Commit | Line | Data |
---|---|---|
b0322a85 CE |
1 | /* |
2 | ** Copyright 1998 - 2011 Double Precision, Inc. See COPYING for | |
3 | ** distribution information. | |
4 | */ | |
5 | ||
6 | #include "rfc822.h" | |
7 | #include <stdio.h> | |
8 | #include <ctype.h> | |
9 | #include <string.h> | |
10 | #include <stdlib.h> | |
11 | #include <errno.h> | |
d50284c4 | 12 | #include <courier-unicode.h> |
b0322a85 CE |
13 | |
14 | #include "rfc822hdr.h" | |
15 | #include "rfc2047.h" | |
b0322a85 CE |
16 | #if LIBIDN |
17 | #include <idna.h> | |
18 | #include <stringprep.h> | |
19 | #endif | |
20 | ||
21 | ||
22 | #define RFC2047_ENCODE_FOLDLENGTH 76 | |
23 | ||
24 | static const char xdigit[]="0123456789ABCDEF"; | |
25 | static const char base64tab[]= | |
26 | "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; | |
27 | ||
28 | static char *a_rfc2047_encode_str(const char *str, const char *charset, | |
29 | int isaddress); | |
30 | ||
31 | static void rfc2047_encode_header_do(const struct rfc822a *a, | |
32 | const char *charset, | |
33 | void (*print_func)(char, void *), | |
34 | void (*print_separator)(const char *, | |
35 | void *), void *ptr) | |
36 | { | |
37 | rfc822_print_common(a, &a_rfc2047_encode_str, charset, | |
38 | print_func, print_separator, ptr); | |
39 | } | |
40 | ||
41 | static char *rfc822_encode_domain_int(const char *pfix, | |
42 | size_t pfix_len, | |
43 | const char *domain) | |
44 | { | |
45 | char *q; | |
46 | ||
47 | #if LIBIDN | |
48 | int err; | |
49 | char *p; | |
50 | size_t s=strlen(domain)+16; | |
51 | char *cpy=malloc(s); | |
52 | ||
53 | if (!cpy) | |
54 | return NULL; | |
55 | ||
56 | /* | |
57 | ** Invalid UTF-8 can make libidn go off the deep end. Add | |
58 | ** padding as a workaround. | |
59 | */ | |
60 | ||
61 | memset(cpy, 0, s); | |
62 | strcpy(cpy, domain); | |
63 | ||
64 | err=idna_to_ascii_8z(cpy, &p, 0); | |
65 | free(cpy); | |
66 | ||
67 | if (err != IDNA_SUCCESS) | |
68 | { | |
69 | errno=EINVAL; | |
70 | return NULL; | |
71 | } | |
72 | #else | |
73 | char *p; | |
74 | ||
75 | p=strdup(domain); | |
76 | ||
77 | if (!p) | |
78 | return NULL; | |
79 | #endif | |
80 | ||
81 | q=malloc(strlen(p)+pfix_len+1); | |
82 | ||
83 | if (!q) | |
84 | { | |
85 | free(p); | |
86 | return NULL; | |
87 | } | |
88 | ||
89 | if (pfix_len) | |
90 | memcpy(q, pfix, pfix_len); | |
91 | ||
92 | strcpy(q + pfix_len, p); | |
93 | free(p); | |
94 | return q; | |
95 | } | |
96 | ||
97 | char *rfc822_encode_domain(const char *address, | |
98 | const char *charset) | |
99 | { | |
d50284c4 | 100 | char *p=unicode_convert_tobuf(address, charset, "utf-8", NULL); |
b0322a85 CE |
101 | char *cp, *q; |
102 | ||
103 | if (!p) | |
104 | return NULL; | |
105 | ||
106 | cp=strchr(p, '@'); | |
107 | ||
108 | if (!cp) | |
109 | { | |
110 | q=rfc822_encode_domain_int("", 0, p); | |
111 | free(p); | |
112 | return q; | |
113 | } | |
114 | ||
115 | ++cp; | |
116 | q=rfc822_encode_domain_int(p, cp-p, cp); | |
117 | free(p); | |
118 | return q; | |
119 | } | |
120 | ||
121 | static char *a_rfc2047_encode_str(const char *str, const char *charset, | |
122 | int isaddress) | |
123 | { | |
124 | size_t l; | |
125 | char *p; | |
126 | ||
127 | if (isaddress) | |
128 | return rfc822_encode_domain(str, charset); | |
129 | ||
130 | for (l=0; str[l]; l++) | |
131 | if (str[l] & 0x80) | |
132 | break; | |
133 | ||
134 | if (str[l] == 0) | |
135 | { | |
136 | size_t n; | |
137 | ||
138 | for (l=0; str[l]; l++) | |
139 | if (strchr(RFC822_SPECIALS, str[l])) | |
140 | break; | |
141 | ||
142 | if (str[l] == 0) | |
143 | return (strdup(str)); | |
144 | ||
145 | for (n=3, l=0; str[l]; l++) | |
146 | { | |
147 | switch (str[l]) { | |
148 | case '"': | |
149 | case '\\': | |
150 | ++n; | |
151 | break; | |
152 | } | |
153 | ||
154 | ++n; | |
155 | } | |
156 | ||
157 | p=malloc(n); | |
158 | ||
159 | if (!p) | |
160 | return NULL; | |
161 | ||
162 | p[0]='"'; | |
163 | ||
164 | for (n=1, l=0; str[l]; l++) | |
165 | { | |
166 | switch (str[l]) { | |
167 | case '"': | |
168 | case '\\': | |
169 | p[n++]='\\'; | |
170 | break; | |
171 | } | |
172 | ||
173 | p[n++]=str[l]; | |
174 | } | |
175 | p[n++]='"'; | |
176 | p[n]=0; | |
177 | ||
178 | return (p); | |
179 | } | |
180 | ||
181 | return rfc2047_encode_str(str, charset, rfc2047_qp_allow_word); | |
182 | } | |
183 | ||
184 | static void count(char c, void *p); | |
185 | static void counts2(const char *c, void *p); | |
186 | static void save(char c, void *p); | |
187 | static void saves2(const char *c, void *p); | |
188 | ||
189 | char *rfc2047_encode_header_addr(const struct rfc822a *a, | |
190 | const char *charset) | |
191 | { | |
192 | size_t l; | |
193 | char *s, *p; | |
194 | ||
195 | l=1; | |
196 | rfc2047_encode_header_do(a, charset, &count, &counts2, &l); | |
197 | if ((s=malloc(l)) == 0) return (0); | |
198 | p=s; | |
199 | rfc2047_encode_header_do(a, charset, &save, &saves2, &p); | |
200 | *p=0; | |
201 | return (s); | |
202 | } | |
203 | ||
204 | ||
205 | char *rfc2047_encode_header_tobuf(const char *name, /* Header name */ | |
206 | const char *header, /* Header's contents */ | |
207 | const char *charset) | |
208 | { | |
209 | if (rfc822hdr_is_addr(name)) | |
210 | { | |
211 | char *s=0; | |
212 | ||
213 | struct rfc822t *t; | |
214 | struct rfc822a *a; | |
215 | ||
216 | if ((t=rfc822t_alloc_new(header, NULL, NULL)) != 0) | |
217 | { | |
218 | if ((a=rfc822a_alloc(t)) != 0) | |
219 | { | |
220 | s=rfc2047_encode_header_addr(a, charset); | |
221 | rfc822a_free(a); | |
222 | } | |
223 | rfc822t_free(t); | |
224 | } | |
225 | return s; | |
226 | } | |
227 | ||
228 | return rfc2047_encode_str(header, charset, rfc2047_qp_allow_word); | |
229 | } | |
230 | ||
231 | static void count(char c, void *p) | |
232 | { | |
233 | ++*(size_t *)p; | |
234 | } | |
235 | ||
236 | static void counts2(const char *c, void *p) | |
237 | { | |
238 | if (*c == ',') | |
239 | count(*c++, p); | |
240 | ||
241 | count('\n', p); | |
242 | count(' ', p); | |
243 | ||
244 | while (*c) count(*c++, p); | |
245 | } | |
246 | ||
247 | static void save(char c, void *p) | |
248 | { | |
249 | **(char **)p=c; | |
250 | ++*(char **)p; | |
251 | } | |
252 | ||
253 | static void saves2(const char *c, void *p) | |
254 | { | |
255 | if (*c == ',') | |
256 | save(*c++, p); | |
257 | ||
258 | save('\n', p); | |
259 | save(' ', p); | |
260 | ||
261 | while (*c) save(*c++, p); | |
262 | } | |
263 | ||
264 | static int encodebase64(const char *ptr, size_t len, const char *charset, | |
265 | int (*qp_allow)(char), | |
266 | int (*func)(const char *, size_t, void *), void *arg) | |
267 | { | |
268 | unsigned char ibuf[3]; | |
269 | char obuf[4]; | |
270 | int rc; | |
271 | ||
272 | if ((rc=(*func)("=?", 2, arg)) || | |
273 | (rc=(*func)(charset, strlen(charset), arg))|| | |
274 | (rc=(*func)("?B?", 3, arg))) | |
275 | return rc; | |
276 | ||
277 | while (len) | |
278 | { | |
279 | size_t n=len > 3 ? 3:len; | |
280 | ||
281 | ibuf[0]= ptr[0]; | |
282 | if (n>1) | |
283 | ibuf[1]=ptr[1]; | |
284 | else | |
285 | ibuf[1]=0; | |
286 | if (n>2) | |
287 | ibuf[2]=ptr[2]; | |
288 | else | |
289 | ibuf[2]=0; | |
290 | ptr += n; | |
291 | len -= n; | |
292 | ||
293 | obuf[0] = base64tab[ ibuf[0] >>2 ]; | |
294 | obuf[1] = base64tab[(ibuf[0] & 0x03)<<4|ibuf[1]>>4]; | |
295 | obuf[2] = base64tab[(ibuf[1] & 0x0F)<<2|ibuf[2]>>6]; | |
296 | obuf[3] = base64tab[ ibuf[2] & 0x3F ]; | |
297 | if (n < 2) | |
298 | obuf[2] = '='; | |
299 | if (n < 3) | |
300 | obuf[3] = '='; | |
301 | ||
302 | if ((rc=(*func)(obuf, 4, arg))) | |
303 | return rc; | |
304 | } | |
305 | ||
306 | if ((rc=(*func)("?=", 2, arg))) | |
307 | return rc; | |
308 | return 0; | |
309 | } | |
310 | ||
311 | #define ISSPACE(i) ((i)=='\t' || (i)=='\r' || (i)=='\n' || (i)==' ') | |
312 | #define DOENCODEWORD(c) \ | |
313 | ((c) < 0x20 || (c) > 0x7F || (c) == '"' || \ | |
314 | (c) == '_' || (c) == '=' || (c) == '?' || !(*qp_allow)((char)c)) | |
315 | ||
316 | /* | |
317 | ** Encode a character stream using quoted-printable encoding. | |
318 | */ | |
319 | static int encodeqp(const char *ptr, size_t len, | |
320 | const char *charset, | |
321 | int (*qp_allow)(char), | |
322 | int (*func)(const char *, size_t, void *), void *arg) | |
323 | { | |
324 | size_t i; | |
325 | int rc; | |
326 | char buf[3]; | |
327 | ||
328 | if ((rc=(*func)("=?", 2, arg)) || | |
329 | (rc=(*func)(charset, strlen(charset), arg))|| | |
330 | (rc=(*func)("?Q?", 3, arg))) | |
331 | return rc; | |
332 | ||
333 | for (i=0; i<len; ++i) | |
334 | { | |
335 | size_t j; | |
336 | ||
337 | for (j=i; j<len; ++j) | |
338 | { | |
339 | if (ptr[j] == ' ' || DOENCODEWORD(ptr[j])) | |
340 | break; | |
341 | } | |
342 | ||
343 | if (j > i) | |
344 | { | |
345 | rc=(*func)(ptr+i, j-i, arg); | |
346 | ||
347 | if (rc) | |
348 | return rc; | |
349 | if (j >= len) | |
350 | break; | |
351 | } | |
352 | i=j; | |
353 | ||
354 | if (ptr[i] == ' ') | |
355 | rc=(*func)("_", 1, arg); | |
356 | else | |
357 | { | |
358 | buf[0]='='; | |
359 | buf[1]=xdigit[ ( ptr[i] >> 4) & 0x0F ]; | |
360 | buf[2]=xdigit[ ptr[i] & 0x0F ]; | |
361 | ||
362 | rc=(*func)(buf, 3, arg); | |
363 | } | |
364 | ||
365 | if (rc) | |
366 | return rc; | |
367 | } | |
368 | ||
369 | return (*func)("?=", 2, arg); | |
370 | } | |
371 | ||
372 | /* | |
373 | ** Calculate whether the next word should be RFC2047-encoded. | |
374 | ** | |
375 | ** Returns 0 if not, 1 if any character in the next word is flagged by | |
376 | ** DOENCODEWORD(). | |
377 | */ | |
378 | ||
0e333c05 | 379 | static int encode_word(const char32_t *uc, |
b0322a85 CE |
380 | size_t ucsize, |
381 | int (*qp_allow)(char), | |
382 | ||
383 | /* | |
384 | ** Points to the starting offset of word in uc. | |
385 | ** At exit, points to the end of the word in uc. | |
386 | */ | |
387 | size_t *word_ptr) | |
388 | { | |
389 | size_t i; | |
390 | int encode=0; | |
391 | ||
392 | for (i=*word_ptr; i<ucsize; ++i) | |
393 | { | |
394 | if (ISSPACE(uc[i])) | |
395 | break; | |
396 | ||
397 | if (DOENCODEWORD(uc[i])) | |
398 | encode=1; | |
399 | } | |
400 | ||
401 | *word_ptr=i; | |
402 | return encode; | |
403 | } | |
404 | ||
405 | /* | |
406 | ** Calculate whether the next sequence of words should be RFC2047-encoded. | |
407 | ** | |
408 | ** Whatever encode_word() returns for the first word, look at the next word | |
409 | ** and keep going as long as encode_word() keeps returning the same value. | |
410 | */ | |
411 | ||
0e333c05 | 412 | static int encode_words(const char32_t *uc, |
b0322a85 CE |
413 | size_t ucsize, |
414 | int (*qp_allow)(char), | |
415 | ||
416 | /* | |
417 | ** Points to the starting offset of words in uc. | |
418 | ** At exit, points to the end of the words in uc. | |
419 | */ | |
420 | ||
421 | size_t *word_ptr) | |
422 | { | |
423 | size_t i= *word_ptr, j, k; | |
424 | ||
425 | int flag=encode_word(uc, ucsize, qp_allow, &i); | |
426 | ||
427 | if (!flag) | |
428 | { | |
429 | *word_ptr=i; | |
430 | return flag; | |
431 | } | |
432 | ||
433 | j=i; | |
434 | ||
435 | while (j < ucsize) | |
436 | { | |
437 | if (ISSPACE(uc[j])) | |
438 | { | |
439 | ++j; | |
440 | continue; | |
441 | } | |
442 | ||
443 | k=j; | |
444 | ||
445 | if (!encode_word(uc, ucsize, qp_allow, &k)) | |
446 | break; | |
447 | i=j=k; | |
448 | } | |
449 | ||
450 | *word_ptr=i; | |
451 | return flag; | |
452 | } | |
453 | ||
454 | /* | |
455 | ** Encode a sequence of words. | |
456 | */ | |
0e333c05 | 457 | static int do_encode_words_method(const char32_t *uc, |
b0322a85 CE |
458 | size_t ucsize, |
459 | const char *charset, | |
460 | int (*qp_allow)(char), | |
461 | size_t offset, | |
462 | int (*encoder)(const char *ptr, size_t len, | |
463 | const char *charset, | |
464 | int (*qp_allow)(char), | |
465 | int (*func)(const char *, | |
466 | size_t, void *), | |
467 | void *arg), | |
468 | int (*func)(const char *, size_t, void *), | |
469 | void *arg) | |
470 | { | |
471 | char *p; | |
472 | size_t psize; | |
473 | int rc; | |
474 | int first=1; | |
475 | ||
476 | while (ucsize) | |
477 | { | |
478 | size_t j; | |
479 | size_t i; | |
480 | ||
481 | if (!first) | |
482 | { | |
483 | rc=(*func)(" ", 1, arg); | |
484 | ||
485 | if (rc) | |
486 | return rc; | |
487 | } | |
488 | first=0; | |
489 | ||
490 | j=(RFC2047_ENCODE_FOLDLENGTH-offset)/2; | |
491 | ||
492 | if (j >= ucsize) | |
493 | j=ucsize; | |
494 | else | |
495 | { | |
496 | /* | |
497 | ** Do not split rfc2047-encoded works across a | |
498 | ** grapheme break. | |
499 | */ | |
500 | ||
501 | for (i=j; i > 0; --i) | |
502 | if (unicode_grapheme_break(uc[i-1], uc[i])) | |
503 | { | |
504 | j=i; | |
505 | break; | |
506 | } | |
507 | } | |
508 | ||
d50284c4 | 509 | if ((rc=unicode_convert_fromu_tobuf(uc, j, charset, |
b0322a85 CE |
510 | &p, &psize, |
511 | NULL)) != 0) | |
512 | return rc; | |
513 | ||
514 | ||
515 | if (psize && p[psize-1] == 0) | |
516 | --psize; | |
517 | ||
518 | rc=(*encoder)(p, psize, charset, qp_allow, | |
519 | func, arg); | |
520 | free(p); | |
521 | if (rc) | |
522 | return rc; | |
523 | offset=0; | |
524 | ucsize -= j; | |
525 | uc += j; | |
526 | } | |
527 | return 0; | |
528 | } | |
529 | ||
530 | static int cnt_conv(const char *dummy, size_t n, void *arg) | |
531 | { | |
532 | *(size_t *)arg += n; | |
533 | return 0; | |
534 | } | |
535 | ||
536 | /* | |
537 | ** Encode, or not encode, words. | |
538 | */ | |
539 | ||
0e333c05 | 540 | static int do_encode_words(const char32_t *uc, |
b0322a85 CE |
541 | size_t ucsize, |
542 | const char *charset, | |
543 | int flag, | |
544 | int (*qp_allow)(char), | |
545 | size_t offset, | |
546 | int (*func)(const char *, size_t, void *), | |
547 | void *arg) | |
548 | { | |
549 | char *p; | |
550 | size_t psize; | |
551 | int rc; | |
552 | size_t b64len, qlen; | |
553 | ||
554 | /* | |
555 | ** Convert from unicode | |
556 | */ | |
557 | ||
d50284c4 | 558 | if ((rc=unicode_convert_fromu_tobuf(uc, ucsize, charset, |
b0322a85 CE |
559 | &p, &psize, |
560 | NULL)) != 0) | |
561 | return rc; | |
562 | ||
563 | if (psize && p[psize-1] == 0) | |
564 | --psize; | |
565 | ||
566 | if (!flag) /* If not converting, then the job is done */ | |
567 | { | |
568 | rc=(*func)(p, psize, arg); | |
569 | free(p); | |
570 | return rc; | |
571 | } | |
572 | free(p); | |
573 | ||
574 | /* | |
575 | ** Try first quoted-printable, then base64, then pick whichever | |
576 | ** one gives the shortest results. | |
577 | */ | |
578 | qlen=0; | |
579 | b64len=0; | |
580 | ||
581 | rc=do_encode_words_method(uc, ucsize, charset, qp_allow, offset, | |
582 | &encodeqp, cnt_conv, &qlen); | |
583 | if (rc) | |
584 | return rc; | |
585 | ||
586 | rc=do_encode_words_method(uc, ucsize, charset, qp_allow, offset, | |
587 | &encodebase64, cnt_conv, &b64len); | |
588 | if (rc) | |
589 | return rc; | |
590 | ||
591 | return do_encode_words_method(uc, ucsize, charset, qp_allow, offset, | |
592 | qlen < b64len ? encodeqp:encodebase64, | |
593 | func, arg); | |
594 | } | |
595 | ||
596 | /* | |
597 | ** RFC2047-encoding pass. | |
598 | */ | |
0e333c05 | 599 | static int rfc2047_encode_callback(const char32_t *uc, |
b0322a85 CE |
600 | size_t ucsize, |
601 | const char *charset, | |
602 | int (*qp_allow)(char), | |
603 | int (*func)(const char *, size_t, void *), | |
604 | void *arg) | |
605 | { | |
606 | int rc; | |
607 | size_t i; | |
608 | int flag; | |
609 | ||
610 | size_t offset=27; /* FIXME: initial offset for line length */ | |
611 | ||
612 | while (ucsize) | |
613 | { | |
614 | /* Pass along all the whitespace */ | |
615 | ||
616 | if (ISSPACE(*uc)) | |
617 | { | |
618 | char c= *uc++; | |
619 | --ucsize; | |
620 | ||
621 | if ((rc=(*func)(&c, 1, arg)) != 0) | |
622 | return rc; | |
623 | continue; | |
624 | } | |
625 | ||
626 | i=0; | |
627 | ||
628 | /* Check if the next word needs to be encoded, or not. */ | |
629 | ||
630 | flag=encode_words(uc, ucsize, qp_allow, &i); | |
631 | ||
632 | /* | |
633 | ** Then proceed to encode, or not encode, the following words. | |
634 | */ | |
635 | ||
636 | if ((rc=do_encode_words(uc, i, charset, flag, | |
637 | qp_allow, offset, | |
638 | func, arg)) != 0) | |
639 | return rc; | |
640 | ||
641 | offset=0; | |
642 | uc += i; | |
643 | ucsize -= i; | |
644 | } | |
645 | ||
646 | return 0; | |
647 | } | |
648 | ||
649 | ||
650 | static int count_char(const char *c, size_t l, void *p) | |
651 | { | |
652 | size_t *i=(size_t *)p; | |
653 | ||
654 | *i += l; | |
655 | return (0); | |
656 | } | |
657 | ||
658 | static int save_char(const char *c, size_t l, void *p) | |
659 | { | |
660 | char **s=(char **)p; | |
661 | ||
662 | memcpy(*s, c, l); | |
663 | *s += l; | |
664 | return (0); | |
665 | } | |
666 | ||
667 | char *rfc2047_encode_str(const char *str, const char *charset, | |
668 | int (*qp_allow)(char c)) | |
669 | { | |
670 | size_t i=1; | |
671 | char *s, *p; | |
0e333c05 | 672 | char32_t *uc; |
b0322a85 CE |
673 | size_t ucsize; |
674 | int err; | |
675 | ||
676 | /* Convert string to unicode */ | |
677 | ||
d50284c4 | 678 | if (unicode_convert_tou_tobuf(str, strlen(str), charset, |
b0322a85 CE |
679 | &uc, &ucsize, &err)) |
680 | return NULL; | |
681 | ||
682 | /* | |
683 | ** Perform two passes: calculate size of the buffer where the | |
684 | ** encoded string gets saved into, then allocate the buffer and | |
685 | ** do a second pass to actually do it. | |
686 | */ | |
687 | ||
688 | if (rfc2047_encode_callback(uc, ucsize, | |
689 | charset, | |
690 | qp_allow, | |
691 | &count_char, &i)) | |
692 | { | |
693 | free(uc); | |
694 | return NULL; | |
695 | } | |
696 | ||
697 | if ((s=malloc(i)) == 0) | |
698 | { | |
699 | free(uc); | |
700 | return NULL; | |
701 | } | |
702 | ||
703 | p=s; | |
704 | (void)rfc2047_encode_callback(uc, ucsize, | |
705 | charset, | |
706 | qp_allow, | |
707 | &save_char, &p); | |
708 | *p=0; | |
709 | free(uc); | |
710 | return (s); | |
711 | } | |
712 | ||
713 | int rfc2047_qp_allow_any(char c) | |
714 | { | |
715 | return 1; | |
716 | } | |
717 | ||
718 | int rfc2047_qp_allow_comment(char c) | |
719 | { | |
720 | if (c == '(' || c == ')' || c == '"') | |
721 | return 0; | |
722 | return 1; | |
723 | } | |
724 | ||
725 | int rfc2047_qp_allow_word(char c) | |
726 | { | |
727 | return strchr(base64tab, c) != NULL || | |
728 | strchr("*-=_", c) != NULL; | |
729 | } |