '=') {
 451                         int d = *in++;
 452                         if (d == '\n' || !d)
 453                                 break; /* drop trailing newline */
 454                         *ot++ = ((hexval(d) << 4) | hexval(*in++));
 455                         continue;
 456                 }
 457                 if (rfc2047 && c == '_') /* rfc2047 4.2 (2) */
 458                         c = 0x20;
 459                 *ot++ = c;
 460         }
 461         *ot = 0;
 462         return 0;
 463 }
 464
 465 static int decode_b_segment(char *in, char *ot, char *ep)
 466 {
 467         /* Decode in..ep, possibly in-place to ot */
 468         int c, pos = 0, acc = 0;
 469
 470         while ((c = *in++) != 0 && (in <= ep)) {
 471                 if (c == '+')
 472                         c = 62;
 473                 else if (c == '/')
 474                         c = 63;
 475                 else if ('A' <= c && c <= 'Z')
 476                         c -= 'A';
 477                 else if ('a' <= c && c <= 'z')
 478                         c -= 'a' - 26;
 479                 else if ('0' <= c && c <= '9')
 480                         c -= '0' - 52;
 481                 else if (c == '=') {
 482                         /* padding is almost like (c == 0), except we do
 483                          * not output NUL resulting only from it;
 484                          * for now we just trust the data.
 485                          */
 486                         c = 0;
 487                 }
 488                 else
 489                         continue; /* garbage */
 490                 switch (pos++) {
 491                 case 0:
 492                         acc = (c << 2);
 493                         break;
 494                 case 1:
 495                         *ot++ = (acc | (c >> 4));
 496                         acc = (c & 15) << 4;
 497                         break;
 498                 case 2:
 499                         *ot++ = (acc | (c >> 2));
 500                         acc = (c & 3) << 6;
 501                         break;
 502                 case 3:
 503                         *ot++ = (acc | c);
 504                         acc = pos = 0;
 505                         break;
 506                 }
 507         }
 508         *ot = 0;
 509         return 0;
 510 }
 511
 512 static void convert_to_utf8(char *line, char *charset)
 513 {
 514         static char latin_one[] = "latin1";
 515         char *input_charset = *charset ? charset : latin_one;
 516         char *out = reencode_string(line, metainfo_charset, input_charset);
 517
 518         if (!out)
 519                 die("cannot convert from %s to %s\n",
 520                     input_charset, metainfo_charset);
 521         strcpy(line, out);
 522         free(out);
 523 }
 524
 525 static int decode_header_bq(char *it)
 526 {
 527         char *in, *out, *ep, *cp, *sp;
 528         char outbuf[1000];
 529         int rfc2047 = 0;
 530
 531         in = it;
 532         out = outbuf;
 533         while ((ep = strstr(in, "=?")) != NULL) {
 534                 int sz, encoding;
 535                 char charset_q[256], piecebuf[256];
 536                 rfc2047 = 1;
 537
 538                 if (in != ep) {
 539                         sz = ep - in;
 540                         memcpy(out, in, sz);
 541                         out += sz;
 542                         in += sz;
 543                 }
 544                 /* E.g.
 545                  * ep : "=?iso-2022-jp?B?GyR...?= foo"
 546                  * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
 547                  */
 548                 ep += 2;
 549                 cp = strchr(ep, '?');
 550                 if (!cp)
 551                         return rfc2047; /* no munging */
 552                 for (sp = ep; sp < cp; sp++)
 553                         charset_q[sp - ep] = tolower(*sp);
 554                 charset_q[cp - ep] = 0;
 555                 encoding = cp[1];
 556                 if (!encoding || cp[2] != '?')
 557                         return rfc2047; /* no munging */
 558                 ep = strstr(cp + 3, "?=");
 559                 if (!ep)
 560                         return rfc2047; /* no munging */
 561                 switch (tolower(encoding)) {
 562                 default:
 563                         return rfc2047; /* no munging */
 564                 case 'b':
 565                         sz = decode_b_segment(cp + 3, piecebuf, ep);
 566                         break;
 567                 case 'q':
 568                         sz = decode_q_segment(cp + 3, piecebuf, ep, 1);
 569                         break;
 570                 }
 571                 if (sz < 0)
 572                         return rfc2047;
 573                 if (metainfo_charset)
 574                         convert_to_utf8(piecebuf, charset_q);
 575                 strcpy(out, piecebuf);
 576                 out += strlen(out);
 577                 in = ep + 2;
 578         }
 579         strcpy(out, in);
 580         strcpy(it, outbuf);
 581         return rfc2047;
 582 }
 583
 584 static void decode_header(char *it)
 585 {
 586
 587         if (decode_header_bq(it))
 588                 return;
 589         /* otherwise "it" is a straight copy of the input.
 590          * This can be binary guck but there is no charset specified.
 591          */
 592         if (metainfo_charset)
 593                 convert_to_utf8(it, "");
 594 }
 595
 596 static void decode_transfer_encoding(char *line)
 597 {
 598         char *ep;
 599
 600         switch (transfer_encoding) {
 601         case TE_QP:
 602                 ep = line + strlen(line);
 603                 decode_q_segment(line, line, ep, 0);
 604                 break;
 605         case TE_BASE64:
 606                 ep = line + strlen(line);
 607                 decode_b_segment(line, line, ep);
 608                 break;
 609         case TE_DONTCARE:
 610                 break;
 611         }
 612 }
 613
 614 static void handle_info(void)
 615 {
 616         char *sub;
 617
 618         sub = cleanup_subject(subject);
 619         cleanup_space(name);
 620         cleanup_space(date);
 621         cleanup_space(email);
 622         cleanup_space(sub);
 623
 624         fprintf(fout, "Author: %s\nEmail: %s\nSubject: %s\nDate: %s\n\n",
 625                name, email, sub, date);
 626 }
 627
 628 /* We are inside message body and have read line[] already.
 629  * Spit out the commit log.
 630  */
 631 static int handle_commit_msg(int *seen)
 632 {
 633         if (!cmitmsg)
 634                 return 0;
 635         do {
 636                 if (!memcmp("diff -", line, 6) ||
 637                     !memcmp("---", line, 3) ||
 638                     !memcmp("Index: ", line, 7))
 639                         break;
 640                 if ((multipart_boundary[0] && is_multipart_boundary(line))) {
 641                         /* We come here when the first part had only
 642                          * the commit message without any patch.  We
 643                          * pretend we have not seen this line yet, and
 644                          * go back to the loop.
 645                          */
 646                         return 1;
 647                 }
 648
 649                 /* Unwrap transfer encoding and optionally
 650                  * normalize the log message to UTF-8.
 651                  */
 652                 decode_transfer_encoding(line);
 653                 if (metainfo_charset)
 654                         convert_to_utf8(line, charset);
 655
 656                 handle_inbody_header(seen, line);
 657                 if (!(*seen & SEEN_PREFIX))
 658                         continue;
 659
 660                 fputs(line, cmitmsg);
 661         } while (fgets(line, sizeof(line), fin) != NULL);
 662         fclose(cmitmsg);
 663         cmitmsg = NULL;
 664         return 0;
 665 }
 666
 667 /* We have done the commit message and have the first
 668  * line of the patch in line[].
 669  */
 670 static void handle_patch(void)
 671 {
 672         do {
 673                 if (multipart_boundary[0] && is_multipart_boundary(line))
 674                         break;
 675                 /* Only unwrap transfer encoding but otherwise do not
 676                  * do anything.  We do *NOT* want UTF-8 conversion
 677                  * here; we are dealing with the user payload.
 678                  */
 679                 decode_transfer_encoding(line);
 680                 fputs(line, patchfile);
 681                 patch_lines++;
 682         } while (fgets(line, sizeof(line), fin) != NULL);
 683 }
 684
 685 /* multipart boundary and transfer encoding are set up for us, and we
 686  * are at the end of the sub header.  do equivalent of handle_body up
 687  * to the next boundary without closing patchfile --- we will expect
 688  * that the first part to contain commit message and a patch, and
 689  * handle other parts as pure patches.
 690  */
 691 static int handle_multipart_one_part(int *seen)
 692 {
 693         int n = 0;
 694
 695         while (fgets(line, sizeof(line), fin) != NULL) {
 696         again:
 697                 n++;
 698                 if (is_multipart_boundary(line))
 699                         break;
 700                 if (handle_commit_msg(seen))
 701                         goto again;
 702                 handle_patch();
 703                 break;
 704         }
 705         if (n == 0)
 706                 return -1;
 707         return 0;
 708 }
 709
 710 static void handle_multipart_body(void)
 711 {
 712         int seen = 0;
 713         int part_num = 0;
 714
 715         /* Skip up to the first boundary */
 716         while (fgets(line, sizeof(line), fin) != NULL)
 717                 if (is_multipart_boundary(line)) {
 718                         part_num = 1;
 719                         break;
 720                 }
 721         if (!part_num)
 722                 return;
 723         /* We are on boundary line.  Start slurping the subhead. */
 724         while (1) {
 725                 int hdr = read_one_header_line(line, sizeof(line), fin);
 726                 if (!hdr) {
 727                         if (handle_multipart_one_part(&seen) < 0)
 728                                 return;
 729                         /* Reset per part headers */
 730                         transfer_encoding = TE_DONTCARE;
 731                         charset[0] = 0;
 732                 }
 733                 else
 734                         check_subheader_line(line);
 735         }
 736         fclose(patchfile);
 737         if (!patch_lines) {
 738                 fprintf(stderr, "No patch found\n");
 739                 exit(1);
 740         }
 741 }
 742
 743 /* Non multipart message */
 744 static void handle_body(void)
 745 {
 746         int seen = 0;
 747
 748         handle_commit_msg(&seen);
 749         handle_patch();
 750         fclose(patchfile);
 751         if (!patch_lines) {
 752                 fprintf(stderr, "No patch found\n");
 753                 exit(1);
 754         }
 755 }
 756
 757 int mailinfo(FILE *in, FILE *out, int ks, const char *encoding,
 758              const char *msg, const char *patch)
 759 {
 760         keep_subject = ks;
 761         metainfo_charset = encoding;
 762         fin = in;
 763         fout = out;
 764
 765         cmitmsg = fopen(msg, "w");
 766         if (!cmitmsg) {
 767                 perror(msg);
 768                 return -1;
 769         }
 770         patchfile = fopen(patch, "w");
 771         if (!patchfile) {
 772                 perror(patch);
 773                 fclose(cmitmsg);
 774                 return -1;
 775         }
 776         while (1) {
 777                 int hdr = read_one_header_line(line, sizeof(line), fin);
 778                 if (!hdr) {
 779                         if (multipart_boundary[0])
 780                                 handle_multipart_body();
 781                         else
 782                                 handle_body();
 783                         handle_info();
 784                         break;
 785                 }
 786                 check_header_line(line);
 787         }
 788
 789         return 0;
 790 }
 791
 792 static const char mailinfo_usage[] =
 793         "git-mailinfo [-k] [-u | --encoding=<encoding>] msg patch <mail >info";
 794
 795 int cmd_mailinfo(int argc, const char **argv, const char *prefix)
 796 {
 797         const char *def_charset;
 798
 799         /* NEEDSWORK: might want to do the optional .git/ directory
 800          * discovery
 801          */
 802         git_config(git_default_config);
 803
 804         def_charset = (git_commit_encoding ? git_commit_encoding : "utf-8");
 805         metainfo_charset = def_charset;
 806
 807         while (1 < argc && argv[1][0] == '-') {
 808                 if (!strcmp(argv[1], "-k"))
 809                         keep_subject = 1;
 810                 else if (!strcmp(argv[1], "-u"))
 811                         metainfo_charset = def_charset;
 812                 else if (!strcmp(argv[1], "-n"))
 813                         metainfo_charset = NULL;
 814                 else if (!prefixcmp(argv[1], "--encoding="))
 815                         metainfo_charset = argv[1] + 11;
 816                 else
 817                         usage(mailinfo_usage);
 818                 argc--; argv++;
 819         }
 820
 821         if (argc != 3)
 822                 usage(mailinfo_usage);
 823
 824         return !!mailinfo(stdin, stdout, keep_subject, metainfo_charset, argv[1], argv[2]);
 825 }