2 * Copyright (C) 1996-2000 Michael R. Elkins <me@mutt.org>
3 * Copyright (C) 2000-2001 Edmund Grimley Evans <edmundo@rano.org>
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA.
35 /* If you are debugging this file, comment out the following line. */
44 #define ENCWORD_LEN_MAX 75
45 #define ENCWORD_LEN_MIN 9 /* strlen ("=?.?.?.?=") */
47 #define HSPACE(x) ((x) == '\0' || (x) == ' ' || (x) == '\t')
49 #define CONTINUATION_BYTE(c) (((c) & 0xc0) == 0x80)
51 extern char RFC822Specials[];
53 typedef size_t (*encoder_t) (char *, ICONV_CONST char *, size_t,
56 static size_t convert_string (ICONV_CONST char *f, size_t flen,
57 const char *from, const char *to,
58 char **t, size_t *tlen)
65 cd = mutt_iconv_open (to, from, 0);
66 if (cd == (iconv_t)(-1))
69 ob = buf = safe_malloc (obl);
70 n = iconv (cd, &f, &flen, &ob, &obl);
71 if (n == (size_t)(-1) || iconv (cd, 0, 0, &ob, &obl) == (size_t)(-1))
83 safe_realloc (&buf, ob - buf + 1);
90 char *mutt_choose_charset (const char *fromcode, const char *charsets,
91 char *u, size_t ulen, char **d, size_t *dlen)
93 char canonical_buff[LONG_STRING];
94 char *e = 0, *tocode = 0;
95 size_t elen = 0, bestn = 0;
98 for (p = charsets; p; p = q ? q + 1 : 0)
105 n = q ? q - p : strlen (p);
108 /* Assume that we never need more than 12 characters of
109 encoded-text to encode a single character. */
110 n > (ENCWORD_LEN_MAX - ENCWORD_LEN_MIN + 2 - 12))
113 t = safe_malloc (n + 1);
117 n = convert_string (u, ulen, fromcode, t, &s, &slen);
118 if (n == (size_t)(-1))
121 if (!tocode || n < bestn)
150 mutt_canonical_charset (canonical_buff, sizeof (canonical_buff), tocode);
151 mutt_str_replace (&tocode, canonical_buff);
156 static size_t b_encoder (char *s, ICONV_CONST char *d, size_t dlen,
161 memcpy (s, "=?", 2), s += 2;
162 memcpy (s, tocode, strlen (tocode)), s += strlen (tocode);
163 memcpy (s, "?B?", 3), s += 3;
170 *s++ = B64Chars[(*d >> 2) & 0x3f];
171 *s++ = B64Chars[(*d & 0x03) << 4];
178 *s++ = B64Chars[(*d >> 2) & 0x3f];
179 *s++ = B64Chars[((*d & 0x03) << 4) | ((d[1] >> 4) & 0x0f)];
180 *s++ = B64Chars[(d[1] & 0x0f) << 2];
186 *s++ = B64Chars[(*d >> 2) & 0x3f];
187 *s++ = B64Chars[((*d & 0x03) << 4) | ((d[1] >> 4) & 0x0f)];
188 *s++ = B64Chars[((d[1] & 0x0f) << 2) | ((d[2] >> 6) & 0x03)];
189 *s++ = B64Chars[d[2] & 0x3f];
193 memcpy (s, "?=", 2), s += 2;
197 static size_t q_encoder (char *s, ICONV_CONST char *d, size_t dlen,
200 char hex[] = "0123456789ABCDEF";
203 memcpy (s, "=?", 2), s += 2;
204 memcpy (s, tocode, strlen (tocode)), s += strlen (tocode);
205 memcpy (s, "?Q?", 3), s += 3;
208 unsigned char c = *d++;
211 else if (c >= 0x7f || c < 0x20 || c == '_' || strchr (MimeSpecials, c))
214 *s++ = hex[(c & 0xf0) >> 4];
215 *s++ = hex[c & 0x0f];
220 memcpy (s, "?=", 2), s += 2;
225 * Return 0 if and set *encoder and *wlen if the data (d, dlen) could
226 * be converted to an encoded word of length *wlen using *encoder.
227 * Otherwise return an upper bound on the maximum length of the data
228 * which could be converted.
229 * The data is converted from fromcode (which must be stateless) to
230 * tocode, unless fromcode is 0, in which case the data is assumed to
231 * be already in tocode, which should be 8-bit and stateless.
233 static size_t try_block (ICONV_CONST char *d, size_t dlen,
234 const char *fromcode, const char *tocode,
235 encoder_t *encoder, size_t *wlen)
237 char buf1[ENCWORD_LEN_MAX - ENCWORD_LEN_MIN + 1];
239 ICONV_CONST char *ib;
242 int count, len, len_b, len_q;
246 cd = mutt_iconv_open (tocode, fromcode, 0);
247 assert (cd != (iconv_t)(-1));
248 ib = d, ibl = dlen, ob = buf1, obl = sizeof (buf1) - strlen (tocode);
249 if (iconv (cd, &ib, &ibl, &ob, &obl) == (size_t)(-1) ||
250 iconv (cd, 0, 0, &ob, &obl) == (size_t)(-1))
252 assert (errno == E2BIG);
255 return (ib - d == dlen) ? dlen : ib - d + 1;
261 if (dlen > sizeof (buf1) - strlen (tocode))
262 return sizeof (buf1) - strlen (tocode) + 1;
263 memcpy (buf1, d, dlen);
268 for (p = buf1; p < ob; p++)
270 unsigned char c = *p;
271 assert (strchr (MimeSpecials, '?'));
272 if (c >= 0x7f || c < 0x20 || *p == '_' ||
273 (c != ' ' && strchr (MimeSpecials, *p)))
277 len = ENCWORD_LEN_MIN - 2 + strlen (tocode);
278 len_b = len + (((ob - buf1) + 2) / 3) * 4;
279 len_q = len + (ob - buf1) + 2 * count;
281 /* Apparently RFC 1468 says to use B encoding for iso-2022-jp. */
282 if (!ascii_strcasecmp (tocode, "ISO-2022-JP"))
283 len_q = ENCWORD_LEN_MAX + 1;
285 if (len_b < len_q && len_b <= ENCWORD_LEN_MAX)
287 *encoder = b_encoder;
291 else if (len_q <= ENCWORD_LEN_MAX)
293 *encoder = q_encoder;
302 * Encode the data (d, dlen) into s using the encoder.
303 * Return the length of the encoded word.
305 static size_t encode_block (char *s, char *d, size_t dlen,
306 const char *fromcode, const char *tocode,
309 char buf1[ENCWORD_LEN_MAX - ENCWORD_LEN_MIN + 1];
311 ICONV_CONST char *ib;
313 size_t ibl, obl, n1, n2;
317 cd = mutt_iconv_open (tocode, fromcode, 0);
318 assert (cd != (iconv_t)(-1));
319 ib = d, ibl = dlen, ob = buf1, obl = sizeof (buf1) - strlen (tocode);
320 n1 = iconv (cd, &ib, &ibl, &ob, &obl);
321 n2 = iconv (cd, 0, 0, &ob, &obl);
322 assert (n1 != (size_t)(-1) && n2 != (size_t)(-1));
324 return (*encoder) (s, buf1, ob - buf1, tocode);
327 return (*encoder) (s, d, dlen, tocode);
331 * Discover how much of the data (d, dlen) can be converted into
332 * a single encoded word. Return how much data can be converted,
333 * and set the length *wlen of the encoded word and *encoder.
334 * We start in column col, which limits the length of the word.
336 static size_t choose_block (char *d, size_t dlen, int col,
337 const char *fromcode, const char *tocode,
338 encoder_t *encoder, size_t *wlen)
341 int utf8 = fromcode && !ascii_strcasecmp (fromcode, "UTF-8");
347 nn = try_block (d, n, fromcode, tocode, encoder, wlen);
348 if (!nn && (col + *wlen <= ENCWORD_LEN_MAX + 1 || n <= 1))
350 n = (nn ? nn : n) - 1;
353 while (n > 1 && CONTINUATION_BYTE(d[n]))
360 * Place the result of RFC-2047-encoding (d, dlen) into the dynamically
361 * allocated buffer (e, elen). The input data is in charset fromcode
362 * and is converted into a charset chosen from charsets.
363 * Return 1 if the conversion to UTF-8 failed, 2 if conversion from UTF-8
364 * failed, otherwise 0. If conversion failed, fromcode is assumed to be
365 * compatible with us-ascii and the original data is used.
366 * The input data is assumed to be a single line starting at column col;
367 * if col is non-zero, the preceding character was a space.
369 static int rfc2047_encode (ICONV_CONST char *d, size_t dlen, int col,
370 const char *fromcode, const char *charsets,
371 char **e, size_t *elen, char *specials)
375 size_t bufpos, buflen;
376 char *u, *t0, *t1, *t;
378 size_t ulen, r, n, wlen;
382 char *icode = "UTF-8";
384 /* Try to convert to UTF-8. */
385 if (convert_string (d, dlen, fromcode, icode, &u, &ulen))
389 u = safe_malloc ((ulen = dlen) + 1);
394 /* Find earliest and latest things we must encode. */
395 s0 = s1 = t0 = t1 = 0;
396 for (t = u; t < u + ulen; t++)
399 (*t == '=' && t[1] == '?' && (t == u || HSPACE(*(t-1)))))
404 else if (specials && strchr (specials, *t))
411 /* If we have something to encode, include RFC822 specials */
412 if (t0 && s0 && s0 < t0)
414 if (t1 && s1 && s1 > t1)
419 /* No encoding is required. */
425 /* Choose target charset. */
429 if ((tocode1 = mutt_choose_charset (icode, charsets, u, ulen, 0, 0)))
435 /* Hack to avoid labelling 8-bit data as us-ascii. */
436 if (!icode && mutt_is_us_ascii (tocode))
437 tocode = "unknown-8bit";
439 /* Adjust t0 for maximum length of line. */
440 t = u + (ENCWORD_LEN_MAX + 1) - col - ENCWORD_LEN_MIN;
445 /* Adjust t0 until we can encode a character after a space. */
448 if (!HSPACE(*(t0-1)))
452 while (t < u + ulen && CONTINUATION_BYTE(*t))
454 if (!try_block (t0, t - t0, icode, tocode, &encoder, &wlen) &&
455 col + (t0 - u) + wlen <= ENCWORD_LEN_MAX + 1)
459 /* Adjust t1 until we can encode a character before a space. */
460 for (; t1 < u + ulen; t1++)
466 while (CONTINUATION_BYTE(*t))
468 if (!try_block (t, t1 - t, icode, tocode, &encoder, &wlen) &&
469 1 + wlen + (u + ulen - t1) <= ENCWORD_LEN_MAX + 1)
473 /* We shall encode the region [t0,t1). */
475 /* Initialise the output buffer with the us-ascii prefix. */
477 buf = safe_malloc (buflen);
479 memcpy (buf, u, t0 - u);
486 /* Find how much we can encode. */
487 n = choose_block (t, t1 - t, col, icode, tocode, &encoder, &wlen);
490 /* See if we can fit the us-ascii suffix, too. */
491 if (col + wlen + (u + ulen - t1) <= ENCWORD_LEN_MAX + 1)
495 while (CONTINUATION_BYTE(t[n]))
500 /* This should only happen in the really stupid case where the
501 only word that needs encoding is one character long, but
502 there is too much us-ascii stuff after it to use a single
503 encoded word. We add the next word to the encoded region
505 assert (t1 < u + ulen);
506 for (t1++; t1 < u + ulen && !HSPACE(*t1); t1++)
510 n = choose_block (t, n, col, icode, tocode, &encoder, &wlen);
513 /* Add to output buffer. */
514 #define LINEBREAK "\n\t"
515 if (bufpos + wlen + strlen (LINEBREAK) > buflen)
517 buflen = bufpos + wlen + strlen (LINEBREAK);
518 safe_realloc (&buf, buflen);
520 r = encode_block (buf + bufpos, t, n, icode, tocode, encoder);
523 memcpy (buf + bufpos, LINEBREAK, strlen (LINEBREAK));
524 bufpos += strlen (LINEBREAK);
532 /* Add last encoded word and us-ascii suffix to buffer. */
533 buflen = bufpos + wlen + (u + ulen - t1);
534 safe_realloc (&buf, buflen + 1);
535 r = encode_block (buf + bufpos, t, t1 - t, icode, tocode, encoder);
538 memcpy (buf + bufpos, t1, u + ulen - t1);
550 void _rfc2047_encode_string (char **pd, int encode_specials, int col)
556 if (!Charset || !*pd)
559 charsets = SendCharset;
560 if (!charsets || !*charsets)
563 rfc2047_encode (*pd, strlen (*pd), col,
564 Charset, charsets, &e, &elen,
565 encode_specials ? RFC822Specials : NULL);
571 void rfc2047_encode_adrlist (ADDRESS *addr, const char *tag)
574 int col = tag ? strlen (tag) + 2 : 32;
579 _rfc2047_encode_string (&ptr->personal, 1, col);
582 _rfc2047_encode_string (&ptr->val, 1, col);
588 static int rfc2047_decode_word (char *d, const char *s, size_t len)
590 const char *pp, *pp1;
593 int enc = 0, count = 0;
594 char *charset = NULL;
596 pd = d0 = safe_malloc (strlen (s));
598 for (pp = s; (pp1 = strchr (pp, '?')); pp = pp1 + 1)
604 /* ignore language specification a la RFC 2231 */
606 if ((t1 = memchr (pp, '*', t - pp)))
608 charset = safe_malloc (t - pp + 1);
609 memcpy (charset, pp, t - pp);
610 charset[t-pp] = '\0';
613 if (toupper ((unsigned char) *pp) == 'Q')
614 enc = ENCQUOTEDPRINTABLE;
615 else if (toupper ((unsigned char) *pp) == 'B')
625 if (enc == ENCQUOTEDPRINTABLE)
627 for (; pp < pp1; pp++)
631 else if (*pp == '=' &&
632 (!(pp[1] & ~127) && hexval(pp[1]) != -1) &&
633 (!(pp[2] & ~127) && hexval(pp[2]) != -1))
635 *pd++ = (hexval(pp[1]) << 4) | hexval(pp[2]);
643 else if (enc == ENCBASE64)
647 for (; pp < pp1; pp++)
651 if ((*pp & ~127) || (c = base64val(*pp)) == -1)
656 *pd++ = b | (c >> k);
672 mutt_convert_string (&d0, charset, Charset, M_ICONV_HOOK_FROM);
673 strfcpy (d, d0, len);
680 * Find the start and end of the first encoded word in the string.
681 * We use the grammar in section 2 of RFC 2047, but the "encoding"
682 * must be B or Q. Also, we don't require the encoded word to be
683 * separated by linear-white-space (section 5(1)).
685 static const char *find_encoded_word (const char *s, const char **x)
690 while ((p = strstr (q, "=?")))
693 0x20 < *q && *q < 0x7f && !strchr ("()<>@,;:\"/[]?.=", *q);
696 if (q[0] != '?' || !strchr ("BbQq", q[1]) || q[2] != '?')
698 for (q = q + 3; 0x20 <= *q && *q < 0x7f && *q != '?'; q++)
700 if (q[0] != '?' || q[1] != '=')
713 /* return length of linear white space */
714 static size_t lwslen (const char *s, size_t n)
722 for (; p < s + n; p++)
723 if (!strchr (" \t\r\n", *p))
725 len = (size_t)(p - s);
728 if (strchr ("\r\n", *(p-1))) /* LWS doesn't end with CRLF */
733 /* return length of linear white space : reverse */
734 static size_t lwsrlen (const char *s, size_t n)
736 const char *p = s + n - 1;
742 if (strchr ("\r\n", *p)) /* LWS doesn't end with CRLF */
746 if (!strchr (" \t\r\n", *p))
748 len = (size_t)(s + n - 1 - p);
754 /* try to decode anything that looks like a valid RFC2047 encoded
755 * header field, ignoring RFC822 parsing rules
757 void rfc2047_decode (char **pd)
761 int found_encoded = 0;
769 dlen = 4 * strlen (s); /* should be enough */
770 d = d0 = safe_malloc (dlen + 1);
772 while (*s && dlen > 0)
774 if (!(p = find_encoded_word (s, &q)))
776 /* no encoded words */
777 if (!option (OPTSTRICTMIME))
780 if (found_encoded && (m = lwslen (s, n)) != 0)
783 *d = ' ', d++, dlen--;
786 if (ascii_strcasecmp (AssumedCharset, "us-ascii"))
791 t = safe_malloc (n + 1);
792 strfcpy (t, s, n + 1);
793 if (mutt_convert_nonmime_string (&t) == 0)
795 tlen = mutt_strlen (t);
796 strncpy (d, t, tlen);
808 strncpy (d, s, dlen);
815 n = (size_t) (p - s);
816 /* ignore spaces between encoded words
817 * and linear white spaces between encoded word and *text */
818 if (!option (OPTSTRICTMIME))
820 if (found_encoded && (m = lwslen (s, n)) != 0)
823 *d = ' ', d++, dlen--;
827 if ((m = n - lwsrlen (s, n)) != 0)
835 *d = ' ', d++, dlen--;
838 else if (!found_encoded || strspn (s, " \t\r\n") != n)
848 rfc2047_decode_word (d, p, dlen);
859 mutt_str_adjust (pd);
862 void rfc2047_decode_adrlist (ADDRESS *a)
867 rfc2047_decode (&a->personal);
869 if (a->val && strstr (a->val, "=?") != NULL)
870 rfc2047_decode (&a->val);