2 * Another stupid program, this one parsing the headers of an
3 * email to figure out authorship and subject
12 static FILE *cmitmsg, *patchfile;
14 static int keep_subject = 0;
15 static int metainfo_utf8 = 0;
16 static char line[1000];
17 static char date[1000];
18 static char name[1000];
19 static char email[1000];
20 static char subject[1000];
23 TE_DONTCARE, TE_QP, TE_BASE64,
25 static char charset[256];
27 static char multipart_boundary[1000];
28 static int multipart_boundary_len;
29 static int patch_lines = 0;
31 static char *sanity_check(char *name, char *email)
33 int len = strlen(name);
34 if (len < 3 || len > 60)
36 if (strchr(name, '@') || strchr(name, '<') || strchr(name, '>'))
41 static int handle_from(char *line)
43 char *at = strchr(line, '@');
50 * If we already have one email, don't take any confusing lines
52 if (*email && strchr(at+1, '@'))
55 /* Pick up the string around '@', possibly delimited with <>
56 * pair; that is the email part. White them out while copying.
70 unsigned char c = *at;
71 if (!c || c == '>' || isspace(c)) {
81 /* The remainder is name. It could be "John Doe <john.doe@xz>"
82 * or "john.doe@xz (John Doe)", but we have whited out the
83 * email part, so trim from both ends, possibly removing
84 * the () pair at the end.
86 at = line + strlen(line);
88 unsigned char c = *--at;
90 at[(c == ')') ? 0 : 1] = 0;
97 unsigned char c = *at;
98 if (!c || !isspace(c)) {
105 at = sanity_check(at, email);
110 static int handle_date(char *line)
116 static int handle_subject(char *line)
118 strcpy(subject, line);
122 /* NOTE NOTE NOTE. We do not claim we do full MIME. We just attempt
123 * to have enough heuristics to grok MIME encoded patches often found
124 * on our mailing lists. For example, we do not even treat header lines
125 * case insensitively.
128 static int slurp_attr(const char *line, const char *name, char *attr)
130 char *ends, *ap = strcasestr(line, name);
144 sz = strcspn(ap, ends);
145 memcpy(attr, ap, sz);
150 static int handle_subcontent_type(char *line)
152 /* We do not want to mess with boundary. Note that we do not
153 * handle nested multipart.
155 if (strcasestr(line, "boundary=")) {
156 fprintf(stderr, "Not handling nested multipart message.\n");
159 slurp_attr(line, "charset=", charset);
162 for (i = 0; (c = charset[i]) != 0; i++)
163 charset[i] = tolower(c);
168 static int handle_content_type(char *line)
170 *multipart_boundary = 0;
171 if (slurp_attr(line, "boundary=", multipart_boundary + 2)) {
172 memcpy(multipart_boundary, "--", 2);
173 multipart_boundary_len = strlen(multipart_boundary);
175 slurp_attr(line, "charset=", charset);
179 static int handle_content_transfer_encoding(char *line)
181 if (strcasestr(line, "base64"))
182 transfer_encoding = TE_BASE64;
183 else if (strcasestr(line, "quoted-printable"))
184 transfer_encoding = TE_QP;
186 transfer_encoding = TE_DONTCARE;
190 static int is_multipart_boundary(const char *line)
192 return (!memcmp(line, multipart_boundary, multipart_boundary_len));
195 static int eatspace(char *line)
197 int len = strlen(line);
198 while (len > 0 && isspace(line[len-1]))
205 #define SEEN_SUBJECT 04
207 /* First lines of body can have From:, Date:, and Subject: */
208 static int handle_inbody_header(int *seen, char *line)
210 if (!memcmp("From:", line, 5) && isspace(line[5])) {
211 if (!(*seen & SEEN_FROM) && handle_from(line+6)) {
216 if (!memcmp("Date:", line, 5) && isspace(line[5])) {
217 if (!(*seen & SEEN_DATE)) {
223 if (!memcmp("Subject:", line, 8) && isspace(line[8])) {
224 if (!(*seen & SEEN_SUBJECT)) {
225 handle_subject(line+9);
226 *seen |= SEEN_SUBJECT;
230 if (!memcmp("[PATCH]", line, 7) && isspace(line[7])) {
231 if (!(*seen & SEEN_SUBJECT)) {
232 handle_subject(line);
233 *seen |= SEEN_SUBJECT;
240 static char *cleanup_subject(char *subject)
249 if (!memcmp("e:", subject+1, 2)) {
254 case ' ': case '\t': case ':':
259 p = strchr(subject, ']');
265 remove = p - subject;
266 if (remove <= len *2) {
276 static void cleanup_space(char *buf)
279 while ((c = *buf) != 0) {
285 int len = strlen(buf);
286 memmove(buf, buf+1, len);
293 typedef int (*header_fn_t)(char *);
300 static void check_header(char *line, int len, struct header_def *header)
304 if (header[0].namelen <= 0) {
305 for (i = 0; header[i].name; i++)
306 header[i].namelen = strlen(header[i].name);
308 for (i = 0; header[i].name; i++) {
309 int len = header[i].namelen;
310 if (!strncasecmp(line, header[i].name, len) &&
311 line[len] == ':' && isspace(line[len + 1])) {
312 header[i].func(line + len + 2);
318 static void check_subheader_line(char *line, int len)
320 static struct header_def header[] = {
321 { "Content-Type", handle_subcontent_type },
322 { "Content-Transfer-Encoding",
323 handle_content_transfer_encoding },
326 check_header(line, len, header);
328 static void check_header_line(char *line, int len)
330 static struct header_def header[] = {
331 { "From", handle_from },
332 { "Date", handle_date },
333 { "Subject", handle_subject },
334 { "Content-Type", handle_content_type },
335 { "Content-Transfer-Encoding",
336 handle_content_transfer_encoding },
339 check_header(line, len, header);
342 static int read_one_header_line(char *line, int sz, FILE *in)
347 if (fgets(line + ofs, sz - ofs, in) == NULL)
349 len = eatspace(line + ofs);
352 peek = fgetc(in); ungetc(peek, in);
353 if (peek == ' ' || peek == '\t') {
354 /* Yuck, 2822 header "folding" */
363 static unsigned hexval(int c)
365 if (c >= '0' && c <= '9')
367 if (c >= 'a' && c <= 'f')
369 if (c >= 'A' && c <= 'F')
374 static int decode_q_segment(char *in, char *ot, char *ep)
377 while ((c = *in++) != 0 && (in <= ep)) {
381 break; /* drop trailing newline */
382 *ot++ = ((hexval(d) << 4) | hexval(*in++));
391 static int decode_b_segment(char *in, char *ot, char *ep)
393 /* Decode in..ep, possibly in-place to ot */
394 int c, pos = 0, acc = 0;
396 while ((c = *in++) != 0 && (in <= ep)) {
401 else if ('A' <= c && c <= 'Z')
403 else if ('a' <= c && c <= 'z')
405 else if ('0' <= c && c <= '9')
408 /* padding is almost like (c == 0), except we do
409 * not output NUL resulting only from it;
410 * for now we just trust the data.
415 continue; /* garbage */
421 *ot++ = (acc | (c >> 4));
425 *ot++ = (acc | (c >> 2));
438 static void convert_to_utf8(char *line, char *charset)
442 size_t insize, outsize, nrc;
443 char outbuf[4096]; /* cheat */
444 iconv_t conv = iconv_open("utf-8", charset);
446 if (conv == (iconv_t) -1) {
447 fprintf(stderr, "cannot convert from %s to utf-8\n",
455 outsize = sizeof(outbuf);
456 nrc = iconv(conv, &in, &insize, &out, &outsize);
458 if (nrc == (size_t) -1)
461 strcpy(line, outbuf);
465 static void decode_header_bq(char *it)
467 char *in, *out, *ep, *cp, *sp;
472 while ((ep = strstr(in, "=?")) != NULL) {
474 char charset_q[256], piecebuf[256];
482 * ep : "=?iso-2022-jp?B?GyR...?= foo"
483 * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
486 cp = strchr(ep, '?');
488 return; /* no munging */
489 for (sp = ep; sp < cp; sp++)
490 charset_q[sp - ep] = tolower(*sp);
491 charset_q[cp - ep] = 0;
493 if (!encoding || cp[2] != '?')
494 return; /* no munging */
495 ep = strstr(cp + 3, "?=");
497 return; /* no munging */
498 switch (tolower(encoding)) {
500 return; /* no munging */
502 sz = decode_b_segment(cp + 3, piecebuf, ep);
505 sz = decode_q_segment(cp + 3, piecebuf, ep);
511 convert_to_utf8(piecebuf, charset_q);
512 strcpy(out, piecebuf);
520 static void decode_transfer_encoding(char *line)
524 switch (transfer_encoding) {
526 ep = line + strlen(line);
527 decode_q_segment(line, line, ep);
530 ep = line + strlen(line);
531 decode_b_segment(line, line, ep);
538 static void handle_info(void)
541 static int done_info = 0;
547 sub = cleanup_subject(subject);
550 cleanup_space(email);
553 /* Unwrap inline B and Q encoding, and optionally
554 * normalize the meta information to utf8.
556 decode_header_bq(name);
557 decode_header_bq(date);
558 decode_header_bq(email);
559 decode_header_bq(sub);
560 printf("Author: %s\nEmail: %s\nSubject: %s\nDate: %s\n\n",
561 name, email, sub, date);
564 /* We are inside message body and have read line[] already.
565 * Spit out the commit log.
567 static int handle_commit_msg(void)
572 if (!memcmp("diff -", line, 6) ||
573 !memcmp("---", line, 3) ||
574 !memcmp("Index: ", line, 7))
576 if ((multipart_boundary[0] && is_multipart_boundary(line))) {
577 /* We come here when the first part had only
578 * the commit message without any patch. We
579 * pretend we have not seen this line yet, and
580 * go back to the loop.
585 /* Unwrap transfer encoding and optionally
586 * normalize the log message to UTF-8.
588 decode_transfer_encoding(line);
590 convert_to_utf8(line, charset);
591 fputs(line, cmitmsg);
592 } while (fgets(line, sizeof(line), stdin) != NULL);
598 /* We have done the commit message and have the first
599 * line of the patch in line[].
601 static void handle_patch(void)
604 if (multipart_boundary[0] && is_multipart_boundary(line))
606 /* Only unwrap transfer encoding but otherwise do not
607 * do anything. We do *NOT* want UTF-8 conversion
608 * here; we are dealing with the user payload.
610 decode_transfer_encoding(line);
611 fputs(line, patchfile);
613 } while (fgets(line, sizeof(line), stdin) != NULL);
616 /* multipart boundary and transfer encoding are set up for us, and we
617 * are at the end of the sub header. do equivalent of handle_body up
618 * to the next boundary without closing patchfile --- we will expect
619 * that the first part to contain commit message and a patch, and
620 * handle other parts as pure patches.
622 static int handle_multipart_one_part(void)
628 while (fgets(line, sizeof(line), stdin) != NULL) {
630 len = eatspace(line);
634 if (is_multipart_boundary(line))
636 if (0 <= seen && handle_inbody_header(&seen, line))
638 seen = -1; /* no more inbody headers */
641 if (handle_commit_msg())
651 static void handle_multipart_body(void)
655 /* Skip up to the first boundary */
656 while (fgets(line, sizeof(line), stdin) != NULL)
657 if (is_multipart_boundary(line)) {
663 /* We are on boundary line. Start slurping the subhead. */
665 int len = read_one_header_line(line, sizeof(line), stdin);
667 if (handle_multipart_one_part() < 0)
671 check_subheader_line(line, len);
675 fprintf(stderr, "No patch found\n");
680 /* Non multipart message */
681 static void handle_body(void)
685 while (fgets(line, sizeof(line), stdin) != NULL) {
686 int len = eatspace(line);
689 if (0 <= seen && handle_inbody_header(&seen, line))
691 seen = -1; /* no more inbody headers */
700 fprintf(stderr, "No patch found\n");
705 static const char mailinfo_usage[] =
706 "git-mailinfo [-k] [-u] msg patch <mail >info";
708 static void usage(void) {
709 fprintf(stderr, "%s\n", mailinfo_usage);
713 int main(int argc, char **argv)
715 while (1 < argc && argv[1][0] == '-') {
716 if (!strcmp(argv[1], "-k"))
718 else if (!strcmp(argv[1], "-u"))
727 cmitmsg = fopen(argv[1], "w");
732 patchfile = fopen(argv[2], "w");
738 int len = read_one_header_line(line, sizeof(line), stdin);
740 if (multipart_boundary[0])
741 handle_multipart_body();
746 check_header_line(line, len);