]>
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 | ||
6b4b013f JT |
503 | /* |
504 | * Returns 1 if the given line or any line beginning with the given line is an | |
505 | * in-body header (that is, check_header will succeed when passed | |
506 | * mi->s_hdr_data). | |
507 | */ | |
508 | static int is_inbody_header(const struct mailinfo *mi, | |
509 | const struct strbuf *line) | |
510 | { | |
511 | int i; | |
512 | for (i = 0; header[i]; i++) | |
513 | if (!mi->s_hdr_data[i] && cmp_header(line, header[i])) | |
514 | return 1; | |
515 | return 0; | |
516 | } | |
517 | ||
c6905e45 JH |
518 | static void decode_transfer_encoding(struct mailinfo *mi, struct strbuf *line) |
519 | { | |
520 | struct strbuf *ret; | |
521 | ||
522 | switch (mi->transfer_encoding) { | |
523 | case TE_QP: | |
524 | ret = decode_q_segment(line, 0); | |
525 | break; | |
526 | case TE_BASE64: | |
527 | ret = decode_b_segment(line); | |
528 | break; | |
529 | case TE_DONTCARE: | |
530 | default: | |
531 | return; | |
532 | } | |
533 | strbuf_reset(line); | |
534 | strbuf_addbuf(line, ret); | |
535 | strbuf_release(ret); | |
536 | free(ret); | |
537 | } | |
538 | ||
539 | static inline int patchbreak(const struct strbuf *line) | |
540 | { | |
541 | size_t i; | |
542 | ||
543 | /* Beginning of a "diff -" header? */ | |
544 | if (starts_with(line->buf, "diff -")) | |
545 | return 1; | |
546 | ||
547 | /* CVS "Index: " line? */ | |
548 | if (starts_with(line->buf, "Index: ")) | |
549 | return 1; | |
550 | ||
551 | /* | |
552 | * "--- <filename>" starts patches without headers | |
553 | * "---<sp>*" is a manual separator | |
554 | */ | |
555 | if (line->len < 4) | |
556 | return 0; | |
557 | ||
558 | if (starts_with(line->buf, "---")) { | |
559 | /* space followed by a filename? */ | |
560 | if (line->buf[3] == ' ' && !isspace(line->buf[4])) | |
561 | return 1; | |
562 | /* Just whitespace? */ | |
563 | for (i = 3; i < line->len; i++) { | |
564 | unsigned char c = line->buf[i]; | |
565 | if (c == '\n') | |
566 | return 1; | |
567 | if (!isspace(c)) | |
568 | break; | |
569 | } | |
570 | return 0; | |
571 | } | |
572 | return 0; | |
573 | } | |
574 | ||
9c5681da | 575 | static int is_scissors_line(const char *line) |
c6905e45 | 576 | { |
9c5681da | 577 | const char *c; |
c6905e45 | 578 | int scissors = 0, gap = 0; |
9c5681da JT |
579 | const char *first_nonblank = NULL, *last_nonblank = NULL; |
580 | int visible, perforation = 0, in_perforation = 0; | |
c6905e45 | 581 | |
9c5681da JT |
582 | for (c = line; *c; c++) { |
583 | if (isspace(*c)) { | |
c6905e45 JH |
584 | if (in_perforation) { |
585 | perforation++; | |
586 | gap++; | |
587 | } | |
588 | continue; | |
589 | } | |
9c5681da JT |
590 | last_nonblank = c; |
591 | if (first_nonblank == NULL) | |
592 | first_nonblank = c; | |
593 | if (*c == '-') { | |
c6905e45 JH |
594 | in_perforation = 1; |
595 | perforation++; | |
596 | continue; | |
597 | } | |
9c5681da JT |
598 | if ((!memcmp(c, ">8", 2) || !memcmp(c, "8<", 2) || |
599 | !memcmp(c, ">%", 2) || !memcmp(c, "%<", 2))) { | |
c6905e45 JH |
600 | in_perforation = 1; |
601 | perforation += 2; | |
602 | scissors += 2; | |
9c5681da | 603 | c++; |
c6905e45 JH |
604 | continue; |
605 | } | |
606 | in_perforation = 0; | |
607 | } | |
608 | ||
609 | /* | |
610 | * The mark must be at least 8 bytes long (e.g. "-- >8 --"). | |
611 | * Even though there can be arbitrary cruft on the same line | |
612 | * (e.g. "cut here"), in order to avoid misidentification, the | |
613 | * perforation must occupy more than a third of the visible | |
614 | * width of the line, and dashes and scissors must occupy more | |
615 | * than half of the perforation. | |
616 | */ | |
617 | ||
9c5681da JT |
618 | if (first_nonblank && last_nonblank) |
619 | visible = last_nonblank - first_nonblank + 1; | |
620 | else | |
621 | visible = 0; | |
c6905e45 JH |
622 | return (scissors && 8 <= visible && |
623 | visible < perforation * 3 && | |
624 | gap * 2 < perforation); | |
625 | } | |
626 | ||
6b4b013f JT |
627 | static void flush_inbody_header_accum(struct mailinfo *mi) |
628 | { | |
629 | if (!mi->inbody_header_accum.len) | |
630 | return; | |
631 | assert(check_header(mi, &mi->inbody_header_accum, mi->s_hdr_data, 0)); | |
632 | strbuf_reset(&mi->inbody_header_accum); | |
633 | } | |
634 | ||
334192b4 JT |
635 | static int check_inbody_header(struct mailinfo *mi, const struct strbuf *line) |
636 | { | |
6b4b013f JT |
637 | if (mi->inbody_header_accum.len && |
638 | (line->buf[0] == ' ' || line->buf[0] == '\t')) { | |
639 | if (mi->use_scissors && is_scissors_line(line->buf)) { | |
640 | /* | |
641 | * This is a scissors line; do not consider this line | |
642 | * as a header continuation line. | |
643 | */ | |
644 | flush_inbody_header_accum(mi); | |
645 | return 0; | |
646 | } | |
647 | strbuf_strip_suffix(&mi->inbody_header_accum, "\n"); | |
648 | strbuf_addbuf(&mi->inbody_header_accum, line); | |
649 | return 1; | |
650 | } | |
651 | ||
652 | flush_inbody_header_accum(mi); | |
653 | ||
334192b4 JT |
654 | if (starts_with(line->buf, ">From") && isspace(line->buf[5])) |
655 | return is_format_patch_separator(line->buf + 1, line->len - 1); | |
656 | if (starts_with(line->buf, "[PATCH]") && isspace(line->buf[7])) { | |
657 | int i; | |
658 | for (i = 0; header[i]; i++) | |
659 | if (!strcmp("Subject", header[i])) { | |
660 | handle_header(&mi->s_hdr_data[i], line); | |
661 | return 1; | |
662 | } | |
663 | return 0; | |
664 | } | |
6b4b013f JT |
665 | if (is_inbody_header(mi, line)) { |
666 | strbuf_addbuf(&mi->inbody_header_accum, line); | |
667 | return 1; | |
668 | } | |
669 | return 0; | |
334192b4 JT |
670 | } |
671 | ||
c6905e45 JH |
672 | static int handle_commit_msg(struct mailinfo *mi, struct strbuf *line) |
673 | { | |
674 | assert(!mi->filter_stage); | |
675 | ||
676 | if (mi->header_stage) { | |
677 | if (!line->len || (line->len == 1 && line->buf[0] == '\n')) | |
678 | return 0; | |
679 | } | |
680 | ||
681 | if (mi->use_inbody_headers && mi->header_stage) { | |
334192b4 | 682 | mi->header_stage = check_inbody_header(mi, line); |
c6905e45 JH |
683 | if (mi->header_stage) |
684 | return 0; | |
685 | } else | |
686 | /* Only trim the first (blank) line of the commit message | |
687 | * when ignoring in-body headers. | |
688 | */ | |
689 | mi->header_stage = 0; | |
690 | ||
691 | /* normalize the log message to UTF-8. */ | |
669b963a | 692 | if (convert_to_utf8(mi, line, mi->charset.buf)) |
6ac617a3 | 693 | return 0; /* mi->input_error already set */ |
c6905e45 | 694 | |
9c5681da | 695 | if (mi->use_scissors && is_scissors_line(line->buf)) { |
c6905e45 JH |
696 | int i; |
697 | ||
698 | strbuf_setlen(&mi->log_message, 0); | |
699 | mi->header_stage = 1; | |
700 | ||
701 | /* | |
702 | * We may have already read "secondary headers"; purge | |
703 | * them to give ourselves a clean restart. | |
704 | */ | |
705 | for (i = 0; header[i]; i++) { | |
706 | if (mi->s_hdr_data[i]) | |
707 | strbuf_release(mi->s_hdr_data[i]); | |
708 | mi->s_hdr_data[i] = NULL; | |
709 | } | |
710 | return 0; | |
711 | } | |
712 | ||
713 | if (patchbreak(line)) { | |
714 | if (mi->message_id) | |
715 | strbuf_addf(&mi->log_message, | |
716 | "Message-Id: %s\n", mi->message_id); | |
717 | return 1; | |
718 | } | |
719 | ||
720 | strbuf_addbuf(&mi->log_message, line); | |
721 | return 0; | |
722 | } | |
723 | ||
724 | static void handle_patch(struct mailinfo *mi, const struct strbuf *line) | |
725 | { | |
726 | fwrite(line->buf, 1, line->len, mi->patchfile); | |
727 | mi->patch_lines++; | |
728 | } | |
729 | ||
730 | static void handle_filter(struct mailinfo *mi, struct strbuf *line) | |
731 | { | |
732 | switch (mi->filter_stage) { | |
733 | case 0: | |
734 | if (!handle_commit_msg(mi, line)) | |
735 | break; | |
736 | mi->filter_stage++; | |
737 | case 1: | |
738 | handle_patch(mi, line); | |
739 | break; | |
740 | } | |
741 | } | |
742 | ||
743 | static int is_rfc2822_header(const struct strbuf *line) | |
744 | { | |
745 | /* | |
746 | * The section that defines the loosest possible | |
747 | * field name is "3.6.8 Optional fields". | |
748 | * | |
749 | * optional-field = field-name ":" unstructured CRLF | |
750 | * field-name = 1*ftext | |
751 | * ftext = %d33-57 / %59-126 | |
752 | */ | |
753 | int ch; | |
754 | char *cp = line->buf; | |
755 | ||
756 | /* Count mbox From headers as headers */ | |
757 | if (starts_with(cp, "From ") || starts_with(cp, ">From ")) | |
758 | return 1; | |
759 | ||
760 | while ((ch = *cp++)) { | |
761 | if (ch == ':') | |
762 | return 1; | |
763 | if ((33 <= ch && ch <= 57) || | |
764 | (59 <= ch && ch <= 126)) | |
765 | continue; | |
766 | break; | |
767 | } | |
768 | return 0; | |
769 | } | |
770 | ||
771 | static int read_one_header_line(struct strbuf *line, FILE *in) | |
772 | { | |
773 | struct strbuf continuation = STRBUF_INIT; | |
774 | ||
775 | /* Get the first part of the line. */ | |
8f309aeb | 776 | if (strbuf_getline_lf(line, in)) |
c6905e45 JH |
777 | return 0; |
778 | ||
779 | /* | |
780 | * Is it an empty line or not a valid rfc2822 header? | |
781 | * If so, stop here, and return false ("not a header") | |
782 | */ | |
783 | strbuf_rtrim(line); | |
784 | if (!line->len || !is_rfc2822_header(line)) { | |
785 | /* Re-add the newline */ | |
786 | strbuf_addch(line, '\n'); | |
787 | return 0; | |
788 | } | |
789 | ||
790 | /* | |
791 | * Now we need to eat all the continuation lines.. | |
792 | * Yuck, 2822 header "folding" | |
793 | */ | |
794 | for (;;) { | |
795 | int peek; | |
796 | ||
797 | peek = fgetc(in); ungetc(peek, in); | |
798 | if (peek != ' ' && peek != '\t') | |
799 | break; | |
8f309aeb | 800 | if (strbuf_getline_lf(&continuation, in)) |
c6905e45 JH |
801 | break; |
802 | continuation.buf[0] = ' '; | |
803 | strbuf_rtrim(&continuation); | |
804 | strbuf_addbuf(line, &continuation); | |
805 | } | |
806 | strbuf_release(&continuation); | |
807 | ||
808 | return 1; | |
809 | } | |
810 | ||
811 | static int find_boundary(struct mailinfo *mi, struct strbuf *line) | |
812 | { | |
8f309aeb | 813 | while (!strbuf_getline_lf(line, mi->input)) { |
c6905e45 JH |
814 | if (*(mi->content_top) && is_multipart_boundary(mi, line)) |
815 | return 1; | |
816 | } | |
817 | return 0; | |
818 | } | |
819 | ||
820 | static int handle_boundary(struct mailinfo *mi, struct strbuf *line) | |
821 | { | |
822 | struct strbuf newline = STRBUF_INIT; | |
823 | ||
824 | strbuf_addch(&newline, '\n'); | |
825 | again: | |
826 | if (line->len >= (*(mi->content_top))->len + 2 && | |
827 | !memcmp(line->buf + (*(mi->content_top))->len, "--", 2)) { | |
828 | /* we hit an end boundary */ | |
829 | /* pop the current boundary off the stack */ | |
830 | strbuf_release(*(mi->content_top)); | |
831 | free(*(mi->content_top)); | |
832 | *(mi->content_top) = NULL; | |
833 | ||
834 | /* technically won't happen as is_multipart_boundary() | |
835 | will fail first. But just in case.. | |
836 | */ | |
837 | if (--mi->content_top < mi->content) { | |
6ac617a3 JH |
838 | error("Detected mismatched boundaries, can't recover"); |
839 | mi->input_error = -1; | |
840 | mi->content_top = mi->content; | |
841 | return 0; | |
c6905e45 JH |
842 | } |
843 | handle_filter(mi, &newline); | |
844 | strbuf_release(&newline); | |
6ac617a3 JH |
845 | if (mi->input_error) |
846 | return 0; | |
c6905e45 JH |
847 | |
848 | /* skip to the next boundary */ | |
849 | if (!find_boundary(mi, line)) | |
850 | return 0; | |
851 | goto again; | |
852 | } | |
853 | ||
854 | /* set some defaults */ | |
855 | mi->transfer_encoding = TE_DONTCARE; | |
856 | strbuf_reset(&mi->charset); | |
857 | ||
858 | /* slurp in this section's info */ | |
859 | while (read_one_header_line(line, mi->input)) | |
860 | check_header(mi, line, mi->p_hdr_data, 0); | |
861 | ||
862 | strbuf_release(&newline); | |
863 | /* replenish line */ | |
8f309aeb | 864 | if (strbuf_getline_lf(line, mi->input)) |
c6905e45 JH |
865 | return 0; |
866 | strbuf_addch(line, '\n'); | |
867 | return 1; | |
868 | } | |
869 | ||
870 | static void handle_body(struct mailinfo *mi, struct strbuf *line) | |
871 | { | |
872 | struct strbuf prev = STRBUF_INIT; | |
873 | ||
874 | /* Skip up to the first boundary */ | |
875 | if (*(mi->content_top)) { | |
876 | if (!find_boundary(mi, line)) | |
877 | goto handle_body_out; | |
878 | } | |
879 | ||
880 | do { | |
881 | /* process any boundary lines */ | |
882 | if (*(mi->content_top) && is_multipart_boundary(mi, line)) { | |
883 | /* flush any leftover */ | |
884 | if (prev.len) { | |
885 | handle_filter(mi, &prev); | |
886 | strbuf_reset(&prev); | |
887 | } | |
888 | if (!handle_boundary(mi, line)) | |
889 | goto handle_body_out; | |
890 | } | |
891 | ||
892 | /* Unwrap transfer encoding */ | |
893 | decode_transfer_encoding(mi, line); | |
894 | ||
895 | switch (mi->transfer_encoding) { | |
896 | case TE_BASE64: | |
897 | case TE_QP: | |
898 | { | |
899 | struct strbuf **lines, **it, *sb; | |
900 | ||
901 | /* Prepend any previous partial lines */ | |
902 | strbuf_insert(line, 0, prev.buf, prev.len); | |
903 | strbuf_reset(&prev); | |
904 | ||
905 | /* | |
906 | * This is a decoded line that may contain | |
907 | * multiple new lines. Pass only one chunk | |
908 | * at a time to handle_filter() | |
909 | */ | |
910 | lines = strbuf_split(line, '\n'); | |
911 | for (it = lines; (sb = *it); it++) { | |
912 | if (*(it + 1) == NULL) /* The last line */ | |
913 | if (sb->buf[sb->len - 1] != '\n') { | |
914 | /* Partial line, save it for later. */ | |
915 | strbuf_addbuf(&prev, sb); | |
916 | break; | |
917 | } | |
918 | handle_filter(mi, sb); | |
919 | } | |
920 | /* | |
921 | * The partial chunk is saved in "prev" and will be | |
922 | * appended by the next iteration of read_line_with_nul(). | |
923 | */ | |
924 | strbuf_list_free(lines); | |
925 | break; | |
926 | } | |
927 | default: | |
928 | handle_filter(mi, line); | |
929 | } | |
930 | ||
6ac617a3 JH |
931 | if (mi->input_error) |
932 | break; | |
c6905e45 JH |
933 | } while (!strbuf_getwholeline(line, mi->input, '\n')); |
934 | ||
6b4b013f JT |
935 | flush_inbody_header_accum(mi); |
936 | ||
c6905e45 JH |
937 | handle_body_out: |
938 | strbuf_release(&prev); | |
939 | } | |
940 | ||
941 | static void output_header_lines(FILE *fout, const char *hdr, const struct strbuf *data) | |
942 | { | |
943 | const char *sp = data->buf; | |
944 | while (1) { | |
945 | char *ep = strchr(sp, '\n'); | |
946 | int len; | |
947 | if (!ep) | |
948 | len = strlen(sp); | |
949 | else | |
950 | len = ep - sp; | |
951 | fprintf(fout, "%s: %.*s\n", hdr, len, sp); | |
952 | if (!ep) | |
953 | break; | |
954 | sp = ep + 1; | |
955 | } | |
956 | } | |
957 | ||
958 | static void handle_info(struct mailinfo *mi) | |
959 | { | |
960 | struct strbuf *hdr; | |
961 | int i; | |
962 | ||
963 | for (i = 0; header[i]; i++) { | |
964 | /* only print inbody headers if we output a patch file */ | |
965 | if (mi->patch_lines && mi->s_hdr_data[i]) | |
966 | hdr = mi->s_hdr_data[i]; | |
967 | else if (mi->p_hdr_data[i]) | |
968 | hdr = mi->p_hdr_data[i]; | |
969 | else | |
970 | continue; | |
971 | ||
972 | if (!strcmp(header[i], "Subject")) { | |
973 | if (!mi->keep_subject) { | |
974 | cleanup_subject(mi, hdr); | |
975 | cleanup_space(hdr); | |
976 | } | |
977 | output_header_lines(mi->output, "Subject", hdr); | |
978 | } else if (!strcmp(header[i], "From")) { | |
979 | cleanup_space(hdr); | |
980 | handle_from(mi, hdr); | |
981 | fprintf(mi->output, "Author: %s\n", mi->name.buf); | |
982 | fprintf(mi->output, "Email: %s\n", mi->email.buf); | |
983 | } else { | |
984 | cleanup_space(hdr); | |
985 | fprintf(mi->output, "%s: %s\n", header[i], hdr->buf); | |
986 | } | |
987 | } | |
988 | fprintf(mi->output, "\n"); | |
989 | } | |
990 | ||
991 | int mailinfo(struct mailinfo *mi, const char *msg, const char *patch) | |
992 | { | |
993 | FILE *cmitmsg; | |
994 | int peek; | |
995 | struct strbuf line = STRBUF_INIT; | |
996 | ||
997 | cmitmsg = fopen(msg, "w"); | |
998 | if (!cmitmsg) { | |
999 | perror(msg); | |
1000 | return -1; | |
1001 | } | |
1002 | mi->patchfile = fopen(patch, "w"); | |
1003 | if (!mi->patchfile) { | |
1004 | perror(patch); | |
1005 | fclose(cmitmsg); | |
1006 | return -1; | |
1007 | } | |
1008 | ||
1009 | mi->p_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(*(mi->p_hdr_data))); | |
1010 | mi->s_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(*(mi->s_hdr_data))); | |
1011 | ||
1012 | do { | |
1013 | peek = fgetc(mi->input); | |
1014 | } while (isspace(peek)); | |
1015 | ungetc(peek, mi->input); | |
1016 | ||
1017 | /* process the email header */ | |
1018 | while (read_one_header_line(&line, mi->input)) | |
1019 | check_header(mi, &line, mi->p_hdr_data, 1); | |
1020 | ||
1021 | handle_body(mi, &line); | |
1022 | fwrite(mi->log_message.buf, 1, mi->log_message.len, cmitmsg); | |
1023 | fclose(cmitmsg); | |
1024 | fclose(mi->patchfile); | |
1025 | ||
1026 | handle_info(mi); | |
1027 | strbuf_release(&line); | |
6ac617a3 | 1028 | return mi->input_error; |
c6905e45 JH |
1029 | } |
1030 | ||
1031 | static int git_mailinfo_config(const char *var, const char *value, void *mi_) | |
1032 | { | |
1033 | struct mailinfo *mi = mi_; | |
1034 | ||
1035 | if (!starts_with(var, "mailinfo.")) | |
1036 | return git_default_config(var, value, NULL); | |
1037 | if (!strcmp(var, "mailinfo.scissors")) { | |
1038 | mi->use_scissors = git_config_bool(var, value); | |
1039 | return 0; | |
1040 | } | |
1041 | /* perhaps others here */ | |
1042 | return 0; | |
1043 | } | |
1044 | ||
1045 | void setup_mailinfo(struct mailinfo *mi) | |
1046 | { | |
1047 | memset(mi, 0, sizeof(*mi)); | |
1048 | strbuf_init(&mi->name, 0); | |
1049 | strbuf_init(&mi->email, 0); | |
1050 | strbuf_init(&mi->charset, 0); | |
1051 | strbuf_init(&mi->log_message, 0); | |
6b4b013f | 1052 | strbuf_init(&mi->inbody_header_accum, 0); |
c6905e45 JH |
1053 | mi->header_stage = 1; |
1054 | mi->use_inbody_headers = 1; | |
1055 | mi->content_top = mi->content; | |
85d9d9dd | 1056 | git_config(git_mailinfo_config, mi); |
c6905e45 JH |
1057 | } |
1058 | ||
1059 | void clear_mailinfo(struct mailinfo *mi) | |
1060 | { | |
1061 | int i; | |
1062 | ||
1063 | strbuf_release(&mi->name); | |
1064 | strbuf_release(&mi->email); | |
1065 | strbuf_release(&mi->charset); | |
6b4b013f | 1066 | strbuf_release(&mi->inbody_header_accum); |
c6905e45 JH |
1067 | free(mi->message_id); |
1068 | ||
1069 | for (i = 0; mi->p_hdr_data[i]; i++) | |
1070 | strbuf_release(mi->p_hdr_data[i]); | |
1071 | free(mi->p_hdr_data); | |
1072 | for (i = 0; mi->s_hdr_data[i]; i++) | |
1073 | strbuf_release(mi->s_hdr_data[i]); | |
1074 | free(mi->s_hdr_data); | |
1075 | ||
1076 | while (mi->content < mi->content_top) { | |
1077 | free(*(mi->content_top)); | |
1078 | mi->content_top--; | |
1079 | } | |
1080 | ||
1081 | strbuf_release(&mi->log_message); | |
1082 | } |