2 * Copyright notice from original mutt:
3 * Copyright (C) 1996-2000 Michael R. Elkins <me@mutt.org>
4 * Copyright (C) 2000-2001 Edmund Grimley Evans <edmundo@rano.org>
6 * This file is part of mutt-ng, see http://www.muttng.org/.
7 * It's licensed under the GNU General Public License,
8 * please see the file GPL in the top level source directory.
26 /* If you are debugging this file, comment out the following line. */
35 #define ENCWORD_LEN_MAX 75
36 #define ENCWORD_LEN_MIN 9 /* mutt_strlen ("=?.?.?.?=") */
38 #define HSPACE(x) ((x) == '\0' || (x) == ' ' || (x) == '\t')
40 #define CONTINUATION_BYTE(c) (((c) & 0xc0) == 0x80)
42 extern char RFC822Specials[];
44 typedef size_t (*encoder_t) (char *, ICONV_CONST char *, size_t,
47 static size_t convert_string (ICONV_CONST char *f, size_t flen,
48 const char *from, const char *to,
49 char **t, size_t * tlen)
56 cd = mutt_iconv_open (to, from, 0);
57 if (cd == (iconv_t) (-1))
60 ob = buf = safe_malloc (obl);
61 n = iconv (cd, &f, &flen, &ob, &obl);
62 if (n == (size_t) (-1) || iconv (cd, 0, 0, &ob, &obl) == (size_t) (-1)) {
73 safe_realloc (&buf, ob - buf + 1);
80 char *mutt_choose_charset (const char *fromcode, const char *charsets,
81 char *u, size_t ulen, char **d, size_t * dlen)
83 char canonical_buff[LONG_STRING];
84 char *e = 0, *tocode = 0;
85 size_t elen = 0, bestn = 0;
88 for (p = charsets; p; p = q ? q + 1 : 0) {
94 n = q ? q - p : mutt_strlen (p);
97 /* Assume that we never need more than 12 characters of
98 encoded-text to encode a single character. */
99 n > (ENCWORD_LEN_MAX - ENCWORD_LEN_MIN + 2 - 12))
102 t = safe_malloc (n + 1);
106 n = convert_string (u, ulen, fromcode, t, &s, &slen);
107 if (n == (size_t) (-1))
110 if (!tocode || n < bestn) {
135 mutt_canonical_charset (canonical_buff, sizeof (canonical_buff), tocode);
136 mutt_str_replace (&tocode, canonical_buff);
141 static size_t b_encoder (char *s, ICONV_CONST char *d, size_t dlen,
146 memcpy (s, "=?", 2), s += 2;
147 memcpy (s, tocode, mutt_strlen (tocode)), s += mutt_strlen (tocode);
148 memcpy (s, "?B?", 3), s += 3;
152 else if (dlen == 1) {
153 *s++ = B64Chars[(*d >> 2) & 0x3f];
154 *s++ = B64Chars[(*d & 0x03) << 4];
159 else if (dlen == 2) {
160 *s++ = B64Chars[(*d >> 2) & 0x3f];
161 *s++ = B64Chars[((*d & 0x03) << 4) | ((d[1] >> 4) & 0x0f)];
162 *s++ = B64Chars[(d[1] & 0x0f) << 2];
167 *s++ = B64Chars[(*d >> 2) & 0x3f];
168 *s++ = B64Chars[((*d & 0x03) << 4) | ((d[1] >> 4) & 0x0f)];
169 *s++ = B64Chars[((d[1] & 0x0f) << 2) | ((d[2] >> 6) & 0x03)];
170 *s++ = B64Chars[d[2] & 0x3f];
174 memcpy (s, "?=", 2), s += 2;
178 static size_t q_encoder (char *s, ICONV_CONST char *d, size_t dlen,
181 char hex[] = "0123456789ABCDEF";
184 memcpy (s, "=?", 2), s += 2;
185 memcpy (s, tocode, mutt_strlen (tocode)), s += mutt_strlen (tocode);
186 memcpy (s, "?Q?", 3), s += 3;
188 unsigned char c = *d++;
192 else if (c >= 0x7f || c < 0x20 || c == '_' || strchr (MimeSpecials, c)) {
194 *s++ = hex[(c & 0xf0) >> 4];
195 *s++ = hex[c & 0x0f];
200 memcpy (s, "?=", 2), s += 2;
205 * Return 0 if and set *encoder and *wlen if the data (d, dlen) could
206 * be converted to an encoded word of length *wlen using *encoder.
207 * Otherwise return an upper bound on the maximum length of the data
208 * which could be converted.
209 * The data is converted from fromcode (which must be stateless) to
210 * tocode, unless fromcode is 0, in which case the data is assumed to
211 * be already in tocode, which should be 8-bit and stateless.
213 static size_t try_block (ICONV_CONST char *d, size_t dlen,
214 const char *fromcode, const char *tocode,
215 encoder_t * encoder, size_t * wlen)
217 char buf1[ENCWORD_LEN_MAX - ENCWORD_LEN_MIN + 1];
219 ICONV_CONST char *ib;
222 int count, len, len_b, len_q;
225 cd = mutt_iconv_open (tocode, fromcode, 0);
226 assert (cd != (iconv_t) (-1));
227 ib = d, ibl = dlen, ob = buf1, obl = sizeof (buf1) - mutt_strlen (tocode);
228 if (iconv (cd, &ib, &ibl, &ob, &obl) == (size_t) (-1) ||
229 iconv (cd, 0, 0, &ob, &obl) == (size_t) (-1)) {
230 assert (errno == E2BIG);
233 return (ib - d == dlen) ? dlen : ib - d + 1;
238 if (dlen > sizeof (buf1) - mutt_strlen (tocode))
239 return sizeof (buf1) - mutt_strlen (tocode) + 1;
240 memcpy (buf1, d, dlen);
245 for (p = buf1; p < ob; p++) {
246 unsigned char c = *p;
248 assert (strchr (MimeSpecials, '?'));
249 if (c >= 0x7f || c < 0x20 || *p == '_' ||
250 (c != ' ' && strchr (MimeSpecials, *p)))
254 len = ENCWORD_LEN_MIN - 2 + mutt_strlen (tocode);
255 len_b = len + (((ob - buf1) + 2) / 3) * 4;
256 len_q = len + (ob - buf1) + 2 * count;
258 /* Apparently RFC 1468 says to use B encoding for iso-2022-jp. */
259 if (!ascii_strcasecmp (tocode, "ISO-2022-JP"))
260 len_q = ENCWORD_LEN_MAX + 1;
262 if (len_b < len_q && len_b <= ENCWORD_LEN_MAX) {
263 *encoder = b_encoder;
267 else if (len_q <= ENCWORD_LEN_MAX) {
268 *encoder = q_encoder;
277 * Encode the data (d, dlen) into s using the encoder.
278 * Return the length of the encoded word.
280 static size_t encode_block (char *s, char *d, size_t dlen,
281 const char *fromcode, const char *tocode,
284 char buf1[ENCWORD_LEN_MAX - ENCWORD_LEN_MIN + 1];
286 ICONV_CONST char *ib;
288 size_t ibl, obl, n1, n2;
291 cd = mutt_iconv_open (tocode, fromcode, 0);
292 assert (cd != (iconv_t) (-1));
293 ib = d, ibl = dlen, ob = buf1, obl = sizeof (buf1) - mutt_strlen (tocode);
294 n1 = iconv (cd, &ib, &ibl, &ob, &obl);
295 n2 = iconv (cd, 0, 0, &ob, &obl);
296 assert (n1 != (size_t) (-1) && n2 != (size_t) (-1));
298 return (*encoder) (s, buf1, ob - buf1, tocode);
301 return (*encoder) (s, d, dlen, tocode);
305 * Discover how much of the data (d, dlen) can be converted into
306 * a single encoded word. Return how much data can be converted,
307 * and set the length *wlen of the encoded word and *encoder.
308 * We start in column col, which limits the length of the word.
310 static size_t choose_block (char *d, size_t dlen, int col,
311 const char *fromcode, const char *tocode,
312 encoder_t * encoder, size_t * wlen)
315 int utf8 = fromcode && !ascii_strcasecmp (fromcode, "UTF-8");
320 nn = try_block (d, n, fromcode, tocode, encoder, wlen);
321 if (!nn && (col + *wlen <= ENCWORD_LEN_MAX + 1 || n <= 1))
323 n = (nn ? nn : n) - 1;
326 while (n > 1 && CONTINUATION_BYTE (d[n]))
333 * Place the result of RFC-2047-encoding (d, dlen) into the dynamically
334 * allocated buffer (e, elen). The input data is in charset fromcode
335 * and is converted into a charset chosen from charsets.
336 * Return 1 if the conversion to UTF-8 failed, 2 if conversion from UTF-8
337 * failed, otherwise 0. If conversion failed, fromcode is assumed to be
338 * compatible with us-ascii and the original data is used.
339 * The input data is assumed to be a single line starting at column col;
340 * if col is non-zero, the preceding character was a space.
342 static int rfc2047_encode (ICONV_CONST char *d, size_t dlen, int col,
343 const char *fromcode, const char *charsets,
344 char **e, size_t * elen, char *specials)
348 size_t bufpos, buflen;
349 char *u, *t0, *t1, *t;
351 size_t ulen, r, n, wlen;
355 char *icode = "UTF-8";
357 /* Try to convert to UTF-8. */
358 if (convert_string (d, dlen, fromcode, icode, &u, &ulen)) {
361 u = safe_malloc ((ulen = dlen) + 1);
366 /* Find earliest and latest things we must encode. */
367 s0 = s1 = t0 = t1 = 0;
368 for (t = u; t < u + ulen; t++) {
370 (*t == '=' && t[1] == '?' && (t == u || HSPACE (*(t - 1))))) {
375 else if (specials && strchr (specials, *t)) {
382 /* If we have something to encode, include RFC822 specials */
383 if (t0 && s0 && s0 < t0)
385 if (t1 && s1 && s1 > t1)
389 /* No encoding is required. */
395 /* Choose target charset. */
398 if ((tocode1 = mutt_choose_charset (icode, charsets, u, ulen, 0, 0)))
404 /* Hack to avoid labelling 8-bit data as us-ascii. */
405 if (!icode && mutt_is_us_ascii (tocode))
406 tocode = "unknown-8bit";
408 /* Adjust t0 for maximum length of line. */
409 t = u + (ENCWORD_LEN_MAX + 1) - col - ENCWORD_LEN_MIN;
416 /* Adjust t0 until we can encode a character after a space. */
417 for (; t0 > u; t0--) {
418 if (!HSPACE (*(t0 - 1)))
422 while (t < u + ulen && CONTINUATION_BYTE (*t))
424 if (!try_block (t0, t - t0, icode, tocode, &encoder, &wlen) &&
425 col + (t0 - u) + wlen <= ENCWORD_LEN_MAX + 1)
429 /* Adjust t1 until we can encode a character before a space. */
430 for (; t1 < u + ulen; t1++) {
435 while (CONTINUATION_BYTE (*t))
437 if (!try_block (t, t1 - t, icode, tocode, &encoder, &wlen) &&
438 1 + wlen + (u + ulen - t1) <= ENCWORD_LEN_MAX + 1)
442 /* We shall encode the region [t0,t1). */
444 /* Initialise the output buffer with the us-ascii prefix. */
446 buf = safe_malloc (buflen);
448 memcpy (buf, u, t0 - u);
454 /* Find how much we can encode. */
455 n = choose_block (t, t1 - t, col, icode, tocode, &encoder, &wlen);
457 /* See if we can fit the us-ascii suffix, too. */
458 if (col + wlen + (u + ulen - t1) <= ENCWORD_LEN_MAX + 1)
462 while (CONTINUATION_BYTE (t[n]))
466 /* This should only happen in the really stupid case where the
467 only word that needs encoding is one character long, but
468 there is too much us-ascii stuff after it to use a single
469 encoded word. We add the next word to the encoded region
471 assert (t1 < u + ulen);
472 for (t1++; t1 < u + ulen && !HSPACE (*t1); t1++);
475 n = choose_block (t, n, col, icode, tocode, &encoder, &wlen);
478 /* Add to output buffer. */
479 #define LINEBREAK "\n\t"
480 if (bufpos + wlen + mutt_strlen (LINEBREAK) > buflen) {
481 buflen = bufpos + wlen + mutt_strlen (LINEBREAK);
482 safe_realloc (&buf, buflen);
484 r = encode_block (buf + bufpos, t, n, icode, tocode, encoder);
487 memcpy (buf + bufpos, LINEBREAK, mutt_strlen (LINEBREAK));
488 bufpos += mutt_strlen (LINEBREAK);
496 /* Add last encoded word and us-ascii suffix to buffer. */
497 buflen = bufpos + wlen + (u + ulen - t1);
498 safe_realloc (&buf, buflen + 1);
499 r = encode_block (buf + bufpos, t, t1 - t, icode, tocode, encoder);
502 memcpy (buf + bufpos, t1, u + ulen - t1);
514 void _rfc2047_encode_string (char **pd, int encode_specials, int col)
520 if (!Charset || !*pd)
523 charsets = SendCharset;
524 if (!charsets || !*charsets)
527 rfc2047_encode (*pd, mutt_strlen (*pd), col,
528 Charset, charsets, &e, &elen,
529 encode_specials ? RFC822Specials : NULL);
535 void rfc2047_encode_adrlist (ADDRESS * addr, const char *tag)
538 int col = tag ? mutt_strlen (tag) + 2 : 32;
542 _rfc2047_encode_string (&ptr->personal, 1, col);
545 _rfc2047_encode_string (&ptr->val, 1, col);
551 static int rfc2047_decode_word (char *d, const char *s, size_t len)
553 const char *pp, *pp1;
556 int enc = 0, count = 0;
557 char *charset = NULL;
559 pd = d0 = safe_malloc (mutt_strlen (s));
561 for (pp = s; (pp1 = strchr (pp, '?')); pp = pp1 + 1) {
565 /* ignore language specification a la RFC 2231 */
567 if ((t1 = memchr (pp, '*', t - pp)))
569 charset = safe_malloc (t - pp + 1);
570 memcpy (charset, pp, t - pp);
571 charset[t - pp] = '\0';
574 if (toupper ((unsigned char) *pp) == 'Q')
575 enc = ENCQUOTEDPRINTABLE;
576 else if (toupper ((unsigned char) *pp) == 'B')
585 if (enc == ENCQUOTEDPRINTABLE) {
586 for (; pp < pp1; pp++) {
589 else if (*pp == '=' &&
590 (!(pp[1] & ~127) && hexval (pp[1]) != -1) &&
591 (!(pp[2] & ~127) && hexval (pp[2]) != -1)) {
592 *pd++ = (hexval (pp[1]) << 4) | hexval (pp[2]);
600 else if (enc == ENCBASE64) {
603 for (; pp < pp1; pp++) {
606 if ((*pp & ~127) || (c = base64val (*pp)) == -1)
610 *pd++ = b | (c >> k);
625 mutt_convert_string (&d0, charset, Charset, M_ICONV_HOOK_FROM);
626 strfcpy (d, d0, len);
633 * Find the start and end of the first encoded word in the string.
634 * We use the grammar in section 2 of RFC 2047, but the "encoding"
635 * must be B or Q. Also, we don't require the encoded word to be
636 * separated by linear-white-space (section 5(1)).
638 static const char *find_encoded_word (const char *s, const char **x)
643 while ((p = strstr (q, "=?"))) {
645 0x20 < *q && *q < 0x7f && !strchr ("()<>@,;:\"/[]?.=", *q); q++);
646 if (q[0] != '?' || !strchr ("BbQq", q[1]) || q[2] != '?')
648 for (q = q + 3; 0x20 <= *q && *q < 0x7f && *q != '?'; q++);
649 if (q[0] != '?' || q[1] != '=') {
661 /* return length of linear white space */
662 static size_t lwslen (const char *s, size_t n)
670 for (; p < s + n; p++)
671 if (!strchr (" \t\r\n", *p)) {
672 len = (size_t) (p - s);
675 if (strchr ("\r\n", *(p - 1))) /* LWS doesn't end with CRLF */
680 /* return length of linear white space : reverse */
681 static size_t lwsrlen (const char *s, size_t n)
683 const char *p = s + n - 1;
689 if (strchr ("\r\n", *p)) /* LWS doesn't end with CRLF */
693 if (!strchr (" \t\r\n", *p)) {
694 len = (size_t) (s + n - 1 - p);
700 /* try to decode anything that looks like a valid RFC2047 encoded
701 * header field, ignoring RFC822 parsing rules
703 void rfc2047_decode (char **pd)
707 int found_encoded = 0;
715 dlen = 4 * mutt_strlen (s); /* should be enough */
716 d = d0 = safe_malloc (dlen + 1);
718 while (*s && dlen > 0) {
719 if (!(p = find_encoded_word (s, &q))) {
720 /* no encoded words */
721 if (!option (OPTSTRICTMIME)) {
723 if (found_encoded && (m = lwslen (s, n)) != 0) {
725 *d = ' ', d++, dlen--;
728 if (ascii_strcasecmp (AssumedCharset, "us-ascii")) {
732 t = safe_malloc (n + 1);
733 strfcpy (t, s, n + 1);
734 if (mutt_convert_nonmime_string (&t) == 0) {
735 tlen = mutt_strlen (t);
736 strncpy (d, t, tlen);
747 strncpy (d, s, dlen);
753 n = (size_t) (p - s);
754 /* ignore spaces between encoded words
755 * and linear white spaces between encoded word and *text */
756 if (!option (OPTSTRICTMIME)) {
757 if (found_encoded && (m = lwslen (s, n)) != 0) {
759 *d = ' ', d++, dlen--;
763 if ((m = n - lwsrlen (s, n)) != 0) {
770 *d = ' ', d++, dlen--;
773 else if (!found_encoded || strspn (s, " \t\r\n") != n) {
782 rfc2047_decode_word (d, p, dlen);
793 mutt_str_adjust (pd);
796 void rfc2047_decode_adrlist (ADDRESS * a)
800 rfc2047_decode (&a->personal);
802 if (a->val && strstr (a->val, "=?") != NULL)
803 rfc2047_decode (&a->val);