2 * This file is part of DisOrder
3 * Copyright (C) 2005, 2007, 2008 Richard Kettlewell
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 3 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, see <http://www.gnu.org/licenses/>.
19 * @brief Support for MIME and allied protocols
34 /** @brief Match whitespace characters */
35 static int whitespace(int c) {
47 /** @brief Match RFC2045 tspecial characters */
48 int mime_tspecial(int c) {
71 /** @brief Match RFC2616 separator characters */
72 int mime_http_separator(int c) {
99 /** @brief Match CRLF */
100 static int iscrlf(const char *ptr) {
101 return ptr[0] == '\r' && ptr[1] == '\n';
104 /** @brief Skip whitespace
105 * @param s Pointer into string
106 * @param rfc822_comments If true, skip RFC822 nested comments
107 * @return Pointer into string after whitespace
109 static const char *skipwhite(const char *s, int rfc822_comments) {
128 case '(': ++depth; break;
129 case ')': --depth; break;
146 /** @brief Test for a word character
147 * @param c Character to test
148 * @param special mime_tspecial() (MIME/RFC2405) or mime_http_separator() (HTTP/RFC2616)
149 * @return 1 if @p c is a word character, else 0
151 static int iswordchar(int c, int (*special)(int)) {
152 return !(c <= ' ' || c > '~' || special(c));
155 /** @brief Parse an RFC1521/RFC2616 word
156 * @param s Pointer to start of word
157 * @param valuep Where to store value
158 * @param special mime_tspecial() (MIME/RFC2405) or mime_http_separator() (HTTP/RFC2616)
159 * @return Pointer just after end of word or NULL if there's no word
161 * A word is a token or a quoted-string.
163 const char *mime_parse_word(const char *s, char **valuep,
164 int (*special)(int)) {
165 struct dynstr value[1];
171 while((c = *s++) != '"') {
177 dynstr_append(value, c);
184 if(!iswordchar((unsigned char)*s, special))
187 while(iswordchar((unsigned char)*s, special))
188 dynstr_append(value, *s++);
190 dynstr_terminate(value);
191 *valuep = value->vec;
195 /** @brief Parse an RFC1521/RFC2616 token
196 * @param s Pointer to start of token
197 * @param valuep Where to store value
198 * @param special mime_tspecial() (MIME/RFC2405) or mime_http_separator() (HTTP/RFC2616)
199 * @return Pointer just after end of token or NULL if there's no token
201 static const char *parsetoken(const char *s, char **valuep,
202 int (*special)(int)) {
205 return mime_parse_word(s, valuep, special);
208 /** @brief Parse a MIME content-type field
209 * @param s Start of field
210 * @param typep Where to store type
211 * @param parametersp Where to store parameter list
212 * @return 0 on success, non-0 on error
214 * See <a href="http://tools.ietf.org/html/rfc2045#section-5">RFC 2045 s5</a>.
216 int mime_content_type(const char *s,
218 struct kvp **parametersp) {
219 struct dynstr type, parametername;
220 struct kvp *parameters = 0;
221 char *parametervalue;
224 if(!(s = skipwhite(s, 1)))
228 while(*s && !mime_tspecial(*s) && !whitespace(*s))
229 dynstr_append(&type, tolower((unsigned char)*s++));
230 if(!(s = skipwhite(s, 1)))
234 dynstr_append(&type, '/');
235 if(!(s = skipwhite(s, 1)))
237 while(*s && !mime_tspecial(*s) && !whitespace(*s))
238 dynstr_append(&type, tolower((unsigned char)*s++));
239 if(!(s = skipwhite(s, 1)))
243 dynstr_init(¶metername);
245 if(!(s = skipwhite(s, 1)))
249 while(*s && !mime_tspecial(*s) && !whitespace(*s))
250 dynstr_append(¶metername, tolower((unsigned char)*s++));
251 if(!(s = skipwhite(s, 1)))
255 if(!(s = skipwhite(s, 1)))
257 if(!(s = mime_parse_word(s, ¶metervalue, mime_tspecial)))
259 if(!(s = skipwhite(s, 1)))
261 dynstr_terminate(¶metername);
262 kvp_set(¶meters, parametername.vec, parametervalue);
264 dynstr_terminate(&type);
266 *parametersp = parameters;
270 /** @brief Parse a MIME message
271 * @param s Start of message
272 * @param callback Called for each header field
273 * @param u Passed to callback
274 * @return Pointer to decoded body (might be in original string), or NULL on error
276 * This does an RFC 822-style parse and honors Content-Transfer-Encoding as
277 * described in <a href="http://tools.ietf.org/html/rfc2045#section-6">RFC 2045
278 * s6</a>. @p callback is called for each header field encountered, in order,
279 * with ASCII characters in the header name forced to lower case.
281 const char *mime_parse(const char *s,
282 int (*callback)(const char *name, const char *value,
285 struct dynstr name, value;
288 while(*s && !iscrlf(s)) {
291 while(*s && !mime_tspecial(*s) && !whitespace(*s))
292 dynstr_append(&name, tolower((unsigned char)*s++));
293 if(!(s = skipwhite(s, 1)))
298 while(*s && !(*s == '\n' && !(s[1] == ' ' || s[1] == '\t'))) {
300 /* Strip leading whitespace */
301 if(value.nvec || !(c == ' ' || c == '\t' || c == '\n' || c == '\r'))
302 dynstr_append(&value, c);
304 /* Strip trailing whitespace */
305 while(value.nvec > 0 && (value.vec[value.nvec - 1] == ' '
306 || value.vec[value.nvec - 1] == '\t'
307 || value.vec[value.nvec - 1] == '\n'
308 || value.vec[value.nvec - 1] == '\r'))
312 dynstr_terminate(&name);
313 dynstr_terminate(&value);
314 if(!strcmp(name.vec, "content-transfer-encoding")) {
315 cte = xstrdup(value.vec);
316 for(p = cte; *p; p++)
317 *p = tolower((unsigned char)*p);
319 if(callback(name.vec, value.vec, u))
325 if(!strcmp(cte, "base64"))
326 return mime_base64(s, 0);
327 if(!strcmp(cte, "quoted-printable"))
329 if(!strcmp(cte, "7bit") || !strcmp(cte, "8bit"))
331 disorder_error(0, "unknown content-transfer-encoding '%s'", cte);
337 /** @brief Match the boundary string */
338 static int isboundary(const char *ptr, const char *boundary, size_t bl) {
339 return (ptr[0] == '-'
341 && !strncmp(ptr + 2, boundary, bl)
342 && (iscrlf(ptr + bl + 2)
343 || (ptr[bl + 2] == '-'
344 && ptr[bl + 3] == '-'
345 && (iscrlf(ptr + bl + 4) || *(ptr + bl + 4) == 0))));
348 /** @brief Match the final boundary string */
349 static int isfinal(const char *ptr, const char *boundary, size_t bl) {
350 return (ptr[0] == '-'
352 && !strncmp(ptr + 2, boundary, bl)
353 && ptr[bl + 2] == '-'
354 && ptr[bl + 3] == '-'
355 && (iscrlf(ptr + bl + 4) || *(ptr + bl + 4) == 0));
358 /** @brief Parse a multipart MIME body
359 * @param s Start of message
360 * @param callback Callback for each part
361 * @param boundary Boundary string
362 * @param u Passed to callback
363 * @return 0 on success, non-0 on error
365 * See <a href="http://tools.ietf.org/html/rfc2046#section-5.1">RFC 2046
366 * s5.1</a>. @p callback is called for each part (not yet decoded in any way)
367 * in succession; you should probably call mime_parse() for each part.
369 int mime_multipart(const char *s,
370 int (*callback)(const char *s, void *u),
371 const char *boundary,
373 size_t bl = strlen(boundary);
374 const char *start, *e;
377 /* We must start with a boundary string */
378 if(!isboundary(s, boundary, bl)) {
379 disorder_error(0, "mime_multipart: first line is not the boundary string");
382 /* Keep going until we hit a final boundary */
383 while(!isfinal(s, boundary, bl)) {
384 s = strstr(s, "\r\n") + 2;
386 while(!isboundary(s, boundary, bl)) {
387 if(!(e = strstr(s, "\r\n"))) {
388 disorder_error(0, "mime_multipart: line does not end CRLF");
393 if((ret = callback(xstrndup(start,
394 s == start ? 0 : s - start - 2),
401 /** @brief Parse an RFC2388-style content-disposition field
402 * @param s Start of field
403 * @param dispositionp Where to store disposition
404 * @param parameternamep Where to store parameter name
405 * @param parametervaluep Where to store parameter value
406 * @return 0 on success, non-0 on error
408 * See <a href="http://tools.ietf.org/html/rfc2388#section-3">RFC 2388 s3</a>
409 * and <a href="http://tools.ietf.org/html/rfc2183">RFC 2183</a>.
411 int mime_rfc2388_content_disposition(const char *s,
413 char **parameternamep,
414 char **parametervaluep) {
415 struct dynstr disposition, parametername;
417 dynstr_init(&disposition);
418 if(!(s = skipwhite(s, 1)))
422 while(*s && !mime_tspecial(*s) && !whitespace(*s))
423 dynstr_append(&disposition, tolower((unsigned char)*s++));
424 if(!(s = skipwhite(s, 1)))
428 dynstr_init(¶metername);
430 if(!(s = skipwhite(s, 1)))
434 while(*s && !mime_tspecial(*s) && !whitespace(*s))
435 dynstr_append(¶metername, tolower((unsigned char)*s++));
436 if(!(s = skipwhite(s, 1)))
440 if(!(s = skipwhite(s, 1)))
442 if(!(s = mime_parse_word(s, parametervaluep, mime_tspecial)))
444 if(!(s = skipwhite(s, 1)))
446 dynstr_terminate(¶metername);
447 *parameternamep = parametername.vec;
449 *parametervaluep = *parameternamep = 0;
450 dynstr_terminate(&disposition);
451 *dispositionp = disposition.vec;
455 /** @brief Convert MIME quoted-printable
456 * @param s Quoted-printable data
457 * @return Decoded data
459 * See <a href="http://tools.ietf.org/html/rfc2045#section-6.7">RFC 2045
462 char *mime_qp(const char *s) {
471 if((a = unhexdigitq(s[0])) != -1
472 && (b = unhexdigitq(s[1])) != -1) {
473 dynstr_append(&d, a * 16 + b);
477 while(*t == ' ' || *t == '\t') ++t;
479 /* soft line break */
488 while(*t == ' ' || *t == '\t') ++t;
490 /* trailing space is always eliminated */
493 dynstr_append(&d, c);
496 dynstr_append(&d, c);
500 dynstr_terminate(&d);
504 /** @brief Match cookie separator characters
506 * This is a subset of the RFC2616 specials, and technically is in breach of
507 * the specification. However rejecting (in particular) slashes is
508 * unreasonably strict and has broken at least one (admittedly somewhat
509 * obscure) browser, so we're more forgiving.
511 static int cookie_separator(int c) {
528 /** @brief Match cookie value separator characters
530 * Same as cookie_separator() but allows for @c = in cookie values.
532 static int cookie_value_separator(int c) {
548 /** @brief Parse a RFC2109 Cookie: header
549 * @param s Header field value
550 * @param cd Where to store result
551 * @return 0 on success, non-0 on error
553 * See <a href="http://tools.ietf.org/html/rfc2109">RFC 2109</a>.
555 int parse_cookie(const char *s,
556 struct cookiedata *cd) {
559 memset(cd, 0, sizeof *cd);
562 /* Skip separators */
563 if(*s == ';' || *s == ',') {
568 if(!(s = parsetoken(s, &n, cookie_separator))) {
569 disorder_error(0, "parse_cookie: cannot parse attribute name");
574 disorder_error(0, "parse_cookie: did not find expected '='");
578 if(!(s = mime_parse_word(s, &v, cookie_value_separator))) {
579 disorder_error(0, "parse_cookie: cannot parse value for '%s'", n);
583 /* Some bit of meta-information */
584 if(!strcmp(n, "$Version"))
586 else if(!strcmp(n, "$Path")) {
587 if(cd->ncookies > 0 && cd->cookies[cd->ncookies-1].path == 0)
588 cd->cookies[cd->ncookies-1].path = v;
590 disorder_error(0, "redundant $Path in Cookie: header");
593 } else if(!strcmp(n, "$Domain")) {
594 if(cd->ncookies > 0 && cd->cookies[cd->ncookies-1].domain == 0)
595 cd->cookies[cd->ncookies-1].domain = v;
597 disorder_error(0, "redundant $Domain in Cookie: header");
602 /* It's a new cookie */
603 cd->cookies = xrealloc(cd->cookies,
604 (cd->ncookies + 1) * sizeof (struct cookie));
605 cd->cookies[cd->ncookies].name = n;
606 cd->cookies[cd->ncookies].value = v;
607 cd->cookies[cd->ncookies].path = 0;
608 cd->cookies[cd->ncookies].domain = 0;
612 if(*s && (*s != ',' && *s != ';')) {
613 disorder_error(0, "missing separator in Cookie: header");
620 /** @brief Find a named cookie
621 * @param cd Parse cookie data
622 * @param name Name of cookie
623 * @return Cookie structure or NULL if not found
625 const struct cookie *find_cookie(const struct cookiedata *cd,
629 for(n = 0; n < cd->ncookies; ++n)
630 if(!strcmp(cd->cookies[n].name, name))
631 return &cd->cookies[n];
635 /** @brief RFC822 quoting
636 * @param s String to quote
637 * @param force If non-0, always quote
638 * @return Possibly quoted string
640 char *quote822(const char *s, int force) {
646 /* See if we need to quote */
647 for(t = s; (c = (unsigned char)*t); ++t) {
648 if(mime_tspecial(c) || mime_http_separator(c) || whitespace(c))
659 dynstr_append(d, '"');
660 for(t = s; (c = (unsigned char)*t); ++t) {
661 if(c == '"' || c == '\\')
662 dynstr_append(d, '\\');
665 dynstr_append(d, '"');
670 /** @brief Return true if @p ptr points at trailing space */
671 static int is_trailing_space(const char *ptr) {
672 if(*ptr == ' ' || *ptr == '\t') {
673 while(*ptr == ' ' || *ptr == '\t')
675 return *ptr == '\n' || *ptr == 0;
680 /** @brief Encoding text as quoted-printable
681 * @param text String to encode
682 * @return Encoded string
684 * See <a href="http://tools.ietf.org/html/rfc2045#section-6.7">RFC2045
687 char *mime_to_qp(const char *text) {
689 int linelength = 0; /* length of current line */
694 * 1. Anything except newline can be replaced with =%02X
695 * 2. Newline, 33-60 and 62-126 stand for themselves (i.e. not '=')
696 * 3. Non-trailing space/tab stand for themselves.
697 * 4. Output lines are limited to 76 chars, with =<newline> being used
698 * as a soft line break
699 * 5. Newlines aren't counted towards the 76 char limit.
702 const int c = (unsigned char)*text;
704 /* Newline stands as itself */
705 dynstr_append(d, '\n');
707 } else if((c >= 33 && c <= 126 && c != '=')
708 || ((c == ' ' || c == '\t')
709 && !is_trailing_space(text))) {
710 /* Things that can stand for themselves */
714 /* Anything else that needs encoding */
715 snprintf(buffer, sizeof buffer, "=%02X", c);
716 dynstr_append_string(d, buffer);
720 if(linelength > 73 && *text && *text != '\n') {
721 /* Next character might overflow 76 character limit if encoded, so we
722 * insert a soft break */
723 dynstr_append_string(d, "=\n");
727 /* Ensure there is a final newline */
729 dynstr_append(d, '\n');
735 /** @brief Encode text
736 * @param text Underlying UTF-8 text
737 * @param charsetp Where to store charset string
738 * @param encodingp Where to store encoding string
739 * @return Encoded text (might be @p text)
741 const char *mime_encode_text(const char *text,
742 const char **charsetp,
743 const char **encodingp) {
746 /* See if there are in fact any non-ASCII characters */
747 for(ptr = text; *ptr; ++ptr)
748 if((unsigned char)*ptr >= 128)
751 /* Plain old ASCII, no encoding required */
752 *charsetp = "us-ascii";
757 *encodingp = "quoted-printable";
758 return mime_to_qp(text);