1 // Output an HTML message, modifying it slightly to make sure it plays nice
2 // with the rest of our web framework.
4 // Copyright (c) 2005-2022 by the citadel.org team
6 // This program is open source software. Use, duplication, or disclosure is subject to the GNU General Public License v3.
11 // Strip surrounding single or double quotes from a string.
12 void stripquotes(char *s) {
24 if (((s[0] == '\"') && (s[len - 1] == '\"')) || ((s[0] == '\'') && (s[len - 1] == '\''))) {
31 // Check to see if a META tag has overridden the declared MIME character set.
33 // charset Character set name (left unchanged if we don't do anything)
34 // meta_http_equiv Content of the "http-equiv" portion of the META tag
35 // meta_content Content of the "content" portion of the META tag
36 void extract_charset_from_meta(char *charset, char *meta_http_equiv, char *meta_content) {
43 if (!meta_http_equiv) {
50 if (strcasecmp(meta_http_equiv, "Content-type")) {
54 ptr = strchr(meta_content, ';');
59 safestrncpy(buf, ++ptr, sizeof buf);
61 if (!strncasecmp(buf, "charset=", 8)) {
62 strcpy(charset, &buf[8]);
64 // The brain-damaged webmail program in Microsoft Exchange declares
65 // a charset of "unicode" when they really mean "UTF-8". GNU iconv
66 // treats "unicode" as an alias for "UTF-16" so we have to manually
67 // fix this here, otherwise messages generated in Exchange webmail
68 // show up as a big pile of weird characters.
69 if (!strcasecmp(charset, "unicode")) {
70 strcpy(charset, "UTF-8");
73 // Remove wandering punctuation
74 if ((ptr = strchr(charset, '\"'))) {
82 // Sanitize and enhance an HTML message for display.
83 // Also convert weird character sets to UTF-8 if necessary.
84 // Also fixup img src="cid:..." type inline images to fetch the image
85 StrBuf *html2html(const char *supplied_charset, int treat_as_wiki, char *roomname, long msgnum, StrBuf *Source) {
91 StrBuf *converted_msg;
92 int buffer_length = 1;
94 int content_length = 0;
99 int script_start_pos = (-1);
103 StrBuf *BodyArea = NULL;
105 iconv_t ic = (iconv_t) (-1);
106 char *ibuf; // Buffer of characters to be converted
107 char *obuf; // Buffer for converted characters
108 size_t ibuflen; // Length of input buffer
109 size_t obuflen; // Length of output buffer
110 char *osav; // Saved pointer to output buffer
112 StrBuf *Target = NewStrBuf();
113 if (Target == NULL) {
117 safestrncpy(charset, supplied_charset, sizeof charset);
118 sprintf(new_window, "<a target=\"%s\" href=", TARGET);
120 content_length = StrLength(Source);
121 msg = (char *) ChrPtr(Source);
122 buffer_length = content_length;
124 // Do a first pass to isolate the message body
127 msgend = &msg[content_length];
129 while (ptr < msgend) {
131 // Advance to next tag
132 ptr = strchr(ptr, '<');
133 if ((ptr == NULL) || (ptr >= msgend))
136 if ((ptr == NULL) || (ptr >= msgend))
139 // Look for META tags. Some messages (particularly in
140 // Asian locales) illegally declare a message's character
141 // set in the HTML instead of in the MIME headers. This
142 // is wrong but we have to work around it anyway.
143 if (!strncasecmp(ptr, "META", 4)) {
149 char *meta_http_equiv;
153 meta_start = &ptr[4];
154 meta_end = strchr(ptr, '>');
155 if ((meta_end != NULL) && (meta_end <= msgend)) {
156 meta_length = meta_end - meta_start + 1;
157 meta = malloc(meta_length + 1);
158 safestrncpy(meta, meta_start, meta_length);
159 meta[meta_length] = 0;
161 if (!strncasecmp(meta, "HTTP-EQUIV=", 11)) {
162 meta_http_equiv = strdup(&meta[11]);
163 spaceptr = strchr(meta_http_equiv, ' ');
164 if (spaceptr != NULL) {
166 meta_content = strdup(++spaceptr);
167 if (!strncasecmp(meta_content, "content=", 8)) {
168 strcpy(meta_content, &meta_content[8]);
169 stripquotes(meta_http_equiv);
170 stripquotes(meta_content);
171 extract_charset_from_meta(charset, meta_http_equiv, meta_content);
175 free(meta_http_equiv);
181 // Any of these tags cause everything up to and including
182 // the tag to be removed.
183 if ( (!strncasecmp(ptr, "HTML", 4))
184 || (!strncasecmp(ptr, "HEAD", 4))
185 || (!strncasecmp(ptr, "/HEAD", 5))
186 || (!strncasecmp(ptr, "BODY", 4))
190 if (!strncasecmp(ptr, "BODY", 4)) {
193 ptr = strchr(ptr, '>');
194 if ((ptr == NULL) || (ptr >= msgend))
196 if ((pBody != NULL) && (ptr - pBody > 4)) {
198 char *cid_start, *cid_end;
202 while ((isspace(*pBody)) && (pBody < ptr))
204 BodyArea = NewStrBufPlain(NULL, ptr - pBody);
207 src = strstr(pBody, "cid:");
211 while ((*cid_end != '"') && !isspace(*cid_end) && (cid_end < ptr))
214 // copy tag and attributes up to src="cid:
215 StrBufAppendBufPlain(BodyArea, pBody, src - pBody, 0);
217 // add in /webcit/mimepart/<msgno>/CID/
218 // trailing / stops dumb URL filters getting excited
219 StrBufAppendPrintf(BodyArea, "/webcit/mimepart/%ld/", msgnum);
220 StrBufAppendBufPlain(BodyArea, cid_start, cid_end - cid_start, 0);
222 if (ptr - cid_end > 0)
223 StrBufAppendBufPlain(BodyArea, cid_end + 1, ptr - cid_end, 0);
226 StrBufAppendBufPlain(BodyArea, pBody, ptr - pBody, 0);
232 if ((ptr == NULL) || (ptr >= msgend))
237 // Any of these tags cause everything including and following
238 // the tag to be removed.
239 if ((!strncasecmp(ptr, "/HTML", 5)) || (!strncasecmp(ptr, "/BODY", 5))) {
247 if (msgstart > msg) {
248 strcpy(msg, msgstart);
251 // Now go through the message, parsing tags as necessary.
252 converted_msg = NewStrBufPlain(NULL, content_length + 8192);
254 // Convert foreign character sets to UTF-8 if necessary
255 if ( (strcasecmp(charset, "us-ascii"))
256 && (strcasecmp(charset, "UTF-8"))
257 && (strcasecmp(charset, ""))
259 syslog(LOG_DEBUG, "Converting %s to UTF-8", charset);
260 ctdl_iconv_open("UTF-8", charset, &ic);
261 if (ic == (iconv_t) (-1)) {
262 syslog(LOG_WARNING, "%s:%d iconv_open() failed: %s", __FILE__, __LINE__, strerror(errno));
265 if (Source == NULL) {
266 if (ic != (iconv_t) (-1)) {
268 ibuflen = content_length;
269 obuflen = content_length + (content_length / 2);
270 obuf = (char *) malloc(obuflen);
272 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
273 content_length = content_length + (content_length / 2) - obuflen;
274 osav[content_length] = 0;
281 if (ic != (iconv_t) (-1)) {
282 StrBuf *Buf = NewStrBufPlain(NULL, StrLength(Source) + 8096);;
283 StrBufConvert(Source, Buf, &ic);
286 msg = (char *) ChrPtr(Source); // TODO: get rid of this.
290 // At this point, the message has been stripped down to
291 // only the content inside the <BODY></BODY> tags, and has
292 // been converted to UTF-8 if it was originally in a foreign
293 // character set. The text is also guaranteed to be null
296 if (converted_msg == NULL) {
297 StrBufAppendPrintf(Target, "Error %d: %s<br>%s:%d", errno, strerror(errno), __FILE__, __LINE__);
301 if (BodyArea != NULL) { // Any attributes that were declared in the <body> tag
302 StrBufAppendBufPlain(converted_msg, HKEY("<div "), 0); // are instead declared in this <div> tag
303 StrBufAppendBuf(converted_msg, BodyArea, 0);
304 StrBufAppendBufPlain(converted_msg, HKEY(">"), 0);
307 msgend = strchr(msg, 0);
308 while (ptr < msgend) {
310 // Try to sanitize the html of any rogue scripts
311 if (!strncasecmp(ptr, "<script", 7)) {
312 if (scriptlevel == 0) {
313 script_start_pos = StrLength(converted_msg);
317 if (!strncasecmp(ptr, "</script", 8)) {
321 // Change mailto: links to WebCit mail, by replacing the
322 // link with one that points back to our mail room. Due to
323 // the way we parse URL's, it'll even handle mailto: links
324 // that have "?subject=" in them.
325 // FIXME change URL syntax for webcit-ng
326 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
327 content_length += 64;
328 StrBufAppendPrintf(converted_msg, "<a href=\"display_enter?force_room=_MAIL_?recp=");
334 // Make external links open in a separate window
335 else if (!strncasecmp(ptr, "<a href=\"", 9)) {
338 if (((strchr(ptr, ':') < strchr(ptr, '/'))) && ((strchr(ptr, '/') < strchr(ptr, '>')))) {
339 // open external links to new window
340 StrBufAppendPrintf(converted_msg, new_window);
343 else if ((treat_as_wiki)
344 && (strncasecmp(ptr, "<a href=\"wiki?", 14))
345 && (strncasecmp(ptr, "<a href=\"dotgoto?", 17))
346 && (strncasecmp(ptr, "<a href=\"knrooms?", 17))
348 content_length += 64;
349 StrBufAppendPrintf(converted_msg, "<a href=\"wiki?go=");
350 //StrBufUrlescAppend(converted_msg, "FIXME ROOM NAME", NULL); // FIXME make compatible with webcit-ng
351 StrBufAppendPrintf(converted_msg, "?page=");
355 StrBufAppendPrintf(converted_msg, "<a href=\"");
360 // Fixup <img src="cid:... ...> to fetch the mime part
361 else if (!strncasecmp(ptr, "<img ", 5)) {
362 char *cid_start, *cid_end;
363 char *tag_end = strchr(ptr, '>');
365 // FIXME - handle this situation (maybe someone opened an <img cid...
366 // and then ended the message)
368 syslog(LOG_DEBUG, "tag_end is null and ptr is:");
369 syslog(LOG_DEBUG, "%s", ptr);
370 syslog(LOG_DEBUG, "Theoretical bytes remaining: %d", (int) (msgend - ptr));
373 src = strstr(ptr, "src=\"cid:");
377 && isspace(*(src - 1))
378 && tag_end && (cid_start = strchr(src, ':'))
379 && (cid_end = strchr(cid_start, '"'))
380 && (cid_end < tag_end)
382 // copy tag and attributes up to src="cid:
383 StrBufAppendBufPlain(converted_msg, ptr, src - ptr, 0);
386 // add in /webcit/mimepart/<msgnum>/CID/
387 // trailing / stops dumb URL filters getting excited
388 StrBufAppendPrintf(converted_msg, " src=\"/ctdl/r/");
389 StrBufXMLEscAppend(converted_msg, NULL, roomname, strlen(roomname), 0);
390 syslog(LOG_DEBUG, "room name is '%s'", roomname);
391 StrBufAppendPrintf(converted_msg, "/%ld/", msgnum);
392 StrBufAppendBufPlain(converted_msg, cid_start, cid_end - cid_start, 0);
393 StrBufAppendBufPlain(converted_msg, "\"", -1, 0);
396 StrBufAppendBufPlain(converted_msg, ptr, tag_end - ptr, 0);
400 // Turn anything that looks like a URL into a real link, as long
401 // as it's not inside a tag already
402 else if ((brak == 0) && (alevel == 0) && ((!strncasecmp(ptr, "http://", 7)) || (!strncasecmp(ptr, "https://", 8)))) {
403 // Find the end of the link
407 strlenptr = strlen(ptr);
408 for (i = 0; i <= strlenptr; ++i) {
426 if ((ptr[i + 2] == ';') ||
427 (ptr[i + 3] == ';') ||
428 (ptr[i + 5] == ';') || (ptr[i + 6] == ';') || (ptr[i + 7] == ';'))
441 linkedchar = ptr[len];
443 // spot for some subject strings tinymce tends to give us.
444 ltreviewptr = strchr(ptr, '<');
445 if (ltreviewptr != NULL) {
447 linklen = ltreviewptr - ptr;
450 nbspreviewptr = strstr(ptr, " ");
451 if (nbspreviewptr != NULL) {
452 // nbspreviewptr = '\0';
453 linklen = nbspreviewptr - ptr;
455 if (ltreviewptr != 0)
458 ptr[len] = linkedchar;
460 content_length += (32 + linklen);
461 StrBufAppendPrintf(converted_msg, "%s\"", new_window);
462 StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
463 StrBufAppendPrintf(converted_msg, "\">");
464 StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
466 StrBufAppendPrintf(converted_msg, "</a>");
470 StrBufAppendBufPlain(converted_msg, ptr, 1, 0);
474 if ((ptr >= msg) && (ptr <= msgend)) {
475 // We need to know when we're inside a tag,
476 // so we don't turn things that look like URL's into
477 // links, when they're already links - or image sources.
478 if ((ptr > msg) && (*(ptr - 1) == '<')) {
481 if ((ptr > msg) && (*(ptr - 1) == '>')) {
483 if ((scriptlevel == 0) && (script_start_pos >= 0)) {
484 StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
485 script_start_pos = (-1);
488 if (!strncasecmp(ptr, "</a>", 3))
493 if (BodyArea != NULL) {
494 StrBufAppendBufPlain(converted_msg, HKEY("</div>"), 0); // Close the div where we declared attributes copied
495 FreeStrBuf(&BodyArea); // from the original <body> tag
498 // uncomment these two lines to override conversion
499 // memcpy(converted_msg, msg, content_length);
500 // output_length = content_length;
502 // Output our big pile of markup
503 StrBufAppendBuf(Target, converted_msg, 0);
505 BAIL: // A little trailing vertical whitespace...
506 StrBufAppendPrintf(Target, "<br>\n");
508 // Now give back the memory
509 FreeStrBuf(&converted_msg);
510 if ((msg != NULL) && (Source == NULL))
516 // Look for URL's embedded in a buffer and make them linkable. We use a
517 // target window in order to keep the Citadel session in its own window.
518 void UrlizeText(StrBuf * Target, StrBuf * Source, StrBuf * WrkBuf) {
519 int len, UrlLen, Offset, TrailerLen;
520 const char *start, *end, *pos;
524 len = StrLength(Source);
525 end = ChrPtr(Source) + len;
526 for (pos = ChrPtr(Source); (pos < end) && (start == NULL); ++pos) {
527 if (!strncasecmp(pos, "http://", 7))
529 else if (!strncasecmp(pos, "ftp://", 6))
534 StrBufAppendBuf(Target, Source, 0);
539 for (pos = ChrPtr(Source) + len; pos > start; --pos) {
559 UrlLen = end - start;
560 StrBufAppendBufPlain(WrkBuf, start, UrlLen, 0);
562 Offset = start - ChrPtr(Source);
564 StrBufAppendBufPlain(Target, ChrPtr(Source), Offset, 0);
565 StrBufAppendPrintf(Target, "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
566 LB, QU, ChrPtr(WrkBuf), QU, QU, TARGET, QU, RB, ChrPtr(WrkBuf), LB, RB);
568 TrailerLen = StrLength(Source) - (end - ChrPtr(Source));
570 StrBufAppendBufPlain(Target, end, TrailerLen, 0);
574 void url(char *buf, size_t bufsize) {
575 int len, UrlLen, Offset, TrailerLen, outpos;
576 char *start, *end, *pos;
583 syslog(LOG_WARNING, "URL: content longer than buffer!");
587 for (pos = buf; (pos < end) && (start == NULL); ++pos) {
588 if (!strncasecmp(pos, "http://", 7))
590 if (!strncasecmp(pos, "ftp://", 6))
597 for (pos = buf + len; pos > start; --pos) {
617 UrlLen = end - start;
618 if (UrlLen > sizeof(urlbuf)) {
619 syslog(LOG_WARNING, "URL: content longer than buffer!");
622 memcpy(urlbuf, start, UrlLen);
623 urlbuf[UrlLen] = '\0';
625 Offset = start - buf;
626 if ((Offset != 0) && (Offset < sizeof(outbuf)))
627 memcpy(outbuf, buf, Offset);
628 outpos = snprintf(&outbuf[Offset], sizeof(outbuf) - Offset,
629 "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c", LB, QU, urlbuf, QU, QU, TARGET, QU, RB, urlbuf, LB, RB);
630 if (outpos >= sizeof(outbuf) - Offset) {
631 syslog(LOG_WARNING, "URL: content longer than buffer!");
635 TrailerLen = len - (end - start);
637 memcpy(outbuf + Offset + outpos, end, TrailerLen);
638 if (Offset + outpos + TrailerLen > bufsize) {
639 syslog(LOG_WARNING, "URL: content longer than buffer!");
642 memcpy(buf, outbuf, Offset + outpos + TrailerLen);
643 *(buf + Offset + outpos + TrailerLen) = '\0';