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 name[1000];
  15 static char email[1000];
  16
  17 static enum  {
  18         TE_DONTCARE, TE_QP, TE_BASE64,
  19 } transfer_encoding;
  20 static enum  {
  21         TYPE_TEXT, TYPE_OTHER,
  22 } message_type;
  23
  24 static char charset[256];
  25 static int patch_lines;
  26 static char **p_hdr_data, **s_hdr_data;
  27
  28 #define MAX_HDR_PARSED 10
  29 #define MAX_BOUNDARIES 5
  30
  31 static char *sanity_check(char *name, char *email)
  32 {
  33         int len = strlen(name);
  34         if (len < 3 || len > 60)
  35                 return email;
  36         if (strchr(name, '@') || strchr(name, '<') || strchr(name, '>'))
  37                 return email;
  38         return name;
  39 }
  40
  41 static int bogus_from(char *line)
  42 {
  43         /* John Doe <johndoe> */
  44         char *bra, *ket, *dst, *cp;
  45
  46         /* This is fallback, so do not bother if we already have an
  47          * e-mail address.
  48          */
  49         if (*email)
  50                 return 0;
  51
  52         bra = strchr(line, '<');
  53         if (!bra)
  54                 return 0;
  55         ket = strchr(bra, '>');
  56         if (!ket)
  57                 return 0;
  58
  59         for (dst = email, cp = bra+1; cp < ket; )
  60                 *dst++ = *cp++;
  61         *dst = 0;
  62         for (cp = line; isspace(*cp); cp++)
  63                 ;
  64         for (bra--; isspace(*bra); bra--)
  65                 *bra = 0;
  66         cp = sanity_check(cp, email);
  67         strcpy(name, cp);
  68         return 1;
  69 }
  70
  71 static int handle_from(char *in_line)
  72 {
  73         char line[1000];
  74         char *at;
  75         char *dst;
  76
  77         strcpy(line, in_line);
  78         at = strchr(line, '@');
  79         if (!at)
  80                 return bogus_from(line);
  81
  82         /*
  83          * If we already have one email, don't take any confusing lines
  84          */
  85         if (*email && strchr(at+1, '@'))
  86                 return 0;
  87
  88         /* Pick up the string around '@', possibly delimited with <>
  89          * pair; that is the email part.  White them out while copying.
  90          */
  91         while (at > line) {
  92                 char c = at[-1];
  93                 if (isspace(c))
  94                         break;
  95                 if (c == '<') {
  96                         at[-1] = ' ';
  97                         break;
  98                 }
  99                 at--;
 100         }
 101         dst = email;
 102         for (;;) {
 103                 unsigned char c = *at;
 104                 if (!c || c == '>' || isspace(c)) {
 105                         if (c == '>')
 106                                 *at = ' ';
 107                         break;
 108                 }
 109                 *at++ = ' ';
 110                 *dst++ = c;
 111         }
 112         *dst++ = 0;
 113
 114         /* The remainder is name.  It could be "John Doe <john.doe@xz>"
 115          * or "john.doe@xz (John Doe)", but we have whited out the
 116          * email part, so trim from both ends, possibly removing
 117          * the () pair at the end.
 118          */
 119         at = line + strlen(line);
 120         while (at > line) {
 121                 unsigned char c = *--at;
 122                 if (!isspace(c)) {
 123                         at[(c == ')') ? 0 : 1] = 0;
 124                         break;
 125                 }
 126         }
 127
 128         at = line;
 129         for (;;) {
 130                 unsigned char c = *at;
 131                 if (!c || !isspace(c)) {
 132                         if (c == '(')
 133                                 at++;
 134                         break;
 135                 }
 136                 at++;
 137         }
 138         at = sanity_check(at, email);
 139         strcpy(name, at);
 140         return 1;
 141 }
 142
 143 static int handle_header(char *line, char *data, int ofs)
 144 {
 145         if (!line || !data)
 146                 return 1;
 147
 148         strcpy(data, line+ofs);
 149
 150         return 0;
 151 }
 152
 153 /* NOTE NOTE NOTE.  We do not claim we do full MIME.  We just attempt
 154  * to have enough heuristics to grok MIME encoded patches often found
 155  * on our mailing lists.  For example, we do not even treat header lines
 156  * case insensitively.
 157  */
 158
 159 static int slurp_attr(const char *line, const char *name, char *attr)
 160 {
 161         const char *ends, *ap = strcasestr(line, name);
 162         size_t sz;
 163
 164         if (!ap) {
 165                 *attr = 0;
 166                 return 0;
 167         }
 168         ap += strlen(name);
 169         if (*ap == '"') {
 170                 ap++;
 171                 ends = "\"";
 172         }
 173         else
 174                 ends = "; \t";
 175         sz = strcspn(ap, ends);
 176         memcpy(attr, ap, sz);
 177         attr[sz] = 0;
 178         return 1;
 179 }
 180
 181 struct content_type {
 182         char *boundary;
 183         int boundary_len;
 184 };
 185
 186 static struct content_type content[MAX_BOUNDARIES];
 187
 188 static struct content_type *content_top = content;
 189
 190 static int handle_content_type(char *line)
 191 {
 192         char boundary[256];
 193
 194         if (strcasestr(line, "text/") == NULL)
 195                  message_type = TYPE_OTHER;
 196         if (slurp_attr(line, "boundary=", boundary + 2)) {
 197                 memcpy(boundary, "--", 2);
 198                 if (content_top++ >= &content[MAX_BOUNDARIES]) {
 199                         fprintf(stderr, "Too many boundaries to handle\n");
 200                         exit(1);
 201                 }
 202                 content_top->boundary_len = strlen(boundary);
 203                 content_top->boundary = xmalloc(content_top->boundary_len+1);
 204                 strcpy(content_top->boundary, boundary);
 205         }
 206         if (slurp_attr(line, "charset=", charset)) {
 207                 int i, c;
 208                 for (i = 0; (c = charset[i]) != 0; i++)
 209                         charset[i] = tolower(c);
 210         }
 211         return 0;
 212 }
 213
 214 static int handle_content_transfer_encoding(char *line)
 215 {
 216         if (strcasestr(line, "base64"))
 217                 transfer_encoding = TE_BASE64;
 218         else if (strcasestr(line, "quoted-printable"))
 219                 transfer_encoding = TE_QP;
 220         else
 221                 transfer_encoding = TE_DONTCARE;
 222         return 0;
 223 }
 224
 225 static int is_multipart_boundary(const char *line)
 226 {
 227         return (!memcmp(line, content_top->boundary, content_top->boundary_len));
 228 }
 229
 230 static int eatspace(char *line)
 231 {
 232         int len = strlen(line);
 233         while (len > 0 && isspace(line[len-1]))
 234                 line[--len] = 0;
 235         return len;
 236 }
 237
 238 static char *cleanup_subject(char *subject)
 239 {
 240         if (keep_subject)
 241                 return subject;
 242         for (;;) {
 243                 char *p;
 244                 int len, remove;
 245                 switch (*subject) {
 246                 case 'r': case 'R':
 247                         if (!memcmp("e:", subject+1, 2)) {
 248                                 subject += 3;
 249                                 continue;
 250                         }
 251                         break;
 252                 case ' ': case '\t': case ':':
 253                         subject++;
 254                         continue;
 255
 256                 case '[':
 257                         p = strchr(subject, ']');
 258                         if (!p) {
 259                                 subject++;
 260                                 continue;
 261                         }
 262                         len = strlen(p);
 263                         remove = p - subject;
 264                         if (remove <= len *2) {
 265                                 subject = p+1;
 266                                 continue;
 267                         }
 268                         break;
 269                 }
 270                 eatspace(subject);
 271                 return subject;
 272         }
 273 }
 274
 275 static void cleanup_space(char *buf)
 276 {
 277         unsigned char c;
 278         while ((c = *buf) != 0) {
 279                 buf++;
 280                 if (isspace(c)) {
 281                         buf[-1] = ' ';
 282                         c = *buf;
 283                         while (isspace(c)) {
 284                                 int len = strlen(buf);
 285                                 memmove(buf, buf+1, len);
 286                                 c = *buf;
 287                         }
 288                 }
 289         }
 290 }
 291
 292 static void decode_header(char *it);
 293 static char *header[MAX_HDR_PARSED] = {
 294         "From","Subject","Date",
 295 };
 296
 297 static int check_header(char *line, char **hdr_data, int overwrite)
 298 {
 299         int i;
 300
 301         /* search for the interesting parts */
 302         for (i = 0; header[i]; i++) {
 303                 int len = strlen(header[i]);
 304                 if ((!hdr_data[i] || overwrite) &&
 305                     !strncasecmp(line, header[i], len) &&
 306                     line[len] == ':' && isspace(line[len + 1])) {
 307                         /* Unwrap inline B and Q encoding, and optionally
 308                          * normalize the meta information to utf8.
 309                          */
 310                         decode_header(line + len + 2);
 311                         hdr_data[i] = xmalloc(1000 * sizeof(char));
 312                         if (! handle_header(line, hdr_data[i], len + 2)) {
 313                                 return 1;
 314                         }
 315                 }
 316         }
 317
 318         /* Content stuff */
 319         if (!strncasecmp(line, "Content-Type", 12) &&
 320                 line[12] == ':' && isspace(line[12 + 1])) {
 321                 decode_header(line + 12 + 2);
 322                 if (! handle_content_type(line)) {
 323                         return 1;
 324                 }
 325         }
 326         if (!strncasecmp(line, "Content-Transfer-Encoding", 25) &&
 327                 line[25] == ':' && isspace(line[25 + 1])) {
 328                 decode_header(line + 25 + 2);
 329                 if (! handle_content_transfer_encoding(line)) {
 330                         return 1;
 331                 }
 332         }
 333
 334         /* for inbody stuff */
 335         if (!memcmp(">From", line, 5) && isspace(line[5]))
 336                 return 1;
 337         if (!memcmp("[PATCH]", line, 7) && isspace(line[7])) {
 338                 for (i = 0; header[i]; i++) {
 339                         if (!memcmp("Subject: ", header[i], 9)) {
 340                                 if (! handle_header(line, hdr_data[i], 0)) {
 341                                         return 1;
 342                                 }
 343                         }
 344                 }
 345         }
 346
 347         /* no match */
 348         return 0;
 349 }
 350
 351 static int is_rfc2822_header(char *line)
 352 {
 353         /*
 354          * The section that defines the loosest possible
 355          * field name is "3.6.8 Optional fields".
 356          *
 357          * optional-field = field-name ":" unstructured CRLF
 358          * field-name = 1*ftext
 359          * ftext = %d33-57 / %59-126
 360          */
 361         int ch;
 362         char *cp = line;
 363
 364         /* Count mbox From headers as headers */
 365         if (!memcmp(line, "From ", 5) || !memcmp(line, ">From ", 6))
 366                 return 1;
 367
 368         while ((ch = *cp++)) {
 369                 if (ch == ':')
 370                         return cp != line;
 371                 if ((33 <= ch && ch <= 57) ||
 372                     (59 <= ch && ch <= 126))
 373                         continue;
 374                 break;
 375         }
 376         return 0;
 377 }
 378
 379 /*
 380  * sz is size of 'line' buffer in bytes.  Must be reasonably
 381  * long enough to hold one physical real-world e-mail line.
 382  */
 383 static int read_one_header_line(char *line, int sz, FILE *in)
 384 {
 385         int len;
 386
 387         /*
 388          * We will read at most (sz-1) bytes and then potentially
 389          * re-add NUL after it.  Accessing line[sz] after this is safe
 390          * and we can allow len to grow up to and including sz.
 391          */
 392         sz--;
 393
 394         /* Get the first part of the line. */
 395         if (!fgets(line, sz, in))
 396                 return 0;
 397
 398         /*
 399          * Is it an empty line or not a valid rfc2822 header?
 400          * If so, stop here, and return false ("not a header")
 401          */
 402         len = eatspace(line);
 403         if (!len || !is_rfc2822_header(line)) {
 404                 /* Re-add the newline */
 405                 line[len] = '\n';
 406                 line[len + 1] = '\0';
 407                 return 0;
 408         }
 409
 410         /*
 411          * Now we need to eat all the continuation lines..
 412          * Yuck, 2822 header "folding"
 413          */
 414         for (;;) {
 415                 int peek, addlen;
 416                 static char continuation[1000];
 417
 418                 peek = fgetc(in); ungetc(peek, in);
 419                 if (peek != ' ' && peek != '\t')
 420                         break;
 421                 if (!fgets(continuation, sizeof(continuation), in))
 422                         break;
 423                 addlen = eatspace(continuation);
 424                 if (len < sz - 1) {
 425                         if (addlen >= sz - len)
 426                                 addlen = sz - len - 1;
 427                         memcpy(line + len, continuation, addlen);
 428                         len += addlen;
 429                 }
 430         }
 431         line[len] = 0;
 432
 433         return 1;
 434 }
 435
 436 static int decode_q_segment(char *in, char *ot, char *ep, int rfc2047)
 437 {
 438         int c;
 439         while ((c = *in++) != 0 && (in <= ep)) {
 440                 if (c == '=') {
 441                         int d = *in++;
 442                         if (d == '\n' || !d)
 443                                 break; /* drop trailing newline */
 444                         *ot++ = ((hexval(d) << 4) | hexval(*in++));
 445                         continue;
 446                 }
 447                 if (rfc2047 && c == '_') /* rfc2047 4.2 (2) */
 448                         c = 0x20;
 449                 *ot++ = c;
 450         }
 451         *ot = 0;
 452         return 0;
 453 }
 454
 455 static int decode_b_segment(char *in, char *ot, char *ep)
 456 {
 457         /* Decode in..ep, possibly in-place to ot */
 458         int c, pos = 0, acc = 0;
 459
 460         while ((c = *in++) != 0 && (in <= ep)) {
 461                 if (c == '+')
 462                         c = 62;
 463                 else if (c == '/')
 464                         c = 63;
 465                 else if ('A' <= c && c <= 'Z')
 466                         c -= 'A';
 467                 else if ('a' <= c && c <= 'z')
 468                         c -= 'a' - 26;
 469                 else if ('0' <= c && c <= '9')
 470                         c -= '0' - 52;
 471                 else if (c == '=') {
 472                         /* padding is almost like (c == 0), except we do
 473                          * not output NUL resulting only from it;
 474                          * for now we just trust the data.
 475                          */
 476                         c = 0;
 477                 }
 478                 else
 479                         continue; /* garbage */
 480                 switch (pos++) {
 481                 case 0:
 482                         acc = (c << 2);
 483                         break;
 484                 case 1:
 485                         *ot++ = (acc | (c >> 4));
 486                         acc = (c & 15) << 4;
 487                         break;
 488                 case 2:
 489                         *ot++ = (acc | (c >> 2));
 490                         acc = (c & 3) << 6;
 491                         break;
 492                 case 3:
 493                         *ot++ = (acc | c);
 494                         acc = pos = 0;
 495                         break;
 496                 }
 497         }
 498         *ot = 0;
 499         return 0;
 500 }
 501
 502 /*
 503  * When there is no known charset, guess.
 504  *
 505  * Right now we assume that if the target is UTF-8 (the default),
 506  * and it already looks like UTF-8 (which includes US-ASCII as its
 507  * subset, of course) then that is what it is and there is nothing
 508  * to do.
 509  *
 510  * Otherwise, we default to assuming it is Latin1 for historical
 511  * reasons.
 512  */
 513 static const char *guess_charset(const char *line, const char *target_charset)
 514 {
 515         if (is_encoding_utf8(target_charset)) {
 516                 if (is_utf8(line))
 517                         return NULL;
 518         }
 519         return "latin1";
 520 }
 521
 522 static void convert_to_utf8(char *line, const char *charset)
 523 {
 524         char *out;
 525
 526         if (!charset || !*charset) {
 527                 charset = guess_charset(line, metainfo_charset);
 528                 if (!charset)
 529                         return;
 530         }
 531
 532         if (!strcmp(metainfo_charset, charset))
 533                 return;
 534         out = reencode_string(line, metainfo_charset, charset);
 535         if (!out)
 536                 die("cannot convert from %s to %s\n",
 537                     charset, metainfo_charset);
 538         strcpy(line, out);
 539         free(out);
 540 }
 541
 542 static int decode_header_bq(char *it)
 543 {
 544         char *in, *out, *ep, *cp, *sp;
 545         char outbuf[1000];
 546         int rfc2047 = 0;
 547
 548         in = it;
 549         out = outbuf;
 550         while ((ep = strstr(in, "=?")) != NULL) {
 551                 int sz, encoding;
 552                 char charset_q[256], piecebuf[256];
 553                 rfc2047 = 1;
 554
 555                 if (in != ep) {
 556                         sz = ep - in;
 557                         memcpy(out, in, sz);
 558                         out += sz;
 559                         in += sz;
 560                 }
 561                 /* E.g.
 562                  * ep : "=?iso-2022-jp?B?GyR...?= foo"
 563                  * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
 564                  */
 565                 ep += 2;
 566                 cp = strchr(ep, '?');
 567                 if (!cp)
 568                         return rfc2047; /* no munging */
 569                 for (sp = ep; sp < cp; sp++)
 570                         charset_q[sp - ep] = tolower(*sp);
 571                 charset_q[cp - ep] = 0;
 572                 encoding = cp[1];
 573                 if (!encoding || cp[2] != '?')
 574                         return rfc2047; /* no munging */
 575                 ep = strstr(cp + 3, "?=");
 576                 if (!ep)
 577                         return rfc2047; /* no munging */
 578                 switch (tolower(encoding)) {
 579                 default:
 580                         return rfc2047; /* no munging */
 581                 case 'b':
 582                         sz = decode_b_segment(cp + 3, piecebuf, ep);
 583                         break;
 584                 case 'q':
 585                         sz = decode_q_segment(cp + 3, piecebuf, ep, 1);
 586                         break;
 587                 }
 588                 if (sz < 0)
 589                         return rfc2047;
 590                 if (metainfo_charset)
 591                         convert_to_utf8(piecebuf, charset_q);
 592                 strcpy(out, piecebuf);
 593                 out += strlen(out);
 594                 in = ep + 2;
 595         }
 596         strcpy(out, in);
 597         strcpy(it, outbuf);
 598         return rfc2047;
 599 }
 600
 601 static void decode_header(char *it)
 602 {
 603
 604         if (decode_header_bq(it))
 605                 return;
 606         /* otherwise "it" is a straight copy of the input.
 607          * This can be binary guck but there is no charset specified.
 608          */
 609         if (metainfo_charset)
 610                 convert_to_utf8(it, "");
 611 }
 612
 613 static void decode_transfer_encoding(char *line)
 614 {
 615         char *ep;
 616
 617         switch (transfer_encoding) {
 618         case TE_QP:
 619                 ep = line + strlen(line);
 620                 decode_q_segment(line, line, ep, 0);
 621                 break;
 622         case TE_BASE64:
 623                 ep = line + strlen(line);
 624                 decode_b_segment(line, line, ep);
 625                 break;
 626         case TE_DONTCARE:
 627                 break;
 628         }
 629 }
 630
 631 static int handle_filter(char *line);
 632
 633 static int find_boundary(void)
 634 {
 635         while(fgets(line, sizeof(line), fin) != NULL) {
 636                 if (is_multipart_boundary(line))
 637                         return 1;
 638         }
 639         return 0;
 640 }
 641
 642 static int handle_boundary(void)
 643 {
 644         char newline[]="\n";
 645 again:
 646         if (!memcmp(line+content_top->boundary_len, "--", 2)) {
 647                 /* we hit an end boundary */
 648                 /* pop the current boundary off the stack */
 649                 free(content_top->boundary);
 650
 651                 /* technically won't happen as is_multipart_boundary()
 652                    will fail first.  But just in case..
 653                  */
 654                 if (content_top-- < content) {
 655                         fprintf(stderr, "Detected mismatched boundaries, "
 656                                         "can't recover\n");
 657                         exit(1);
 658                 }
 659                 handle_filter(newline);
 660
 661                 /* skip to the next boundary */
 662                 if (!find_boundary())
 663                         return 0;
 664                 goto again;
 665         }
 666
 667         /* set some defaults */
 668         transfer_encoding = TE_DONTCARE;
 669         charset[0] = 0;
 670         message_type = TYPE_TEXT;
 671
 672         /* slurp in this section's info */
 673         while (read_one_header_line(line, sizeof(line), fin))
 674                 check_header(line, p_hdr_data, 0);
 675
 676         /* eat the blank line after section info */
 677         return (fgets(line, sizeof(line), fin) != NULL);
 678 }
 679
 680 static inline int patchbreak(const char *line)
 681 {
 682         /* Beginning of a "diff -" header? */
 683         if (!memcmp("diff -", line, 6))
 684                 return 1;
 685
 686         /* CVS "Index: " line? */
 687         if (!memcmp("Index: ", line, 7))
 688                 return 1;
 689
 690         /*
 691          * "--- <filename>" starts patches without headers
 692          * "---<sp>*" is a manual separator
 693          */
 694         if (!memcmp("---", line, 3)) {
 695                 line += 3;
 696                 /* space followed by a filename? */
 697                 if (line[0] == ' ' && !isspace(line[1]))
 698                         return 1;
 699                 /* Just whitespace? */
 700                 for (;;) {
 701                         unsigned char c = *line++;
 702                         if (c == '\n')
 703                                 return 1;
 704                         if (!isspace(c))
 705                                 break;
 706                 }
 707                 return 0;
 708         }
 709         return 0;
 710 }
 711
 712
 713 static int handle_commit_msg(char *line)
 714 {
 715         static int still_looking = 1;
 716
 717         if (!cmitmsg)
 718                 return 0;
 719
 720         if (still_looking) {
 721                 char *cp = line;
 722                 if (isspace(*line)) {
 723                         for (cp = line + 1; *cp; cp++) {
 724                                 if (!isspace(*cp))
 725                                         break;
 726                         }
 727                         if (!*cp)
 728                                 return 0;
 729                 }
 730                 if ((still_looking = check_header(cp, s_hdr_data, 0)) != 0)
 731                         return 0;
 732         }
 733
 734         /* normalize the log message to UTF-8. */
 735         if (metainfo_charset)
 736                 convert_to_utf8(line, charset);
 737
 738         if (patchbreak(line)) {
 739                 fclose(cmitmsg);
 740                 cmitmsg = NULL;
 741                 return 1;
 742         }
 743
 744         fputs(line, cmitmsg);
 745         return 0;
 746 }
 747
 748 static int handle_patch(char *line)
 749 {
 750         fputs(line, patchfile);
 751         patch_lines++;
 752         return 0;
 753 }
 754
 755 static int handle_filter(char *line)
 756 {
 757         static int filter = 0;
 758
 759         /* filter tells us which part we left off on
 760          * a non-zero return indicates we hit a filter point
 761          */
 762         switch (filter) {
 763         case 0:
 764                 if (!handle_commit_msg(line))
 765                         break;
 766                 filter++;
 767         case 1:
 768                 if (!handle_patch(line))
 769                         break;
 770                 filter++;
 771         default:
 772                 return 1;
 773         }
 774
 775         return 0;
 776 }
 777
 778 static void handle_body(void)
 779 {
 780         int rc = 0;
 781         static char newline[2000];
 782         static char *np = newline;
 783
 784         /* Skip up to the first boundary */
 785         if (content_top->boundary) {
 786                 if (!find_boundary())
 787                         return;
 788         }
 789
 790         do {
 791                 /* process any boundary lines */
 792                 if (content_top->boundary && is_multipart_boundary(line)) {
 793                         /* flush any leftover */
 794                         if ((transfer_encoding == TE_BASE64)  &&
 795                             (np != newline)) {
 796                                 handle_filter(newline);
 797                         }
 798                         if (!handle_boundary())
 799                                 return;
 800                 }
 801
 802                 /* Unwrap transfer encoding */
 803                 decode_transfer_encoding(line);
 804
 805                 switch (transfer_encoding) {
 806                 case TE_BASE64:
 807                 {
 808                         char *op = line;
 809
 810                         /* binary data most likely doesn't have newlines */
 811                         if (message_type != TYPE_TEXT) {
 812                                 rc = handle_filter(line);
 813                                 break;
 814                         }
 815
 816                         /* this is a decoded line that may contain
 817                          * multiple new lines.  Pass only one chunk
 818                          * at a time to handle_filter()
 819                          */
 820
 821                         do {
 822                                 while (*op != '\n' && *op != 0)
 823                                         *np++ = *op++;
 824                                 *np = *op;
 825                                 if (*np != 0) {
 826                                         /* should be sitting on a new line */
 827                                         *(++np) = 0;
 828                                         op++;
 829                                         rc = handle_filter(newline);
 830                                         np = newline;
 831                                 }
 832                         } while (*op != 0);
 833                         /* the partial chunk is saved in newline and
 834                          * will be appended by the next iteration of fgets
 835                          */
 836                         break;
 837                 }
 838                 default:
 839                         rc = handle_filter(line);
 840                 }
 841                 if (rc)
 842                         /* nothing left to filter */
 843                         break;
 844         } while (fgets(line, sizeof(line), fin));
 845
 846         return;
 847 }
 848
 849 static void handle_info(void)
 850 {
 851         char *sub;
 852         char *hdr;
 853         int i;
 854
 855         for (i = 0; header[i]; i++) {
 856
 857                 /* only print inbody headers if we output a patch file */
 858                 if (patch_lines && s_hdr_data[i])
 859                         hdr = s_hdr_data[i];
 860                 else if (p_hdr_data[i])
 861                         hdr = p_hdr_data[i];
 862                 else
 863                         continue;
 864
 865                 if (!memcmp(header[i], "Subject", 7)) {
 866                         sub = cleanup_subject(hdr);
 867                         cleanup_space(sub);
 868                         fprintf(fout, "Subject: %s\n", sub);
 869                 } else if (!memcmp(header[i], "From", 4)) {
 870                         handle_from(hdr);
 871                         fprintf(fout, "Author: %s\n", name);
 872                         fprintf(fout, "Email: %s\n", email);
 873                 } else {
 874                         cleanup_space(hdr);
 875                         fprintf(fout, "%s: %s\n", header[i], hdr);
 876                 }
 877         }
 878         fprintf(fout, "\n");
 879 }
 880
 881 static int mailinfo(FILE *in, FILE *out, int ks, const char *encoding,
 882                     const char *msg, const char *patch)
 883 {
 884         keep_subject = ks;
 885         metainfo_charset = encoding;
 886         fin = in;
 887         fout = out;
 888
 889         cmitmsg = fopen(msg, "w");
 890         if (!cmitmsg) {
 891                 perror(msg);
 892                 return -1;
 893         }
 894         patchfile = fopen(patch, "w");
 895         if (!patchfile) {
 896                 perror(patch);
 897                 fclose(cmitmsg);
 898                 return -1;
 899         }
 900
 901         p_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(char *));
 902         s_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(char *));
 903
 904         /* process the email header */
 905         while (read_one_header_line(line, sizeof(line), fin))
 906                 check_header(line, p_hdr_data, 1);
 907
 908         handle_body();
 909         handle_info();
 910
 911         return 0;
 912 }
 913
 914 static const char mailinfo_usage[] =
 915         "git-mailinfo [-k] [-u | --encoding=<encoding>] msg patch <mail >info";
 916
 917 int cmd_mailinfo(int argc, const char **argv, const char *prefix)
 918 {
 919         const char *def_charset;
 920
 921         /* NEEDSWORK: might want to do the optional .git/ directory
 922          * discovery
 923          */
 924         git_config(git_default_config);
 925
 926         def_charset = (git_commit_encoding ? git_commit_encoding : "utf-8");
 927         metainfo_charset = def_charset;
 928
 929         while (1 < argc && argv[1][0] == '-') {
 930                 if (!strcmp(argv[1], "-k"))
 931                         keep_subject = 1;
 932                 else if (!strcmp(argv[1], "-u"))
 933                         metainfo_charset = def_charset;
 934                 else if (!strcmp(argv[1], "-n"))
 935                         metainfo_charset = NULL;
 936                 else if (!prefixcmp(argv[1], "--encoding="))
 937                         metainfo_charset = argv[1] + 11;
 938                 else
 939                         usage(mailinfo_usage);
 940                 argc--; argv++;
 941         }
 942
 943         if (argc != 3)
 944                 usage(mailinfo_usage);
 945
 946         return !!mailinfo(stdin, stdout, keep_subject, metainfo_charset, argv[1], argv[2]);
 947 }