]>
Commit | Line | Data |
---|---|---|
c6905e45 JH |
1 | #include "cache.h" |
2 | #include "utf8.h" | |
3 | #include "strbuf.h" | |
4 | #include "mailinfo.h" | |
5 | ||
6 | static void cleanup_space(struct strbuf *sb) | |
7 | { | |
8 | size_t pos, cnt; | |
9 | for (pos = 0; pos < sb->len; pos++) { | |
10 | if (isspace(sb->buf[pos])) { | |
11 | sb->buf[pos] = ' '; | |
12 | for (cnt = 0; isspace(sb->buf[pos + cnt + 1]); cnt++); | |
13 | strbuf_remove(sb, pos + 1, cnt); | |
14 | } | |
15 | } | |
16 | } | |
17 | ||
18 | static void get_sane_name(struct strbuf *out, struct strbuf *name, struct strbuf *email) | |
19 | { | |
20 | struct strbuf *src = name; | |
21 | if (name->len < 3 || 60 < name->len || strchr(name->buf, '@') || | |
22 | strchr(name->buf, '<') || strchr(name->buf, '>')) | |
23 | src = email; | |
24 | else if (name == out) | |
25 | return; | |
26 | strbuf_reset(out); | |
27 | strbuf_addbuf(out, src); | |
28 | } | |
29 | ||
30 | static 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 | ||
57 | static void handle_from(struct mailinfo *mi, const struct strbuf *from) | |
58 | { | |
59 | char *at; | |
60 | size_t el; | |
61 | struct strbuf f; | |
62 | ||
63 | strbuf_init(&f, from->len); | |
64 | strbuf_addbuf(&f, from); | |
65 | ||
66 | at = strchr(f.buf, '@'); | |
67 | if (!at) { | |
68 | parse_bogus_from(mi, from); | |
69 | return; | |
70 | } | |
71 | ||
72 | /* | |
73 | * If we already have one email, don't take any confusing lines | |
74 | */ | |
75 | if (mi->email.len && strchr(at + 1, '@')) { | |
76 | strbuf_release(&f); | |
77 | return; | |
78 | } | |
79 | ||
80 | /* Pick up the string around '@', possibly delimited with <> | |
81 | * pair; that is the email part. | |
82 | */ | |
83 | while (at > f.buf) { | |
84 | char c = at[-1]; | |
85 | if (isspace(c)) | |
86 | break; | |
87 | if (c == '<') { | |
88 | at[-1] = ' '; | |
89 | break; | |
90 | } | |
91 | at--; | |
92 | } | |
93 | el = strcspn(at, " \n\t\r\v\f>"); | |
94 | strbuf_reset(&mi->email); | |
95 | strbuf_add(&mi->email, at, el); | |
96 | strbuf_remove(&f, at - f.buf, el + (at[el] ? 1 : 0)); | |
97 | ||
98 | /* The remainder is name. It could be | |
99 | * | |
100 | * - "John Doe <john.doe@xz>" (a), or | |
101 | * - "john.doe@xz (John Doe)" (b), or | |
102 | * - "John (zzz) Doe <john.doe@xz> (Comment)" (c) | |
103 | * | |
104 | * but we have removed the email part, so | |
105 | * | |
106 | * - remove extra spaces which could stay after email (case 'c'), and | |
107 | * - trim from both ends, possibly removing the () pair at the end | |
108 | * (cases 'a' and 'b'). | |
109 | */ | |
110 | cleanup_space(&f); | |
111 | strbuf_trim(&f); | |
112 | if (f.buf[0] == '(' && f.len && f.buf[f.len - 1] == ')') { | |
113 | strbuf_remove(&f, 0, 1); | |
114 | strbuf_setlen(&f, f.len - 1); | |
115 | } | |
116 | ||
117 | get_sane_name(&mi->name, &f, &mi->email); | |
118 | strbuf_release(&f); | |
119 | } | |
120 | ||
121 | static void handle_header(struct strbuf **out, const struct strbuf *line) | |
122 | { | |
123 | if (!*out) { | |
124 | *out = xmalloc(sizeof(struct strbuf)); | |
125 | strbuf_init(*out, line->len); | |
126 | } else | |
127 | strbuf_reset(*out); | |
128 | ||
129 | strbuf_addbuf(*out, line); | |
130 | } | |
131 | ||
132 | /* NOTE NOTE NOTE. We do not claim we do full MIME. We just attempt | |
133 | * to have enough heuristics to grok MIME encoded patches often found | |
134 | * on our mailing lists. For example, we do not even treat header lines | |
135 | * case insensitively. | |
136 | */ | |
137 | ||
138 | static int slurp_attr(const char *line, const char *name, struct strbuf *attr) | |
139 | { | |
140 | const char *ends, *ap = strcasestr(line, name); | |
141 | size_t sz; | |
142 | ||
143 | strbuf_setlen(attr, 0); | |
144 | if (!ap) | |
145 | return 0; | |
146 | ap += strlen(name); | |
147 | if (*ap == '"') { | |
148 | ap++; | |
149 | ends = "\""; | |
150 | } | |
151 | else | |
152 | ends = "; \t"; | |
153 | sz = strcspn(ap, ends); | |
154 | strbuf_add(attr, ap, sz); | |
155 | return 1; | |
156 | } | |
157 | ||
158 | static void handle_content_type(struct mailinfo *mi, struct strbuf *line) | |
159 | { | |
160 | struct strbuf *boundary = xmalloc(sizeof(struct strbuf)); | |
161 | strbuf_init(boundary, line->len); | |
162 | ||
163 | if (slurp_attr(line->buf, "boundary=", boundary)) { | |
164 | strbuf_insert(boundary, 0, "--", 2); | |
165 | if (++mi->content_top >= &mi->content[MAX_BOUNDARIES]) { | |
6ac617a3 JH |
166 | error("Too many boundaries to handle"); |
167 | mi->input_error = -1; | |
168 | mi->content_top = &mi->content[MAX_BOUNDARIES] - 1; | |
169 | return; | |
c6905e45 JH |
170 | } |
171 | *(mi->content_top) = boundary; | |
172 | boundary = NULL; | |
173 | } | |
174 | slurp_attr(line->buf, "charset=", &mi->charset); | |
175 | ||
176 | if (boundary) { | |
177 | strbuf_release(boundary); | |
178 | free(boundary); | |
179 | } | |
180 | } | |
181 | ||
c6905e45 JH |
182 | static void handle_content_transfer_encoding(struct mailinfo *mi, |
183 | const struct strbuf *line) | |
184 | { | |
185 | if (strcasestr(line->buf, "base64")) | |
186 | mi->transfer_encoding = TE_BASE64; | |
187 | else if (strcasestr(line->buf, "quoted-printable")) | |
188 | mi->transfer_encoding = TE_QP; | |
189 | else | |
190 | mi->transfer_encoding = TE_DONTCARE; | |
191 | } | |
192 | ||
193 | static int is_multipart_boundary(struct mailinfo *mi, const struct strbuf *line) | |
194 | { | |
195 | struct strbuf *content_top = *(mi->content_top); | |
196 | ||
197 | return ((content_top->len <= line->len) && | |
198 | !memcmp(line->buf, content_top->buf, content_top->len)); | |
199 | } | |
200 | ||
201 | static void cleanup_subject(struct mailinfo *mi, struct strbuf *subject) | |
202 | { | |
203 | size_t at = 0; | |
204 | ||
205 | while (at < subject->len) { | |
206 | char *pos; | |
207 | size_t remove; | |
208 | ||
209 | switch (subject->buf[at]) { | |
210 | case 'r': case 'R': | |
211 | if (subject->len <= at + 3) | |
212 | break; | |
213 | if ((subject->buf[at + 1] == 'e' || | |
214 | subject->buf[at + 1] == 'E') && | |
215 | subject->buf[at + 2] == ':') { | |
216 | strbuf_remove(subject, at, 3); | |
217 | continue; | |
218 | } | |
219 | at++; | |
220 | break; | |
221 | case ' ': case '\t': case ':': | |
222 | strbuf_remove(subject, at, 1); | |
223 | continue; | |
224 | case '[': | |
225 | pos = strchr(subject->buf + at, ']'); | |
226 | if (!pos) | |
227 | break; | |
228 | remove = pos - subject->buf + at + 1; | |
229 | if (!mi->keep_non_patch_brackets_in_subject || | |
230 | (7 <= remove && | |
231 | memmem(subject->buf + at, remove, "PATCH", 5))) | |
232 | strbuf_remove(subject, at, remove); | |
233 | else { | |
234 | at += remove; | |
235 | /* | |
236 | * If the input had a space after the ], keep | |
237 | * it. We don't bother with finding the end of | |
238 | * the space, since we later normalize it | |
239 | * anyway. | |
240 | */ | |
241 | if (isspace(subject->buf[at])) | |
242 | at += 1; | |
243 | } | |
244 | continue; | |
245 | } | |
246 | break; | |
247 | } | |
248 | strbuf_trim(subject); | |
249 | } | |
250 | ||
251 | #define MAX_HDR_PARSED 10 | |
252 | static const char *header[MAX_HDR_PARSED] = { | |
253 | "From","Subject","Date", | |
254 | }; | |
255 | ||
256 | static inline int cmp_header(const struct strbuf *line, const char *hdr) | |
257 | { | |
258 | int len = strlen(hdr); | |
259 | return !strncasecmp(line->buf, hdr, len) && line->len > len && | |
260 | line->buf[len] == ':' && isspace(line->buf[len + 1]); | |
261 | } | |
262 | ||
263 | static int is_format_patch_separator(const char *line, int len) | |
264 | { | |
265 | static const char SAMPLE[] = | |
266 | "From e6807f3efca28b30decfecb1732a56c7db1137ee Mon Sep 17 00:00:00 2001\n"; | |
267 | const char *cp; | |
268 | ||
269 | if (len != strlen(SAMPLE)) | |
270 | return 0; | |
271 | if (!skip_prefix(line, "From ", &cp)) | |
272 | return 0; | |
273 | if (strspn(cp, "0123456789abcdef") != 40) | |
274 | return 0; | |
275 | cp += 40; | |
276 | return !memcmp(SAMPLE + (cp - line), cp, strlen(SAMPLE) - (cp - line)); | |
277 | } | |
278 | ||
279 | static struct strbuf *decode_q_segment(const struct strbuf *q_seg, int rfc2047) | |
280 | { | |
281 | const char *in = q_seg->buf; | |
282 | int c; | |
283 | struct strbuf *out = xmalloc(sizeof(struct strbuf)); | |
284 | strbuf_init(out, q_seg->len); | |
285 | ||
286 | while ((c = *in++) != 0) { | |
287 | if (c == '=') { | |
288 | int d = *in++; | |
289 | if (d == '\n' || !d) | |
290 | break; /* drop trailing newline */ | |
291 | strbuf_addch(out, (hexval(d) << 4) | hexval(*in++)); | |
292 | continue; | |
293 | } | |
294 | if (rfc2047 && c == '_') /* rfc2047 4.2 (2) */ | |
295 | c = 0x20; | |
296 | strbuf_addch(out, c); | |
297 | } | |
298 | return out; | |
299 | } | |
300 | ||
301 | static struct strbuf *decode_b_segment(const struct strbuf *b_seg) | |
302 | { | |
303 | /* Decode in..ep, possibly in-place to ot */ | |
304 | int c, pos = 0, acc = 0; | |
305 | const char *in = b_seg->buf; | |
306 | struct strbuf *out = xmalloc(sizeof(struct strbuf)); | |
307 | strbuf_init(out, b_seg->len); | |
308 | ||
309 | while ((c = *in++) != 0) { | |
310 | if (c == '+') | |
311 | c = 62; | |
312 | else if (c == '/') | |
313 | c = 63; | |
314 | else if ('A' <= c && c <= 'Z') | |
315 | c -= 'A'; | |
316 | else if ('a' <= c && c <= 'z') | |
317 | c -= 'a' - 26; | |
318 | else if ('0' <= c && c <= '9') | |
319 | c -= '0' - 52; | |
320 | else | |
321 | continue; /* garbage */ | |
322 | switch (pos++) { | |
323 | case 0: | |
324 | acc = (c << 2); | |
325 | break; | |
326 | case 1: | |
327 | strbuf_addch(out, (acc | (c >> 4))); | |
328 | acc = (c & 15) << 4; | |
329 | break; | |
330 | case 2: | |
331 | strbuf_addch(out, (acc | (c >> 2))); | |
332 | acc = (c & 3) << 6; | |
333 | break; | |
334 | case 3: | |
335 | strbuf_addch(out, (acc | c)); | |
336 | acc = pos = 0; | |
337 | break; | |
338 | } | |
339 | } | |
340 | return out; | |
341 | } | |
342 | ||
669b963a JH |
343 | static int convert_to_utf8(struct mailinfo *mi, |
344 | struct strbuf *line, const char *charset) | |
c6905e45 JH |
345 | { |
346 | char *out; | |
347 | ||
348 | if (!mi->metainfo_charset || !charset || !*charset) | |
669b963a | 349 | return 0; |
c6905e45 JH |
350 | |
351 | if (same_encoding(mi->metainfo_charset, charset)) | |
669b963a | 352 | return 0; |
c6905e45 | 353 | out = reencode_string(line->buf, mi->metainfo_charset, charset); |
6ac617a3 JH |
354 | if (!out) { |
355 | mi->input_error = -1; | |
669b963a JH |
356 | return error("cannot convert from %s to %s", |
357 | charset, mi->metainfo_charset); | |
6ac617a3 | 358 | } |
c6905e45 | 359 | strbuf_attach(line, out, strlen(out), strlen(out)); |
669b963a | 360 | return 0; |
c6905e45 JH |
361 | } |
362 | ||
363 | static void decode_header(struct mailinfo *mi, struct strbuf *it) | |
364 | { | |
365 | char *in, *ep, *cp; | |
366 | struct strbuf outbuf = STRBUF_INIT, *dec; | |
367 | struct strbuf charset_q = STRBUF_INIT, piecebuf = STRBUF_INIT; | |
6ac617a3 | 368 | int found_error = 1; /* pessimism */ |
c6905e45 JH |
369 | |
370 | in = it->buf; | |
371 | while (in - it->buf <= it->len && (ep = strstr(in, "=?")) != NULL) { | |
372 | int encoding; | |
373 | strbuf_reset(&charset_q); | |
374 | strbuf_reset(&piecebuf); | |
375 | ||
376 | if (in != ep) { | |
377 | /* | |
378 | * We are about to process an encoded-word | |
379 | * that begins at ep, but there is something | |
380 | * before the encoded word. | |
381 | */ | |
382 | char *scan; | |
383 | for (scan = in; scan < ep; scan++) | |
384 | if (!isspace(*scan)) | |
385 | break; | |
386 | ||
387 | if (scan != ep || in == it->buf) { | |
388 | /* | |
389 | * We should not lose that "something", | |
390 | * unless we have just processed an | |
391 | * encoded-word, and there is only LWS | |
392 | * before the one we are about to process. | |
393 | */ | |
394 | strbuf_add(&outbuf, in, ep - in); | |
395 | } | |
396 | } | |
397 | /* E.g. | |
398 | * ep : "=?iso-2022-jp?B?GyR...?= foo" | |
399 | * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz" | |
400 | */ | |
401 | ep += 2; | |
402 | ||
403 | if (ep - it->buf >= it->len || !(cp = strchr(ep, '?'))) | |
404 | goto release_return; | |
405 | ||
406 | if (cp + 3 - it->buf > it->len) | |
407 | goto release_return; | |
408 | strbuf_add(&charset_q, ep, cp - ep); | |
409 | ||
410 | encoding = cp[1]; | |
411 | if (!encoding || cp[2] != '?') | |
412 | goto release_return; | |
413 | ep = strstr(cp + 3, "?="); | |
414 | if (!ep) | |
415 | goto release_return; | |
416 | strbuf_add(&piecebuf, cp + 3, ep - cp - 3); | |
417 | switch (tolower(encoding)) { | |
418 | default: | |
419 | goto release_return; | |
420 | case 'b': | |
421 | dec = decode_b_segment(&piecebuf); | |
422 | break; | |
423 | case 'q': | |
424 | dec = decode_q_segment(&piecebuf, 1); | |
425 | break; | |
426 | } | |
669b963a JH |
427 | if (convert_to_utf8(mi, dec, charset_q.buf)) |
428 | goto release_return; | |
c6905e45 JH |
429 | |
430 | strbuf_addbuf(&outbuf, dec); | |
431 | strbuf_release(dec); | |
432 | free(dec); | |
433 | in = ep + 2; | |
434 | } | |
435 | strbuf_addstr(&outbuf, in); | |
436 | strbuf_reset(it); | |
437 | strbuf_addbuf(it, &outbuf); | |
6ac617a3 | 438 | found_error = 0; |
c6905e45 JH |
439 | release_return: |
440 | strbuf_release(&outbuf); | |
441 | strbuf_release(&charset_q); | |
442 | strbuf_release(&piecebuf); | |
6ac617a3 JH |
443 | |
444 | if (found_error) | |
445 | mi->input_error = -1; | |
c6905e45 JH |
446 | } |
447 | ||
448 | static int check_header(struct mailinfo *mi, | |
449 | const struct strbuf *line, | |
450 | struct strbuf *hdr_data[], int overwrite) | |
451 | { | |
452 | int i, ret = 0, len; | |
453 | struct strbuf sb = STRBUF_INIT; | |
454 | ||
455 | /* search for the interesting parts */ | |
456 | for (i = 0; header[i]; i++) { | |
457 | int len = strlen(header[i]); | |
458 | if ((!hdr_data[i] || overwrite) && cmp_header(line, header[i])) { | |
459 | /* Unwrap inline B and Q encoding, and optionally | |
460 | * normalize the meta information to utf8. | |
461 | */ | |
462 | strbuf_add(&sb, line->buf + len + 2, line->len - len - 2); | |
463 | decode_header(mi, &sb); | |
464 | handle_header(&hdr_data[i], &sb); | |
465 | ret = 1; | |
466 | goto check_header_out; | |
467 | } | |
468 | } | |
469 | ||
470 | /* Content stuff */ | |
471 | if (cmp_header(line, "Content-Type")) { | |
472 | len = strlen("Content-Type: "); | |
473 | strbuf_add(&sb, line->buf + len, line->len - len); | |
474 | decode_header(mi, &sb); | |
475 | strbuf_insert(&sb, 0, "Content-Type: ", len); | |
476 | handle_content_type(mi, &sb); | |
477 | ret = 1; | |
478 | goto check_header_out; | |
479 | } | |
480 | if (cmp_header(line, "Content-Transfer-Encoding")) { | |
481 | len = strlen("Content-Transfer-Encoding: "); | |
482 | strbuf_add(&sb, line->buf + len, line->len - len); | |
483 | decode_header(mi, &sb); | |
484 | handle_content_transfer_encoding(mi, &sb); | |
485 | ret = 1; | |
486 | goto check_header_out; | |
487 | } | |
488 | if (cmp_header(line, "Message-Id")) { | |
489 | len = strlen("Message-Id: "); | |
490 | strbuf_add(&sb, line->buf + len, line->len - len); | |
491 | decode_header(mi, &sb); | |
ecf30b23 RS |
492 | if (mi->add_message_id) |
493 | mi->message_id = strbuf_detach(&sb, NULL); | |
c6905e45 JH |
494 | ret = 1; |
495 | goto check_header_out; | |
496 | } | |
497 | ||
c6905e45 JH |
498 | check_header_out: |
499 | strbuf_release(&sb); | |
500 | return ret; | |
501 | } | |
502 | ||
503 | static void decode_transfer_encoding(struct mailinfo *mi, struct strbuf *line) | |
504 | { | |
505 | struct strbuf *ret; | |
506 | ||
507 | switch (mi->transfer_encoding) { | |
508 | case TE_QP: | |
509 | ret = decode_q_segment(line, 0); | |
510 | break; | |
511 | case TE_BASE64: | |
512 | ret = decode_b_segment(line); | |
513 | break; | |
514 | case TE_DONTCARE: | |
515 | default: | |
516 | return; | |
517 | } | |
518 | strbuf_reset(line); | |
519 | strbuf_addbuf(line, ret); | |
520 | strbuf_release(ret); | |
521 | free(ret); | |
522 | } | |
523 | ||
524 | static inline int patchbreak(const struct strbuf *line) | |
525 | { | |
526 | size_t i; | |
527 | ||
528 | /* Beginning of a "diff -" header? */ | |
529 | if (starts_with(line->buf, "diff -")) | |
530 | return 1; | |
531 | ||
532 | /* CVS "Index: " line? */ | |
533 | if (starts_with(line->buf, "Index: ")) | |
534 | return 1; | |
535 | ||
536 | /* | |
537 | * "--- <filename>" starts patches without headers | |
538 | * "---<sp>*" is a manual separator | |
539 | */ | |
540 | if (line->len < 4) | |
541 | return 0; | |
542 | ||
543 | if (starts_with(line->buf, "---")) { | |
544 | /* space followed by a filename? */ | |
545 | if (line->buf[3] == ' ' && !isspace(line->buf[4])) | |
546 | return 1; | |
547 | /* Just whitespace? */ | |
548 | for (i = 3; i < line->len; i++) { | |
549 | unsigned char c = line->buf[i]; | |
550 | if (c == '\n') | |
551 | return 1; | |
552 | if (!isspace(c)) | |
553 | break; | |
554 | } | |
555 | return 0; | |
556 | } | |
557 | return 0; | |
558 | } | |
559 | ||
9c5681da | 560 | static int is_scissors_line(const char *line) |
c6905e45 | 561 | { |
9c5681da | 562 | const char *c; |
c6905e45 | 563 | int scissors = 0, gap = 0; |
9c5681da JT |
564 | const char *first_nonblank = NULL, *last_nonblank = NULL; |
565 | int visible, perforation = 0, in_perforation = 0; | |
c6905e45 | 566 | |
9c5681da JT |
567 | for (c = line; *c; c++) { |
568 | if (isspace(*c)) { | |
c6905e45 JH |
569 | if (in_perforation) { |
570 | perforation++; | |
571 | gap++; | |
572 | } | |
573 | continue; | |
574 | } | |
9c5681da JT |
575 | last_nonblank = c; |
576 | if (first_nonblank == NULL) | |
577 | first_nonblank = c; | |
578 | if (*c == '-') { | |
c6905e45 JH |
579 | in_perforation = 1; |
580 | perforation++; | |
581 | continue; | |
582 | } | |
9c5681da JT |
583 | if ((!memcmp(c, ">8", 2) || !memcmp(c, "8<", 2) || |
584 | !memcmp(c, ">%", 2) || !memcmp(c, "%<", 2))) { | |
c6905e45 JH |
585 | in_perforation = 1; |
586 | perforation += 2; | |
587 | scissors += 2; | |
9c5681da | 588 | c++; |
c6905e45 JH |
589 | continue; |
590 | } | |
591 | in_perforation = 0; | |
592 | } | |
593 | ||
594 | /* | |
595 | * The mark must be at least 8 bytes long (e.g. "-- >8 --"). | |
596 | * Even though there can be arbitrary cruft on the same line | |
597 | * (e.g. "cut here"), in order to avoid misidentification, the | |
598 | * perforation must occupy more than a third of the visible | |
599 | * width of the line, and dashes and scissors must occupy more | |
600 | * than half of the perforation. | |
601 | */ | |
602 | ||
9c5681da JT |
603 | if (first_nonblank && last_nonblank) |
604 | visible = last_nonblank - first_nonblank + 1; | |
605 | else | |
606 | visible = 0; | |
c6905e45 JH |
607 | return (scissors && 8 <= visible && |
608 | visible < perforation * 3 && | |
609 | gap * 2 < perforation); | |
610 | } | |
611 | ||
334192b4 JT |
612 | static int check_inbody_header(struct mailinfo *mi, const struct strbuf *line) |
613 | { | |
614 | if (starts_with(line->buf, ">From") && isspace(line->buf[5])) | |
615 | return is_format_patch_separator(line->buf + 1, line->len - 1); | |
616 | if (starts_with(line->buf, "[PATCH]") && isspace(line->buf[7])) { | |
617 | int i; | |
618 | for (i = 0; header[i]; i++) | |
619 | if (!strcmp("Subject", header[i])) { | |
620 | handle_header(&mi->s_hdr_data[i], line); | |
621 | return 1; | |
622 | } | |
623 | return 0; | |
624 | } | |
625 | return check_header(mi, line, mi->s_hdr_data, 0); | |
626 | } | |
627 | ||
c6905e45 JH |
628 | static int handle_commit_msg(struct mailinfo *mi, struct strbuf *line) |
629 | { | |
630 | assert(!mi->filter_stage); | |
631 | ||
632 | if (mi->header_stage) { | |
633 | if (!line->len || (line->len == 1 && line->buf[0] == '\n')) | |
634 | return 0; | |
635 | } | |
636 | ||
637 | if (mi->use_inbody_headers && mi->header_stage) { | |
334192b4 | 638 | mi->header_stage = check_inbody_header(mi, line); |
c6905e45 JH |
639 | if (mi->header_stage) |
640 | return 0; | |
641 | } else | |
642 | /* Only trim the first (blank) line of the commit message | |
643 | * when ignoring in-body headers. | |
644 | */ | |
645 | mi->header_stage = 0; | |
646 | ||
647 | /* normalize the log message to UTF-8. */ | |
669b963a | 648 | if (convert_to_utf8(mi, line, mi->charset.buf)) |
6ac617a3 | 649 | return 0; /* mi->input_error already set */ |
c6905e45 | 650 | |
9c5681da | 651 | if (mi->use_scissors && is_scissors_line(line->buf)) { |
c6905e45 JH |
652 | int i; |
653 | ||
654 | strbuf_setlen(&mi->log_message, 0); | |
655 | mi->header_stage = 1; | |
656 | ||
657 | /* | |
658 | * We may have already read "secondary headers"; purge | |
659 | * them to give ourselves a clean restart. | |
660 | */ | |
661 | for (i = 0; header[i]; i++) { | |
662 | if (mi->s_hdr_data[i]) | |
663 | strbuf_release(mi->s_hdr_data[i]); | |
664 | mi->s_hdr_data[i] = NULL; | |
665 | } | |
666 | return 0; | |
667 | } | |
668 | ||
669 | if (patchbreak(line)) { | |
670 | if (mi->message_id) | |
671 | strbuf_addf(&mi->log_message, | |
672 | "Message-Id: %s\n", mi->message_id); | |
673 | return 1; | |
674 | } | |
675 | ||
676 | strbuf_addbuf(&mi->log_message, line); | |
677 | return 0; | |
678 | } | |
679 | ||
680 | static void handle_patch(struct mailinfo *mi, const struct strbuf *line) | |
681 | { | |
682 | fwrite(line->buf, 1, line->len, mi->patchfile); | |
683 | mi->patch_lines++; | |
684 | } | |
685 | ||
686 | static void handle_filter(struct mailinfo *mi, struct strbuf *line) | |
687 | { | |
688 | switch (mi->filter_stage) { | |
689 | case 0: | |
690 | if (!handle_commit_msg(mi, line)) | |
691 | break; | |
692 | mi->filter_stage++; | |
693 | case 1: | |
694 | handle_patch(mi, line); | |
695 | break; | |
696 | } | |
697 | } | |
698 | ||
699 | static int is_rfc2822_header(const struct strbuf *line) | |
700 | { | |
701 | /* | |
702 | * The section that defines the loosest possible | |
703 | * field name is "3.6.8 Optional fields". | |
704 | * | |
705 | * optional-field = field-name ":" unstructured CRLF | |
706 | * field-name = 1*ftext | |
707 | * ftext = %d33-57 / %59-126 | |
708 | */ | |
709 | int ch; | |
710 | char *cp = line->buf; | |
711 | ||
712 | /* Count mbox From headers as headers */ | |
713 | if (starts_with(cp, "From ") || starts_with(cp, ">From ")) | |
714 | return 1; | |
715 | ||
716 | while ((ch = *cp++)) { | |
717 | if (ch == ':') | |
718 | return 1; | |
719 | if ((33 <= ch && ch <= 57) || | |
720 | (59 <= ch && ch <= 126)) | |
721 | continue; | |
722 | break; | |
723 | } | |
724 | return 0; | |
725 | } | |
726 | ||
727 | static int read_one_header_line(struct strbuf *line, FILE *in) | |
728 | { | |
729 | struct strbuf continuation = STRBUF_INIT; | |
730 | ||
731 | /* Get the first part of the line. */ | |
8f309aeb | 732 | if (strbuf_getline_lf(line, in)) |
c6905e45 JH |
733 | return 0; |
734 | ||
735 | /* | |
736 | * Is it an empty line or not a valid rfc2822 header? | |
737 | * If so, stop here, and return false ("not a header") | |
738 | */ | |
739 | strbuf_rtrim(line); | |
740 | if (!line->len || !is_rfc2822_header(line)) { | |
741 | /* Re-add the newline */ | |
742 | strbuf_addch(line, '\n'); | |
743 | return 0; | |
744 | } | |
745 | ||
746 | /* | |
747 | * Now we need to eat all the continuation lines.. | |
748 | * Yuck, 2822 header "folding" | |
749 | */ | |
750 | for (;;) { | |
751 | int peek; | |
752 | ||
753 | peek = fgetc(in); ungetc(peek, in); | |
754 | if (peek != ' ' && peek != '\t') | |
755 | break; | |
8f309aeb | 756 | if (strbuf_getline_lf(&continuation, in)) |
c6905e45 JH |
757 | break; |
758 | continuation.buf[0] = ' '; | |
759 | strbuf_rtrim(&continuation); | |
760 | strbuf_addbuf(line, &continuation); | |
761 | } | |
762 | strbuf_release(&continuation); | |
763 | ||
764 | return 1; | |
765 | } | |
766 | ||
767 | static int find_boundary(struct mailinfo *mi, struct strbuf *line) | |
768 | { | |
8f309aeb | 769 | while (!strbuf_getline_lf(line, mi->input)) { |
c6905e45 JH |
770 | if (*(mi->content_top) && is_multipart_boundary(mi, line)) |
771 | return 1; | |
772 | } | |
773 | return 0; | |
774 | } | |
775 | ||
776 | static int handle_boundary(struct mailinfo *mi, struct strbuf *line) | |
777 | { | |
778 | struct strbuf newline = STRBUF_INIT; | |
779 | ||
780 | strbuf_addch(&newline, '\n'); | |
781 | again: | |
782 | if (line->len >= (*(mi->content_top))->len + 2 && | |
783 | !memcmp(line->buf + (*(mi->content_top))->len, "--", 2)) { | |
784 | /* we hit an end boundary */ | |
785 | /* pop the current boundary off the stack */ | |
786 | strbuf_release(*(mi->content_top)); | |
787 | free(*(mi->content_top)); | |
788 | *(mi->content_top) = NULL; | |
789 | ||
790 | /* technically won't happen as is_multipart_boundary() | |
791 | will fail first. But just in case.. | |
792 | */ | |
793 | if (--mi->content_top < mi->content) { | |
6ac617a3 JH |
794 | error("Detected mismatched boundaries, can't recover"); |
795 | mi->input_error = -1; | |
796 | mi->content_top = mi->content; | |
797 | return 0; | |
c6905e45 JH |
798 | } |
799 | handle_filter(mi, &newline); | |
800 | strbuf_release(&newline); | |
6ac617a3 JH |
801 | if (mi->input_error) |
802 | return 0; | |
c6905e45 JH |
803 | |
804 | /* skip to the next boundary */ | |
805 | if (!find_boundary(mi, line)) | |
806 | return 0; | |
807 | goto again; | |
808 | } | |
809 | ||
810 | /* set some defaults */ | |
811 | mi->transfer_encoding = TE_DONTCARE; | |
812 | strbuf_reset(&mi->charset); | |
813 | ||
814 | /* slurp in this section's info */ | |
815 | while (read_one_header_line(line, mi->input)) | |
816 | check_header(mi, line, mi->p_hdr_data, 0); | |
817 | ||
818 | strbuf_release(&newline); | |
819 | /* replenish line */ | |
8f309aeb | 820 | if (strbuf_getline_lf(line, mi->input)) |
c6905e45 JH |
821 | return 0; |
822 | strbuf_addch(line, '\n'); | |
823 | return 1; | |
824 | } | |
825 | ||
826 | static void handle_body(struct mailinfo *mi, struct strbuf *line) | |
827 | { | |
828 | struct strbuf prev = STRBUF_INIT; | |
829 | ||
830 | /* Skip up to the first boundary */ | |
831 | if (*(mi->content_top)) { | |
832 | if (!find_boundary(mi, line)) | |
833 | goto handle_body_out; | |
834 | } | |
835 | ||
836 | do { | |
837 | /* process any boundary lines */ | |
838 | if (*(mi->content_top) && is_multipart_boundary(mi, line)) { | |
839 | /* flush any leftover */ | |
840 | if (prev.len) { | |
841 | handle_filter(mi, &prev); | |
842 | strbuf_reset(&prev); | |
843 | } | |
844 | if (!handle_boundary(mi, line)) | |
845 | goto handle_body_out; | |
846 | } | |
847 | ||
848 | /* Unwrap transfer encoding */ | |
849 | decode_transfer_encoding(mi, line); | |
850 | ||
851 | switch (mi->transfer_encoding) { | |
852 | case TE_BASE64: | |
853 | case TE_QP: | |
854 | { | |
855 | struct strbuf **lines, **it, *sb; | |
856 | ||
857 | /* Prepend any previous partial lines */ | |
858 | strbuf_insert(line, 0, prev.buf, prev.len); | |
859 | strbuf_reset(&prev); | |
860 | ||
861 | /* | |
862 | * This is a decoded line that may contain | |
863 | * multiple new lines. Pass only one chunk | |
864 | * at a time to handle_filter() | |
865 | */ | |
866 | lines = strbuf_split(line, '\n'); | |
867 | for (it = lines; (sb = *it); it++) { | |
868 | if (*(it + 1) == NULL) /* The last line */ | |
869 | if (sb->buf[sb->len - 1] != '\n') { | |
870 | /* Partial line, save it for later. */ | |
871 | strbuf_addbuf(&prev, sb); | |
872 | break; | |
873 | } | |
874 | handle_filter(mi, sb); | |
875 | } | |
876 | /* | |
877 | * The partial chunk is saved in "prev" and will be | |
878 | * appended by the next iteration of read_line_with_nul(). | |
879 | */ | |
880 | strbuf_list_free(lines); | |
881 | break; | |
882 | } | |
883 | default: | |
884 | handle_filter(mi, line); | |
885 | } | |
886 | ||
6ac617a3 JH |
887 | if (mi->input_error) |
888 | break; | |
c6905e45 JH |
889 | } while (!strbuf_getwholeline(line, mi->input, '\n')); |
890 | ||
891 | handle_body_out: | |
892 | strbuf_release(&prev); | |
893 | } | |
894 | ||
895 | static void output_header_lines(FILE *fout, const char *hdr, const struct strbuf *data) | |
896 | { | |
897 | const char *sp = data->buf; | |
898 | while (1) { | |
899 | char *ep = strchr(sp, '\n'); | |
900 | int len; | |
901 | if (!ep) | |
902 | len = strlen(sp); | |
903 | else | |
904 | len = ep - sp; | |
905 | fprintf(fout, "%s: %.*s\n", hdr, len, sp); | |
906 | if (!ep) | |
907 | break; | |
908 | sp = ep + 1; | |
909 | } | |
910 | } | |
911 | ||
912 | static void handle_info(struct mailinfo *mi) | |
913 | { | |
914 | struct strbuf *hdr; | |
915 | int i; | |
916 | ||
917 | for (i = 0; header[i]; i++) { | |
918 | /* only print inbody headers if we output a patch file */ | |
919 | if (mi->patch_lines && mi->s_hdr_data[i]) | |
920 | hdr = mi->s_hdr_data[i]; | |
921 | else if (mi->p_hdr_data[i]) | |
922 | hdr = mi->p_hdr_data[i]; | |
923 | else | |
924 | continue; | |
925 | ||
926 | if (!strcmp(header[i], "Subject")) { | |
927 | if (!mi->keep_subject) { | |
928 | cleanup_subject(mi, hdr); | |
929 | cleanup_space(hdr); | |
930 | } | |
931 | output_header_lines(mi->output, "Subject", hdr); | |
932 | } else if (!strcmp(header[i], "From")) { | |
933 | cleanup_space(hdr); | |
934 | handle_from(mi, hdr); | |
935 | fprintf(mi->output, "Author: %s\n", mi->name.buf); | |
936 | fprintf(mi->output, "Email: %s\n", mi->email.buf); | |
937 | } else { | |
938 | cleanup_space(hdr); | |
939 | fprintf(mi->output, "%s: %s\n", header[i], hdr->buf); | |
940 | } | |
941 | } | |
942 | fprintf(mi->output, "\n"); | |
943 | } | |
944 | ||
945 | int mailinfo(struct mailinfo *mi, const char *msg, const char *patch) | |
946 | { | |
947 | FILE *cmitmsg; | |
948 | int peek; | |
949 | struct strbuf line = STRBUF_INIT; | |
950 | ||
951 | cmitmsg = fopen(msg, "w"); | |
952 | if (!cmitmsg) { | |
953 | perror(msg); | |
954 | return -1; | |
955 | } | |
956 | mi->patchfile = fopen(patch, "w"); | |
957 | if (!mi->patchfile) { | |
958 | perror(patch); | |
959 | fclose(cmitmsg); | |
960 | return -1; | |
961 | } | |
962 | ||
963 | mi->p_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(*(mi->p_hdr_data))); | |
964 | mi->s_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(*(mi->s_hdr_data))); | |
965 | ||
966 | do { | |
967 | peek = fgetc(mi->input); | |
968 | } while (isspace(peek)); | |
969 | ungetc(peek, mi->input); | |
970 | ||
971 | /* process the email header */ | |
972 | while (read_one_header_line(&line, mi->input)) | |
973 | check_header(mi, &line, mi->p_hdr_data, 1); | |
974 | ||
975 | handle_body(mi, &line); | |
976 | fwrite(mi->log_message.buf, 1, mi->log_message.len, cmitmsg); | |
977 | fclose(cmitmsg); | |
978 | fclose(mi->patchfile); | |
979 | ||
980 | handle_info(mi); | |
981 | strbuf_release(&line); | |
6ac617a3 | 982 | return mi->input_error; |
c6905e45 JH |
983 | } |
984 | ||
985 | static int git_mailinfo_config(const char *var, const char *value, void *mi_) | |
986 | { | |
987 | struct mailinfo *mi = mi_; | |
988 | ||
989 | if (!starts_with(var, "mailinfo.")) | |
990 | return git_default_config(var, value, NULL); | |
991 | if (!strcmp(var, "mailinfo.scissors")) { | |
992 | mi->use_scissors = git_config_bool(var, value); | |
993 | return 0; | |
994 | } | |
995 | /* perhaps others here */ | |
996 | return 0; | |
997 | } | |
998 | ||
999 | void setup_mailinfo(struct mailinfo *mi) | |
1000 | { | |
1001 | memset(mi, 0, sizeof(*mi)); | |
1002 | strbuf_init(&mi->name, 0); | |
1003 | strbuf_init(&mi->email, 0); | |
1004 | strbuf_init(&mi->charset, 0); | |
1005 | strbuf_init(&mi->log_message, 0); | |
1006 | mi->header_stage = 1; | |
1007 | mi->use_inbody_headers = 1; | |
1008 | mi->content_top = mi->content; | |
85d9d9dd | 1009 | git_config(git_mailinfo_config, mi); |
c6905e45 JH |
1010 | } |
1011 | ||
1012 | void clear_mailinfo(struct mailinfo *mi) | |
1013 | { | |
1014 | int i; | |
1015 | ||
1016 | strbuf_release(&mi->name); | |
1017 | strbuf_release(&mi->email); | |
1018 | strbuf_release(&mi->charset); | |
1019 | free(mi->message_id); | |
1020 | ||
1021 | for (i = 0; mi->p_hdr_data[i]; i++) | |
1022 | strbuf_release(mi->p_hdr_data[i]); | |
1023 | free(mi->p_hdr_data); | |
1024 | for (i = 0; mi->s_hdr_data[i]; i++) | |
1025 | strbuf_release(mi->s_hdr_data[i]); | |
1026 | free(mi->s_hdr_data); | |
1027 | ||
1028 | while (mi->content < mi->content_top) { | |
1029 | free(*(mi->content_top)); | |
1030 | mi->content_top--; | |
1031 | } | |
1032 | ||
1033 | strbuf_release(&mi->log_message); | |
1034 | } |