4 * This file contains functions which handle the mapping of Internet addresses
5 * to users on the Citadel system.
17 #include <sys/types.h>
19 #if TIME_WITH_SYS_TIME
20 # include <sys/time.h>
24 # include <sys/time.h>
33 #include <libcitadel.h>
36 #include "sysdep_decls.h"
37 #include "citserver.h"
41 #include "internet_addressing.h"
44 #include "parsedate.h"
57 * Wrapper around iconv_open()
58 * Our version adds aliases for non-standard Microsoft charsets
59 * such as 'MS950', aliasing them to names like 'CP950'
61 * tocode Target encoding
62 * fromcode Source encoding
64 iconv_t ctdl_iconv_open(const char *tocode, const char *fromcode)
66 iconv_t ic = (iconv_t)(-1) ;
67 ic = iconv_open(tocode, fromcode);
68 if (ic == (iconv_t)(-1) ) {
69 char alias_fromcode[64];
70 if ( (strlen(fromcode) == 5) && (!strncasecmp(fromcode, "MS", 2)) ) {
71 safestrncpy(alias_fromcode, fromcode, sizeof alias_fromcode);
72 alias_fromcode[0] = 'C';
73 alias_fromcode[1] = 'P';
74 ic = iconv_open(tocode, alias_fromcode);
82 inline char *FindNextEnd (char *bptr)
85 /* Find the next ?Q? */
86 end = strchr(bptr + 2, '?');
87 if (end == NULL) return NULL;
88 if (((*(end + 1) == 'B') || (*(end + 1) == 'Q')) &&
89 (*(end + 2) == '?')) {
90 /* skip on to the end of the cluster, the next ?= */
91 end = strstr(end + 3, "?=");
94 /* sort of half valid encoding, try to find an end. */
95 end = strstr(bptr, "?=");
100 * Handle subjects with RFC2047 encoding such as:
101 * =?koi8-r?B?78bP0s3Mxc7JxSDXz9rE1dvO2c3JINvB0sHNySDP?=
103 void utf8ify_rfc822_string(char *buf) {
104 char *start, *end, *next, *nextend, *ptr;
109 iconv_t ic = (iconv_t)(-1) ;
110 char *ibuf; /**< Buffer of characters to be converted */
111 char *obuf; /**< Buffer for converted characters */
112 size_t ibuflen; /**< Length of input buffer */
113 size_t obuflen; /**< Length of output buffer */
114 char *isav; /**< Saved pointer to input buffer */
115 char *osav; /**< Saved pointer to output buffer */
118 int illegal_non_rfc2047_encoding = 0;
120 /* Sometimes, badly formed messages contain strings which were simply
121 * written out directly in some foreign character set instead of
122 * using RFC2047 encoding. This is illegal but we will attempt to
123 * handle it anyway by converting from a user-specified default
124 * charset to UTF-8 if we see any nonprintable characters.
127 for (i=0; i<len; ++i) {
128 if ((buf[i] < 32) || (buf[i] > 126)) {
129 illegal_non_rfc2047_encoding = 1;
130 i = len; ///< take a shortcut, it won't be more than one.
133 if (illegal_non_rfc2047_encoding) {
134 const char *default_header_charset = "iso-8859-1";
135 if ( (strcasecmp(default_header_charset, "UTF-8")) && (strcasecmp(default_header_charset, "us-ascii")) ) {
136 ic = ctdl_iconv_open("UTF-8", default_header_charset);
137 if (ic != (iconv_t)(-1) ) {
140 safestrncpy(ibuf, buf, 1024);
141 ibuflen = strlen(ibuf);
143 obuf = (char *) malloc(obuflen);
145 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
146 osav[1024-obuflen] = 0;
155 /* pre evaluate the first pair */
156 nextend = end = NULL;
158 start = strstr(buf, "=?");
160 end = FindNextEnd (start);
162 while ((start != NULL) && (end != NULL))
164 next = strstr(end, "=?");
166 nextend = FindNextEnd(next);
170 /* did we find two partitions */
171 if ((next != NULL) &&
175 while ((ptr < next) &&
181 /* did we find a gab just filled with blanks? */
186 len - (next - start));
188 /* now terminate the gab at the end */
189 delta = (next - end) - 2;
193 /* move next to its new location. */
198 /* our next-pair is our new first pair now. */
203 /* Now we handle foreign character sets properly encoded
206 while (start=strstr(buf, "=?"), end=FindNextEnd((start != NULL)? start : buf),
207 ((start != NULL) && (end != NULL) && (end > start)) )
209 extract_token(charset, start, 1, '?', sizeof charset);
210 extract_token(encoding, start, 2, '?', sizeof encoding);
211 extract_token(istr, start, 3, '?', sizeof istr);
215 if (!strcasecmp(encoding, "B")) { /**< base64 */
216 ibuflen = CtdlDecodeBase64(ibuf, istr, strlen(istr));
218 else if (!strcasecmp(encoding, "Q")) { /**< quoted-printable */
226 if (istr[pos] == '_') istr[pos] = ' ';
230 ibuflen = CtdlDecodeQuotedPrintable(ibuf, istr, len);
233 strcpy(ibuf, istr); /**< unknown encoding */
234 ibuflen = strlen(istr);
237 ic = ctdl_iconv_open("UTF-8", charset);
238 if (ic != (iconv_t)(-1) ) {
240 obuf = (char *) malloc(obuflen);
242 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
243 osav[1024-obuflen] = 0;
248 remove_token(end, 0, '?');
249 remove_token(end, 0, '?');
250 remove_token(end, 0, '?');
251 remove_token(end, 0, '?');
252 strcpy(end, &end[1]);
254 snprintf(newbuf, sizeof newbuf, "%s%s%s", buf, osav, end);
263 remove_token(end, 0, '?');
264 remove_token(end, 0, '?');
265 remove_token(end, 0, '?');
266 remove_token(end, 0, '?');
267 strcpy(end, &end[1]);
269 snprintf(newbuf, sizeof newbuf, "%s(unreadable)%s", buf, end);
276 * Since spammers will go to all sorts of absurd lengths to get their
277 * messages through, there are LOTS of corrupt headers out there.
278 * So, prevent a really badly formed RFC2047 header from throwing
279 * this function into an infinite loop.
282 if (passes > 20) return;
287 inline void utf8ify_rfc822_string(char *a){};
298 char *inetcfg = NULL;
299 struct spamstrings_t *spamstrings = NULL;
303 * Return nonzero if the supplied name is an alias for this host.
305 int CtdlHostAlias(char *fqdn) {
309 char host[256], type[256];
311 if (fqdn == NULL) return(hostalias_nomatch);
312 if (IsEmptyStr(fqdn)) return(hostalias_nomatch);
313 if (!strcasecmp(fqdn, "localhost")) return(hostalias_localhost);
314 if (!strcasecmp(fqdn, config.c_fqdn)) return(hostalias_localhost);
315 if (!strcasecmp(fqdn, config.c_nodename)) return(hostalias_localhost);
316 if (inetcfg == NULL) return(hostalias_nomatch);
318 config_lines = num_tokens(inetcfg, '\n');
319 for (i=0; i<config_lines; ++i) {
320 extract_token(buf, inetcfg, i, '\n', sizeof buf);
321 extract_token(host, buf, 0, '|', sizeof host);
322 extract_token(type, buf, 1, '|', sizeof type);
324 if ( (!strcasecmp(type, "localhost"))
325 && (!strcasecmp(fqdn, host)))
326 return(hostalias_localhost);
328 if ( (!strcasecmp(type, "directory"))
329 && (!strcasecmp(fqdn, host)))
330 return(hostalias_directory);
332 if ( (!strcasecmp(type, "masqdomain"))
333 && (!strcasecmp(fqdn, host)))
334 return(hostalias_masq);
338 return(hostalias_nomatch);
348 * Return 0 if a given string fuzzy-matches a Citadel user account
350 * FIXME ... this needs to be updated to handle aliases.
352 int fuzzy_match(struct ctdluser *us, char *matchstring) {
356 if ( (!strncasecmp(matchstring, "cit", 3))
357 && (atol(&matchstring[3]) == us->usernum)) {
361 len = strlen(matchstring);
362 for (a=0; !IsEmptyStr(&us->fullname[a]); ++a) {
363 if (!strncasecmp(&us->fullname[a],
373 * Unfold a multi-line field into a single line, removing multi-whitespaces
375 void unfold_rfc822_field(char *field) {
379 striplt(field); /* remove leading/trailing whitespace */
381 /* convert non-space whitespace to spaces, and remove double blanks */
382 for (i=0; i<strlen(field); ++i) {
383 if (field[i]=='\"') quote = 1 - quote;
385 if (isspace(field[i])) field[i] = ' ';
386 while (isspace(field[i]) && isspace(field[i+1])) {
387 strcpy(&field[i+1], &field[i+2]);
396 * Split an RFC822-style address into userid, host, and full name
399 void process_rfc822_addr(const char *rfc822, char *user, char *node, char *name)
404 strcpy(node, config.c_fqdn);
407 if (rfc822 == NULL) return;
409 /* extract full name - first, it's From minus <userid> */
410 strcpy(name, rfc822);
411 stripout(name, '<', '>');
413 /* strip anything to the left of a bang */
414 while ((!IsEmptyStr(name)) && (haschar(name, '!') > 0))
415 strcpy(name, &name[1]);
417 /* and anything to the right of a @ or % */
418 for (a = 0; a < strlen(name); ++a) {
425 /* but if there are parentheses, that changes the rules... */
426 if ((haschar(rfc822, '(') == 1) && (haschar(rfc822, ')') == 1)) {
427 strcpy(name, rfc822);
428 stripallbut(name, '(', ')');
431 /* but if there are a set of quotes, that supersedes everything */
432 if (haschar(rfc822, 34) == 2) {
433 strcpy(name, rfc822);
434 while ((!IsEmptyStr(name)) && (name[0] != 34)) {
435 strcpy(&name[0], &name[1]);
437 strcpy(&name[0], &name[1]);
438 for (a = 0; a < strlen(name); ++a)
442 /* extract user id */
443 strcpy(user, rfc822);
445 /* first get rid of anything in parens */
446 stripout(user, '(', ')');
448 /* if there's a set of angle brackets, strip it down to that */
449 if ((haschar(user, '<') == 1) && (haschar(user, '>') == 1)) {
450 stripallbut(user, '<', '>');
453 /* strip anything to the left of a bang */
454 while ((!IsEmptyStr(user)) && (haschar(user, '!') > 0))
455 strcpy(user, &user[1]);
457 /* and anything to the right of a @ or % */
458 for (a = 0; a < strlen(user); ++a) {
466 /* extract node name */
467 strcpy(node, rfc822);
469 /* first get rid of anything in parens */
470 stripout(node, '(', ')');
472 /* if there's a set of angle brackets, strip it down to that */
473 if ((haschar(node, '<') == 1) && (haschar(node, '>') == 1)) {
474 stripallbut(node, '<', '>');
477 /* If no node specified, tack ours on instead */
479 (haschar(node, '@')==0)
480 && (haschar(node, '%')==0)
481 && (haschar(node, '!')==0)
483 strcpy(node, config.c_nodename);
488 /* strip anything to the left of a @ */
489 while ((!IsEmptyStr(node)) && (haschar(node, '@') > 0))
490 strcpy(node, &node[1]);
492 /* strip anything to the left of a % */
493 while ((!IsEmptyStr(node)) && (haschar(node, '%') > 0))
494 strcpy(node, &node[1]);
496 /* reduce multiple system bang paths to node!user */
497 while ((!IsEmptyStr(node)) && (haschar(node, '!') > 1))
498 strcpy(node, &node[1]);
500 /* now get rid of the user portion of a node!user string */
501 for (a = 0; a < strlen(node); ++a)
506 /* strip leading and trailing spaces in all strings */
511 /* If we processed a string that had the address in angle brackets
512 * but no name outside the brackets, we now have an empty name. In
513 * this case, use the user portion of the address as the name.
515 if ((IsEmptyStr(name)) && (!IsEmptyStr(user))) {
523 * convert_field() is a helper function for convert_internet_message().
524 * Given start/end positions for an rfc822 field, it converts it to a Citadel
525 * field if it wants to, and unfolds it if necessary.
527 * Returns 1 if the field was converted and inserted into the Citadel message
528 * structure, implying that the source field should be removed from the
531 int convert_field(struct CtdlMessage *msg, int beg, int end) {
544 rfc822 = msg->cm_fields['M']; /* M field contains rfc822 text */
545 for (i = end; i >= beg; --i) {
546 if (rfc822[i] == ':') colonpos = i;
549 if (colonpos < 0) return(0); /* no colon? not a valid header line */
551 key = malloc((end - beg) + 2);
552 safestrncpy(key, &rfc822[beg], (end-beg)+1);
553 key[colonpos - beg] = 0;
554 value = &key[(colonpos - beg) + 1];
555 unfold_rfc822_field(value);
558 * Here's the big rfc822-to-citadel loop.
561 /* Date/time is converted into a unix timestamp. If the conversion
562 * fails, we replace it with the time the message arrived locally.
564 if (!strcasecmp(key, "Date")) {
565 parsed_date = parsedate(value);
566 if (parsed_date < 0L) parsed_date = time(NULL);
567 snprintf(buf, sizeof buf, "%ld", (long)parsed_date );
568 if (msg->cm_fields['T'] == NULL)
569 msg->cm_fields['T'] = strdup(buf);
573 else if (!strcasecmp(key, "From")) {
574 process_rfc822_addr(value, user, node, name);
575 CtdlLogPrintf(CTDL_DEBUG, "Converted to <%s@%s> (%s)\n", user, node, name);
576 snprintf(addr, sizeof addr, "%s@%s", user, node);
577 if (msg->cm_fields['A'] == NULL)
578 msg->cm_fields['A'] = strdup(name);
580 if (msg->cm_fields['F'] == NULL)
581 msg->cm_fields['F'] = strdup(addr);
585 else if (!strcasecmp(key, "Subject")) {
586 if (msg->cm_fields['U'] == NULL)
587 msg->cm_fields['U'] = strdup(value);
591 else if (!strcasecmp(key, "To")) {
592 if (msg->cm_fields['R'] == NULL)
593 msg->cm_fields['R'] = strdup(value);
597 else if (!strcasecmp(key, "CC")) {
598 if (msg->cm_fields['Y'] == NULL)
599 msg->cm_fields['Y'] = strdup(value);
603 else if (!strcasecmp(key, "Message-ID")) {
604 if (msg->cm_fields['I'] != NULL) {
605 CtdlLogPrintf(CTDL_WARNING, "duplicate message id\n");
608 if (msg->cm_fields['I'] == NULL) {
609 msg->cm_fields['I'] = strdup(value);
611 /* Strip angle brackets */
612 while (haschar(msg->cm_fields['I'], '<') > 0) {
613 strcpy(&msg->cm_fields['I'][0],
614 &msg->cm_fields['I'][1]);
616 for (i = 0; i<strlen(msg->cm_fields['I']); ++i)
617 if (msg->cm_fields['I'][i] == '>')
618 msg->cm_fields['I'][i] = 0;
624 else if (!strcasecmp(key, "Return-Path")) {
625 if (msg->cm_fields['P'] == NULL)
626 msg->cm_fields['P'] = strdup(value);
630 else if (!strcasecmp(key, "Envelope-To")) {
631 if (msg->cm_fields['V'] == NULL)
632 msg->cm_fields['V'] = strdup(value);
636 else if (!strcasecmp(key, "References")) {
637 if (msg->cm_fields['W'] != NULL) {
638 free(msg->cm_fields['W']);
640 msg->cm_fields['W'] = strdup(value);
644 else if (!strcasecmp(key, "In-reply-to")) {
645 if (msg->cm_fields['W'] == NULL) { /* References: supersedes In-reply-to: */
646 msg->cm_fields['W'] = strdup(value);
653 /* Clean up and move on. */
654 free(key); /* Don't free 'value', it's actually the same buffer */
660 * Convert RFC822 references format (References) to Citadel references format (Weferences)
662 void convert_references_to_wefewences(char *str) {
663 int bracket_nesting = 0;
665 char *moveptr = NULL;
672 if (bracket_nesting < 0) bracket_nesting = 0;
674 if ((ch == '>') && (bracket_nesting == 0) && (*(ptr+1)) && (ptr>str) ) {
678 else if (bracket_nesting > 0) {
684 *moveptr = *(moveptr+1);
688 if (ch == '<') ++bracket_nesting;
695 * Convert an RFC822 message (headers + body) to a CtdlMessage structure.
696 * NOTE: the supplied buffer becomes part of the CtdlMessage structure, and
697 * will be deallocated when CtdlFreeMessage() is called. Therefore, the
698 * supplied buffer should be DEREFERENCED. It should not be freed or used
701 struct CtdlMessage *convert_internet_message(char *rfc822) {
703 struct CtdlMessage *msg;
704 int pos, beg, end, msglen;
709 msg = malloc(sizeof(struct CtdlMessage));
710 if (msg == NULL) return msg;
712 memset(msg, 0, sizeof(struct CtdlMessage));
713 msg->cm_magic = CTDLMESSAGE_MAGIC; /* self check */
714 msg->cm_anon_type = 0; /* never anonymous */
715 msg->cm_format_type = FMT_RFC822; /* internet message */
716 msg->cm_fields['M'] = rfc822;
723 /* Locate beginning and end of field, keeping in mind that
724 * some fields might be multiline
729 msglen = strlen(rfc822);
730 while ( (end < 0) && (done == 0) ) {
732 if ((rfc822[pos]=='\n')
733 && (!isspace(rfc822[pos+1]))) {
737 /* done with headers? */
738 if ( (rfc822[pos]=='\n')
739 && ( (rfc822[pos+1]=='\n')
740 ||(rfc822[pos+1]=='\r')) ) {
745 if (pos >= (msglen-1) ) {
754 /* At this point we have a field. Are we interested in it? */
755 converted = convert_field(msg, beg, end);
757 /* Strip the field out of the RFC822 header if we used it */
759 strcpy(&rfc822[beg], &rfc822[pos]);
763 /* If we've hit the end of the message, bail out */
764 if (pos > strlen(rfc822)) done = 1;
767 /* Follow-up sanity checks... */
769 /* If there's no timestamp on this message, set it to now. */
770 if (msg->cm_fields['T'] == NULL) {
771 snprintf(buf, sizeof buf, "%ld", (long)time(NULL));
772 msg->cm_fields['T'] = strdup(buf);
775 /* If a W (references, or rather, Wefewences) field is present, we
776 * have to convert it from RFC822 format to Citadel format.
778 if (msg->cm_fields['W'] != NULL) {
779 convert_references_to_wefewences(msg->cm_fields['W']);
788 * Look for a particular header field in an RFC822 message text. If the
789 * requested field is found, it is unfolded (if necessary) and returned to
790 * the caller. The field name is stripped out, leaving only its contents.
791 * The caller is responsible for freeing the returned buffer. If the requested
792 * field is not present, or anything else goes wrong, it returns NULL.
794 char *rfc822_fetch_field(char *rfc822, char *fieldname) {
795 char *fieldbuf = NULL;
796 char *end_of_headers;
802 /* Should never happen, but sometimes we get stupid */
803 if (rfc822 == NULL) return(NULL);
804 if (fieldname == NULL) return(NULL);
806 snprintf(fieldhdr, sizeof fieldhdr, "%s:", fieldname);
808 /* Locate the end of the headers, so we don't run past that point */
809 end_of_headers = bmstrcasestr(rfc822, "\n\r\n");
810 if (end_of_headers == NULL) {
811 end_of_headers = bmstrcasestr(rfc822, "\n\n");
813 if (end_of_headers == NULL) return (NULL);
815 field_start = bmstrcasestr(rfc822, fieldhdr);
816 if (field_start == NULL) return(NULL);
817 if (field_start > end_of_headers) return(NULL);
819 fieldbuf = malloc(SIZ);
820 strcpy(fieldbuf, "");
823 ptr = memreadline(ptr, fieldbuf, SIZ-strlen(fieldbuf) );
824 while ( (isspace(ptr[0])) && (ptr < end_of_headers) ) {
825 strcat(fieldbuf, " ");
826 cont = &fieldbuf[strlen(fieldbuf)];
827 ptr = memreadline(ptr, cont, SIZ-strlen(fieldbuf) );
831 strcpy(fieldbuf, &fieldbuf[strlen(fieldhdr)]);
839 /*****************************************************************************
840 * DIRECTORY MANAGEMENT FUNCTIONS *
841 *****************************************************************************/
844 * Generate the index key for an Internet e-mail address to be looked up
847 void directory_key(char *key, char *addr) {
851 for (i=0; !IsEmptyStr(&addr[i]); ++i) {
852 if (!isspace(addr[i])) {
853 key[keylen++] = tolower(addr[i]);
858 CtdlLogPrintf(CTDL_DEBUG, "Directory key is <%s>\n", key);
863 /* Return nonzero if the supplied address is in a domain we keep in
866 int IsDirectory(char *addr, int allow_masq_domains) {
870 extract_token(domain, addr, 1, '@', sizeof domain);
873 h = CtdlHostAlias(domain);
875 if ( (h == hostalias_masq) && allow_masq_domains)
878 if ( (h == hostalias_localhost) || (h == hostalias_directory) ) {
888 * Initialize the directory database (erasing anything already there)
890 void CtdlDirectoryInit(void) {
891 cdb_trunc(CDB_DIRECTORY);
896 * Add an Internet e-mail address to the directory for a user
898 void CtdlDirectoryAddUser(char *internet_addr, char *citadel_addr) {
901 CtdlLogPrintf(CTDL_DEBUG, "Dir: %s --> %s\n",
902 internet_addr, citadel_addr);
903 if (IsDirectory(internet_addr, 0) == 0) return;
905 directory_key(key, internet_addr);
907 cdb_store(CDB_DIRECTORY, key, strlen(key),
908 citadel_addr, strlen(citadel_addr)+1 );
913 * Delete an Internet e-mail address from the directory.
915 * (NOTE: we don't actually use or need the citadel_addr variable; it's merely
916 * here because the callback API expects to be able to send it.)
918 void CtdlDirectoryDelUser(char *internet_addr, char *citadel_addr) {
921 directory_key(key, internet_addr);
922 cdb_delete(CDB_DIRECTORY, key, strlen(key) );
927 * Look up an Internet e-mail address in the directory.
928 * On success: returns 0, and Citadel address stored in 'target'
929 * On failure: returns nonzero
931 int CtdlDirectoryLookup(char *target, char *internet_addr, size_t targbuflen) {
932 struct cdbdata *cdbrec;
935 /* Dump it in there unchanged, just for kicks */
936 safestrncpy(target, internet_addr, targbuflen);
938 /* Only do lookups for addresses with hostnames in them */
939 if (num_tokens(internet_addr, '@') != 2) return(-1);
941 /* Only do lookups for domains in the directory */
942 if (IsDirectory(internet_addr, 0) == 0) return(-1);
944 directory_key(key, internet_addr);
945 cdbrec = cdb_fetch(CDB_DIRECTORY, key, strlen(key) );
946 if (cdbrec != NULL) {
947 safestrncpy(target, cdbrec->ptr, targbuflen);
957 * Harvest any email addresses that someone might want to have in their
958 * "collected addresses" book.
960 char *harvest_collected_addresses(struct CtdlMessage *msg) {
963 char user[256], node[256], name[256];
968 if (msg == NULL) return(NULL);
972 if (msg->cm_fields['A'] != NULL) {
973 strcat(addr, msg->cm_fields['A']);
975 if (msg->cm_fields['F'] != NULL) {
977 strcat(addr, msg->cm_fields['F']);
979 if (IsDirectory(msg->cm_fields['F'], 0)) {
984 if (is_harvestable) {
991 if (coll == NULL) return(NULL);
993 /* Scan both the R (To) and Y (CC) fields */
994 for (i = 0; i < 2; ++i) {
995 if (i == 0) field = 'R' ;
996 if (i == 1) field = 'Y' ;
998 if (msg->cm_fields[field] != NULL) {
999 for (j=0; j<num_tokens(msg->cm_fields[field], ','); ++j) {
1000 extract_token(addr, msg->cm_fields[field], j, ',', sizeof addr);
1001 utf8ify_rfc822_string(addr);
1002 process_rfc822_addr(addr, user, node, name);
1003 h = CtdlHostAlias(node);
1004 if ( (h != hostalias_localhost) && (h != hostalias_directory) ) {
1005 coll = realloc(coll, strlen(coll) + strlen(addr) + 4);
1006 if (coll == NULL) return(NULL);
1007 if (!IsEmptyStr(coll)) {
1017 if (IsEmptyStr(coll)) {