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 #define _GNU_SOURCE
   6 #include <stdio.h>
   7 #include <stdlib.h>
   8 #include <string.h>
   9 #include <ctype.h>
  10 #ifndef NO_ICONV
  11 #include <iconv.h>
  12 #endif
  13 #include "git-compat-util.h"
  14 #include "cache.h"
  15 #include "builtin.h"
  16
  17 static FILE *cmitmsg, *patchfile, *fin, *fout;
  18
  19 static int keep_subject;
  20 static const char *metainfo_charset;
  21 static char line[1000];
  22 static char date[1000];
  23 static char name[1000];
  24 static char email[1000];
  25 static char subject[1000];
  26
  27 static enum  {
  28         TE_DONTCARE, TE_QP, TE_BASE64,
  29 } transfer_encoding;
  30 static char charset[256];
  31
  32 static char multipart_boundary[1000];
  33 static int multipart_boundary_len;
  34 static int patch_lines;
  35
  36 static char *sanity_check(char *name, char *email)
  37 {
  38         int len = strlen(name);
  39         if (len < 3 || len > 60)
  40                 return email;
  41         if (strchr(name, '@') || strchr(name, '<') || strchr(name, '>'))
  42                 return email;
  43         return name;
  44 }
  45
  46 static int bogus_from(char *line)
  47 {
  48         /* John Doe <johndoe> */
  49         char *bra, *ket, *dst, *cp;
  50
  51         /* This is fallback, so do not bother if we already have an
  52          * e-mail address.
  53          */
  54         if (*email)
  55                 return 0;
  56
  57         bra = strchr(line, '<');
  58         if (!bra)
  59                 return 0;
  60         ket = strchr(bra, '>');
  61         if (!ket)
  62                 return 0;
  63
  64         for (dst = email, cp = bra+1; cp < ket; )
  65                 *dst++ = *cp++;
  66         *dst = 0;
  67         for (cp = line; isspace(*cp); cp++)
  68                 ;
  69         for (bra--; isspace(*bra); bra--)
  70                 *bra = 0;
  71         cp = sanity_check(cp, email);
  72         strcpy(name, cp);
  73         return 1;
  74 }
  75
  76 static int handle_from(char *in_line)
  77 {
  78         char line[1000];
  79         char *at;
  80         char *dst;
  81
  82         strcpy(line, in_line);
  83         at = strchr(line, '@');
  84         if (!at)
  85                 return bogus_from(line);
  86
  87         /*
  88          * If we already have one email, don't take any confusing lines
  89          */
  90         if (*email && strchr(at+1, '@'))
  91                 return 0;
  92
  93         /* Pick up the string around '@', possibly delimited with <>
  94          * pair; that is the email part.  White them out while copying.
  95          */
  96         while (at > line) {
  97                 char c = at[-1];
  98                 if (isspace(c))
  99                         break;
 100                 if (c == '<') {
 101                         at[-1] = ' ';
 102                         break;
 103                 }
 104                 at--;
 105         }
 106         dst = email;
 107         for (;;) {
 108                 unsigned char c = *at;
 109                 if (!c || c == '>' || isspace(c)) {
 110                         if (c == '>')
 111                                 *at = ' ';
 112                         break;
 113                 }
 114                 *at++ = ' ';
 115                 *dst++ = c;
 116         }
 117         *dst++ = 0;
 118
 119         /* The remainder is name.  It could be "John Doe <john.doe@xz>"
 120          * or "john.doe@xz (John Doe)", but we have whited out the
 121          * email part, so trim from both ends, possibly removing
 122          * the () pair at the end.
 123          */
 124         at = line + strlen(line);
 125         while (at > line) {
 126                 unsigned char c = *--at;
 127                 if (!isspace(c)) {
 128                         at[(c == ')') ? 0 : 1] = 0;
 129                         break;
 130                 }
 131         }
 132
 133         at = line;
 134         for (;;) {
 135                 unsigned char c = *at;
 136                 if (!c || !isspace(c)) {
 137                         if (c == '(')
 138                                 at++;
 139                         break;
 140                 }
 141                 at++;
 142         }
 143         at = sanity_check(at, email);
 144         strcpy(name, at);
 145         return 1;
 146 }
 147
 148 static int handle_date(char *line)
 149 {
 150         strcpy(date, line);
 151         return 0;
 152 }
 153
 154 static int handle_subject(char *line)
 155 {
 156         strcpy(subject, line);
 157         return 0;
 158 }
 159
 160 /* NOTE NOTE NOTE.  We do not claim we do full MIME.  We just attempt
 161  * to have enough heuristics to grok MIME encoded patches often found
 162  * on our mailing lists.  For example, we do not even treat header lines
 163  * case insensitively.
 164  */
 165
 166 static int slurp_attr(const char *line, const char *name, char *attr)
 167 {
 168         const char *ends, *ap = strcasestr(line, name);
 169         size_t sz;
 170
 171         if (!ap) {
 172                 *attr = 0;
 173                 return 0;
 174         }
 175         ap += strlen(name);
 176         if (*ap == '"') {
 177                 ap++;
 178                 ends = "\"";
 179         }
 180         else
 181                 ends = "; \t";
 182         sz = strcspn(ap, ends);
 183         memcpy(attr, ap, sz);
 184         attr[sz] = 0;
 185         return 1;
 186 }
 187
 188 static int handle_subcontent_type(char *line)
 189 {
 190         /* We do not want to mess with boundary.  Note that we do not
 191          * handle nested multipart.
 192          */
 193         if (strcasestr(line, "boundary=")) {
 194                 fprintf(stderr, "Not handling nested multipart message.\n");
 195                 exit(1);
 196         }
 197         slurp_attr(line, "charset=", charset);
 198         if (*charset) {
 199                 int i, c;
 200                 for (i = 0; (c = charset[i]) != 0; i++)
 201                         charset[i] = tolower(c);
 202         }
 203         return 0;
 204 }
 205
 206 static int handle_content_type(char *line)
 207 {
 208         *multipart_boundary = 0;
 209         if (slurp_attr(line, "boundary=", multipart_boundary + 2)) {
 210                 memcpy(multipart_boundary, "--", 2);
 211                 multipart_boundary_len = strlen(multipart_boundary);
 212         }
 213         slurp_attr(line, "charset=", charset);
 214         return 0;
 215 }
 216
 217 static int handle_content_transfer_encoding(char *line)
 218 {
 219         if (strcasestr(line, "base64"))
 220                 transfer_encoding = TE_BASE64;
 221         else if (strcasestr(line, "quoted-printable"))
 222                 transfer_encoding = TE_QP;
 223         else
 224                 transfer_encoding = TE_DONTCARE;
 225         return 0;
 226 }
 227
 228 static int is_multipart_boundary(const char *line)
 229 {
 230         return (!memcmp(line, multipart_boundary, multipart_boundary_len));
 231 }
 232
 233 static int eatspace(char *line)
 234 {
 235         int len = strlen(line);
 236         while (len > 0 && isspace(line[len-1]))
 237                 line[--len] = 0;
 238         return len;
 239 }
 240
 241 #define SEEN_FROM 01
 242 #define SEEN_DATE 02
 243 #define SEEN_SUBJECT 04
 244 #define SEEN_BOGUS_UNIX_FROM 010
 245 #define SEEN_PREFIX  020
 246
 247 /* First lines of body can have From:, Date:, and Subject: or empty */
 248 static void handle_inbody_header(int *seen, char *line)
 249 {
 250         if (*seen & SEEN_PREFIX)
 251                 return;
 252         if (isspace(*line)) {
 253                 char *cp;
 254                 for (cp = line + 1; *cp; cp++) {
 255                         if (!isspace(*cp))
 256                                 break;
 257                 }
 258                 if (!*cp)
 259                         return;
 260         }
 261         if (!memcmp(">From", line, 5) && isspace(line[5])) {
 262                 if (!(*seen & SEEN_BOGUS_UNIX_FROM)) {
 263                         *seen |= SEEN_BOGUS_UNIX_FROM;
 264                         return;
 265                 }
 266         }
 267         if (!memcmp("From:", line, 5) && isspace(line[5])) {
 268                 if (!(*seen & SEEN_FROM) && handle_from(line+6)) {
 269                         *seen |= SEEN_FROM;
 270                         return;
 271                 }
 272         }
 273         if (!memcmp("Date:", line, 5) && isspace(line[5])) {
 274                 if (!(*seen & SEEN_DATE)) {
 275                         handle_date(line+6);
 276                         *seen |= SEEN_DATE;
 277                         return;
 278                 }
 279         }
 280         if (!memcmp("Subject:", line, 8) && isspace(line[8])) {
 281                 if (!(*seen & SEEN_SUBJECT)) {
 282                         handle_subject(line+9);
 283                         *seen |= SEEN_SUBJECT;
 284                         return;
 285                 }
 286         }
 287         if (!memcmp("[PATCH]", line, 7) && isspace(line[7])) {
 288                 if (!(*seen & SEEN_SUBJECT)) {
 289                         handle_subject(line);
 290                         *seen |= SEEN_SUBJECT;
 291                         return;
 292                 }
 293         }
 294         *seen |= SEEN_PREFIX;
 295 }
 296
 297 static char *cleanup_subject(char *subject)
 298 {
 299         if (keep_subject)
 300                 return subject;
 301         for (;;) {
 302                 char *p;
 303                 int len, remove;
 304                 switch (*subject) {
 305                 case 'r': case 'R':
 306                         if (!memcmp("e:", subject+1, 2)) {
 307                                 subject +=3;
 308                                 continue;
 309                         }
 310                         break;
 311                 case ' ': case '\t': case ':':
 312                         subject++;
 313                         continue;
 314
 315                 case '[':
 316                         p = strchr(subject, ']');
 317                         if (!p) {
 318                                 subject++;
 319                                 continue;
 320                         }
 321                         len = strlen(p);
 322                         remove = p - subject;
 323                         if (remove <= len *2) {
 324                                 subject = p+1;
 325                                 continue;
 326                         }
 327                         break;
 328                 }
 329                 eatspace(subject);
 330                 return subject;
 331         }
 332 }
 333
 334 static void cleanup_space(char *buf)
 335 {
 336         unsigned char c;
 337         while ((c = *buf) != 0) {
 338                 buf++;
 339                 if (isspace(c)) {
 340                         buf[-1] = ' ';
 341                         c = *buf;
 342                         while (isspace(c)) {
 343                                 int len = strlen(buf);
 344                                 memmove(buf, buf+1, len);
 345                                 c = *buf;
 346                         }
 347                 }
 348         }
 349 }
 350
 351 static void decode_header(char *it);
 352 typedef int (*header_fn_t)(char *);
 353 struct header_def {
 354         const char *name;
 355         header_fn_t func;
 356         int namelen;
 357 };
 358
 359 static void check_header(char *line, struct header_def *header)
 360 {
 361         int i;
 362
 363         if (header[0].namelen <= 0) {
 364                 for (i = 0; header[i].name; i++)
 365                         header[i].namelen = strlen(header[i].name);
 366         }
 367         for (i = 0; header[i].name; i++) {
 368                 int len = header[i].namelen;
 369                 if (!strncasecmp(line, header[i].name, len) &&
 370                     line[len] == ':' && isspace(line[len + 1])) {
 371                         /* Unwrap inline B and Q encoding, and optionally
 372                          * normalize the meta information to utf8.
 373                          */
 374                         decode_header(line + len + 2);
 375                         header[i].func(line + len + 2);
 376                         break;
 377                 }
 378         }
 379 }
 380
 381 static void check_subheader_line(char *line)
 382 {
 383         static struct header_def header[] = {
 384                 { "Content-Type", handle_subcontent_type },
 385                 { "Content-Transfer-Encoding",
 386                   handle_content_transfer_encoding },
 387                 { NULL },
 388         };
 389         check_header(line, header);
 390 }
 391 static void check_header_line(char *line)
 392 {
 393         static struct header_def header[] = {
 394                 { "From", handle_from },
 395                 { "Date", handle_date },
 396                 { "Subject", handle_subject },
 397                 { "Content-Type", handle_content_type },
 398                 { "Content-Transfer-Encoding",
 399                   handle_content_transfer_encoding },
 400                 { NULL },
 401         };
 402         check_header(line, header);
 403 }
 404
 405 static int is_rfc2822_header(char *line)
 406 {
 407         /*
 408          * The section that defines the loosest possible
 409          * field name is "3.6.8 Optional fields".
 410          *
 411          * optional-field = field-name ":" unstructured CRLF
 412          * field-name = 1*ftext
 413          * ftext = %d33-57 / %59-126
 414          */
 415         int ch;
 416         char *cp = line;
 417         while ((ch = *cp++)) {
 418                 if (ch == ':')
 419                         return cp != line;
 420                 if ((33 <= ch && ch <= 57) ||
 421                     (59 <= ch && ch <= 126))
 422                         continue;
 423                 break;
 424         }
 425         return 0;
 426 }
 427
 428 static int read_one_header_line(char *line, int sz, FILE *in)
 429 {
 430         int ofs = 0;
 431         while (ofs < sz) {
 432                 int peek, len;
 433                 if (fgets(line + ofs, sz - ofs, in) == NULL)
 434                         break;
 435                 len = eatspace(line + ofs);
 436                 if ((len == 0) || !is_rfc2822_header(line)) {
 437                         /* Re-add the newline */
 438                         line[ofs + len] = '\n';
 439                         line[ofs + len + 1] = '\0';
 440                         break;
 441                 }
 442                 ofs += len;
 443                 /* Yuck, 2822 header "folding" */
 444                 peek = fgetc(in); ungetc(peek, in);
 445                 if (peek != ' ' && peek != '\t')
 446                         break;
 447         }
 448         /* Count mbox From headers as headers */
 449         if (!ofs && (!memcmp(line, "From ", 5) || !memcmp(line, ">From ", 6)))
 450                 ofs = 1;
 451         return ofs;
 452 }
 453
 454 static int decode_q_segment(char *in, char *ot, char *ep, int rfc2047)
 455 {
 456         int c;
 457         while ((c = *in++) != 0 && (in <= ep)) {
 458                 if (c == '=') {
 459                         int d = *in++;
 460                         if (d == '\n' || !d)
 461                                 break; /* drop trailing newline */
 462                         *ot++ = ((hexval(d) << 4) | hexval(*in++));
 463                         continue;
 464                 }
 465                 if (rfc2047 && c == '_') /* rfc2047 4.2 (2) */
 466                         c = 0x20;
 467                 *ot++ = c;
 468         }
 469         *ot = 0;
 470         return 0;
 471 }
 472
 473 static int decode_b_segment(char *in, char *ot, char *ep)
 474 {
 475         /* Decode in..ep, possibly in-place to ot */
 476         int c, pos = 0, acc = 0;
 477
 478         while ((c = *in++) != 0 && (in <= ep)) {
 479                 if (c == '+')
 480                         c = 62;
 481                 else if (c == '/')
 482                         c = 63;
 483                 else if ('A' <= c && c <= 'Z')
 484                         c -= 'A';
 485                 else if ('a' <= c && c <= 'z')
 486                         c -= 'a' - 26;
 487                 else if ('0' <= c && c <= '9')
 488                         c -= '0' - 52;
 489                 else if (c == '=') {
 490                         /* padding is almost like (c == 0), except we do
 491                          * not output NUL resulting only from it;
 492                          * for now we just trust the data.
 493                          */
 494                         c = 0;
 495                 }
 496                 else
 497                         continue; /* garbage */
 498                 switch (pos++) {
 499                 case 0:
 500                         acc = (c << 2);
 501                         break;
 502                 case 1:
 503                         *ot++ = (acc | (c >> 4));
 504                         acc = (c & 15) << 4;
 505                         break;
 506                 case 2:
 507                         *ot++ = (acc | (c >> 2));
 508                         acc = (c & 3) << 6;
 509                         break;
 510                 case 3:
 511                         *ot++ = (acc | c);
 512                         acc = pos = 0;
 513                         break;
 514                 }
 515         }
 516         *ot = 0;
 517         return 0;
 518 }
 519
 520 static void convert_to_utf8(char *line, char *charset)
 521 {
 522 #ifndef NO_ICONV
 523         char *in, *out;
 524         size_t insize, outsize, nrc;
 525         char outbuf[4096]; /* cheat */
 526         static char latin_one[] = "latin1";
 527         char *input_charset = *charset ? charset : latin_one;
 528         iconv_t conv = iconv_open(metainfo_charset, input_charset);
 529
 530         if (conv == (iconv_t) -1) {
 531                 static int warned_latin1_once = 0;
 532                 if (input_charset != latin_one) {
 533                         fprintf(stderr, "cannot convert from %s to %s\n",
 534                                 input_charset, metainfo_charset);
 535                         *charset = 0;
 536                 }
 537                 else if (!warned_latin1_once) {
 538                         warned_latin1_once = 1;
 539                         fprintf(stderr, "tried to convert from %s to %s, "
 540                                 "but your iconv does not work with it.\n",
 541                                 input_charset, metainfo_charset);
 542                 }
 543                 return;
 544         }
 545         in = line;
 546         insize = strlen(in);
 547         out = outbuf;
 548         outsize = sizeof(outbuf);
 549         nrc = iconv(conv, &in, &insize, &out, &outsize);
 550         iconv_close(conv);
 551         if (nrc == (size_t) -1)
 552                 return;
 553         *out = 0;
 554         strcpy(line, outbuf);
 555 #endif
 556 }
 557
 558 static int decode_header_bq(char *it)
 559 {
 560         char *in, *out, *ep, *cp, *sp;
 561         char outbuf[1000];
 562         int rfc2047 = 0;
 563
 564         in = it;
 565         out = outbuf;
 566         while ((ep = strstr(in, "=?")) != NULL) {
 567                 int sz, encoding;
 568                 char charset_q[256], piecebuf[256];
 569                 rfc2047 = 1;
 570
 571                 if (in != ep) {
 572                         sz = ep - in;
 573                         memcpy(out, in, sz);
 574                         out += sz;
 575                         in += sz;
 576                 }
 577                 /* E.g.
 578                  * ep : "=?iso-2022-jp?B?GyR...?= foo"
 579                  * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
 580                  */
 581                 ep += 2;
 582                 cp = strchr(ep, '?');
 583                 if (!cp)
 584                         return rfc2047; /* no munging */
 585                 for (sp = ep; sp < cp; sp++)
 586                         charset_q[sp - ep] = tolower(*sp);
 587                 charset_q[cp - ep] = 0;
 588                 encoding = cp[1];
 589                 if (!encoding || cp[2] != '?')
 590                         return rfc2047; /* no munging */
 591                 ep = strstr(cp + 3, "?=");
 592                 if (!ep)
 593                         return rfc2047; /* no munging */
 594                 switch (tolower(encoding)) {
 595                 default:
 596                         return rfc2047; /* no munging */
 597                 case 'b':
 598                         sz = decode_b_segment(cp + 3, piecebuf, ep);
 599                         break;
 600                 case 'q':
 601                         sz = decode_q_segment(cp + 3, piecebuf, ep, 1);
 602                         break;
 603                 }
 604                 if (sz < 0)
 605                         return rfc2047;
 606                 if (metainfo_charset)
 607                         convert_to_utf8(piecebuf, charset_q);
 608                 strcpy(out, piecebuf);
 609                 out += strlen(out);
 610                 in = ep + 2;
 611         }
 612         strcpy(out, in);
 613         strcpy(it, outbuf);
 614         return rfc2047;
 615 }
 616
 617 static void decode_header(char *it)
 618 {
 619
 620         if (decode_header_bq(it))
 621                 return;
 622         /* otherwise "it" is a straight copy of the input.
 623          * This can be binary guck but there is no charset specified.
 624          */
 625         if (metainfo_charset)
 626                 convert_to_utf8(it, "");
 627 }
 628
 629 static void decode_transfer_encoding(char *line)
 630 {
 631         char *ep;
 632
 633         switch (transfer_encoding) {
 634         case TE_QP:
 635                 ep = line + strlen(line);
 636                 decode_q_segment(line, line, ep, 0);
 637                 break;
 638         case TE_BASE64:
 639                 ep = line + strlen(line);
 640                 decode_b_segment(line, line, ep);
 641                 break;
 642         case TE_DONTCARE:
 643                 break;
 644         }
 645 }
 646
 647 static void handle_info(void)
 648 {
 649         char *sub;
 650
 651         sub = cleanup_subject(subject);
 652         cleanup_space(name);
 653         cleanup_space(date);
 654         cleanup_space(email);
 655         cleanup_space(sub);
 656
 657         fprintf(fout, "Author: %s\nEmail: %s\nSubject: %s\nDate: %s\n\n",
 658                name, email, sub, date);
 659 }
 660
 661 /* We are inside message body and have read line[] already.
 662  * Spit out the commit log.
 663  */
 664 static int handle_commit_msg(int *seen)
 665 {
 666         if (!cmitmsg)
 667                 return 0;
 668         do {
 669                 if (!memcmp("diff -", line, 6) ||
 670                     !memcmp("---", line, 3) ||
 671                     !memcmp("Index: ", line, 7))
 672                         break;
 673                 if ((multipart_boundary[0] && is_multipart_boundary(line))) {
 674                         /* We come here when the first part had only
 675                          * the commit message without any patch.  We
 676                          * pretend we have not seen this line yet, and
 677                          * go back to the loop.
 678                          */
 679                         return 1;
 680                 }
 681
 682                 /* Unwrap transfer encoding and optionally
 683                  * normalize the log message to UTF-8.
 684                  */
 685                 decode_transfer_encoding(line);
 686                 if (metainfo_charset)
 687                         convert_to_utf8(line, charset);
 688
 689                 handle_inbody_header(seen, line);
 690                 if (!(*seen & SEEN_PREFIX))
 691                         continue;
 692
 693                 fputs(line, cmitmsg);
 694         } while (fgets(line, sizeof(line), fin) != NULL);
 695         fclose(cmitmsg);
 696         cmitmsg = NULL;
 697         return 0;
 698 }
 699
 700 /* We have done the commit message and have the first
 701  * line of the patch in line[].
 702  */
 703 static void handle_patch(void)
 704 {
 705         do {
 706                 if (multipart_boundary[0] && is_multipart_boundary(line))
 707                         break;
 708                 /* Only unwrap transfer encoding but otherwise do not
 709                  * do anything.  We do *NOT* want UTF-8 conversion
 710                  * here; we are dealing with the user payload.
 711                  */
 712                 decode_transfer_encoding(line);
 713                 fputs(line, patchfile);
 714                 patch_lines++;
 715         } while (fgets(line, sizeof(line), fin) != NULL);
 716 }
 717
 718 /* multipart boundary and transfer encoding are set up for us, and we
 719  * are at the end of the sub header.  do equivalent of handle_body up
 720  * to the next boundary without closing patchfile --- we will expect
 721  * that the first part to contain commit message and a patch, and
 722  * handle other parts as pure patches.
 723  */
 724 static int handle_multipart_one_part(int *seen)
 725 {
 726         int n = 0;
 727
 728         while (fgets(line, sizeof(line), fin) != NULL) {
 729         again:
 730                 n++;
 731                 if (is_multipart_boundary(line))
 732                         break;
 733                 if (handle_commit_msg(seen))
 734                         goto again;
 735                 handle_patch();
 736                 break;
 737         }
 738         if (n == 0)
 739                 return -1;
 740         return 0;
 741 }
 742
 743 static void handle_multipart_body(void)
 744 {
 745         int seen = 0;
 746         int part_num = 0;
 747
 748         /* Skip up to the first boundary */
 749         while (fgets(line, sizeof(line), fin) != NULL)
 750                 if (is_multipart_boundary(line)) {
 751                         part_num = 1;
 752                         break;
 753                 }
 754         if (!part_num)
 755                 return;
 756         /* We are on boundary line.  Start slurping the subhead. */
 757         while (1) {
 758                 int hdr = read_one_header_line(line, sizeof(line), fin);
 759                 if (!hdr) {
 760                         if (handle_multipart_one_part(&seen) < 0)
 761                                 return;
 762                         /* Reset per part headers */
 763                         transfer_encoding = TE_DONTCARE;
 764                         charset[0] = 0;
 765                 }
 766                 else
 767                         check_subheader_line(line);
 768         }
 769         fclose(patchfile);
 770         if (!patch_lines) {
 771                 fprintf(stderr, "No patch found\n");
 772                 exit(1);
 773         }
 774 }
 775
 776 /* Non multipart message */
 777 static void handle_body(void)
 778 {
 779         int seen = 0;
 780
 781         handle_commit_msg(&seen);
 782         handle_patch();
 783         fclose(patchfile);
 784         if (!patch_lines) {
 785                 fprintf(stderr, "No patch found\n");
 786                 exit(1);
 787         }
 788 }
 789
 790 int mailinfo(FILE *in, FILE *out, int ks, const char *encoding,
 791              const char *msg, const char *patch)
 792 {
 793         keep_subject = ks;
 794         metainfo_charset = encoding;
 795         fin = in;
 796         fout = out;
 797
 798         cmitmsg = fopen(msg, "w");
 799         if (!cmitmsg) {
 800                 perror(msg);
 801                 return -1;
 802         }
 803         patchfile = fopen(patch, "w");
 804         if (!patchfile) {
 805                 perror(patch);
 806                 fclose(cmitmsg);
 807                 return -1;
 808         }
 809         while (1) {
 810                 int hdr = read_one_header_line(line, sizeof(line), fin);
 811                 if (!hdr) {
 812                         if (multipart_boundary[0])
 813                                 handle_multipart_body();
 814                         else
 815                                 handle_body();
 816                         handle_info();
 817                         break;
 818                 }
 819                 check_header_line(line);
 820         }
 821
 822         return 0;
 823 }
 824
 825 static const char mailinfo_usage[] =
 826         "git-mailinfo [-k] [-u | --encoding=<encoding>] msg patch <mail >info";
 827
 828 int cmd_mailinfo(int argc, const char **argv, const char *prefix)
 829 {
 830         /* NEEDSWORK: might want to do the optional .git/ directory
 831          * discovery
 832          */
 833         git_config(git_default_config);
 834
 835         while (1 < argc && argv[1][0] == '-') {
 836                 if (!strcmp(argv[1], "-k"))
 837                         keep_subject = 1;
 838                 else if (!strcmp(argv[1], "-u"))
 839                         metainfo_charset = git_commit_encoding;
 840                 else if (!strncmp(argv[1], "--encoding=", 11))
 841                         metainfo_charset = argv[1] + 11;
 842                 else
 843                         usage(mailinfo_usage);
 844                 argc--; argv++;
 845         }
 846
 847         if (argc != 3)
 848                 usage(mailinfo_usage);
 849
 850         return !!mailinfo(stdin, stdout, keep_subject, metainfo_charset, argv[1], argv[2]);
 851 }