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.
15 #include <lib-lib/mem.h>
16 #include <lib-lib/str.h>
17 #include <lib-lib/ascii.h>
19 #include <lib-mime/mime.h>
33 /* If you are debugging this file, comment out the following line. */
42 #define ENCWORD_LEN_MAX 75
43 #define ENCWORD_LEN_MIN 9 /* m_strlen("=?.?.?.?=") */
45 #define HSPACE(x) ((x) == '\0' || (x) == ' ' || (x) == '\t')
47 #define CONTINUATION_BYTE(c) (((c) & 0xc0) == 0x80)
49 typedef size_t (*encoder_t) (char *, const char *, size_t,
52 static size_t convert_string (const char *f, size_t flen,
53 const char *from, const char *to,
54 char **t, size_t * tlen)
61 cd = mutt_iconv_open (to, from, 0);
62 if (cd == (iconv_t) (-1))
65 ob = buf = p_new(char, obl);
66 n = my_iconv(cd, &f, &flen, &ob, &obl);
67 if (n == (size_t) (-1) || my_iconv(cd, 0, 0, &ob, &obl) == (size_t) (-1)) {
78 p_realloc(&buf, ob - buf + 1);
85 char *mutt_choose_charset (const char *fromcode, const char *charsets,
86 char *u, size_t ulen, char **d, size_t * dlen)
88 char canonical_buff[LONG_STRING];
89 char *e = 0, *tocode = 0;
90 size_t elen = 0, bestn = 0;
93 for (p = charsets; p; p = q ? q + 1 : 0) {
99 n = q ? q - p : m_strlen(p);
102 /* Assume that we never need more than 12 characters of
103 encoded-text to encode a single character. */
104 n > (ENCWORD_LEN_MAX - ENCWORD_LEN_MIN + 2 - 12))
109 n = convert_string (u, ulen, fromcode, t, &s, &slen);
110 if (n == (size_t) (-1))
113 if (!tocode || n < bestn) {
138 mutt_canonical_charset (canonical_buff, sizeof (canonical_buff), tocode);
139 str_replace (&tocode, canonical_buff);
144 static size_t b_encoder (char *s, const char *d, size_t dlen,
149 memcpy (s, "=?", 2), s += 2;
150 memcpy (s, tocode, m_strlen(tocode)), s += m_strlen(tocode);
151 memcpy (s, "?B?", 3), s += 3;
155 else if (dlen == 1) {
156 *s++ = __m_b64chars[(*d >> 2) & 0x3f];
157 *s++ = __m_b64chars[(*d & 0x03) << 4];
162 else if (dlen == 2) {
163 *s++ = __m_b64chars[(*d >> 2) & 0x3f];
164 *s++ = __m_b64chars[((*d & 0x03) << 4) | ((d[1] >> 4) & 0x0f)];
165 *s++ = __m_b64chars[(d[1] & 0x0f) << 2];
170 *s++ = __m_b64chars[(*d >> 2) & 0x3f];
171 *s++ = __m_b64chars[((*d & 0x03) << 4) | ((d[1] >> 4) & 0x0f)];
172 *s++ = __m_b64chars[((d[1] & 0x0f) << 2) | ((d[2] >> 6) & 0x03)];
173 *s++ = __m_b64chars[d[2] & 0x3f];
177 memcpy (s, "?=", 2), s += 2;
181 static size_t q_encoder (char *s, const char *d, size_t dlen,
184 char hex[] = "0123456789ABCDEF";
187 memcpy (s, "=?", 2), s += 2;
188 memcpy (s, tocode, m_strlen(tocode)), s += m_strlen(tocode);
189 memcpy (s, "?Q?", 3), s += 3;
191 unsigned char c = *d++;
195 else if (c >= 0x7f || c < 0x20 || c == '_' || strchr (MimeSpecials, c)) {
197 *s++ = hex[(c & 0xf0) >> 4];
198 *s++ = hex[c & 0x0f];
203 memcpy (s, "?=", 2), s += 2;
208 * Return 0 if and set *encoder and *wlen if the data (d, dlen) could
209 * be converted to an encoded word of length *wlen using *encoder.
210 * Otherwise return an upper bound on the maximum length of the data
211 * which could be converted.
212 * The data is converted from fromcode (which must be stateless) to
213 * tocode, unless fromcode is 0, in which case the data is assumed to
214 * be already in tocode, which should be 8-bit and stateless.
216 static size_t try_block (const char *d, size_t dlen,
217 const char *fromcode, const char *tocode,
218 encoder_t * encoder, size_t * wlen)
220 char buf1[ENCWORD_LEN_MAX - ENCWORD_LEN_MIN + 1];
225 int count, len, len_b, len_q;
228 cd = mutt_iconv_open (tocode, fromcode, 0);
229 assert (cd != (iconv_t) (-1));
230 ib = d, ibl = dlen, ob = buf1, obl = sizeof (buf1) - m_strlen(tocode);
231 if (my_iconv(cd, &ib, &ibl, &ob, &obl) == (size_t) (-1) ||
232 my_iconv(cd, 0, 0, &ob, &obl) == (size_t) (-1)) {
233 assert (errno == E2BIG);
236 return (ib - d == dlen) ? dlen : ib - d + 1;
241 if (dlen > sizeof (buf1) - m_strlen(tocode))
242 return sizeof (buf1) - m_strlen(tocode) + 1;
243 memcpy (buf1, d, dlen);
248 for (p = buf1; p < ob; p++) {
249 unsigned char c = *p;
251 assert (strchr (MimeSpecials, '?'));
252 if (c >= 0x7f || c < 0x20 || *p == '_' ||
253 (c != ' ' && strchr (MimeSpecials, *p)))
257 len = ENCWORD_LEN_MIN - 2 + m_strlen(tocode);
258 len_b = len + (((ob - buf1) + 2) / 3) * 4;
259 len_q = len + (ob - buf1) + 2 * count;
261 /* Apparently RFC 1468 says to use B encoding for iso-2022-jp. */
262 if (!ascii_strcasecmp (tocode, "ISO-2022-JP"))
263 len_q = ENCWORD_LEN_MAX + 1;
265 if (len_b < len_q && len_b <= ENCWORD_LEN_MAX) {
266 *encoder = b_encoder;
270 else if (len_q <= ENCWORD_LEN_MAX) {
271 *encoder = q_encoder;
280 * Encode the data (d, dlen) into s using the encoder.
281 * Return the length of the encoded word.
283 static size_t encode_block (char *s, char *d, size_t dlen,
284 const char *fromcode, const char *tocode,
287 char buf1[ENCWORD_LEN_MAX - ENCWORD_LEN_MIN + 1];
291 size_t ibl, obl, n1, n2;
294 cd = mutt_iconv_open (tocode, fromcode, 0);
295 assert (cd != (iconv_t) (-1));
296 ib = d, ibl = dlen, ob = buf1, obl = sizeof (buf1) - m_strlen(tocode);
297 n1 = my_iconv(cd, &ib, &ibl, &ob, &obl);
298 n2 = my_iconv(cd, 0, 0, &ob, &obl);
299 assert (n1 != (size_t) (-1) && n2 != (size_t) (-1));
301 return (*encoder) (s, buf1, ob - buf1, tocode);
304 return (*encoder) (s, d, dlen, tocode);
308 * Discover how much of the data (d, dlen) can be converted into
309 * a single encoded word. Return how much data can be converted,
310 * and set the length *wlen of the encoded word and *encoder.
311 * We start in column col, which limits the length of the word.
313 static size_t choose_block (char *d, size_t dlen, int col,
314 const char *fromcode, const char *tocode,
315 encoder_t * encoder, size_t * wlen)
318 int utf8 = fromcode && !ascii_strcasecmp (fromcode, "UTF-8");
323 nn = try_block (d, n, fromcode, tocode, encoder, wlen);
324 if (!nn && (col + *wlen <= ENCWORD_LEN_MAX + 1 || n <= 1))
326 n = (nn ? nn : n) - 1;
329 while (n > 1 && CONTINUATION_BYTE (d[n]))
336 * Place the result of RFC-2047-encoding (d, dlen) into the dynamically
337 * allocated buffer (e, elen). The input data is in charset fromcode
338 * and is converted into a charset chosen from charsets.
339 * Return 1 if the conversion to UTF-8 failed, 2 if conversion from UTF-8
340 * failed, otherwise 0. If conversion failed, fromcode is assumed to be
341 * compatible with us-ascii and the original data is used.
342 * The input data is assumed to be a single line starting at column col;
343 * if col is non-zero, the preceding character was a space.
345 static int rfc2047_encode (const char *d, size_t dlen, int col,
346 const char *fromcode, const char *charsets,
347 char **e, size_t * elen, char *specials)
351 size_t bufpos, buflen;
352 char *u, *t0, *t1, *t;
354 size_t ulen, r, n, wlen;
358 const char *icode = "UTF-8";
360 /* Try to convert to UTF-8. */
361 if (convert_string (d, dlen, fromcode, icode, &u, &ulen)) {
364 u = p_dupstr(d, ulen = dlen);
367 /* Find earliest and latest things we must encode. */
368 s0 = s1 = t0 = t1 = 0;
369 for (t = u; t < u + ulen; t++) {
371 (*t == '=' && t[1] == '?' && (t == u || HSPACE (*(t - 1))))) {
376 else if (specials && strchr (specials, *t)) {
383 /* If we have something to encode, include RFC822 specials */
384 if (t0 && s0 && s0 < t0)
386 if (t1 && s1 && s1 > t1)
390 /* No encoding is required. */
396 /* Choose target charset. */
399 if ((tocode1 = mutt_choose_charset (icode, charsets, u, ulen, 0, 0)))
405 /* Hack to avoid labelling 8-bit data as us-ascii. */
406 if (!icode && mutt_is_us_ascii (tocode))
407 tocode = "unknown-8bit";
409 /* Adjust t0 for maximum length of line. */
410 t = u + (ENCWORD_LEN_MAX + 1) - col - ENCWORD_LEN_MIN;
417 /* Adjust t0 until we can encode a character after a space. */
418 for (; t0 > u; t0--) {
419 if (!HSPACE (*(t0 - 1)))
423 while (t < u + ulen && CONTINUATION_BYTE (*t))
425 if (!try_block (t0, t - t0, icode, tocode, &encoder, &wlen) &&
426 col + (t0 - u) + wlen <= ENCWORD_LEN_MAX + 1)
430 /* Adjust t1 until we can encode a character before a space. */
431 for (; t1 < u + ulen; t1++) {
436 while (CONTINUATION_BYTE (*t))
438 if (!try_block (t, t1 - t, icode, tocode, &encoder, &wlen) &&
439 1 + wlen + (u + ulen - t1) <= ENCWORD_LEN_MAX + 1)
443 /* We shall encode the region [t0,t1). */
445 /* Initialise the output buffer with the us-ascii prefix. */
447 buf = p_new(char, buflen);
449 memcpy (buf, u, t0 - u);
455 /* Find how much we can encode. */
456 n = choose_block (t, t1 - t, col, icode, tocode, &encoder, &wlen);
458 /* See if we can fit the us-ascii suffix, too. */
459 if (col + wlen + (u + ulen - t1) <= ENCWORD_LEN_MAX + 1)
463 while (CONTINUATION_BYTE (t[n]))
467 /* This should only happen in the really stupid case where the
468 only word that needs encoding is one character long, but
469 there is too much us-ascii stuff after it to use a single
470 encoded word. We add the next word to the encoded region
472 assert (t1 < u + ulen);
473 for (t1++; t1 < u + ulen && !HSPACE (*t1); t1++);
476 n = choose_block (t, n, col, icode, tocode, &encoder, &wlen);
479 /* Add to output buffer. */
480 #define LINEBREAK "\n\t"
481 if (bufpos + wlen + m_strlen(LINEBREAK) > buflen) {
482 buflen = bufpos + wlen + m_strlen(LINEBREAK);
483 p_realloc(&buf, buflen);
485 r = encode_block (buf + bufpos, t, n, icode, tocode, encoder);
488 memcpy (buf + bufpos, LINEBREAK, m_strlen(LINEBREAK));
489 bufpos += m_strlen(LINEBREAK);
497 /* Add last encoded word and us-ascii suffix to buffer. */
498 buflen = bufpos + wlen + (u + ulen - t1);
499 p_realloc(&buf, buflen + 1);
500 r = encode_block (buf + bufpos, t, t1 - t, icode, tocode, encoder);
503 memcpy (buf + bufpos, t1, u + ulen - t1);
515 void _rfc2047_encode_string (char **pd, int encode_specials, int col)
519 const char *charsets;
521 if (!Charset || !*pd)
524 charsets = SendCharset;
525 if (!charsets || !*charsets)
528 rfc2047_encode (*pd, m_strlen(*pd), col,
529 Charset, charsets, &e, &elen,
530 encode_specials ? RFC822Specials : NULL);
536 void rfc2047_encode_adrlist (address_t * addr, const char *tag)
538 address_t *ptr = addr;
539 int col = tag ? m_strlen(tag) + 2 : 32;
543 _rfc2047_encode_string (&ptr->personal, 1, col);
548 static int rfc2047_decode_word (char *d, const char *s, size_t len)
550 const char *pp, *pp1;
553 int enc = 0, count = 0;
554 char *charset = NULL;
556 pd = d0 = p_new(char, m_strlen(s));
558 for (pp = s; (pp1 = strchr (pp, '?')); pp = pp1 + 1) {
562 /* ignore language specification a la RFC 2231 */
564 if ((t1 = memchr (pp, '*', t - pp)))
566 charset = p_dupstr(pp, t - pp);
569 if (toupper ((unsigned char) *pp) == 'Q')
570 enc = ENCQUOTEDPRINTABLE;
571 else if (toupper ((unsigned char) *pp) == 'B')
580 if (enc == ENCQUOTEDPRINTABLE) {
581 for (; pp < pp1; pp++) {
584 else if (*pp == '=' && hexval(pp[1]) >= 0 && hexval(pp[2]) >= 0) {
585 *pd++ = (hexval (pp[1]) << 4) | hexval (pp[2]);
593 else if (enc == ENCBASE64) {
596 for (; pp < pp1; pp++) {
599 if ((c = base64val(*pp)) < 0)
603 *pd++ = b | (c >> k);
618 mutt_convert_string (&d0, charset, Charset, M_ICONV_HOOK_FROM);
619 m_strcpy(d, len, d0);
626 * Find the start and end of the first encoded word in the string.
627 * We use the grammar in section 2 of RFC 2047, but the "encoding"
628 * must be B or Q. Also, we don't require the encoded word to be
629 * separated by linear-white-space (section 5(1)).
631 static const char *find_encoded_word (const char *s, const char **x)
636 while ((p = strstr (q, "=?"))) {
638 0x20 < *q && *q < 0x7f && !strchr ("()<>@,;:\"/[]?.=", *q); q++);
639 if (q[0] != '?' || !strchr ("BbQq", q[1]) || q[2] != '?')
641 for (q = q + 3; 0x20 <= *q && *q < 0x7f && *q != '?'; q++);
642 if (q[0] != '?' || q[1] != '=') {
654 /* return length of linear white space */
655 static size_t lwslen (const char *s, size_t n)
663 for (; p < s + n; p++)
664 if (!strchr (" \t\r\n", *p)) {
665 len = (size_t) (p - s);
668 if (strchr ("\r\n", *(p - 1))) /* LWS doesn't end with CRLF */
673 /* return length of linear white space : reverse */
674 static size_t lwsrlen (const char *s, size_t n)
676 const char *p = s + n - 1;
682 if (strchr ("\r\n", *p)) /* LWS doesn't end with CRLF */
686 if (!strchr (" \t\r\n", *p)) {
687 len = (size_t) (s + n - 1 - p);
693 /* try to decode anything that looks like a valid RFC2047 encoded
694 * header field, ignoring RFC822 parsing rules
696 void rfc2047_decode (char **pd)
700 int found_encoded = 0;
708 dlen = 4 * m_strlen(s); /* should be enough */
709 d = d0 = p_new(char, dlen + 1);
711 while (*s && dlen > 0) {
712 if (!(p = find_encoded_word (s, &q))) {
713 /* no encoded words */
714 if (!option (OPTSTRICTMIME)) {
716 if (found_encoded && (m = lwslen (s, n)) != 0) {
718 *d = ' ', d++, dlen--;
721 if (ascii_strcasecmp (AssumedCharset, "us-ascii")) {
726 if (mutt_convert_nonmime_string (&t) == 0) {
728 strncpy (d, t, tlen);
739 strncpy (d, s, dlen);
745 n = (size_t) (p - s);
746 /* ignore spaces between encoded words
747 * and linear white spaces between encoded word and *text */
748 if (!option (OPTSTRICTMIME)) {
749 if (found_encoded && (m = lwslen (s, n)) != 0) {
751 *d = ' ', d++, dlen--;
755 if ((m = n - lwsrlen (s, n)) != 0) {
762 *d = ' ', d++, dlen--;
765 else if (!found_encoded || strspn (s, " \t\r\n") != n) {
774 rfc2047_decode_word (d, p, dlen);
788 void rfc2047_decode_adrlist (address_t * a)
792 rfc2047_decode (&a->personal);
797 void rfc2047_decode_envelope (ENVELOPE* e) {
802 /* do RFC2047 decoding */
803 rfc2047_decode_adrlist (e->from);
804 rfc2047_decode_adrlist (e->to);
805 rfc2047_decode_adrlist (e->cc);
806 rfc2047_decode_adrlist (e->bcc);
807 rfc2047_decode_adrlist (e->reply_to);
808 rfc2047_decode_adrlist (e->mail_followup_to);
809 rfc2047_decode_adrlist (e->return_path);
810 rfc2047_decode_adrlist (e->sender);
813 rfc2047_decode (&e->subject);
814 mutt_adjust_subject (e);