]> git.ipfire.org Git - thirdparty/git.git/blame - mailinfo.c
reftable: read reftable files
[thirdparty/git.git] / mailinfo.c
CommitLineData
c6905e45 1#include "cache.h"
b2141fc1 2#include "config.h"
c6905e45
JH
3#include "utf8.h"
4#include "strbuf.h"
5#include "mailinfo.h"
6
7static void cleanup_space(struct strbuf *sb)
8{
9 size_t pos, cnt;
10 for (pos = 0; pos < sb->len; pos++) {
11 if (isspace(sb->buf[pos])) {
12 sb->buf[pos] = ' ';
13 for (cnt = 0; isspace(sb->buf[pos + cnt + 1]); cnt++);
14 strbuf_remove(sb, pos + 1, cnt);
15 }
16 }
17}
18
19static void get_sane_name(struct strbuf *out, struct strbuf *name, struct strbuf *email)
20{
21 struct strbuf *src = name;
72ee47ce 22 if (!name->len || 60 < name->len || strpbrk(name->buf, "@<>"))
c6905e45
JH
23 src = email;
24 else if (name == out)
25 return;
26 strbuf_reset(out);
27 strbuf_addbuf(out, src);
28}
29
30static void parse_bogus_from(struct mailinfo *mi, const struct strbuf *line)
31{
32 /* John Doe <johndoe> */
33
34 char *bra, *ket;
35 /* This is fallback, so do not bother if we already have an
36 * e-mail address.
37 */
38 if (mi->email.len)
39 return;
40
41 bra = strchr(line->buf, '<');
42 if (!bra)
43 return;
44 ket = strchr(bra, '>');
45 if (!ket)
46 return;
47
48 strbuf_reset(&mi->email);
49 strbuf_add(&mi->email, bra + 1, ket - bra - 1);
50
51 strbuf_reset(&mi->name);
52 strbuf_add(&mi->name, line->buf, bra - line->buf);
53 strbuf_trim(&mi->name);
54 get_sane_name(&mi->name, &mi->name, &mi->email);
55}
56
f357e5de
KD
57static const char *unquote_comment(struct strbuf *outbuf, const char *in)
58{
59 int c;
64127575 60 int take_next_literally = 0;
f357e5de
KD
61
62 strbuf_addch(outbuf, '(');
63
64 while ((c = *in++) != 0) {
64127575
VS
65 if (take_next_literally == 1) {
66 take_next_literally = 0;
f357e5de
KD
67 } else {
68 switch (c) {
69 case '\\':
64127575 70 take_next_literally = 1;
f357e5de
KD
71 continue;
72 case '(':
73 in = unquote_comment(outbuf, in);
74 continue;
75 case ')':
76 strbuf_addch(outbuf, ')');
77 return in;
78 }
79 }
80
81 strbuf_addch(outbuf, c);
82 }
83
84 return in;
85}
86
87static const char *unquote_quoted_string(struct strbuf *outbuf, const char *in)
88{
89 int c;
64127575 90 int take_next_literally = 0;
f357e5de
KD
91
92 while ((c = *in++) != 0) {
64127575
VS
93 if (take_next_literally == 1) {
94 take_next_literally = 0;
f357e5de
KD
95 } else {
96 switch (c) {
97 case '\\':
64127575 98 take_next_literally = 1;
f357e5de
KD
99 continue;
100 case '"':
101 return in;
102 }
103 }
104
105 strbuf_addch(outbuf, c);
106 }
107
108 return in;
109}
110
111static void unquote_quoted_pair(struct strbuf *line)
112{
113 struct strbuf outbuf;
114 const char *in = line->buf;
115 int c;
116
117 strbuf_init(&outbuf, line->len);
118
119 while ((c = *in++) != 0) {
120 switch (c) {
121 case '"':
122 in = unquote_quoted_string(&outbuf, in);
123 continue;
124 case '(':
125 in = unquote_comment(&outbuf, in);
126 continue;
127 }
128
129 strbuf_addch(&outbuf, c);
130 }
131
132 strbuf_swap(&outbuf, line);
133 strbuf_release(&outbuf);
134
135}
136
c6905e45
JH
137static void handle_from(struct mailinfo *mi, const struct strbuf *from)
138{
139 char *at;
140 size_t el;
141 struct strbuf f;
142
143 strbuf_init(&f, from->len);
144 strbuf_addbuf(&f, from);
145
f357e5de
KD
146 unquote_quoted_pair(&f);
147
c6905e45
JH
148 at = strchr(f.buf, '@');
149 if (!at) {
150 parse_bogus_from(mi, from);
11fa5e2a 151 goto out;
c6905e45
JH
152 }
153
154 /*
155 * If we already have one email, don't take any confusing lines
156 */
11fa5e2a
RS
157 if (mi->email.len && strchr(at + 1, '@'))
158 goto out;
c6905e45
JH
159
160 /* Pick up the string around '@', possibly delimited with <>
161 * pair; that is the email part.
162 */
163 while (at > f.buf) {
164 char c = at[-1];
165 if (isspace(c))
166 break;
167 if (c == '<') {
168 at[-1] = ' ';
169 break;
170 }
171 at--;
172 }
173 el = strcspn(at, " \n\t\r\v\f>");
174 strbuf_reset(&mi->email);
175 strbuf_add(&mi->email, at, el);
176 strbuf_remove(&f, at - f.buf, el + (at[el] ? 1 : 0));
177
178 /* The remainder is name. It could be
179 *
180 * - "John Doe <john.doe@xz>" (a), or
181 * - "john.doe@xz (John Doe)" (b), or
182 * - "John (zzz) Doe <john.doe@xz> (Comment)" (c)
183 *
184 * but we have removed the email part, so
185 *
186 * - remove extra spaces which could stay after email (case 'c'), and
187 * - trim from both ends, possibly removing the () pair at the end
188 * (cases 'a' and 'b').
189 */
190 cleanup_space(&f);
191 strbuf_trim(&f);
192 if (f.buf[0] == '(' && f.len && f.buf[f.len - 1] == ')') {
193 strbuf_remove(&f, 0, 1);
194 strbuf_setlen(&f, f.len - 1);
195 }
196
197 get_sane_name(&mi->name, &f, &mi->email);
11fa5e2a 198out:
c6905e45
JH
199 strbuf_release(&f);
200}
201
202static void handle_header(struct strbuf **out, const struct strbuf *line)
203{
204 if (!*out) {
205 *out = xmalloc(sizeof(struct strbuf));
206 strbuf_init(*out, line->len);
207 } else
208 strbuf_reset(*out);
209
210 strbuf_addbuf(*out, line);
211}
212
213/* NOTE NOTE NOTE. We do not claim we do full MIME. We just attempt
214 * to have enough heuristics to grok MIME encoded patches often found
215 * on our mailing lists. For example, we do not even treat header lines
216 * case insensitively.
217 */
218
219static int slurp_attr(const char *line, const char *name, struct strbuf *attr)
220{
221 const char *ends, *ap = strcasestr(line, name);
222 size_t sz;
223
224 strbuf_setlen(attr, 0);
225 if (!ap)
226 return 0;
227 ap += strlen(name);
228 if (*ap == '"') {
229 ap++;
230 ends = "\"";
231 }
232 else
233 ends = "; \t";
234 sz = strcspn(ap, ends);
235 strbuf_add(attr, ap, sz);
236 return 1;
237}
238
3aa4d81f
RS
239static int has_attr_value(const char *line, const char *name, const char *value)
240{
241 struct strbuf sb = STRBUF_INIT;
242 int rc = slurp_attr(line, name, &sb) && !strcasecmp(sb.buf, value);
243 strbuf_release(&sb);
244 return rc;
245}
246
c6905e45
JH
247static void handle_content_type(struct mailinfo *mi, struct strbuf *line)
248{
249 struct strbuf *boundary = xmalloc(sizeof(struct strbuf));
250 strbuf_init(boundary, line->len);
251
3aa4d81f
RS
252 mi->format_flowed = has_attr_value(line->buf, "format=", "flowed");
253 mi->delsp = has_attr_value(line->buf, "delsp=", "yes");
254
c6905e45 255 if (slurp_attr(line->buf, "boundary=", boundary)) {
a91cc7fa 256 strbuf_insertstr(boundary, 0, "--");
c6905e45 257 if (++mi->content_top >= &mi->content[MAX_BOUNDARIES]) {
6ac617a3
JH
258 error("Too many boundaries to handle");
259 mi->input_error = -1;
260 mi->content_top = &mi->content[MAX_BOUNDARIES] - 1;
261 return;
c6905e45
JH
262 }
263 *(mi->content_top) = boundary;
264 boundary = NULL;
265 }
266 slurp_attr(line->buf, "charset=", &mi->charset);
267
268 if (boundary) {
269 strbuf_release(boundary);
270 free(boundary);
271 }
272}
273
c6905e45
JH
274static void handle_content_transfer_encoding(struct mailinfo *mi,
275 const struct strbuf *line)
276{
277 if (strcasestr(line->buf, "base64"))
278 mi->transfer_encoding = TE_BASE64;
279 else if (strcasestr(line->buf, "quoted-printable"))
280 mi->transfer_encoding = TE_QP;
281 else
282 mi->transfer_encoding = TE_DONTCARE;
283}
284
285static int is_multipart_boundary(struct mailinfo *mi, const struct strbuf *line)
286{
287 struct strbuf *content_top = *(mi->content_top);
288
289 return ((content_top->len <= line->len) &&
290 !memcmp(line->buf, content_top->buf, content_top->len));
291}
292
293static void cleanup_subject(struct mailinfo *mi, struct strbuf *subject)
294{
295 size_t at = 0;
296
297 while (at < subject->len) {
298 char *pos;
299 size_t remove;
300
301 switch (subject->buf[at]) {
302 case 'r': case 'R':
303 if (subject->len <= at + 3)
304 break;
305 if ((subject->buf[at + 1] == 'e' ||
306 subject->buf[at + 1] == 'E') &&
307 subject->buf[at + 2] == ':') {
308 strbuf_remove(subject, at, 3);
309 continue;
310 }
311 at++;
312 break;
313 case ' ': case '\t': case ':':
314 strbuf_remove(subject, at, 1);
315 continue;
316 case '[':
317 pos = strchr(subject->buf + at, ']');
318 if (!pos)
319 break;
320 remove = pos - subject->buf + at + 1;
321 if (!mi->keep_non_patch_brackets_in_subject ||
322 (7 <= remove &&
323 memmem(subject->buf + at, remove, "PATCH", 5)))
324 strbuf_remove(subject, at, remove);
325 else {
326 at += remove;
327 /*
328 * If the input had a space after the ], keep
329 * it. We don't bother with finding the end of
330 * the space, since we later normalize it
331 * anyway.
332 */
333 if (isspace(subject->buf[at]))
334 at += 1;
335 }
336 continue;
337 }
338 break;
339 }
340 strbuf_trim(subject);
341}
342
343#define MAX_HDR_PARSED 10
344static const char *header[MAX_HDR_PARSED] = {
345 "From","Subject","Date",
346};
347
f447d029
JK
348static inline int skip_header(const struct strbuf *line, const char *hdr,
349 const char **outval)
c6905e45 350{
f447d029
JK
351 const char *val;
352 if (!skip_iprefix(line->buf, hdr, &val) ||
ffbea181 353 *val++ != ':')
f447d029 354 return 0;
ffbea181
JK
355 while (isspace(*val))
356 val++;
f447d029
JK
357 *outval = val;
358 return 1;
c6905e45
JH
359}
360
361static int is_format_patch_separator(const char *line, int len)
362{
363 static const char SAMPLE[] =
364 "From e6807f3efca28b30decfecb1732a56c7db1137ee Mon Sep 17 00:00:00 2001\n";
365 const char *cp;
366
367 if (len != strlen(SAMPLE))
368 return 0;
369 if (!skip_prefix(line, "From ", &cp))
370 return 0;
371 if (strspn(cp, "0123456789abcdef") != 40)
372 return 0;
373 cp += 40;
374 return !memcmp(SAMPLE + (cp - line), cp, strlen(SAMPLE) - (cp - line));
375}
376
377static struct strbuf *decode_q_segment(const struct strbuf *q_seg, int rfc2047)
378{
379 const char *in = q_seg->buf;
380 int c;
381 struct strbuf *out = xmalloc(sizeof(struct strbuf));
382 strbuf_init(out, q_seg->len);
383
384 while ((c = *in++) != 0) {
385 if (c == '=') {
c8cf423e 386 int ch, d = *in;
c6905e45
JH
387 if (d == '\n' || !d)
388 break; /* drop trailing newline */
c8cf423e
RS
389 ch = hex2chr(in);
390 if (ch >= 0) {
391 strbuf_addch(out, ch);
392 in += 2;
393 continue;
394 }
395 /* garbage -- fall through */
c6905e45
JH
396 }
397 if (rfc2047 && c == '_') /* rfc2047 4.2 (2) */
398 c = 0x20;
399 strbuf_addch(out, c);
400 }
401 return out;
402}
403
404static struct strbuf *decode_b_segment(const struct strbuf *b_seg)
405{
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);
411
412 while ((c = *in++) != 0) {
413 if (c == '+')
414 c = 62;
415 else if (c == '/')
416 c = 63;
417 else if ('A' <= c && c <= 'Z')
418 c -= 'A';
419 else if ('a' <= c && c <= 'z')
420 c -= 'a' - 26;
421 else if ('0' <= c && c <= '9')
422 c -= '0' - 52;
423 else
424 continue; /* garbage */
425 switch (pos++) {
426 case 0:
427 acc = (c << 2);
428 break;
429 case 1:
430 strbuf_addch(out, (acc | (c >> 4)));
431 acc = (c & 15) << 4;
432 break;
433 case 2:
434 strbuf_addch(out, (acc | (c >> 2)));
435 acc = (c & 3) << 6;
436 break;
437 case 3:
438 strbuf_addch(out, (acc | c));
439 acc = pos = 0;
440 break;
441 }
442 }
443 return out;
444}
445
669b963a
JH
446static int convert_to_utf8(struct mailinfo *mi,
447 struct strbuf *line, const char *charset)
c6905e45
JH
448{
449 char *out;
2a2ff603 450 size_t out_len;
c6905e45
JH
451
452 if (!mi->metainfo_charset || !charset || !*charset)
669b963a 453 return 0;
c6905e45
JH
454
455 if (same_encoding(mi->metainfo_charset, charset))
669b963a 456 return 0;
2a2ff603
ĐTCD
457 out = reencode_string_len(line->buf, line->len,
458 mi->metainfo_charset, charset, &out_len);
6ac617a3
JH
459 if (!out) {
460 mi->input_error = -1;
669b963a
JH
461 return error("cannot convert from %s to %s",
462 charset, mi->metainfo_charset);
6ac617a3 463 }
2a2ff603 464 strbuf_attach(line, out, out_len, out_len);
669b963a 465 return 0;
c6905e45
JH
466}
467
468static void decode_header(struct mailinfo *mi, struct strbuf *it)
469{
470 char *in, *ep, *cp;
471 struct strbuf outbuf = STRBUF_INIT, *dec;
472 struct strbuf charset_q = STRBUF_INIT, piecebuf = STRBUF_INIT;
6ac617a3 473 int found_error = 1; /* pessimism */
c6905e45
JH
474
475 in = it->buf;
476 while (in - it->buf <= it->len && (ep = strstr(in, "=?")) != NULL) {
477 int encoding;
478 strbuf_reset(&charset_q);
479 strbuf_reset(&piecebuf);
480
481 if (in != ep) {
482 /*
483 * We are about to process an encoded-word
484 * that begins at ep, but there is something
485 * before the encoded word.
486 */
487 char *scan;
488 for (scan = in; scan < ep; scan++)
489 if (!isspace(*scan))
490 break;
491
492 if (scan != ep || in == it->buf) {
493 /*
494 * We should not lose that "something",
495 * unless we have just processed an
496 * encoded-word, and there is only LWS
497 * before the one we are about to process.
498 */
499 strbuf_add(&outbuf, in, ep - in);
500 }
501 }
502 /* E.g.
503 * ep : "=?iso-2022-jp?B?GyR...?= foo"
504 * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
505 */
506 ep += 2;
507
508 if (ep - it->buf >= it->len || !(cp = strchr(ep, '?')))
509 goto release_return;
510
511 if (cp + 3 - it->buf > it->len)
512 goto release_return;
513 strbuf_add(&charset_q, ep, cp - ep);
514
515 encoding = cp[1];
516 if (!encoding || cp[2] != '?')
517 goto release_return;
518 ep = strstr(cp + 3, "?=");
519 if (!ep)
520 goto release_return;
521 strbuf_add(&piecebuf, cp + 3, ep - cp - 3);
522 switch (tolower(encoding)) {
523 default:
524 goto release_return;
525 case 'b':
526 dec = decode_b_segment(&piecebuf);
527 break;
528 case 'q':
529 dec = decode_q_segment(&piecebuf, 1);
530 break;
531 }
669b963a
JH
532 if (convert_to_utf8(mi, dec, charset_q.buf))
533 goto release_return;
c6905e45
JH
534
535 strbuf_addbuf(&outbuf, dec);
536 strbuf_release(dec);
537 free(dec);
538 in = ep + 2;
539 }
540 strbuf_addstr(&outbuf, in);
541 strbuf_reset(it);
542 strbuf_addbuf(it, &outbuf);
6ac617a3 543 found_error = 0;
c6905e45
JH
544release_return:
545 strbuf_release(&outbuf);
546 strbuf_release(&charset_q);
547 strbuf_release(&piecebuf);
6ac617a3
JH
548
549 if (found_error)
550 mi->input_error = -1;
c6905e45
JH
551}
552
f696a2b1
JK
553/*
554 * Returns true if "line" contains a header matching "hdr", in which case "val"
555 * will contain the value of the header with any RFC2047 B and Q encoding
556 * unwrapped, and optionally normalize the meta information to utf8.
557 */
558static int parse_header(const struct strbuf *line,
559 const char *hdr,
560 struct mailinfo *mi,
561 struct strbuf *val)
562{
563 const char *val_str;
564
565 if (!skip_header(line, hdr, &val_str))
566 return 0;
567 strbuf_addstr(val, val_str);
568 decode_header(mi, val);
569 return 1;
570}
571
c6905e45
JH
572static int check_header(struct mailinfo *mi,
573 const struct strbuf *line,
574 struct strbuf *hdr_data[], int overwrite)
575{
f447d029 576 int i, ret = 0;
c6905e45
JH
577 struct strbuf sb = STRBUF_INIT;
578
579 /* search for the interesting parts */
580 for (i = 0; header[i]; i++) {
f447d029 581 if ((!hdr_data[i] || overwrite) &&
f696a2b1 582 parse_header(line, header[i], mi, &sb)) {
c6905e45
JH
583 handle_header(&hdr_data[i], &sb);
584 ret = 1;
585 goto check_header_out;
586 }
587 }
588
589 /* Content stuff */
f696a2b1 590 if (parse_header(line, "Content-Type", mi, &sb)) {
c6905e45
JH
591 handle_content_type(mi, &sb);
592 ret = 1;
593 goto check_header_out;
594 }
f696a2b1 595 if (parse_header(line, "Content-Transfer-Encoding", mi, &sb)) {
c6905e45
JH
596 handle_content_transfer_encoding(mi, &sb);
597 ret = 1;
598 goto check_header_out;
599 }
f696a2b1 600 if (parse_header(line, "Message-Id", mi, &sb)) {
ecf30b23
RS
601 if (mi->add_message_id)
602 mi->message_id = strbuf_detach(&sb, NULL);
c6905e45
JH
603 ret = 1;
604 goto check_header_out;
605 }
606
c6905e45
JH
607check_header_out:
608 strbuf_release(&sb);
609 return ret;
610}
611
6b4b013f
JT
612/*
613 * Returns 1 if the given line or any line beginning with the given line is an
614 * in-body header (that is, check_header will succeed when passed
615 * mi->s_hdr_data).
616 */
617static int is_inbody_header(const struct mailinfo *mi,
618 const struct strbuf *line)
619{
620 int i;
f447d029 621 const char *val;
6b4b013f 622 for (i = 0; header[i]; i++)
f447d029 623 if (!mi->s_hdr_data[i] && skip_header(line, header[i], &val))
6b4b013f
JT
624 return 1;
625 return 0;
626}
627
c6905e45
JH
628static void decode_transfer_encoding(struct mailinfo *mi, struct strbuf *line)
629{
630 struct strbuf *ret;
631
632 switch (mi->transfer_encoding) {
633 case TE_QP:
634 ret = decode_q_segment(line, 0);
635 break;
636 case TE_BASE64:
637 ret = decode_b_segment(line);
638 break;
639 case TE_DONTCARE:
640 default:
641 return;
642 }
643 strbuf_reset(line);
644 strbuf_addbuf(line, ret);
645 strbuf_release(ret);
646 free(ret);
647}
648
649static inline int patchbreak(const struct strbuf *line)
650{
651 size_t i;
652
653 /* Beginning of a "diff -" header? */
654 if (starts_with(line->buf, "diff -"))
655 return 1;
656
657 /* CVS "Index: " line? */
658 if (starts_with(line->buf, "Index: "))
659 return 1;
660
661 /*
662 * "--- <filename>" starts patches without headers
663 * "---<sp>*" is a manual separator
664 */
665 if (line->len < 4)
666 return 0;
667
668 if (starts_with(line->buf, "---")) {
669 /* space followed by a filename? */
670 if (line->buf[3] == ' ' && !isspace(line->buf[4]))
671 return 1;
672 /* Just whitespace? */
673 for (i = 3; i < line->len; i++) {
674 unsigned char c = line->buf[i];
675 if (c == '\n')
676 return 1;
677 if (!isspace(c))
678 break;
679 }
680 return 0;
681 }
682 return 0;
683}
684
9c5681da 685static int is_scissors_line(const char *line)
c6905e45 686{
9c5681da 687 const char *c;
c6905e45 688 int scissors = 0, gap = 0;
9c5681da
JT
689 const char *first_nonblank = NULL, *last_nonblank = NULL;
690 int visible, perforation = 0, in_perforation = 0;
c6905e45 691
9c5681da
JT
692 for (c = line; *c; c++) {
693 if (isspace(*c)) {
c6905e45
JH
694 if (in_perforation) {
695 perforation++;
696 gap++;
697 }
698 continue;
699 }
9c5681da
JT
700 last_nonblank = c;
701 if (first_nonblank == NULL)
702 first_nonblank = c;
703 if (*c == '-') {
c6905e45
JH
704 in_perforation = 1;
705 perforation++;
706 continue;
707 }
4184cbd6
AR
708 if (starts_with(c, ">8") || starts_with(c, "8<") ||
709 starts_with(c, ">%") || starts_with(c, "%<")) {
c6905e45
JH
710 in_perforation = 1;
711 perforation += 2;
712 scissors += 2;
9c5681da 713 c++;
c6905e45
JH
714 continue;
715 }
716 in_perforation = 0;
717 }
718
719 /*
720 * The mark must be at least 8 bytes long (e.g. "-- >8 --").
721 * Even though there can be arbitrary cruft on the same line
722 * (e.g. "cut here"), in order to avoid misidentification, the
723 * perforation must occupy more than a third of the visible
724 * width of the line, and dashes and scissors must occupy more
725 * than half of the perforation.
726 */
727
9c5681da
JT
728 if (first_nonblank && last_nonblank)
729 visible = last_nonblank - first_nonblank + 1;
730 else
731 visible = 0;
c6905e45
JH
732 return (scissors && 8 <= visible &&
733 visible < perforation * 3 &&
734 gap * 2 < perforation);
735}
736
6b4b013f
JT
737static void flush_inbody_header_accum(struct mailinfo *mi)
738{
739 if (!mi->inbody_header_accum.len)
740 return;
08414938 741 if (!check_header(mi, &mi->inbody_header_accum, mi->s_hdr_data, 0))
033abf97 742 BUG("inbody_header_accum, if not empty, must always contain a valid in-body header");
6b4b013f
JT
743 strbuf_reset(&mi->inbody_header_accum);
744}
745
334192b4
JT
746static int check_inbody_header(struct mailinfo *mi, const struct strbuf *line)
747{
6b4b013f
JT
748 if (mi->inbody_header_accum.len &&
749 (line->buf[0] == ' ' || line->buf[0] == '\t')) {
750 if (mi->use_scissors && is_scissors_line(line->buf)) {
751 /*
752 * This is a scissors line; do not consider this line
753 * as a header continuation line.
754 */
755 flush_inbody_header_accum(mi);
756 return 0;
757 }
758 strbuf_strip_suffix(&mi->inbody_header_accum, "\n");
759 strbuf_addbuf(&mi->inbody_header_accum, line);
760 return 1;
761 }
762
763 flush_inbody_header_accum(mi);
764
334192b4
JT
765 if (starts_with(line->buf, ">From") && isspace(line->buf[5]))
766 return is_format_patch_separator(line->buf + 1, line->len - 1);
767 if (starts_with(line->buf, "[PATCH]") && isspace(line->buf[7])) {
768 int i;
769 for (i = 0; header[i]; i++)
770 if (!strcmp("Subject", header[i])) {
771 handle_header(&mi->s_hdr_data[i], line);
772 return 1;
773 }
774 return 0;
775 }
6b4b013f
JT
776 if (is_inbody_header(mi, line)) {
777 strbuf_addbuf(&mi->inbody_header_accum, line);
778 return 1;
779 }
780 return 0;
334192b4
JT
781}
782
c6905e45
JH
783static int handle_commit_msg(struct mailinfo *mi, struct strbuf *line)
784{
785 assert(!mi->filter_stage);
786
787 if (mi->header_stage) {
fd1062e5
LT
788 if (!line->len || (line->len == 1 && line->buf[0] == '\n')) {
789 if (mi->inbody_header_accum.len) {
790 flush_inbody_header_accum(mi);
791 mi->header_stage = 0;
792 }
c6905e45 793 return 0;
fd1062e5 794 }
c6905e45
JH
795 }
796
797 if (mi->use_inbody_headers && mi->header_stage) {
334192b4 798 mi->header_stage = check_inbody_header(mi, line);
c6905e45
JH
799 if (mi->header_stage)
800 return 0;
801 } else
802 /* Only trim the first (blank) line of the commit message
803 * when ignoring in-body headers.
804 */
805 mi->header_stage = 0;
806
807 /* normalize the log message to UTF-8. */
669b963a 808 if (convert_to_utf8(mi, line, mi->charset.buf))
6ac617a3 809 return 0; /* mi->input_error already set */
c6905e45 810
9c5681da 811 if (mi->use_scissors && is_scissors_line(line->buf)) {
c6905e45
JH
812 int i;
813
814 strbuf_setlen(&mi->log_message, 0);
815 mi->header_stage = 1;
816
817 /*
818 * We may have already read "secondary headers"; purge
819 * them to give ourselves a clean restart.
820 */
821 for (i = 0; header[i]; i++) {
822 if (mi->s_hdr_data[i])
823 strbuf_release(mi->s_hdr_data[i]);
f3a96807 824 FREE_AND_NULL(mi->s_hdr_data[i]);
c6905e45
JH
825 }
826 return 0;
827 }
828
829 if (patchbreak(line)) {
830 if (mi->message_id)
831 strbuf_addf(&mi->log_message,
832 "Message-Id: %s\n", mi->message_id);
833 return 1;
834 }
835
836 strbuf_addbuf(&mi->log_message, line);
837 return 0;
838}
839
840static void handle_patch(struct mailinfo *mi, const struct strbuf *line)
841{
842 fwrite(line->buf, 1, line->len, mi->patchfile);
843 mi->patch_lines++;
844}
845
846static void handle_filter(struct mailinfo *mi, struct strbuf *line)
847{
848 switch (mi->filter_stage) {
849 case 0:
850 if (!handle_commit_msg(mi, line))
851 break;
852 mi->filter_stage++;
1cf01a34 853 /* fallthrough */
c6905e45
JH
854 case 1:
855 handle_patch(mi, line);
856 break;
857 }
858}
859
860static int is_rfc2822_header(const struct strbuf *line)
861{
862 /*
863 * The section that defines the loosest possible
864 * field name is "3.6.8 Optional fields".
865 *
866 * optional-field = field-name ":" unstructured CRLF
867 * field-name = 1*ftext
868 * ftext = %d33-57 / %59-126
869 */
870 int ch;
871 char *cp = line->buf;
872
873 /* Count mbox From headers as headers */
874 if (starts_with(cp, "From ") || starts_with(cp, ">From "))
875 return 1;
876
877 while ((ch = *cp++)) {
878 if (ch == ':')
879 return 1;
880 if ((33 <= ch && ch <= 57) ||
881 (59 <= ch && ch <= 126))
882 continue;
883 break;
884 }
885 return 0;
886}
887
888static int read_one_header_line(struct strbuf *line, FILE *in)
889{
890 struct strbuf continuation = STRBUF_INIT;
891
892 /* Get the first part of the line. */
8f309aeb 893 if (strbuf_getline_lf(line, in))
c6905e45
JH
894 return 0;
895
896 /*
897 * Is it an empty line or not a valid rfc2822 header?
898 * If so, stop here, and return false ("not a header")
899 */
900 strbuf_rtrim(line);
901 if (!line->len || !is_rfc2822_header(line)) {
902 /* Re-add the newline */
903 strbuf_addch(line, '\n');
904 return 0;
905 }
906
907 /*
908 * Now we need to eat all the continuation lines..
909 * Yuck, 2822 header "folding"
910 */
911 for (;;) {
912 int peek;
913
f0733c13
JS
914 peek = fgetc(in);
915 if (peek == EOF)
916 break;
917 ungetc(peek, in);
c6905e45
JH
918 if (peek != ' ' && peek != '\t')
919 break;
8f309aeb 920 if (strbuf_getline_lf(&continuation, in))
c6905e45
JH
921 break;
922 continuation.buf[0] = ' ';
923 strbuf_rtrim(&continuation);
924 strbuf_addbuf(line, &continuation);
925 }
926 strbuf_release(&continuation);
927
928 return 1;
929}
930
931static int find_boundary(struct mailinfo *mi, struct strbuf *line)
932{
8f309aeb 933 while (!strbuf_getline_lf(line, mi->input)) {
c6905e45
JH
934 if (*(mi->content_top) && is_multipart_boundary(mi, line))
935 return 1;
936 }
937 return 0;
938}
939
940static int handle_boundary(struct mailinfo *mi, struct strbuf *line)
941{
942 struct strbuf newline = STRBUF_INIT;
943
944 strbuf_addch(&newline, '\n');
945again:
946 if (line->len >= (*(mi->content_top))->len + 2 &&
947 !memcmp(line->buf + (*(mi->content_top))->len, "--", 2)) {
948 /* we hit an end boundary */
949 /* pop the current boundary off the stack */
950 strbuf_release(*(mi->content_top));
6a83d902 951 FREE_AND_NULL(*(mi->content_top));
c6905e45
JH
952
953 /* technically won't happen as is_multipart_boundary()
954 will fail first. But just in case..
955 */
956 if (--mi->content_top < mi->content) {
6ac617a3
JH
957 error("Detected mismatched boundaries, can't recover");
958 mi->input_error = -1;
959 mi->content_top = mi->content;
400cd6bf 960 strbuf_release(&newline);
6ac617a3 961 return 0;
c6905e45
JH
962 }
963 handle_filter(mi, &newline);
964 strbuf_release(&newline);
6ac617a3
JH
965 if (mi->input_error)
966 return 0;
c6905e45
JH
967
968 /* skip to the next boundary */
969 if (!find_boundary(mi, line))
970 return 0;
971 goto again;
972 }
973
974 /* set some defaults */
975 mi->transfer_encoding = TE_DONTCARE;
976 strbuf_reset(&mi->charset);
977
978 /* slurp in this section's info */
979 while (read_one_header_line(line, mi->input))
980 check_header(mi, line, mi->p_hdr_data, 0);
981
982 strbuf_release(&newline);
983 /* replenish line */
8f309aeb 984 if (strbuf_getline_lf(line, mi->input))
c6905e45
JH
985 return 0;
986 strbuf_addch(line, '\n');
987 return 1;
988}
989
3aa4d81f
RS
990static void handle_filter_flowed(struct mailinfo *mi, struct strbuf *line,
991 struct strbuf *prev)
992{
993 size_t len = line->len;
994 const char *rest;
995
996 if (!mi->format_flowed) {
0b689562
ĐTCD
997 if (len >= 2 &&
998 line->buf[len - 2] == '\r' &&
999 line->buf[len - 1] == '\n') {
1000 mi->have_quoted_cr = 1;
133a4fda
ĐTCD
1001 if (mi->quoted_cr == quoted_cr_strip) {
1002 strbuf_setlen(line, len - 2);
1003 strbuf_addch(line, '\n');
1004 len--;
1005 }
0b689562 1006 }
3aa4d81f
RS
1007 handle_filter(mi, line);
1008 return;
1009 }
1010
1011 if (line->buf[len - 1] == '\n') {
1012 len--;
1013 if (len && line->buf[len - 1] == '\r')
1014 len--;
1015 }
1016
1017 /* Keep signature separator as-is. */
1018 if (skip_prefix(line->buf, "-- ", &rest) && rest - line->buf == len) {
1019 if (prev->len) {
1020 handle_filter(mi, prev);
1021 strbuf_reset(prev);
1022 }
1023 handle_filter(mi, line);
1024 return;
1025 }
1026
1027 /* Unstuff space-stuffed line. */
1028 if (len && line->buf[0] == ' ') {
1029 strbuf_remove(line, 0, 1);
1030 len--;
1031 }
1032
1033 /* Save flowed line for later, but without the soft line break. */
1034 if (len && line->buf[len - 1] == ' ') {
1035 strbuf_add(prev, line->buf, len - !!mi->delsp);
1036 return;
1037 }
1038
1039 /* Prepend any previous partial lines */
1040 strbuf_insert(line, 0, prev->buf, prev->len);
1041 strbuf_reset(prev);
1042
1043 handle_filter(mi, line);
1044}
1045
0b689562
ĐTCD
1046static void summarize_quoted_cr(struct mailinfo *mi)
1047{
f1aa2994
ĐTCD
1048 if (mi->have_quoted_cr &&
1049 mi->quoted_cr == quoted_cr_warn)
0b689562
ĐTCD
1050 warning(_("quoted CRLF detected"));
1051}
1052
c6905e45
JH
1053static void handle_body(struct mailinfo *mi, struct strbuf *line)
1054{
1055 struct strbuf prev = STRBUF_INIT;
1056
1057 /* Skip up to the first boundary */
1058 if (*(mi->content_top)) {
1059 if (!find_boundary(mi, line))
1060 goto handle_body_out;
1061 }
1062
1063 do {
1064 /* process any boundary lines */
1065 if (*(mi->content_top) && is_multipart_boundary(mi, line)) {
1066 /* flush any leftover */
1067 if (prev.len) {
1068 handle_filter(mi, &prev);
1069 strbuf_reset(&prev);
1070 }
0b689562
ĐTCD
1071 summarize_quoted_cr(mi);
1072 mi->have_quoted_cr = 0;
c6905e45
JH
1073 if (!handle_boundary(mi, line))
1074 goto handle_body_out;
1075 }
1076
1077 /* Unwrap transfer encoding */
1078 decode_transfer_encoding(mi, line);
1079
1080 switch (mi->transfer_encoding) {
1081 case TE_BASE64:
1082 case TE_QP:
1083 {
1084 struct strbuf **lines, **it, *sb;
1085
1086 /* Prepend any previous partial lines */
1087 strbuf_insert(line, 0, prev.buf, prev.len);
1088 strbuf_reset(&prev);
1089
1090 /*
1091 * This is a decoded line that may contain
1092 * multiple new lines. Pass only one chunk
1093 * at a time to handle_filter()
1094 */
1095 lines = strbuf_split(line, '\n');
1096 for (it = lines; (sb = *it); it++) {
1097 if (*(it + 1) == NULL) /* The last line */
1098 if (sb->buf[sb->len - 1] != '\n') {
1099 /* Partial line, save it for later. */
1100 strbuf_addbuf(&prev, sb);
1101 break;
1102 }
3aa4d81f 1103 handle_filter_flowed(mi, sb, &prev);
c6905e45
JH
1104 }
1105 /*
1106 * The partial chunk is saved in "prev" and will be
1107 * appended by the next iteration of read_line_with_nul().
1108 */
1109 strbuf_list_free(lines);
1110 break;
1111 }
1112 default:
3aa4d81f 1113 handle_filter_flowed(mi, line, &prev);
c6905e45
JH
1114 }
1115
6ac617a3
JH
1116 if (mi->input_error)
1117 break;
c6905e45
JH
1118 } while (!strbuf_getwholeline(line, mi->input, '\n'));
1119
3aa4d81f
RS
1120 if (prev.len)
1121 handle_filter(mi, &prev);
0b689562 1122 summarize_quoted_cr(mi);
3aa4d81f 1123
6b4b013f
JT
1124 flush_inbody_header_accum(mi);
1125
c6905e45
JH
1126handle_body_out:
1127 strbuf_release(&prev);
1128}
1129
1130static void output_header_lines(FILE *fout, const char *hdr, const struct strbuf *data)
1131{
1132 const char *sp = data->buf;
1133 while (1) {
1134 char *ep = strchr(sp, '\n');
1135 int len;
1136 if (!ep)
1137 len = strlen(sp);
1138 else
1139 len = ep - sp;
1140 fprintf(fout, "%s: %.*s\n", hdr, len, sp);
1141 if (!ep)
1142 break;
1143 sp = ep + 1;
1144 }
1145}
1146
1147static void handle_info(struct mailinfo *mi)
1148{
1149 struct strbuf *hdr;
1150 int i;
1151
1152 for (i = 0; header[i]; i++) {
1153 /* only print inbody headers if we output a patch file */
1154 if (mi->patch_lines && mi->s_hdr_data[i])
1155 hdr = mi->s_hdr_data[i];
1156 else if (mi->p_hdr_data[i])
1157 hdr = mi->p_hdr_data[i];
1158 else
1159 continue;
1160
39199974
ĐTCD
1161 if (memchr(hdr->buf, '\0', hdr->len)) {
1162 error("a NUL byte in '%s' is not allowed.", header[i]);
1163 mi->input_error = -1;
1164 }
1165
c6905e45
JH
1166 if (!strcmp(header[i], "Subject")) {
1167 if (!mi->keep_subject) {
1168 cleanup_subject(mi, hdr);
1169 cleanup_space(hdr);
1170 }
1171 output_header_lines(mi->output, "Subject", hdr);
1172 } else if (!strcmp(header[i], "From")) {
1173 cleanup_space(hdr);
1174 handle_from(mi, hdr);
1175 fprintf(mi->output, "Author: %s\n", mi->name.buf);
1176 fprintf(mi->output, "Email: %s\n", mi->email.buf);
1177 } else {
1178 cleanup_space(hdr);
1179 fprintf(mi->output, "%s: %s\n", header[i], hdr->buf);
1180 }
1181 }
1182 fprintf(mi->output, "\n");
1183}
1184
1185int mailinfo(struct mailinfo *mi, const char *msg, const char *patch)
1186{
1187 FILE *cmitmsg;
1188 int peek;
1189 struct strbuf line = STRBUF_INIT;
1190
1191 cmitmsg = fopen(msg, "w");
1192 if (!cmitmsg) {
1193 perror(msg);
1194 return -1;
1195 }
1196 mi->patchfile = fopen(patch, "w");
1197 if (!mi->patchfile) {
1198 perror(patch);
1199 fclose(cmitmsg);
1200 return -1;
1201 }
1202
1203 mi->p_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(*(mi->p_hdr_data)));
1204 mi->s_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(*(mi->s_hdr_data)));
1205
1206 do {
1207 peek = fgetc(mi->input);
f0733c13
JS
1208 if (peek == EOF) {
1209 fclose(cmitmsg);
1210 return error("empty patch: '%s'", patch);
1211 }
c6905e45
JH
1212 } while (isspace(peek));
1213 ungetc(peek, mi->input);
1214
1215 /* process the email header */
1216 while (read_one_header_line(&line, mi->input))
1217 check_header(mi, &line, mi->p_hdr_data, 1);
1218
1219 handle_body(mi, &line);
1220 fwrite(mi->log_message.buf, 1, mi->log_message.len, cmitmsg);
1221 fclose(cmitmsg);
1222 fclose(mi->patchfile);
1223
1224 handle_info(mi);
1225 strbuf_release(&line);
6ac617a3 1226 return mi->input_error;
c6905e45
JH
1227}
1228
f1aa2994
ĐTCD
1229int mailinfo_parse_quoted_cr_action(const char *actionstr, int *action)
1230{
1231 if (!strcmp(actionstr, "nowarn"))
1232 *action = quoted_cr_nowarn;
1233 else if (!strcmp(actionstr, "warn"))
1234 *action = quoted_cr_warn;
133a4fda
ĐTCD
1235 else if (!strcmp(actionstr, "strip"))
1236 *action = quoted_cr_strip;
f1aa2994
ĐTCD
1237 else
1238 return -1;
1239 return 0;
1240}
1241
c6905e45
JH
1242static int git_mailinfo_config(const char *var, const char *value, void *mi_)
1243{
1244 struct mailinfo *mi = mi_;
1245
1246 if (!starts_with(var, "mailinfo."))
1247 return git_default_config(var, value, NULL);
1248 if (!strcmp(var, "mailinfo.scissors")) {
1249 mi->use_scissors = git_config_bool(var, value);
1250 return 0;
1251 }
f1aa2994
ĐTCD
1252 if (!strcmp(var, "mailinfo.quotedcr")) {
1253 if (mailinfo_parse_quoted_cr_action(value, &mi->quoted_cr) != 0)
1254 return error(_("bad action '%s' for '%s'"), value, var);
1255 return 0;
1256 }
c6905e45
JH
1257 /* perhaps others here */
1258 return 0;
1259}
1260
1261void setup_mailinfo(struct mailinfo *mi)
1262{
1263 memset(mi, 0, sizeof(*mi));
1264 strbuf_init(&mi->name, 0);
1265 strbuf_init(&mi->email, 0);
1266 strbuf_init(&mi->charset, 0);
1267 strbuf_init(&mi->log_message, 0);
6b4b013f 1268 strbuf_init(&mi->inbody_header_accum, 0);
f1aa2994 1269 mi->quoted_cr = quoted_cr_warn;
c6905e45
JH
1270 mi->header_stage = 1;
1271 mi->use_inbody_headers = 1;
1272 mi->content_top = mi->content;
85d9d9dd 1273 git_config(git_mailinfo_config, mi);
c6905e45
JH
1274}
1275
1276void clear_mailinfo(struct mailinfo *mi)
1277{
c6905e45
JH
1278 strbuf_release(&mi->name);
1279 strbuf_release(&mi->email);
1280 strbuf_release(&mi->charset);
6b4b013f 1281 strbuf_release(&mi->inbody_header_accum);
c6905e45
JH
1282 free(mi->message_id);
1283
f3a96807
AH
1284 strbuf_list_free(mi->p_hdr_data);
1285 strbuf_list_free(mi->s_hdr_data);
c6905e45
JH
1286
1287 while (mi->content < mi->content_top) {
1288 free(*(mi->content_top));
1289 mi->content_top--;
1290 }
1291
1292 strbuf_release(&mi->log_message);
1293}