freshlybakedca.ke
Git fork
1#define DISABLE_SIGN_COMPARE_WARNINGS
2
3#include "git-compat-util.h"
4#include "config.h"
5#include "environment.h"
6#include "gettext.h"
7#include "hex-ll.h"
8#include "utf8.h"
9#include "strbuf.h"
10#include "mailinfo.h"
11
12static void cleanup_space(struct strbuf *sb)
13{
14 size_t pos, cnt;
15 for (pos = 0; pos < sb->len; pos++) {
16 if (isspace(sb->buf[pos])) {
17 sb->buf[pos] = ' ';
18 for (cnt = 0; isspace(sb->buf[pos + cnt + 1]); cnt++);
19 strbuf_remove(sb, pos + 1, cnt);
20 }
21 }
22}
23
24static void get_sane_name(struct strbuf *out, struct strbuf *name, struct strbuf *email)
25{
26 struct strbuf *src = name;
27 if (!name->len || 60 < name->len || strpbrk(name->buf, "@<>"))
28 src = email;
29 else if (name == out)
30 return;
31 strbuf_reset(out);
32 strbuf_addbuf(out, src);
33}
34
35static void parse_bogus_from(struct mailinfo *mi, const struct strbuf *line)
36{
37 /* John Doe <johndoe> */
38
39 char *bra, *ket;
40 /* This is fallback, so do not bother if we already have an
41 * e-mail address.
42 */
43 if (mi->email.len)
44 return;
45
46 bra = strchr(line->buf, '<');
47 if (!bra)
48 return;
49 ket = strchr(bra, '>');
50 if (!ket)
51 return;
52
53 strbuf_reset(&mi->email);
54 strbuf_add(&mi->email, bra + 1, ket - bra - 1);
55
56 strbuf_reset(&mi->name);
57 strbuf_add(&mi->name, line->buf, bra - line->buf);
58 strbuf_trim(&mi->name);
59 get_sane_name(&mi->name, &mi->name, &mi->email);
60}
61
62static const char *unquote_comment(struct strbuf *outbuf, const char *in)
63{
64 int take_next_literally = 0;
65 int depth = 1;
66
67 strbuf_addch(outbuf, '(');
68
69 while (*in) {
70 int c = *in++;
71 if (take_next_literally == 1) {
72 take_next_literally = 0;
73 } else {
74 switch (c) {
75 case '\\':
76 take_next_literally = 1;
77 continue;
78 case '(':
79 strbuf_addch(outbuf, '(');
80 depth++;
81 continue;
82 case ')':
83 strbuf_addch(outbuf, ')');
84 if (!--depth)
85 return in;
86 continue;
87 }
88 }
89
90 strbuf_addch(outbuf, c);
91 }
92
93 return in;
94}
95
96static const char *unquote_quoted_string(struct strbuf *outbuf, const char *in)
97{
98 int take_next_literally = 0;
99
100 while (*in) {
101 int c = *in++;
102 if (take_next_literally == 1) {
103 take_next_literally = 0;
104 } else {
105 switch (c) {
106 case '\\':
107 take_next_literally = 1;
108 continue;
109 case '"':
110 return in;
111 }
112 }
113
114 strbuf_addch(outbuf, c);
115 }
116
117 return in;
118}
119
120static void unquote_quoted_pair(struct strbuf *line)
121{
122 struct strbuf outbuf;
123 const char *in = line->buf;
124 int c;
125
126 strbuf_init(&outbuf, line->len);
127
128 while ((c = *in++) != 0) {
129 switch (c) {
130 case '"':
131 in = unquote_quoted_string(&outbuf, in);
132 continue;
133 case '(':
134 in = unquote_comment(&outbuf, in);
135 continue;
136 }
137
138 strbuf_addch(&outbuf, c);
139 }
140
141 strbuf_swap(&outbuf, line);
142 strbuf_release(&outbuf);
143
144}
145
146static void handle_from(struct mailinfo *mi, const struct strbuf *from)
147{
148 char *at;
149 size_t el;
150 struct strbuf f;
151
152 strbuf_init(&f, from->len);
153 strbuf_addbuf(&f, from);
154
155 unquote_quoted_pair(&f);
156
157 at = strchr(f.buf, '@');
158 if (!at) {
159 parse_bogus_from(mi, from);
160 goto out;
161 }
162
163 /*
164 * If we already have one email, don't take any confusing lines
165 */
166 if (mi->email.len && strchr(at + 1, '@'))
167 goto out;
168
169 /* Pick up the string around '@', possibly delimited with <>
170 * pair; that is the email part.
171 */
172 while (at > f.buf) {
173 char c = at[-1];
174 if (isspace(c))
175 break;
176 if (c == '<') {
177 at[-1] = ' ';
178 break;
179 }
180 at--;
181 }
182 el = strcspn(at, " \n\t\r\v\f>");
183 strbuf_reset(&mi->email);
184 strbuf_add(&mi->email, at, el);
185 strbuf_remove(&f, at - f.buf, el + (at[el] ? 1 : 0));
186
187 /* The remainder is name. It could be
188 *
189 * - "John Doe <john.doe@xz>" (a), or
190 * - "john.doe@xz (John Doe)" (b), or
191 * - "John (zzz) Doe <john.doe@xz> (Comment)" (c)
192 *
193 * but we have removed the email part, so
194 *
195 * - remove extra spaces which could stay after email (case 'c'), and
196 * - trim from both ends, possibly removing the () pair at the end
197 * (cases 'a' and 'b').
198 */
199 cleanup_space(&f);
200 strbuf_trim(&f);
201 if (f.buf[0] == '(' && f.len && f.buf[f.len - 1] == ')') {
202 strbuf_remove(&f, 0, 1);
203 strbuf_setlen(&f, f.len - 1);
204 }
205
206 get_sane_name(&mi->name, &f, &mi->email);
207out:
208 strbuf_release(&f);
209}
210
211static void handle_header(struct strbuf **out, const struct strbuf *line)
212{
213 if (!*out) {
214 *out = xmalloc(sizeof(struct strbuf));
215 strbuf_init(*out, line->len);
216 } else
217 strbuf_reset(*out);
218
219 strbuf_addbuf(*out, line);
220}
221
222/* NOTE NOTE NOTE. We do not claim we do full MIME. We just attempt
223 * to have enough heuristics to grok MIME encoded patches often found
224 * on our mailing lists. For example, we do not even treat header lines
225 * case insensitively.
226 */
227
228static int slurp_attr(const char *line, const char *name, struct strbuf *attr)
229{
230 const char *ends, *ap = strcasestr(line, name);
231 size_t sz;
232
233 strbuf_setlen(attr, 0);
234 if (!ap)
235 return 0;
236 ap += strlen(name);
237 if (*ap == '"') {
238 ap++;
239 ends = "\"";
240 }
241 else
242 ends = "; \t";
243 sz = strcspn(ap, ends);
244 strbuf_add(attr, ap, sz);
245 return 1;
246}
247
248static int has_attr_value(const char *line, const char *name, const char *value)
249{
250 struct strbuf sb = STRBUF_INIT;
251 int rc = slurp_attr(line, name, &sb) && !strcasecmp(sb.buf, value);
252 strbuf_release(&sb);
253 return rc;
254}
255
256static void handle_content_type(struct mailinfo *mi, struct strbuf *line)
257{
258 struct strbuf *boundary = xmalloc(sizeof(struct strbuf));
259 strbuf_init(boundary, line->len);
260
261 mi->format_flowed = has_attr_value(line->buf, "format=", "flowed");
262 mi->delsp = has_attr_value(line->buf, "delsp=", "yes");
263
264 if (slurp_attr(line->buf, "boundary=", boundary)) {
265 strbuf_insertstr(boundary, 0, "--");
266 if (++mi->content_top >= &mi->content[MAX_BOUNDARIES]) {
267 error("Too many boundaries to handle");
268 mi->input_error = -1;
269 mi->content_top = &mi->content[MAX_BOUNDARIES] - 1;
270 strbuf_release(boundary);
271 free(boundary);
272 return;
273 }
274 *(mi->content_top) = boundary;
275 boundary = NULL;
276 }
277 slurp_attr(line->buf, "charset=", &mi->charset);
278
279 if (boundary) {
280 strbuf_release(boundary);
281 free(boundary);
282 }
283}
284
285static void handle_content_transfer_encoding(struct mailinfo *mi,
286 const struct strbuf *line)
287{
288 if (strcasestr(line->buf, "base64"))
289 mi->transfer_encoding = TE_BASE64;
290 else if (strcasestr(line->buf, "quoted-printable"))
291 mi->transfer_encoding = TE_QP;
292 else
293 mi->transfer_encoding = TE_DONTCARE;
294}
295
296static int is_multipart_boundary(struct mailinfo *mi, const struct strbuf *line)
297{
298 struct strbuf *content_top = *(mi->content_top);
299
300 return ((content_top->len <= line->len) &&
301 !memcmp(line->buf, content_top->buf, content_top->len));
302}
303
304static void cleanup_subject(struct mailinfo *mi, struct strbuf *subject)
305{
306 size_t at = 0;
307
308 while (at < subject->len) {
309 char *pos;
310 size_t remove;
311
312 switch (subject->buf[at]) {
313 case 'r': case 'R':
314 if (subject->len <= at + 3)
315 break;
316 if ((subject->buf[at + 1] == 'e' ||
317 subject->buf[at + 1] == 'E') &&
318 subject->buf[at + 2] == ':') {
319 strbuf_remove(subject, at, 3);
320 continue;
321 }
322 at++;
323 break;
324 case ' ': case '\t': case ':':
325 strbuf_remove(subject, at, 1);
326 continue;
327 case '[':
328 pos = strchr(subject->buf + at, ']');
329 if (!pos)
330 break;
331 remove = pos - (subject->buf + at) + 1;
332 if (!mi->keep_non_patch_brackets_in_subject ||
333 (7 <= remove &&
334 memmem(subject->buf + at, remove, "PATCH", 5)))
335 strbuf_remove(subject, at, remove);
336 else {
337 at += remove;
338 /*
339 * If the input had a space after the ], keep
340 * it. We don't bother with finding the end of
341 * the space, since we later normalize it
342 * anyway.
343 */
344 if (isspace(subject->buf[at]))
345 at += 1;
346 }
347 continue;
348 }
349 break;
350 }
351 strbuf_trim(subject);
352}
353
354static const char * const header[] = {
355 "From", "Subject", "Date",
356};
357
358static inline int skip_header(const struct strbuf *line, const char *hdr,
359 const char **outval)
360{
361 const char *val;
362 if (!skip_iprefix(line->buf, hdr, &val) ||
363 *val++ != ':')
364 return 0;
365 while (isspace(*val))
366 val++;
367 *outval = val;
368 return 1;
369}
370
371static int is_format_patch_separator(const char *line, int len)
372{
373 static const char SAMPLE[] =
374 "From e6807f3efca28b30decfecb1732a56c7db1137ee Mon Sep 17 00:00:00 2001\n";
375 const char *cp;
376
377 if (len != strlen(SAMPLE))
378 return 0;
379 if (!skip_prefix(line, "From ", &cp))
380 return 0;
381 if (strspn(cp, "0123456789abcdef") != 40)
382 return 0;
383 cp += 40;
384 return !memcmp(SAMPLE + (cp - line), cp, strlen(SAMPLE) - (cp - line));
385}
386
387static int decode_q_segment(struct strbuf *out, const struct strbuf *q_seg,
388 int rfc2047)
389{
390 const char *in = q_seg->buf;
391 int c;
392 strbuf_grow(out, q_seg->len);
393
394 while ((c = *in++) != 0) {
395 if (c == '=') {
396 int ch, d = *in;
397 if (d == '\n' || !d)
398 break; /* drop trailing newline */
399 ch = hex2chr(in);
400 if (ch >= 0) {
401 strbuf_addch(out, ch);
402 in += 2;
403 continue;
404 }
405 /* garbage -- fall through */
406 }
407 if (rfc2047 && c == '_') /* rfc2047 4.2 (2) */
408 c = 0x20;
409 strbuf_addch(out, c);
410 }
411 return 0;
412}
413
414static int decode_b_segment(struct strbuf *out, const struct strbuf *b_seg)
415{
416 /* Decode in..ep, possibly in-place to ot */
417 int c, pos = 0, acc = 0;
418 const char *in = b_seg->buf;
419 strbuf_grow(out, b_seg->len);
420
421 while ((c = *in++) != 0) {
422 if (c == '+')
423 c = 62;
424 else if (c == '/')
425 c = 63;
426 else if ('A' <= c && c <= 'Z')
427 c -= 'A';
428 else if ('a' <= c && c <= 'z')
429 c -= 'a' - 26;
430 else if ('0' <= c && c <= '9')
431 c -= '0' - 52;
432 else
433 continue; /* garbage */
434 switch (pos++) {
435 case 0:
436 acc = (c << 2);
437 break;
438 case 1:
439 strbuf_addch(out, (acc | (c >> 4)));
440 acc = (c & 15) << 4;
441 break;
442 case 2:
443 strbuf_addch(out, (acc | (c >> 2)));
444 acc = (c & 3) << 6;
445 break;
446 case 3:
447 strbuf_addch(out, (acc | c));
448 acc = pos = 0;
449 break;
450 }
451 }
452 return 0;
453}
454
455static int convert_to_utf8(struct mailinfo *mi,
456 struct strbuf *line, const char *charset)
457{
458 char *out;
459 size_t out_len;
460
461 if (!mi->metainfo_charset || !charset || !*charset)
462 return 0;
463
464 if (same_encoding(mi->metainfo_charset, charset))
465 return 0;
466 out = reencode_string_len(line->buf, line->len,
467 mi->metainfo_charset, charset, &out_len);
468 if (!out) {
469 mi->input_error = -1;
470 return error("cannot convert from %s to %s",
471 charset, mi->metainfo_charset);
472 }
473 strbuf_attach(line, out, out_len, out_len);
474 return 0;
475}
476
477static void decode_header(struct mailinfo *mi, struct strbuf *it)
478{
479 char *in, *ep, *cp;
480 struct strbuf outbuf = STRBUF_INIT, dec = STRBUF_INIT;
481 struct strbuf charset_q = STRBUF_INIT, piecebuf = STRBUF_INIT;
482 int found_error = 1; /* pessimism */
483
484 in = it->buf;
485 while (in - it->buf <= it->len && (ep = strstr(in, "=?")) != NULL) {
486 int encoding;
487 strbuf_reset(&charset_q);
488 strbuf_reset(&piecebuf);
489
490 if (in != ep) {
491 /*
492 * We are about to process an encoded-word
493 * that begins at ep, but there is something
494 * before the encoded word.
495 */
496 char *scan;
497 for (scan = in; scan < ep; scan++)
498 if (!isspace(*scan))
499 break;
500
501 if (scan != ep || in == it->buf) {
502 /*
503 * We should not lose that "something",
504 * unless we have just processed an
505 * encoded-word, and there is only LWS
506 * before the one we are about to process.
507 */
508 strbuf_add(&outbuf, in, ep - in);
509 }
510 }
511 /* E.g.
512 * ep : "=?iso-2022-jp?B?GyR...?= foo"
513 * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
514 */
515 ep += 2;
516
517 if (ep - it->buf >= it->len || !(cp = strchr(ep, '?')))
518 goto release_return;
519
520 if (cp + 3 - it->buf > it->len)
521 goto release_return;
522 strbuf_add(&charset_q, ep, cp - ep);
523
524 encoding = cp[1];
525 if (!encoding || cp[2] != '?')
526 goto release_return;
527 ep = strstr(cp + 3, "?=");
528 if (!ep)
529 goto release_return;
530 strbuf_add(&piecebuf, cp + 3, ep - cp - 3);
531 switch (tolower(encoding)) {
532 default:
533 goto release_return;
534 case 'b':
535 if ((found_error = decode_b_segment(&dec, &piecebuf)))
536 goto release_return;
537 break;
538 case 'q':
539 if ((found_error = decode_q_segment(&dec, &piecebuf, 1)))
540 goto release_return;
541 break;
542 }
543 if (convert_to_utf8(mi, &dec, charset_q.buf))
544 goto release_return;
545
546 strbuf_addbuf(&outbuf, &dec);
547 strbuf_release(&dec);
548 in = ep + 2;
549 }
550 strbuf_addstr(&outbuf, in);
551 strbuf_reset(it);
552 strbuf_addbuf(it, &outbuf);
553 found_error = 0;
554release_return:
555 strbuf_release(&outbuf);
556 strbuf_release(&charset_q);
557 strbuf_release(&piecebuf);
558 strbuf_release(&dec);
559
560 if (found_error)
561 mi->input_error = -1;
562}
563
564/*
565 * Returns true if "line" contains a header matching "hdr", in which case "val"
566 * will contain the value of the header with any RFC2047 B and Q encoding
567 * unwrapped, and optionally normalize the meta information to utf8.
568 */
569static int parse_header(const struct strbuf *line,
570 const char *hdr,
571 struct mailinfo *mi,
572 struct strbuf *val)
573{
574 const char *val_str;
575
576 if (!skip_header(line, hdr, &val_str))
577 return 0;
578 strbuf_addstr(val, val_str);
579 decode_header(mi, val);
580 return 1;
581}
582
583static int check_header(struct mailinfo *mi,
584 const struct strbuf *line,
585 struct strbuf *hdr_data[], int overwrite)
586{
587 int i, ret = 0;
588 struct strbuf sb = STRBUF_INIT;
589
590 /* search for the interesting parts */
591 for (i = 0; i < ARRAY_SIZE(header); i++) {
592 if ((!hdr_data[i] || overwrite) &&
593 parse_header(line, header[i], mi, &sb)) {
594 handle_header(&hdr_data[i], &sb);
595 ret = 1;
596 goto check_header_out;
597 }
598 }
599
600 /* Content stuff */
601 if (parse_header(line, "Content-Type", mi, &sb)) {
602 handle_content_type(mi, &sb);
603 ret = 1;
604 goto check_header_out;
605 }
606 if (parse_header(line, "Content-Transfer-Encoding", mi, &sb)) {
607 handle_content_transfer_encoding(mi, &sb);
608 ret = 1;
609 goto check_header_out;
610 }
611 if (parse_header(line, "Message-ID", mi, &sb)) {
612 if (mi->add_message_id)
613 mi->message_id = strbuf_detach(&sb, NULL);
614 ret = 1;
615 goto check_header_out;
616 }
617
618check_header_out:
619 strbuf_release(&sb);
620 return ret;
621}
622
623/*
624 * Returns 1 if the given line or any line beginning with the given line is an
625 * in-body header (that is, check_header will succeed when passed
626 * mi->s_hdr_data).
627 */
628static int is_inbody_header(const struct mailinfo *mi,
629 const struct strbuf *line)
630{
631 int i;
632 const char *val;
633 for (i = 0; i < ARRAY_SIZE(header); i++)
634 if (!mi->s_hdr_data[i] && skip_header(line, header[i], &val))
635 return 1;
636 return 0;
637}
638
639static void decode_transfer_encoding(struct mailinfo *mi, struct strbuf *line)
640{
641 struct strbuf ret = STRBUF_INIT;
642
643 switch (mi->transfer_encoding) {
644 case TE_QP:
645 decode_q_segment(&ret, line, 0);
646 break;
647 case TE_BASE64:
648 decode_b_segment(&ret, line);
649 break;
650 case TE_DONTCARE:
651 default:
652 return;
653 }
654 strbuf_reset(line);
655 strbuf_addbuf(line, &ret);
656 strbuf_release(&ret);
657}
658
659static inline int patchbreak(const struct strbuf *line)
660{
661 size_t i;
662
663 /* Beginning of a "diff -" header? */
664 if (starts_with(line->buf, "diff -"))
665 return 1;
666
667 /* CVS "Index: " line? */
668 if (starts_with(line->buf, "Index: "))
669 return 1;
670
671 /*
672 * "--- <filename>" starts patches without headers
673 * "---<sp>*" is a manual separator
674 */
675 if (line->len < 4)
676 return 0;
677
678 if (starts_with(line->buf, "---")) {
679 /* space followed by a filename? */
680 if (line->buf[3] == ' ' && !isspace(line->buf[4]))
681 return 1;
682 /* Just whitespace? */
683 for (i = 3; i < line->len; i++) {
684 unsigned char c = line->buf[i];
685 if (c == '\n')
686 return 1;
687 if (!isspace(c))
688 break;
689 }
690 return 0;
691 }
692 return 0;
693}
694
695static int is_scissors_line(const char *line)
696{
697 const char *c;
698 int scissors = 0, gap = 0;
699 const char *first_nonblank = NULL, *last_nonblank = NULL;
700 int visible, perforation = 0, in_perforation = 0;
701
702 for (c = line; *c; c++) {
703 if (isspace(*c)) {
704 if (in_perforation) {
705 perforation++;
706 gap++;
707 }
708 continue;
709 }
710 last_nonblank = c;
711 if (!first_nonblank)
712 first_nonblank = c;
713 if (*c == '-') {
714 in_perforation = 1;
715 perforation++;
716 continue;
717 }
718 if (starts_with(c, ">8") || starts_with(c, "8<") ||
719 starts_with(c, ">%") || starts_with(c, "%<")) {
720 in_perforation = 1;
721 perforation += 2;
722 scissors += 2;
723 c++;
724 continue;
725 }
726 in_perforation = 0;
727 }
728
729 /*
730 * The mark must be at least 8 bytes long (e.g. "-- >8 --").
731 * Even though there can be arbitrary cruft on the same line
732 * (e.g. "cut here"), in order to avoid misidentification, the
733 * perforation must occupy more than a third of the visible
734 * width of the line, and dashes and scissors must occupy more
735 * than half of the perforation.
736 */
737
738 if (first_nonblank && last_nonblank)
739 visible = last_nonblank - first_nonblank + 1;
740 else
741 visible = 0;
742 return (scissors && 8 <= visible &&
743 visible < perforation * 3 &&
744 gap * 2 < perforation);
745}
746
747static void flush_inbody_header_accum(struct mailinfo *mi)
748{
749 if (!mi->inbody_header_accum.len)
750 return;
751 if (!check_header(mi, &mi->inbody_header_accum, mi->s_hdr_data, 0))
752 BUG("inbody_header_accum, if not empty, must always contain a valid in-body header");
753 strbuf_reset(&mi->inbody_header_accum);
754}
755
756static int check_inbody_header(struct mailinfo *mi, const struct strbuf *line)
757{
758 if (mi->inbody_header_accum.len &&
759 (line->buf[0] == ' ' || line->buf[0] == '\t')) {
760 if (mi->use_scissors && is_scissors_line(line->buf)) {
761 /*
762 * This is a scissors line; do not consider this line
763 * as a header continuation line.
764 */
765 flush_inbody_header_accum(mi);
766 return 0;
767 }
768 strbuf_strip_suffix(&mi->inbody_header_accum, "\n");
769 strbuf_addbuf(&mi->inbody_header_accum, line);
770 return 1;
771 }
772
773 flush_inbody_header_accum(mi);
774
775 if (starts_with(line->buf, ">From") && isspace(line->buf[5]))
776 return is_format_patch_separator(line->buf + 1, line->len - 1);
777 if (starts_with(line->buf, "[PATCH]") && isspace(line->buf[7])) {
778 int i;
779 for (i = 0; i < ARRAY_SIZE(header); i++)
780 if (!strcmp("Subject", header[i])) {
781 handle_header(&mi->s_hdr_data[i], line);
782 return 1;
783 }
784 return 0;
785 }
786 if (is_inbody_header(mi, line)) {
787 strbuf_addbuf(&mi->inbody_header_accum, line);
788 return 1;
789 }
790 return 0;
791}
792
793static int handle_commit_msg(struct mailinfo *mi, struct strbuf *line)
794{
795 assert(!mi->filter_stage);
796
797 if (mi->header_stage) {
798 if (!line->len || (line->len == 1 && line->buf[0] == '\n')) {
799 if (mi->inbody_header_accum.len) {
800 flush_inbody_header_accum(mi);
801 mi->header_stage = 0;
802 }
803 return 0;
804 }
805 }
806
807 if (mi->use_inbody_headers && mi->header_stage) {
808 mi->header_stage = check_inbody_header(mi, line);
809 if (mi->header_stage)
810 return 0;
811 } else
812 /* Only trim the first (blank) line of the commit message
813 * when ignoring in-body headers.
814 */
815 mi->header_stage = 0;
816
817 /* normalize the log message to UTF-8. */
818 if (convert_to_utf8(mi, line, mi->charset.buf))
819 return 0; /* mi->input_error already set */
820
821 if (mi->use_scissors && is_scissors_line(line->buf)) {
822 int i;
823
824 strbuf_setlen(&mi->log_message, 0);
825 mi->header_stage = 1;
826
827 /*
828 * We may have already read "secondary headers"; purge
829 * them to give ourselves a clean restart.
830 */
831 for (i = 0; i < ARRAY_SIZE(header); i++) {
832 if (mi->s_hdr_data[i])
833 strbuf_release(mi->s_hdr_data[i]);
834 FREE_AND_NULL(mi->s_hdr_data[i]);
835 }
836 return 0;
837 }
838
839 if (patchbreak(line)) {
840 if (mi->message_id)
841 strbuf_addf(&mi->log_message,
842 "Message-ID: %s\n", mi->message_id);
843 return 1;
844 }
845
846 strbuf_addbuf(&mi->log_message, line);
847 return 0;
848}
849
850static void handle_patch(struct mailinfo *mi, const struct strbuf *line)
851{
852 fwrite(line->buf, 1, line->len, mi->patchfile);
853 mi->patch_lines++;
854}
855
856static void handle_filter(struct mailinfo *mi, struct strbuf *line)
857{
858 switch (mi->filter_stage) {
859 case 0:
860 if (!handle_commit_msg(mi, line))
861 break;
862 mi->filter_stage++;
863 /* fallthrough */
864 case 1:
865 handle_patch(mi, line);
866 break;
867 }
868}
869
870static int is_rfc2822_header(const struct strbuf *line)
871{
872 /*
873 * The section that defines the loosest possible
874 * field name is "3.6.8 Optional fields".
875 *
876 * optional-field = field-name ":" unstructured CRLF
877 * field-name = 1*ftext
878 * ftext = %d33-57 / %59-126
879 */
880 int ch;
881 char *cp = line->buf;
882
883 /* Count mbox From headers as headers */
884 if (starts_with(cp, "From ") || starts_with(cp, ">From "))
885 return 1;
886
887 while ((ch = *cp++)) {
888 if (ch == ':')
889 return 1;
890 if ((33 <= ch && ch <= 57) ||
891 (59 <= ch && ch <= 126))
892 continue;
893 break;
894 }
895 return 0;
896}
897
898static int read_one_header_line(struct strbuf *line, FILE *in)
899{
900 struct strbuf continuation = STRBUF_INIT;
901
902 /* Get the first part of the line. */
903 if (strbuf_getline_lf(line, in))
904 return 0;
905
906 /*
907 * Is it an empty line or not a valid rfc2822 header?
908 * If so, stop here, and return false ("not a header")
909 */
910 strbuf_rtrim(line);
911 if (!line->len || !is_rfc2822_header(line)) {
912 /* Re-add the newline */
913 strbuf_addch(line, '\n');
914 return 0;
915 }
916
917 /*
918 * Now we need to eat all the continuation lines..
919 * Yuck, 2822 header "folding"
920 */
921 for (;;) {
922 int peek;
923
924 peek = fgetc(in);
925 if (peek == EOF)
926 break;
927 ungetc(peek, in);
928 if (peek != ' ' && peek != '\t')
929 break;
930 if (strbuf_getline_lf(&continuation, in))
931 break;
932 continuation.buf[0] = ' ';
933 strbuf_rtrim(&continuation);
934 strbuf_addbuf(line, &continuation);
935 }
936 strbuf_release(&continuation);
937
938 return 1;
939}
940
941static int find_boundary(struct mailinfo *mi, struct strbuf *line)
942{
943 while (!strbuf_getline_lf(line, mi->input)) {
944 if (*(mi->content_top) && is_multipart_boundary(mi, line))
945 return 1;
946 }
947 return 0;
948}
949
950static int handle_boundary(struct mailinfo *mi, struct strbuf *line)
951{
952 struct strbuf newline = STRBUF_INIT;
953
954 strbuf_addch(&newline, '\n');
955again:
956 if (line->len >= (*(mi->content_top))->len + 2 &&
957 !memcmp(line->buf + (*(mi->content_top))->len, "--", 2)) {
958 /* we hit an end boundary */
959 /* pop the current boundary off the stack */
960 strbuf_release(*(mi->content_top));
961 FREE_AND_NULL(*(mi->content_top));
962
963 /* technically won't happen as is_multipart_boundary()
964 will fail first. But just in case..
965 */
966 if (--mi->content_top < mi->content) {
967 error("Detected mismatched boundaries, can't recover");
968 mi->input_error = -1;
969 mi->content_top = mi->content;
970 strbuf_release(&newline);
971 return 0;
972 }
973 handle_filter(mi, &newline);
974 strbuf_release(&newline);
975 if (mi->input_error)
976 return 0;
977
978 /* skip to the next boundary */
979 if (!find_boundary(mi, line))
980 return 0;
981 goto again;
982 }
983
984 /* set some defaults */
985 mi->transfer_encoding = TE_DONTCARE;
986 strbuf_reset(&mi->charset);
987
988 /* slurp in this section's info */
989 while (read_one_header_line(line, mi->input))
990 check_header(mi, line, mi->p_hdr_data, 0);
991
992 strbuf_release(&newline);
993 /* replenish line */
994 if (strbuf_getline_lf(line, mi->input))
995 return 0;
996 strbuf_addch(line, '\n');
997 return 1;
998}
999
1000static void handle_filter_flowed(struct mailinfo *mi, struct strbuf *line,
1001 struct strbuf *prev)
1002{
1003 size_t len = line->len;
1004 const char *rest;
1005
1006 if (!mi->format_flowed) {
1007 if (len >= 2 &&
1008 line->buf[len - 2] == '\r' &&
1009 line->buf[len - 1] == '\n') {
1010 mi->have_quoted_cr = 1;
1011 if (mi->quoted_cr == quoted_cr_strip) {
1012 strbuf_setlen(line, len - 2);
1013 strbuf_addch(line, '\n');
1014 len--;
1015 }
1016 }
1017 handle_filter(mi, line);
1018 return;
1019 }
1020
1021 if (line->buf[len - 1] == '\n') {
1022 len--;
1023 if (len && line->buf[len - 1] == '\r')
1024 len--;
1025 }
1026
1027 /* Keep signature separator as-is. */
1028 if (skip_prefix(line->buf, "-- ", &rest) && rest - line->buf == len) {
1029 if (prev->len) {
1030 handle_filter(mi, prev);
1031 strbuf_reset(prev);
1032 }
1033 handle_filter(mi, line);
1034 return;
1035 }
1036
1037 /* Unstuff space-stuffed line. */
1038 if (len && line->buf[0] == ' ') {
1039 strbuf_remove(line, 0, 1);
1040 len--;
1041 }
1042
1043 /* Save flowed line for later, but without the soft line break. */
1044 if (len && line->buf[len - 1] == ' ') {
1045 strbuf_add(prev, line->buf, len - !!mi->delsp);
1046 return;
1047 }
1048
1049 /* Prepend any previous partial lines */
1050 strbuf_insert(line, 0, prev->buf, prev->len);
1051 strbuf_reset(prev);
1052
1053 handle_filter(mi, line);
1054}
1055
1056static void summarize_quoted_cr(struct mailinfo *mi)
1057{
1058 if (mi->have_quoted_cr &&
1059 mi->quoted_cr == quoted_cr_warn)
1060 warning(_("quoted CRLF detected"));
1061}
1062
1063static void handle_body(struct mailinfo *mi, struct strbuf *line)
1064{
1065 struct strbuf prev = STRBUF_INIT;
1066
1067 /* Skip up to the first boundary */
1068 if (*(mi->content_top)) {
1069 if (!find_boundary(mi, line))
1070 goto handle_body_out;
1071 }
1072
1073 do {
1074 /* process any boundary lines */
1075 if (*(mi->content_top) && is_multipart_boundary(mi, line)) {
1076 /* flush any leftover */
1077 if (prev.len) {
1078 handle_filter(mi, &prev);
1079 strbuf_reset(&prev);
1080 }
1081 summarize_quoted_cr(mi);
1082 mi->have_quoted_cr = 0;
1083 if (!handle_boundary(mi, line))
1084 goto handle_body_out;
1085 }
1086
1087 /* Unwrap transfer encoding */
1088 decode_transfer_encoding(mi, line);
1089
1090 switch (mi->transfer_encoding) {
1091 case TE_BASE64:
1092 case TE_QP:
1093 {
1094 struct strbuf **lines, **it, *sb;
1095
1096 /* Prepend any previous partial lines */
1097 strbuf_insert(line, 0, prev.buf, prev.len);
1098 strbuf_reset(&prev);
1099
1100 /*
1101 * This is a decoded line that may contain
1102 * multiple new lines. Pass only one chunk
1103 * at a time to handle_filter()
1104 */
1105 lines = strbuf_split(line, '\n');
1106 for (it = lines; (sb = *it); it++) {
1107 if (!*(it + 1)) /* The last line */
1108 if (sb->buf[sb->len - 1] != '\n') {
1109 /* Partial line, save it for later. */
1110 strbuf_addbuf(&prev, sb);
1111 break;
1112 }
1113 handle_filter_flowed(mi, sb, &prev);
1114 }
1115 /*
1116 * The partial chunk is saved in "prev" and will be
1117 * appended by the next iteration of read_line_with_nul().
1118 */
1119 strbuf_list_free(lines);
1120 break;
1121 }
1122 default:
1123 handle_filter_flowed(mi, line, &prev);
1124 }
1125
1126 if (mi->input_error)
1127 break;
1128 } while (!strbuf_getwholeline(line, mi->input, '\n'));
1129
1130 if (prev.len)
1131 handle_filter(mi, &prev);
1132 summarize_quoted_cr(mi);
1133
1134 flush_inbody_header_accum(mi);
1135
1136handle_body_out:
1137 strbuf_release(&prev);
1138}
1139
1140static void output_header_lines(FILE *fout, const char *hdr, const struct strbuf *data)
1141{
1142 const char *sp = data->buf;
1143 while (1) {
1144 char *ep = strchr(sp, '\n');
1145 int len;
1146 if (!ep)
1147 len = strlen(sp);
1148 else
1149 len = ep - sp;
1150 fprintf(fout, "%s: %.*s\n", hdr, len, sp);
1151 if (!ep)
1152 break;
1153 sp = ep + 1;
1154 }
1155}
1156
1157static void handle_info(struct mailinfo *mi)
1158{
1159 struct strbuf *hdr;
1160 int i;
1161
1162 for (i = 0; i < ARRAY_SIZE(header); i++) {
1163 /* only print inbody headers if we output a patch file */
1164 if (mi->patch_lines && mi->s_hdr_data[i])
1165 hdr = mi->s_hdr_data[i];
1166 else if (mi->p_hdr_data[i])
1167 hdr = mi->p_hdr_data[i];
1168 else
1169 continue;
1170
1171 if (memchr(hdr->buf, '\0', hdr->len)) {
1172 error("a NUL byte in '%s' is not allowed.", header[i]);
1173 mi->input_error = -1;
1174 }
1175
1176 if (!strcmp(header[i], "Subject")) {
1177 if (!mi->keep_subject) {
1178 cleanup_subject(mi, hdr);
1179 cleanup_space(hdr);
1180 }
1181 output_header_lines(mi->output, "Subject", hdr);
1182 } else if (!strcmp(header[i], "From")) {
1183 cleanup_space(hdr);
1184 handle_from(mi, hdr);
1185 fprintf(mi->output, "Author: %s\n", mi->name.buf);
1186 fprintf(mi->output, "Email: %s\n", mi->email.buf);
1187 } else {
1188 cleanup_space(hdr);
1189 fprintf(mi->output, "%s: %s\n", header[i], hdr->buf);
1190 }
1191 }
1192 fprintf(mi->output, "\n");
1193}
1194
1195int mailinfo(struct mailinfo *mi, const char *msg, const char *patch)
1196{
1197 FILE *cmitmsg;
1198 int peek;
1199 struct strbuf line = STRBUF_INIT;
1200
1201 cmitmsg = fopen(msg, "w");
1202 if (!cmitmsg) {
1203 perror(msg);
1204 return -1;
1205 }
1206 mi->patchfile = fopen(patch, "w");
1207 if (!mi->patchfile) {
1208 perror(patch);
1209 fclose(cmitmsg);
1210 return -1;
1211 }
1212
1213 mi->p_hdr_data = xcalloc(ARRAY_SIZE(header), sizeof(*(mi->p_hdr_data)));
1214 mi->s_hdr_data = xcalloc(ARRAY_SIZE(header), sizeof(*(mi->s_hdr_data)));
1215
1216 do {
1217 peek = fgetc(mi->input);
1218 if (peek == EOF) {
1219 fclose(cmitmsg);
1220 return error("empty patch: '%s'", patch);
1221 }
1222 } while (isspace(peek));
1223 ungetc(peek, mi->input);
1224
1225 /* process the email header */
1226 while (read_one_header_line(&line, mi->input))
1227 check_header(mi, &line, mi->p_hdr_data, 1);
1228
1229 handle_body(mi, &line);
1230 fwrite(mi->log_message.buf, 1, mi->log_message.len, cmitmsg);
1231 fclose(cmitmsg);
1232 fclose(mi->patchfile);
1233
1234 handle_info(mi);
1235 strbuf_release(&line);
1236 return mi->input_error;
1237}
1238
1239int mailinfo_parse_quoted_cr_action(const char *actionstr, int *action)
1240{
1241 if (!strcmp(actionstr, "nowarn"))
1242 *action = quoted_cr_nowarn;
1243 else if (!strcmp(actionstr, "warn"))
1244 *action = quoted_cr_warn;
1245 else if (!strcmp(actionstr, "strip"))
1246 *action = quoted_cr_strip;
1247 else
1248 return -1;
1249 return 0;
1250}
1251
1252static int git_mailinfo_config(const char *var, const char *value,
1253 const struct config_context *ctx, void *mi_)
1254{
1255 struct mailinfo *mi = mi_;
1256
1257 if (!starts_with(var, "mailinfo."))
1258 return git_default_config(var, value, ctx, NULL);
1259 if (!strcmp(var, "mailinfo.scissors")) {
1260 mi->use_scissors = git_config_bool(var, value);
1261 return 0;
1262 }
1263 if (!strcmp(var, "mailinfo.quotedcr")) {
1264 if (!value)
1265 return config_error_nonbool(var);
1266 if (mailinfo_parse_quoted_cr_action(value, &mi->quoted_cr) != 0)
1267 return error(_("bad action '%s' for '%s'"), value, var);
1268 return 0;
1269 }
1270 /* perhaps others here */
1271 return 0;
1272}
1273
1274void setup_mailinfo(struct repository *r, struct mailinfo *mi)
1275{
1276 memset(mi, 0, sizeof(*mi));
1277 strbuf_init(&mi->name, 0);
1278 strbuf_init(&mi->email, 0);
1279 strbuf_init(&mi->charset, 0);
1280 strbuf_init(&mi->log_message, 0);
1281 strbuf_init(&mi->inbody_header_accum, 0);
1282 mi->quoted_cr = quoted_cr_warn;
1283 mi->header_stage = 1;
1284 mi->use_inbody_headers = 1;
1285 mi->content_top = mi->content;
1286 repo_config(r, git_mailinfo_config, mi);
1287}
1288
1289void clear_mailinfo(struct mailinfo *mi)
1290{
1291 strbuf_release(&mi->name);
1292 strbuf_release(&mi->email);
1293 strbuf_release(&mi->charset);
1294 strbuf_release(&mi->inbody_header_accum);
1295 free(mi->message_id);
1296
1297 for (size_t i = 0; i < ARRAY_SIZE(header); i++) {
1298 if (!mi->p_hdr_data[i])
1299 continue;
1300 strbuf_release(mi->p_hdr_data[i]);
1301 free(mi->p_hdr_data[i]);
1302 }
1303 free(mi->p_hdr_data);
1304
1305 for (size_t i = 0; i < ARRAY_SIZE(header); i++) {
1306 if (!mi->s_hdr_data[i])
1307 continue;
1308 strbuf_release(mi->s_hdr_data[i]);
1309 free(mi->s_hdr_data[i]);
1310 }
1311 free(mi->s_hdr_data);
1312
1313 while (mi->content < mi->content_top) {
1314 free(*(mi->content_top));
1315 mi->content_top--;
1316 }
1317
1318 strbuf_release(&mi->log_message);
1319}