Andrew's git / gitweb.git / builtin-mailinfo.c
?
builtin-mailinfo.con commit tree_entry_interesting(): allow it to say "everything is interesting" (1d848f6)
   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
   9static FILE *cmitmsg, *patchfile, *fin, *fout;
  10
  11static int keep_subject;
  12static const char *metainfo_charset;
  13static char line[1000];
  14static char name[1000];
  15static char email[1000];
  16
  17static enum  {
  18        TE_DONTCARE, TE_QP, TE_BASE64,
  19} transfer_encoding;
  20static enum  {
  21        TYPE_TEXT, TYPE_OTHER,
  22} message_type;
  23
  24static char charset[256];
  25static int patch_lines;
  26static char **p_hdr_data, **s_hdr_data;
  27
  28#define MAX_HDR_PARSED 10
  29#define MAX_BOUNDARIES 5
  30
  31static 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
  41static 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
  71static 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
 143static 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
 159static 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
 181struct content_type {
 182        char *boundary;
 183        int boundary_len;
 184};
 185
 186static struct content_type content[MAX_BOUNDARIES];
 187
 188static struct content_type *content_top = content;
 189
 190static 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
 214static 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
 225static int is_multipart_boundary(const char *line)
 226{
 227        return (!memcmp(line, content_top->boundary, content_top->boundary_len));
 228}
 229
 230static 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
 238static 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
 275static 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
 292static void decode_header(char *it);
 293static char *header[MAX_HDR_PARSED] = {
 294        "From","Subject","Date",
 295};
 296
 297static int check_header(char *line, char **hdr_data)
 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] &&
 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
 351static 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 */
 383static 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
 436static 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
 455static 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
 502static void convert_to_utf8(char *line, const char *charset)
 503{
 504        static const char latin_one[] = "latin1";
 505        const char *input_charset = *charset ? charset : latin_one;
 506        char *out = reencode_string(line, metainfo_charset, input_charset);
 507
 508        if (!out)
 509                die("cannot convert from %s to %s\n",
 510                    input_charset, metainfo_charset);
 511        strcpy(line, out);
 512        free(out);
 513}
 514
 515static int decode_header_bq(char *it)
 516{
 517        char *in, *out, *ep, *cp, *sp;
 518        char outbuf[1000];
 519        int rfc2047 = 0;
 520
 521        in = it;
 522        out = outbuf;
 523        while ((ep = strstr(in, "=?")) != NULL) {
 524                int sz, encoding;
 525                char charset_q[256], piecebuf[256];
 526                rfc2047 = 1;
 527
 528                if (in != ep) {
 529                        sz = ep - in;
 530                        memcpy(out, in, sz);
 531                        out += sz;
 532                        in += sz;
 533                }
 534                /* E.g.
 535                 * ep : "=?iso-2022-jp?B?GyR...?= foo"
 536                 * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
 537                 */
 538                ep += 2;
 539                cp = strchr(ep, '?');
 540                if (!cp)
 541                        return rfc2047; /* no munging */
 542                for (sp = ep; sp < cp; sp++)
 543                        charset_q[sp - ep] = tolower(*sp);
 544                charset_q[cp - ep] = 0;
 545                encoding = cp[1];
 546                if (!encoding || cp[2] != '?')
 547                        return rfc2047; /* no munging */
 548                ep = strstr(cp + 3, "?=");
 549                if (!ep)
 550                        return rfc2047; /* no munging */
 551                switch (tolower(encoding)) {
 552                default:
 553                        return rfc2047; /* no munging */
 554                case 'b':
 555                        sz = decode_b_segment(cp + 3, piecebuf, ep);
 556                        break;
 557                case 'q':
 558                        sz = decode_q_segment(cp + 3, piecebuf, ep, 1);
 559                        break;
 560                }
 561                if (sz < 0)
 562                        return rfc2047;
 563                if (metainfo_charset)
 564                        convert_to_utf8(piecebuf, charset_q);
 565                strcpy(out, piecebuf);
 566                out += strlen(out);
 567                in = ep + 2;
 568        }
 569        strcpy(out, in);
 570        strcpy(it, outbuf);
 571        return rfc2047;
 572}
 573
 574static void decode_header(char *it)
 575{
 576
 577        if (decode_header_bq(it))
 578                return;
 579        /* otherwise "it" is a straight copy of the input.
 580         * This can be binary guck but there is no charset specified.
 581         */
 582        if (metainfo_charset)
 583                convert_to_utf8(it, "");
 584}
 585
 586static void decode_transfer_encoding(char *line)
 587{
 588        char *ep;
 589
 590        switch (transfer_encoding) {
 591        case TE_QP:
 592                ep = line + strlen(line);
 593                decode_q_segment(line, line, ep, 0);
 594                break;
 595        case TE_BASE64:
 596                ep = line + strlen(line);
 597                decode_b_segment(line, line, ep);
 598                break;
 599        case TE_DONTCARE:
 600                break;
 601        }
 602}
 603
 604static int handle_filter(char *line);
 605
 606static int find_boundary(void)
 607{
 608        while(fgets(line, sizeof(line), fin) != NULL) {
 609                if (is_multipart_boundary(line))
 610                        return 1;
 611        }
 612        return 0;
 613}
 614
 615static int handle_boundary(void)
 616{
 617again:
 618        if (!memcmp(line+content_top->boundary_len, "--", 2)) {
 619                /* we hit an end boundary */
 620                /* pop the current boundary off the stack */
 621                free(content_top->boundary);
 622
 623                /* technically won't happen as is_multipart_boundary()
 624                   will fail first.  But just in case..
 625                 */
 626                if (content_top-- < content) {
 627                        fprintf(stderr, "Detected mismatched boundaries, "
 628                                        "can't recover\n");
 629                        exit(1);
 630                }
 631                handle_filter("\n");
 632
 633                /* skip to the next boundary */
 634                if (!find_boundary())
 635                        return 0;
 636                goto again;
 637        }
 638
 639        /* set some defaults */
 640        transfer_encoding = TE_DONTCARE;
 641        charset[0] = 0;
 642        message_type = TYPE_TEXT;
 643
 644        /* slurp in this section's info */
 645        while (read_one_header_line(line, sizeof(line), fin))
 646                check_header(line, p_hdr_data);
 647
 648        /* eat the blank line after section info */
 649        return (fgets(line, sizeof(line), fin) != NULL);
 650}
 651
 652static inline int patchbreak(const char *line)
 653{
 654        /* Beginning of a "diff -" header? */
 655        if (!memcmp("diff -", line, 6))
 656                return 1;
 657
 658        /* CVS "Index: " line? */
 659        if (!memcmp("Index: ", line, 7))
 660                return 1;
 661
 662        /*
 663         * "--- <filename>" starts patches without headers
 664         * "---<sp>*" is a manual separator
 665         */
 666        if (!memcmp("---", line, 3)) {
 667                line += 3;
 668                /* space followed by a filename? */
 669                if (line[0] == ' ' && !isspace(line[1]))
 670                        return 1;
 671                /* Just whitespace? */
 672                for (;;) {
 673                        unsigned char c = *line++;
 674                        if (c == '\n')
 675                                return 1;
 676                        if (!isspace(c))
 677                                break;
 678                }
 679                return 0;
 680        }
 681        return 0;
 682}
 683
 684
 685static int handle_commit_msg(char *line)
 686{
 687        static int still_looking = 1;
 688
 689        if (!cmitmsg)
 690                return 0;
 691
 692        if (still_looking) {
 693                char *cp = line;
 694                if (isspace(*line)) {
 695                        for (cp = line + 1; *cp; cp++) {
 696                                if (!isspace(*cp))
 697                                        break;
 698                        }
 699                        if (!*cp)
 700                                return 0;
 701                }
 702                if ((still_looking = check_header(cp, s_hdr_data)) != 0)
 703                        return 0;
 704        }
 705
 706        if (patchbreak(line)) {
 707                fclose(cmitmsg);
 708                cmitmsg = NULL;
 709                return 1;
 710        }
 711
 712        fputs(line, cmitmsg);
 713        return 0;
 714}
 715
 716static int handle_patch(char *line)
 717{
 718        fputs(line, patchfile);
 719        patch_lines++;
 720        return 0;
 721}
 722
 723static int handle_filter(char *line)
 724{
 725        static int filter = 0;
 726
 727        /* filter tells us which part we left off on
 728         * a non-zero return indicates we hit a filter point
 729         */
 730        switch (filter) {
 731        case 0:
 732                if (!handle_commit_msg(line))
 733                        break;
 734                filter++;
 735        case 1:
 736                if (!handle_patch(line))
 737                        break;
 738                filter++;
 739        default:
 740                return 1;
 741        }
 742
 743        return 0;
 744}
 745
 746static void handle_body(void)
 747{
 748        int rc = 0;
 749        static char newline[2000];
 750        static char *np = newline;
 751
 752        /* Skip up to the first boundary */
 753        if (content_top->boundary) {
 754                if (!find_boundary())
 755                        return;
 756        }
 757
 758        do {
 759                /* process any boundary lines */
 760                if (content_top->boundary && is_multipart_boundary(line)) {
 761                        /* flush any leftover */
 762                        if ((transfer_encoding == TE_BASE64)  &&
 763                            (np != newline)) {
 764                                handle_filter(newline);
 765                        }
 766                        if (!handle_boundary())
 767                                return;
 768                }
 769
 770                /* Unwrap transfer encoding and optionally
 771                 * normalize the log message to UTF-8.
 772                 */
 773                decode_transfer_encoding(line);
 774                if (metainfo_charset)
 775                        convert_to_utf8(line, charset);
 776
 777                switch (transfer_encoding) {
 778                case TE_BASE64:
 779                {
 780                        char *op = line;
 781
 782                        /* binary data most likely doesn't have newlines */
 783                        if (message_type != TYPE_TEXT) {
 784                                rc = handle_filter(line);
 785                                break;
 786                        }
 787
 788                        /* this is a decoded line that may contain
 789                         * multiple new lines.  Pass only one chunk
 790                         * at a time to handle_filter()
 791                         */
 792
 793                        do {
 794                                while (*op != '\n' && *op != 0)
 795                                        *np++ = *op++;
 796                                *np = *op;
 797                                if (*np != 0) {
 798                                        /* should be sitting on a new line */
 799                                        *(++np) = 0;
 800                                        op++;
 801                                        rc = handle_filter(newline);
 802                                        np = newline;
 803                                }
 804                        } while (*op != 0);
 805                        /* the partial chunk is saved in newline and
 806                         * will be appended by the next iteration of fgets
 807                         */
 808                        break;
 809                }
 810                default:
 811                        rc = handle_filter(line);
 812                }
 813                if (rc)
 814                        /* nothing left to filter */
 815                        break;
 816        } while (fgets(line, sizeof(line), fin));
 817
 818        return;
 819}
 820
 821static void handle_info(void)
 822{
 823        char *sub;
 824        char *hdr;
 825        int i;
 826
 827        for (i = 0; header[i]; i++) {
 828
 829                /* only print inbody headers if we output a patch file */
 830                if (patch_lines && s_hdr_data[i])
 831                        hdr = s_hdr_data[i];
 832                else if (p_hdr_data[i])
 833                        hdr = p_hdr_data[i];
 834                else
 835                        continue;
 836
 837                if (!memcmp(header[i], "Subject", 7)) {
 838                        sub = cleanup_subject(hdr);
 839                        cleanup_space(sub);
 840                        fprintf(fout, "Subject: %s\n", sub);
 841                } else if (!memcmp(header[i], "From", 4)) {
 842                        handle_from(hdr);
 843                        fprintf(fout, "Author: %s\n", name);
 844                        fprintf(fout, "Email: %s\n", email);
 845                } else {
 846                        cleanup_space(hdr);
 847                        fprintf(fout, "%s: %s\n", header[i], hdr);
 848                }
 849        }
 850        fprintf(fout, "\n");
 851}
 852
 853int mailinfo(FILE *in, FILE *out, int ks, const char *encoding,
 854             const char *msg, const char *patch)
 855{
 856        keep_subject = ks;
 857        metainfo_charset = encoding;
 858        fin = in;
 859        fout = out;
 860
 861        cmitmsg = fopen(msg, "w");
 862        if (!cmitmsg) {
 863                perror(msg);
 864                return -1;
 865        }
 866        patchfile = fopen(patch, "w");
 867        if (!patchfile) {
 868                perror(patch);
 869                fclose(cmitmsg);
 870                return -1;
 871        }
 872
 873        p_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(char *));
 874        s_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(char *));
 875
 876        /* process the email header */
 877        while (read_one_header_line(line, sizeof(line), fin))
 878                check_header(line, p_hdr_data);
 879
 880        handle_body();
 881        handle_info();
 882
 883        return 0;
 884}
 885
 886static const char mailinfo_usage[] =
 887        "git-mailinfo [-k] [-u | --encoding=<encoding>] msg patch <mail >info";
 888
 889int cmd_mailinfo(int argc, const char **argv, const char *prefix)
 890{
 891        const char *def_charset;
 892
 893        /* NEEDSWORK: might want to do the optional .git/ directory
 894         * discovery
 895         */
 896        git_config(git_default_config);
 897
 898        def_charset = (git_commit_encoding ? git_commit_encoding : "utf-8");
 899        metainfo_charset = def_charset;
 900
 901        while (1 < argc && argv[1][0] == '-') {
 902                if (!strcmp(argv[1], "-k"))
 903                        keep_subject = 1;
 904                else if (!strcmp(argv[1], "-u"))
 905                        metainfo_charset = def_charset;
 906                else if (!strcmp(argv[1], "-n"))
 907                        metainfo_charset = NULL;
 908                else if (!prefixcmp(argv[1], "--encoding="))
 909                        metainfo_charset = argv[1] + 11;
 910                else
 911                        usage(mailinfo_usage);
 912                argc--; argv++;
 913        }
 914
 915        if (argc != 3)
 916                usage(mailinfo_usage);
 917
 918        return !!mailinfo(stdin, stdout, keep_subject, metainfo_charset, argv[1], argv[2]);
 919}