7 static void cleanup_space(struct strbuf
*sb
)
10 for (pos
= 0; pos
< sb
->len
; pos
++) {
11 if (isspace(sb
->buf
[pos
])) {
13 for (cnt
= 0; isspace(sb
->buf
[pos
+ cnt
+ 1]); cnt
++);
14 strbuf_remove(sb
, pos
+ 1, cnt
);
19 static void get_sane_name(struct strbuf
*out
, struct strbuf
*name
, struct strbuf
*email
)
21 struct strbuf
*src
= name
;
22 if (name
->len
< 3 || 60 < name
->len
|| strchr(name
->buf
, '@') ||
23 strchr(name
->buf
, '<') || strchr(name
->buf
, '>'))
28 strbuf_addbuf(out
, src
);
31 static void parse_bogus_from(struct mailinfo
*mi
, const struct strbuf
*line
)
33 /* John Doe <johndoe> */
36 /* This is fallback, so do not bother if we already have an
42 bra
= strchr(line
->buf
, '<');
45 ket
= strchr(bra
, '>');
49 strbuf_reset(&mi
->email
);
50 strbuf_add(&mi
->email
, bra
+ 1, ket
- bra
- 1);
52 strbuf_reset(&mi
->name
);
53 strbuf_add(&mi
->name
, line
->buf
, bra
- line
->buf
);
54 strbuf_trim(&mi
->name
);
55 get_sane_name(&mi
->name
, &mi
->name
, &mi
->email
);
58 static const char *unquote_comment(struct strbuf
*outbuf
, const char *in
)
61 int take_next_literally
= 0;
63 strbuf_addch(outbuf
, '(');
65 while ((c
= *in
++) != 0) {
66 if (take_next_literally
== 1) {
67 take_next_literally
= 0;
71 take_next_literally
= 1;
74 in
= unquote_comment(outbuf
, in
);
77 strbuf_addch(outbuf
, ')');
82 strbuf_addch(outbuf
, c
);
88 static const char *unquote_quoted_string(struct strbuf
*outbuf
, const char *in
)
91 int take_next_literally
= 0;
93 while ((c
= *in
++) != 0) {
94 if (take_next_literally
== 1) {
95 take_next_literally
= 0;
99 take_next_literally
= 1;
106 strbuf_addch(outbuf
, c
);
112 static void unquote_quoted_pair(struct strbuf
*line
)
114 struct strbuf outbuf
;
115 const char *in
= line
->buf
;
118 strbuf_init(&outbuf
, line
->len
);
120 while ((c
= *in
++) != 0) {
123 in
= unquote_quoted_string(&outbuf
, in
);
126 in
= unquote_comment(&outbuf
, in
);
130 strbuf_addch(&outbuf
, c
);
133 strbuf_swap(&outbuf
, line
);
134 strbuf_release(&outbuf
);
138 static void handle_from(struct mailinfo
*mi
, const struct strbuf
*from
)
144 strbuf_init(&f
, from
->len
);
145 strbuf_addbuf(&f
, from
);
147 unquote_quoted_pair(&f
);
149 at
= strchr(f
.buf
, '@');
151 parse_bogus_from(mi
, from
);
156 * If we already have one email, don't take any confusing lines
158 if (mi
->email
.len
&& strchr(at
+ 1, '@'))
161 /* Pick up the string around '@', possibly delimited with <>
162 * pair; that is the email part.
174 el
= strcspn(at
, " \n\t\r\v\f>");
175 strbuf_reset(&mi
->email
);
176 strbuf_add(&mi
->email
, at
, el
);
177 strbuf_remove(&f
, at
- f
.buf
, el
+ (at
[el
] ? 1 : 0));
179 /* The remainder is name. It could be
181 * - "John Doe <john.doe@xz>" (a), or
182 * - "john.doe@xz (John Doe)" (b), or
183 * - "John (zzz) Doe <john.doe@xz> (Comment)" (c)
185 * but we have removed the email part, so
187 * - remove extra spaces which could stay after email (case 'c'), and
188 * - trim from both ends, possibly removing the () pair at the end
189 * (cases 'a' and 'b').
193 if (f
.buf
[0] == '(' && f
.len
&& f
.buf
[f
.len
- 1] == ')') {
194 strbuf_remove(&f
, 0, 1);
195 strbuf_setlen(&f
, f
.len
- 1);
198 get_sane_name(&mi
->name
, &f
, &mi
->email
);
203 static void handle_header(struct strbuf
**out
, const struct strbuf
*line
)
206 *out
= xmalloc(sizeof(struct strbuf
));
207 strbuf_init(*out
, line
->len
);
211 strbuf_addbuf(*out
, line
);
214 /* NOTE NOTE NOTE. We do not claim we do full MIME. We just attempt
215 * to have enough heuristics to grok MIME encoded patches often found
216 * on our mailing lists. For example, we do not even treat header lines
217 * case insensitively.
220 static int slurp_attr(const char *line
, const char *name
, struct strbuf
*attr
)
222 const char *ends
, *ap
= strcasestr(line
, name
);
225 strbuf_setlen(attr
, 0);
235 sz
= strcspn(ap
, ends
);
236 strbuf_add(attr
, ap
, sz
);
240 static int has_attr_value(const char *line
, const char *name
, const char *value
)
242 struct strbuf sb
= STRBUF_INIT
;
243 int rc
= slurp_attr(line
, name
, &sb
) && !strcasecmp(sb
.buf
, value
);
248 static void handle_content_type(struct mailinfo
*mi
, struct strbuf
*line
)
250 struct strbuf
*boundary
= xmalloc(sizeof(struct strbuf
));
251 strbuf_init(boundary
, line
->len
);
253 mi
->format_flowed
= has_attr_value(line
->buf
, "format=", "flowed");
254 mi
->delsp
= has_attr_value(line
->buf
, "delsp=", "yes");
256 if (slurp_attr(line
->buf
, "boundary=", boundary
)) {
257 strbuf_insertstr(boundary
, 0, "--");
258 if (++mi
->content_top
>= &mi
->content
[MAX_BOUNDARIES
]) {
259 error("Too many boundaries to handle");
260 mi
->input_error
= -1;
261 mi
->content_top
= &mi
->content
[MAX_BOUNDARIES
] - 1;
264 *(mi
->content_top
) = boundary
;
267 slurp_attr(line
->buf
, "charset=", &mi
->charset
);
270 strbuf_release(boundary
);
275 static void handle_content_transfer_encoding(struct mailinfo
*mi
,
276 const struct strbuf
*line
)
278 if (strcasestr(line
->buf
, "base64"))
279 mi
->transfer_encoding
= TE_BASE64
;
280 else if (strcasestr(line
->buf
, "quoted-printable"))
281 mi
->transfer_encoding
= TE_QP
;
283 mi
->transfer_encoding
= TE_DONTCARE
;
286 static int is_multipart_boundary(struct mailinfo
*mi
, const struct strbuf
*line
)
288 struct strbuf
*content_top
= *(mi
->content_top
);
290 return ((content_top
->len
<= line
->len
) &&
291 !memcmp(line
->buf
, content_top
->buf
, content_top
->len
));
294 static void cleanup_subject(struct mailinfo
*mi
, struct strbuf
*subject
)
298 while (at
< subject
->len
) {
302 switch (subject
->buf
[at
]) {
304 if (subject
->len
<= at
+ 3)
306 if ((subject
->buf
[at
+ 1] == 'e' ||
307 subject
->buf
[at
+ 1] == 'E') &&
308 subject
->buf
[at
+ 2] == ':') {
309 strbuf_remove(subject
, at
, 3);
314 case ' ': case '\t': case ':':
315 strbuf_remove(subject
, at
, 1);
318 pos
= strchr(subject
->buf
+ at
, ']');
321 remove
= pos
- subject
->buf
+ at
+ 1;
322 if (!mi
->keep_non_patch_brackets_in_subject
||
324 memmem(subject
->buf
+ at
, remove
, "PATCH", 5)))
325 strbuf_remove(subject
, at
, remove
);
329 * If the input had a space after the ], keep
330 * it. We don't bother with finding the end of
331 * the space, since we later normalize it
334 if (isspace(subject
->buf
[at
]))
341 strbuf_trim(subject
);
344 #define MAX_HDR_PARSED 10
345 static const char *header
[MAX_HDR_PARSED
] = {
346 "From","Subject","Date",
349 static inline int skip_header(const struct strbuf
*line
, const char *hdr
,
353 if (!skip_iprefix(line
->buf
, hdr
, &val
) ||
361 static int is_format_patch_separator(const char *line
, int len
)
363 static const char SAMPLE
[] =
364 "From e6807f3efca28b30decfecb1732a56c7db1137ee Mon Sep 17 00:00:00 2001\n";
367 if (len
!= strlen(SAMPLE
))
369 if (!skip_prefix(line
, "From ", &cp
))
371 if (strspn(cp
, "0123456789abcdef") != 40)
374 return !memcmp(SAMPLE
+ (cp
- line
), cp
, strlen(SAMPLE
) - (cp
- line
));
377 static struct strbuf
*decode_q_segment(const struct strbuf
*q_seg
, int rfc2047
)
379 const char *in
= q_seg
->buf
;
381 struct strbuf
*out
= xmalloc(sizeof(struct strbuf
));
382 strbuf_init(out
, q_seg
->len
);
384 while ((c
= *in
++) != 0) {
388 break; /* drop trailing newline */
391 strbuf_addch(out
, ch
);
395 /* garbage -- fall through */
397 if (rfc2047
&& c
== '_') /* rfc2047 4.2 (2) */
399 strbuf_addch(out
, c
);
404 static struct strbuf
*decode_b_segment(const struct strbuf
*b_seg
)
406 /* Decode in..ep, possibly in-place to ot */
407 int c
, pos
= 0, acc
= 0;
408 const char *in
= b_seg
->buf
;
409 struct strbuf
*out
= xmalloc(sizeof(struct strbuf
));
410 strbuf_init(out
, b_seg
->len
);
412 while ((c
= *in
++) != 0) {
417 else if ('A' <= c
&& c
<= 'Z')
419 else if ('a' <= c
&& c
<= 'z')
421 else if ('0' <= c
&& c
<= '9')
424 continue; /* garbage */
430 strbuf_addch(out
, (acc
| (c
>> 4)));
434 strbuf_addch(out
, (acc
| (c
>> 2)));
438 strbuf_addch(out
, (acc
| c
));
446 static int convert_to_utf8(struct mailinfo
*mi
,
447 struct strbuf
*line
, const char *charset
)
451 if (!mi
->metainfo_charset
|| !charset
|| !*charset
)
454 if (same_encoding(mi
->metainfo_charset
, charset
))
456 out
= reencode_string(line
->buf
, mi
->metainfo_charset
, charset
);
458 mi
->input_error
= -1;
459 return error("cannot convert from %s to %s",
460 charset
, mi
->metainfo_charset
);
462 strbuf_attach(line
, out
, strlen(out
), strlen(out
));
466 static void decode_header(struct mailinfo
*mi
, struct strbuf
*it
)
469 struct strbuf outbuf
= STRBUF_INIT
, *dec
;
470 struct strbuf charset_q
= STRBUF_INIT
, piecebuf
= STRBUF_INIT
;
471 int found_error
= 1; /* pessimism */
474 while (in
- it
->buf
<= it
->len
&& (ep
= strstr(in
, "=?")) != NULL
) {
476 strbuf_reset(&charset_q
);
477 strbuf_reset(&piecebuf
);
481 * We are about to process an encoded-word
482 * that begins at ep, but there is something
483 * before the encoded word.
486 for (scan
= in
; scan
< ep
; scan
++)
490 if (scan
!= ep
|| in
== it
->buf
) {
492 * We should not lose that "something",
493 * unless we have just processed an
494 * encoded-word, and there is only LWS
495 * before the one we are about to process.
497 strbuf_add(&outbuf
, in
, ep
- in
);
501 * ep : "=?iso-2022-jp?B?GyR...?= foo"
502 * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
506 if (ep
- it
->buf
>= it
->len
|| !(cp
= strchr(ep
, '?')))
509 if (cp
+ 3 - it
->buf
> it
->len
)
511 strbuf_add(&charset_q
, ep
, cp
- ep
);
514 if (!encoding
|| cp
[2] != '?')
516 ep
= strstr(cp
+ 3, "?=");
519 strbuf_add(&piecebuf
, cp
+ 3, ep
- cp
- 3);
520 switch (tolower(encoding
)) {
524 dec
= decode_b_segment(&piecebuf
);
527 dec
= decode_q_segment(&piecebuf
, 1);
530 if (convert_to_utf8(mi
, dec
, charset_q
.buf
))
533 strbuf_addbuf(&outbuf
, dec
);
538 strbuf_addstr(&outbuf
, in
);
540 strbuf_addbuf(it
, &outbuf
);
543 strbuf_release(&outbuf
);
544 strbuf_release(&charset_q
);
545 strbuf_release(&piecebuf
);
548 mi
->input_error
= -1;
551 static int check_header(struct mailinfo
*mi
,
552 const struct strbuf
*line
,
553 struct strbuf
*hdr_data
[], int overwrite
)
556 struct strbuf sb
= STRBUF_INIT
;
559 /* search for the interesting parts */
560 for (i
= 0; header
[i
]; i
++) {
561 if ((!hdr_data
[i
] || overwrite
) &&
562 skip_header(line
, header
[i
], &val
)) {
563 /* Unwrap inline B and Q encoding, and optionally
564 * normalize the meta information to utf8.
566 strbuf_addstr(&sb
, val
);
567 decode_header(mi
, &sb
);
568 handle_header(&hdr_data
[i
], &sb
);
570 goto check_header_out
;
575 if (skip_header(line
, "Content-Type", &val
)) {
576 strbuf_addstr(&sb
, val
);
577 decode_header(mi
, &sb
);
578 handle_content_type(mi
, &sb
);
580 goto check_header_out
;
582 if (skip_header(line
, "Content-Transfer-Encoding", &val
)) {
583 strbuf_addstr(&sb
, val
);
584 decode_header(mi
, &sb
);
585 handle_content_transfer_encoding(mi
, &sb
);
587 goto check_header_out
;
589 if (skip_header(line
, "Message-Id", &val
)) {
590 strbuf_addstr(&sb
, val
);
591 decode_header(mi
, &sb
);
592 if (mi
->add_message_id
)
593 mi
->message_id
= strbuf_detach(&sb
, NULL
);
595 goto check_header_out
;
604 * Returns 1 if the given line or any line beginning with the given line is an
605 * in-body header (that is, check_header will succeed when passed
608 static int is_inbody_header(const struct mailinfo
*mi
,
609 const struct strbuf
*line
)
613 for (i
= 0; header
[i
]; i
++)
614 if (!mi
->s_hdr_data
[i
] && skip_header(line
, header
[i
], &val
))
619 static void decode_transfer_encoding(struct mailinfo
*mi
, struct strbuf
*line
)
623 switch (mi
->transfer_encoding
) {
625 ret
= decode_q_segment(line
, 0);
628 ret
= decode_b_segment(line
);
635 strbuf_addbuf(line
, ret
);
640 static inline int patchbreak(const struct strbuf
*line
)
644 /* Beginning of a "diff -" header? */
645 if (starts_with(line
->buf
, "diff -"))
648 /* CVS "Index: " line? */
649 if (starts_with(line
->buf
, "Index: "))
653 * "--- <filename>" starts patches without headers
654 * "---<sp>*" is a manual separator
659 if (starts_with(line
->buf
, "---")) {
660 /* space followed by a filename? */
661 if (line
->buf
[3] == ' ' && !isspace(line
->buf
[4]))
663 /* Just whitespace? */
664 for (i
= 3; i
< line
->len
; i
++) {
665 unsigned char c
= line
->buf
[i
];
676 static int is_scissors_line(const char *line
)
679 int scissors
= 0, gap
= 0;
680 const char *first_nonblank
= NULL
, *last_nonblank
= NULL
;
681 int visible
, perforation
= 0, in_perforation
= 0;
683 for (c
= line
; *c
; c
++) {
685 if (in_perforation
) {
692 if (first_nonblank
== NULL
)
699 if ((!memcmp(c
, ">8", 2) || !memcmp(c
, "8<", 2) ||
700 !memcmp(c
, ">%", 2) || !memcmp(c
, "%<", 2))) {
711 * The mark must be at least 8 bytes long (e.g. "-- >8 --").
712 * Even though there can be arbitrary cruft on the same line
713 * (e.g. "cut here"), in order to avoid misidentification, the
714 * perforation must occupy more than a third of the visible
715 * width of the line, and dashes and scissors must occupy more
716 * than half of the perforation.
719 if (first_nonblank
&& last_nonblank
)
720 visible
= last_nonblank
- first_nonblank
+ 1;
723 return (scissors
&& 8 <= visible
&&
724 visible
< perforation
* 3 &&
725 gap
* 2 < perforation
);
728 static void flush_inbody_header_accum(struct mailinfo
*mi
)
730 if (!mi
->inbody_header_accum
.len
)
732 if (!check_header(mi
, &mi
->inbody_header_accum
, mi
->s_hdr_data
, 0))
733 BUG("inbody_header_accum, if not empty, must always contain a valid in-body header");
734 strbuf_reset(&mi
->inbody_header_accum
);
737 static int check_inbody_header(struct mailinfo
*mi
, const struct strbuf
*line
)
739 if (mi
->inbody_header_accum
.len
&&
740 (line
->buf
[0] == ' ' || line
->buf
[0] == '\t')) {
741 if (mi
->use_scissors
&& is_scissors_line(line
->buf
)) {
743 * This is a scissors line; do not consider this line
744 * as a header continuation line.
746 flush_inbody_header_accum(mi
);
749 strbuf_strip_suffix(&mi
->inbody_header_accum
, "\n");
750 strbuf_addbuf(&mi
->inbody_header_accum
, line
);
754 flush_inbody_header_accum(mi
);
756 if (starts_with(line
->buf
, ">From") && isspace(line
->buf
[5]))
757 return is_format_patch_separator(line
->buf
+ 1, line
->len
- 1);
758 if (starts_with(line
->buf
, "[PATCH]") && isspace(line
->buf
[7])) {
760 for (i
= 0; header
[i
]; i
++)
761 if (!strcmp("Subject", header
[i
])) {
762 handle_header(&mi
->s_hdr_data
[i
], line
);
767 if (is_inbody_header(mi
, line
)) {
768 strbuf_addbuf(&mi
->inbody_header_accum
, line
);
774 static int handle_commit_msg(struct mailinfo
*mi
, struct strbuf
*line
)
776 assert(!mi
->filter_stage
);
778 if (mi
->header_stage
) {
779 if (!line
->len
|| (line
->len
== 1 && line
->buf
[0] == '\n')) {
780 if (mi
->inbody_header_accum
.len
) {
781 flush_inbody_header_accum(mi
);
782 mi
->header_stage
= 0;
788 if (mi
->use_inbody_headers
&& mi
->header_stage
) {
789 mi
->header_stage
= check_inbody_header(mi
, line
);
790 if (mi
->header_stage
)
793 /* Only trim the first (blank) line of the commit message
794 * when ignoring in-body headers.
796 mi
->header_stage
= 0;
798 /* normalize the log message to UTF-8. */
799 if (convert_to_utf8(mi
, line
, mi
->charset
.buf
))
800 return 0; /* mi->input_error already set */
802 if (mi
->use_scissors
&& is_scissors_line(line
->buf
)) {
805 strbuf_setlen(&mi
->log_message
, 0);
806 mi
->header_stage
= 1;
809 * We may have already read "secondary headers"; purge
810 * them to give ourselves a clean restart.
812 for (i
= 0; header
[i
]; i
++) {
813 if (mi
->s_hdr_data
[i
])
814 strbuf_release(mi
->s_hdr_data
[i
]);
815 mi
->s_hdr_data
[i
] = NULL
;
820 if (patchbreak(line
)) {
822 strbuf_addf(&mi
->log_message
,
823 "Message-Id: %s\n", mi
->message_id
);
827 strbuf_addbuf(&mi
->log_message
, line
);
831 static void handle_patch(struct mailinfo
*mi
, const struct strbuf
*line
)
833 fwrite(line
->buf
, 1, line
->len
, mi
->patchfile
);
837 static void handle_filter(struct mailinfo
*mi
, struct strbuf
*line
)
839 switch (mi
->filter_stage
) {
841 if (!handle_commit_msg(mi
, line
))
846 handle_patch(mi
, line
);
851 static int is_rfc2822_header(const struct strbuf
*line
)
854 * The section that defines the loosest possible
855 * field name is "3.6.8 Optional fields".
857 * optional-field = field-name ":" unstructured CRLF
858 * field-name = 1*ftext
859 * ftext = %d33-57 / %59-126
862 char *cp
= line
->buf
;
864 /* Count mbox From headers as headers */
865 if (starts_with(cp
, "From ") || starts_with(cp
, ">From "))
868 while ((ch
= *cp
++)) {
871 if ((33 <= ch
&& ch
<= 57) ||
872 (59 <= ch
&& ch
<= 126))
879 static int read_one_header_line(struct strbuf
*line
, FILE *in
)
881 struct strbuf continuation
= STRBUF_INIT
;
883 /* Get the first part of the line. */
884 if (strbuf_getline_lf(line
, in
))
888 * Is it an empty line or not a valid rfc2822 header?
889 * If so, stop here, and return false ("not a header")
892 if (!line
->len
|| !is_rfc2822_header(line
)) {
893 /* Re-add the newline */
894 strbuf_addch(line
, '\n');
899 * Now we need to eat all the continuation lines..
900 * Yuck, 2822 header "folding"
909 if (peek
!= ' ' && peek
!= '\t')
911 if (strbuf_getline_lf(&continuation
, in
))
913 continuation
.buf
[0] = ' ';
914 strbuf_rtrim(&continuation
);
915 strbuf_addbuf(line
, &continuation
);
917 strbuf_release(&continuation
);
922 static int find_boundary(struct mailinfo
*mi
, struct strbuf
*line
)
924 while (!strbuf_getline_lf(line
, mi
->input
)) {
925 if (*(mi
->content_top
) && is_multipart_boundary(mi
, line
))
931 static int handle_boundary(struct mailinfo
*mi
, struct strbuf
*line
)
933 struct strbuf newline
= STRBUF_INIT
;
935 strbuf_addch(&newline
, '\n');
937 if (line
->len
>= (*(mi
->content_top
))->len
+ 2 &&
938 !memcmp(line
->buf
+ (*(mi
->content_top
))->len
, "--", 2)) {
939 /* we hit an end boundary */
940 /* pop the current boundary off the stack */
941 strbuf_release(*(mi
->content_top
));
942 FREE_AND_NULL(*(mi
->content_top
));
944 /* technically won't happen as is_multipart_boundary()
945 will fail first. But just in case..
947 if (--mi
->content_top
< mi
->content
) {
948 error("Detected mismatched boundaries, can't recover");
949 mi
->input_error
= -1;
950 mi
->content_top
= mi
->content
;
951 strbuf_release(&newline
);
954 handle_filter(mi
, &newline
);
955 strbuf_release(&newline
);
959 /* skip to the next boundary */
960 if (!find_boundary(mi
, line
))
965 /* set some defaults */
966 mi
->transfer_encoding
= TE_DONTCARE
;
967 strbuf_reset(&mi
->charset
);
969 /* slurp in this section's info */
970 while (read_one_header_line(line
, mi
->input
))
971 check_header(mi
, line
, mi
->p_hdr_data
, 0);
973 strbuf_release(&newline
);
975 if (strbuf_getline_lf(line
, mi
->input
))
977 strbuf_addch(line
, '\n');
981 static void handle_filter_flowed(struct mailinfo
*mi
, struct strbuf
*line
,
984 size_t len
= line
->len
;
987 if (!mi
->format_flowed
) {
988 handle_filter(mi
, line
);
992 if (line
->buf
[len
- 1] == '\n') {
994 if (len
&& line
->buf
[len
- 1] == '\r')
998 /* Keep signature separator as-is. */
999 if (skip_prefix(line
->buf
, "-- ", &rest
) && rest
- line
->buf
== len
) {
1001 handle_filter(mi
, prev
);
1004 handle_filter(mi
, line
);
1008 /* Unstuff space-stuffed line. */
1009 if (len
&& line
->buf
[0] == ' ') {
1010 strbuf_remove(line
, 0, 1);
1014 /* Save flowed line for later, but without the soft line break. */
1015 if (len
&& line
->buf
[len
- 1] == ' ') {
1016 strbuf_add(prev
, line
->buf
, len
- !!mi
->delsp
);
1020 /* Prepend any previous partial lines */
1021 strbuf_insert(line
, 0, prev
->buf
, prev
->len
);
1024 handle_filter(mi
, line
);
1027 static void handle_body(struct mailinfo
*mi
, struct strbuf
*line
)
1029 struct strbuf prev
= STRBUF_INIT
;
1031 /* Skip up to the first boundary */
1032 if (*(mi
->content_top
)) {
1033 if (!find_boundary(mi
, line
))
1034 goto handle_body_out
;
1038 /* process any boundary lines */
1039 if (*(mi
->content_top
) && is_multipart_boundary(mi
, line
)) {
1040 /* flush any leftover */
1042 handle_filter(mi
, &prev
);
1043 strbuf_reset(&prev
);
1045 if (!handle_boundary(mi
, line
))
1046 goto handle_body_out
;
1049 /* Unwrap transfer encoding */
1050 decode_transfer_encoding(mi
, line
);
1052 switch (mi
->transfer_encoding
) {
1056 struct strbuf
**lines
, **it
, *sb
;
1058 /* Prepend any previous partial lines */
1059 strbuf_insert(line
, 0, prev
.buf
, prev
.len
);
1060 strbuf_reset(&prev
);
1063 * This is a decoded line that may contain
1064 * multiple new lines. Pass only one chunk
1065 * at a time to handle_filter()
1067 lines
= strbuf_split(line
, '\n');
1068 for (it
= lines
; (sb
= *it
); it
++) {
1069 if (*(it
+ 1) == NULL
) /* The last line */
1070 if (sb
->buf
[sb
->len
- 1] != '\n') {
1071 /* Partial line, save it for later. */
1072 strbuf_addbuf(&prev
, sb
);
1075 handle_filter_flowed(mi
, sb
, &prev
);
1078 * The partial chunk is saved in "prev" and will be
1079 * appended by the next iteration of read_line_with_nul().
1081 strbuf_list_free(lines
);
1085 handle_filter_flowed(mi
, line
, &prev
);
1088 if (mi
->input_error
)
1090 } while (!strbuf_getwholeline(line
, mi
->input
, '\n'));
1093 handle_filter(mi
, &prev
);
1095 flush_inbody_header_accum(mi
);
1098 strbuf_release(&prev
);
1101 static void output_header_lines(FILE *fout
, const char *hdr
, const struct strbuf
*data
)
1103 const char *sp
= data
->buf
;
1105 char *ep
= strchr(sp
, '\n');
1111 fprintf(fout
, "%s: %.*s\n", hdr
, len
, sp
);
1118 static void handle_info(struct mailinfo
*mi
)
1123 for (i
= 0; header
[i
]; i
++) {
1124 /* only print inbody headers if we output a patch file */
1125 if (mi
->patch_lines
&& mi
->s_hdr_data
[i
])
1126 hdr
= mi
->s_hdr_data
[i
];
1127 else if (mi
->p_hdr_data
[i
])
1128 hdr
= mi
->p_hdr_data
[i
];
1132 if (!strcmp(header
[i
], "Subject")) {
1133 if (!mi
->keep_subject
) {
1134 cleanup_subject(mi
, hdr
);
1137 output_header_lines(mi
->output
, "Subject", hdr
);
1138 } else if (!strcmp(header
[i
], "From")) {
1140 handle_from(mi
, hdr
);
1141 fprintf(mi
->output
, "Author: %s\n", mi
->name
.buf
);
1142 fprintf(mi
->output
, "Email: %s\n", mi
->email
.buf
);
1145 fprintf(mi
->output
, "%s: %s\n", header
[i
], hdr
->buf
);
1148 fprintf(mi
->output
, "\n");
1151 int mailinfo(struct mailinfo
*mi
, const char *msg
, const char *patch
)
1155 struct strbuf line
= STRBUF_INIT
;
1157 cmitmsg
= fopen(msg
, "w");
1162 mi
->patchfile
= fopen(patch
, "w");
1163 if (!mi
->patchfile
) {
1169 mi
->p_hdr_data
= xcalloc(MAX_HDR_PARSED
, sizeof(*(mi
->p_hdr_data
)));
1170 mi
->s_hdr_data
= xcalloc(MAX_HDR_PARSED
, sizeof(*(mi
->s_hdr_data
)));
1173 peek
= fgetc(mi
->input
);
1176 return error("empty patch: '%s'", patch
);
1178 } while (isspace(peek
));
1179 ungetc(peek
, mi
->input
);
1181 /* process the email header */
1182 while (read_one_header_line(&line
, mi
->input
))
1183 check_header(mi
, &line
, mi
->p_hdr_data
, 1);
1185 handle_body(mi
, &line
);
1186 fwrite(mi
->log_message
.buf
, 1, mi
->log_message
.len
, cmitmsg
);
1188 fclose(mi
->patchfile
);
1191 strbuf_release(&line
);
1192 return mi
->input_error
;
1195 static int git_mailinfo_config(const char *var
, const char *value
, void *mi_
)
1197 struct mailinfo
*mi
= mi_
;
1199 if (!starts_with(var
, "mailinfo."))
1200 return git_default_config(var
, value
, NULL
);
1201 if (!strcmp(var
, "mailinfo.scissors")) {
1202 mi
->use_scissors
= git_config_bool(var
, value
);
1205 /* perhaps others here */
1209 void setup_mailinfo(struct mailinfo
*mi
)
1211 memset(mi
, 0, sizeof(*mi
));
1212 strbuf_init(&mi
->name
, 0);
1213 strbuf_init(&mi
->email
, 0);
1214 strbuf_init(&mi
->charset
, 0);
1215 strbuf_init(&mi
->log_message
, 0);
1216 strbuf_init(&mi
->inbody_header_accum
, 0);
1217 mi
->header_stage
= 1;
1218 mi
->use_inbody_headers
= 1;
1219 mi
->content_top
= mi
->content
;
1220 git_config(git_mailinfo_config
, mi
);
1223 void clear_mailinfo(struct mailinfo
*mi
)
1227 strbuf_release(&mi
->name
);
1228 strbuf_release(&mi
->email
);
1229 strbuf_release(&mi
->charset
);
1230 strbuf_release(&mi
->inbody_header_accum
);
1231 free(mi
->message_id
);
1234 for (i
= 0; mi
->p_hdr_data
[i
]; i
++)
1235 strbuf_release(mi
->p_hdr_data
[i
]);
1236 free(mi
->p_hdr_data
);
1238 for (i
= 0; mi
->s_hdr_data
[i
]; i
++)
1239 strbuf_release(mi
->s_hdr_data
[i
]);
1240 free(mi
->s_hdr_data
);
1242 while (mi
->content
< mi
->content_top
) {
1243 free(*(mi
->content_top
));
1247 strbuf_release(&mi
->log_message
);