2 * Another stupid program, this one parsing the headers of an
3 * email to figure out authorship and subject
13 #include "git-compat-util.h"
16 static FILE *cmitmsg
, *patchfile
;
18 static int keep_subject
= 0;
19 static char *metainfo_charset
= NULL
;
20 static char line
[1000];
21 static char date
[1000];
22 static char name
[1000];
23 static char email
[1000];
24 static char subject
[1000];
27 TE_DONTCARE
, TE_QP
, TE_BASE64
,
29 static char charset
[256];
31 static char multipart_boundary
[1000];
32 static int multipart_boundary_len
;
33 static int patch_lines
= 0;
35 static char *sanity_check(char *name
, char *email
)
37 int len
= strlen(name
);
38 if (len
< 3 || len
> 60)
40 if (strchr(name
, '@') || strchr(name
, '<') || strchr(name
, '>'))
45 static int bogus_from(char *line
)
47 /* John Doe <johndoe> */
48 char *bra
, *ket
, *dst
, *cp
;
50 /* This is fallback, so do not bother if we already have an
56 bra
= strchr(line
, '<');
59 ket
= strchr(bra
, '>');
63 for (dst
= email
, cp
= bra
+1; cp
< ket
; )
66 for (cp
= line
; isspace(*cp
); cp
++)
68 for (bra
--; isspace(*bra
); bra
--)
70 cp
= sanity_check(cp
, email
);
75 static int handle_from(char *in_line
)
81 strcpy(line
, in_line
);
82 at
= strchr(line
, '@');
84 return bogus_from(line
);
87 * If we already have one email, don't take any confusing lines
89 if (*email
&& strchr(at
+1, '@'))
92 /* Pick up the string around '@', possibly delimited with <>
93 * pair; that is the email part. White them out while copying.
107 unsigned char c
= *at
;
108 if (!c
|| c
== '>' || isspace(c
)) {
118 /* The remainder is name. It could be "John Doe <john.doe@xz>"
119 * or "john.doe@xz (John Doe)", but we have whited out the
120 * email part, so trim from both ends, possibly removing
121 * the () pair at the end.
123 at
= line
+ strlen(line
);
125 unsigned char c
= *--at
;
127 at
[(c
== ')') ? 0 : 1] = 0;
134 unsigned char c
= *at
;
135 if (!c
|| !isspace(c
)) {
142 at
= sanity_check(at
, email
);
147 static int handle_date(char *line
)
153 static int handle_subject(char *line
)
155 strcpy(subject
, line
);
159 /* NOTE NOTE NOTE. We do not claim we do full MIME. We just attempt
160 * to have enough heuristics to grok MIME encoded patches often found
161 * on our mailing lists. For example, we do not even treat header lines
162 * case insensitively.
165 static int slurp_attr(const char *line
, const char *name
, char *attr
)
167 char *ends
, *ap
= strcasestr(line
, name
);
181 sz
= strcspn(ap
, ends
);
182 memcpy(attr
, ap
, sz
);
187 static int handle_subcontent_type(char *line
)
189 /* We do not want to mess with boundary. Note that we do not
190 * handle nested multipart.
192 if (strcasestr(line
, "boundary=")) {
193 fprintf(stderr
, "Not handling nested multipart message.\n");
196 slurp_attr(line
, "charset=", charset
);
199 for (i
= 0; (c
= charset
[i
]) != 0; i
++)
200 charset
[i
] = tolower(c
);
205 static int handle_content_type(char *line
)
207 *multipart_boundary
= 0;
208 if (slurp_attr(line
, "boundary=", multipart_boundary
+ 2)) {
209 memcpy(multipart_boundary
, "--", 2);
210 multipart_boundary_len
= strlen(multipart_boundary
);
212 slurp_attr(line
, "charset=", charset
);
216 static int handle_content_transfer_encoding(char *line
)
218 if (strcasestr(line
, "base64"))
219 transfer_encoding
= TE_BASE64
;
220 else if (strcasestr(line
, "quoted-printable"))
221 transfer_encoding
= TE_QP
;
223 transfer_encoding
= TE_DONTCARE
;
227 static int is_multipart_boundary(const char *line
)
229 return (!memcmp(line
, multipart_boundary
, multipart_boundary_len
));
232 static int eatspace(char *line
)
234 int len
= strlen(line
);
235 while (len
> 0 && isspace(line
[len
-1]))
242 #define SEEN_SUBJECT 04
243 #define SEEN_BOGUS_UNIX_FROM 010
244 #define SEEN_PREFIX 020
246 /* First lines of body can have From:, Date:, and Subject: or empty */
247 static void handle_inbody_header(int *seen
, char *line
)
249 if (*seen
& SEEN_PREFIX
)
251 if (isspace(*line
)) {
253 for (cp
= line
+ 1; *cp
; cp
++) {
260 if (!memcmp(">From", line
, 5) && isspace(line
[5])) {
261 if (!(*seen
& SEEN_BOGUS_UNIX_FROM
)) {
262 *seen
|= SEEN_BOGUS_UNIX_FROM
;
266 if (!memcmp("From:", line
, 5) && isspace(line
[5])) {
267 if (!(*seen
& SEEN_FROM
) && handle_from(line
+6)) {
272 if (!memcmp("Date:", line
, 5) && isspace(line
[5])) {
273 if (!(*seen
& SEEN_DATE
)) {
279 if (!memcmp("Subject:", line
, 8) && isspace(line
[8])) {
280 if (!(*seen
& SEEN_SUBJECT
)) {
281 handle_subject(line
+9);
282 *seen
|= SEEN_SUBJECT
;
286 if (!memcmp("[PATCH]", line
, 7) && isspace(line
[7])) {
287 if (!(*seen
& SEEN_SUBJECT
)) {
288 handle_subject(line
);
289 *seen
|= SEEN_SUBJECT
;
293 *seen
|= SEEN_PREFIX
;
296 static char *cleanup_subject(char *subject
)
305 if (!memcmp("e:", subject
+1, 2)) {
310 case ' ': case '\t': case ':':
315 p
= strchr(subject
, ']');
321 remove
= p
- subject
;
322 if (remove
<= len
*2) {
333 static void cleanup_space(char *buf
)
336 while ((c
= *buf
) != 0) {
342 int len
= strlen(buf
);
343 memmove(buf
, buf
+1, len
);
350 static void decode_header_bq(char *it
);
351 typedef int (*header_fn_t
)(char *);
358 static void check_header(char *line
, struct header_def
*header
)
362 if (header
[0].namelen
<= 0) {
363 for (i
= 0; header
[i
].name
; i
++)
364 header
[i
].namelen
= strlen(header
[i
].name
);
366 for (i
= 0; header
[i
].name
; i
++) {
367 int len
= header
[i
].namelen
;
368 if (!strncasecmp(line
, header
[i
].name
, len
) &&
369 line
[len
] == ':' && isspace(line
[len
+ 1])) {
370 /* Unwrap inline B and Q encoding, and optionally
371 * normalize the meta information to utf8.
373 decode_header_bq(line
+ len
+ 2);
374 header
[i
].func(line
+ len
+ 2);
380 static void check_subheader_line(char *line
)
382 static struct header_def header
[] = {
383 { "Content-Type", handle_subcontent_type
},
384 { "Content-Transfer-Encoding",
385 handle_content_transfer_encoding
},
388 check_header(line
, header
);
390 static void check_header_line(char *line
)
392 static struct header_def header
[] = {
393 { "From", handle_from
},
394 { "Date", handle_date
},
395 { "Subject", handle_subject
},
396 { "Content-Type", handle_content_type
},
397 { "Content-Transfer-Encoding",
398 handle_content_transfer_encoding
},
401 check_header(line
, header
);
404 static int is_rfc2822_header(char *line
)
407 * The section that defines the loosest possible
408 * field name is "3.6.8 Optional fields".
410 * optional-field = field-name ":" unstructured CRLF
411 * field-name = 1*ftext
412 * ftext = %d33-57 / %59-126
416 while ((ch
= *cp
++)) {
419 if ((33 <= ch
&& ch
<= 57) ||
420 (59 <= ch
&& ch
<= 126))
427 static int read_one_header_line(char *line
, int sz
, FILE *in
)
432 if (fgets(line
+ ofs
, sz
- ofs
, in
) == NULL
)
434 len
= eatspace(line
+ ofs
);
435 if ((len
== 0) || !is_rfc2822_header(line
)) {
436 /* Re-add the newline */
437 line
[ofs
+ len
] = '\n';
438 line
[ofs
+ len
+ 1] = '\0';
442 /* Yuck, 2822 header "folding" */
443 peek
= fgetc(in
); ungetc(peek
, in
);
444 if (peek
!= ' ' && peek
!= '\t')
447 /* Count mbox From headers as headers */
448 if (!ofs
&& !memcmp(line
, "From ", 5))
453 static unsigned hexval(int c
)
455 if (c
>= '0' && c
<= '9')
457 if (c
>= 'a' && c
<= 'f')
459 if (c
>= 'A' && c
<= 'F')
464 static int decode_q_segment(char *in
, char *ot
, char *ep
, int rfc2047
)
467 while ((c
= *in
++) != 0 && (in
<= ep
)) {
471 break; /* drop trailing newline */
472 *ot
++ = ((hexval(d
) << 4) | hexval(*in
++));
475 if (rfc2047
&& c
== '_') /* rfc2047 4.2 (2) */
483 static int decode_b_segment(char *in
, char *ot
, char *ep
)
485 /* Decode in..ep, possibly in-place to ot */
486 int c
, pos
= 0, acc
= 0;
488 while ((c
= *in
++) != 0 && (in
<= ep
)) {
493 else if ('A' <= c
&& c
<= 'Z')
495 else if ('a' <= c
&& c
<= 'z')
497 else if ('0' <= c
&& c
<= '9')
500 /* padding is almost like (c == 0), except we do
501 * not output NUL resulting only from it;
502 * for now we just trust the data.
507 continue; /* garbage */
513 *ot
++ = (acc
| (c
>> 4));
517 *ot
++ = (acc
| (c
>> 2));
530 static void convert_to_utf8(char *line
, char *charset
)
534 size_t insize
, outsize
, nrc
;
535 char outbuf
[4096]; /* cheat */
536 static char latin_one
[] = "latin1";
537 char *input_charset
= *charset
? charset
: latin_one
;
538 iconv_t conv
= iconv_open(metainfo_charset
, input_charset
);
540 if (conv
== (iconv_t
) -1) {
541 static int warned_latin1_once
= 0;
542 if (input_charset
!= latin_one
) {
543 fprintf(stderr
, "cannot convert from %s to %s\n",
544 input_charset
, metainfo_charset
);
547 else if (!warned_latin1_once
) {
548 warned_latin1_once
= 1;
549 fprintf(stderr
, "tried to convert from %s to %s, "
550 "but your iconv does not work with it.\n",
551 input_charset
, metainfo_charset
);
558 outsize
= sizeof(outbuf
);
559 nrc
= iconv(conv
, &in
, &insize
, &out
, &outsize
);
561 if (nrc
== (size_t) -1)
564 strcpy(line
, outbuf
);
568 static void decode_header_bq(char *it
)
570 char *in
, *out
, *ep
, *cp
, *sp
;
575 while ((ep
= strstr(in
, "=?")) != NULL
) {
577 char charset_q
[256], piecebuf
[256];
585 * ep : "=?iso-2022-jp?B?GyR...?= foo"
586 * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
589 cp
= strchr(ep
, '?');
591 return; /* no munging */
592 for (sp
= ep
; sp
< cp
; sp
++)
593 charset_q
[sp
- ep
] = tolower(*sp
);
594 charset_q
[cp
- ep
] = 0;
596 if (!encoding
|| cp
[2] != '?')
597 return; /* no munging */
598 ep
= strstr(cp
+ 3, "?=");
600 return; /* no munging */
601 switch (tolower(encoding
)) {
603 return; /* no munging */
605 sz
= decode_b_segment(cp
+ 3, piecebuf
, ep
);
608 sz
= decode_q_segment(cp
+ 3, piecebuf
, ep
, 1);
613 if (metainfo_charset
)
614 convert_to_utf8(piecebuf
, charset_q
);
615 strcpy(out
, piecebuf
);
623 static void decode_transfer_encoding(char *line
)
627 switch (transfer_encoding
) {
629 ep
= line
+ strlen(line
);
630 decode_q_segment(line
, line
, ep
, 0);
633 ep
= line
+ strlen(line
);
634 decode_b_segment(line
, line
, ep
);
641 static void handle_info(void)
645 sub
= cleanup_subject(subject
);
648 cleanup_space(email
);
651 printf("Author: %s\nEmail: %s\nSubject: %s\nDate: %s\n\n",
652 name
, email
, sub
, date
);
655 /* We are inside message body and have read line[] already.
656 * Spit out the commit log.
658 static int handle_commit_msg(int *seen
)
663 if (!memcmp("diff -", line
, 6) ||
664 !memcmp("---", line
, 3) ||
665 !memcmp("Index: ", line
, 7))
667 if ((multipart_boundary
[0] && is_multipart_boundary(line
))) {
668 /* We come here when the first part had only
669 * the commit message without any patch. We
670 * pretend we have not seen this line yet, and
671 * go back to the loop.
676 /* Unwrap transfer encoding and optionally
677 * normalize the log message to UTF-8.
679 decode_transfer_encoding(line
);
680 if (metainfo_charset
)
681 convert_to_utf8(line
, charset
);
683 handle_inbody_header(seen
, line
);
684 if (!(*seen
& SEEN_PREFIX
))
687 fputs(line
, cmitmsg
);
688 } while (fgets(line
, sizeof(line
), stdin
) != NULL
);
694 /* We have done the commit message and have the first
695 * line of the patch in line[].
697 static void handle_patch(void)
700 if (multipart_boundary
[0] && is_multipart_boundary(line
))
702 /* Only unwrap transfer encoding but otherwise do not
703 * do anything. We do *NOT* want UTF-8 conversion
704 * here; we are dealing with the user payload.
706 decode_transfer_encoding(line
);
707 fputs(line
, patchfile
);
709 } while (fgets(line
, sizeof(line
), stdin
) != NULL
);
712 /* multipart boundary and transfer encoding are set up for us, and we
713 * are at the end of the sub header. do equivalent of handle_body up
714 * to the next boundary without closing patchfile --- we will expect
715 * that the first part to contain commit message and a patch, and
716 * handle other parts as pure patches.
718 static int handle_multipart_one_part(int *seen
)
722 while (fgets(line
, sizeof(line
), stdin
) != NULL
) {
725 if (is_multipart_boundary(line
))
727 if (handle_commit_msg(seen
))
737 static void handle_multipart_body(void)
742 /* Skip up to the first boundary */
743 while (fgets(line
, sizeof(line
), stdin
) != NULL
)
744 if (is_multipart_boundary(line
)) {
750 /* We are on boundary line. Start slurping the subhead. */
752 int hdr
= read_one_header_line(line
, sizeof(line
), stdin
);
754 if (handle_multipart_one_part(&seen
) < 0)
756 /* Reset per part headers */
757 transfer_encoding
= TE_DONTCARE
;
761 check_subheader_line(line
);
765 fprintf(stderr
, "No patch found\n");
770 /* Non multipart message */
771 static void handle_body(void)
775 handle_commit_msg(&seen
);
779 fprintf(stderr
, "No patch found\n");
784 static const char mailinfo_usage
[] =
785 "git-mailinfo [-k] [-u | --encoding=<encoding>] msg patch <mail >info";
787 int main(int argc
, char **argv
)
789 /* NEEDSWORK: might want to do the optional .git/ directory
792 git_config(git_default_config
);
794 while (1 < argc
&& argv
[1][0] == '-') {
795 if (!strcmp(argv
[1], "-k"))
797 else if (!strcmp(argv
[1], "-u"))
798 metainfo_charset
= git_commit_encoding
;
799 else if (!strncmp(argv
[1], "--encoding=", 11))
800 metainfo_charset
= argv
[1] + 11;
802 usage(mailinfo_usage
);
807 usage(mailinfo_usage
);
808 cmitmsg
= fopen(argv
[1], "w");
813 patchfile
= fopen(argv
[2], "w");
819 int hdr
= read_one_header_line(line
, sizeof(line
), stdin
);
821 if (multipart_boundary
[0])
822 handle_multipart_body();
828 check_header_line(line
);