webcit/html2html.c

   1 /*
   2  * Output an HTML message, modifying it slightly to make sure it plays nice
   3  * with the rest of our web framework.
   4  *
   5  * Copyright (c) 2005-2012 by the citadel.org team
   6  *
   7  * This program is open source software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License, version 3.
   9  *
  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.
  14  */
  15
  16 #include "webcit.h"
  17 #include "webserver.h"
  18
  19
  20 /*
  21  * Strip surrounding single or double quotes from a string.
  22  */
  23 void stripquotes(char *s)
  24 {
  25         int len;
  26
  27         if (!s) return;
  28
  29         len = strlen(s);
  30         if (len < 2) return;
  31
  32         if ( ( (s[0] == '\"') && (s[len-1] == '\"') ) || ( (s[0] == '\'') && (s[len-1] == '\'') ) ) {
  33                 s[len-1] = 0;
  34                 strcpy(s, &s[1]);
  35         }
  36 }
  37
  38
  39 /*
  40  * Check to see if a META tag has overridden the declared MIME character set.
  41  *
  42  * charset              Character set name (left unchanged if we don't do anything)
  43  * meta_http_equiv      Content of the "http-equiv" portion of the META tag
  44  * meta_content         Content of the "content" portion of the META tag
  45  */
  46 void extract_charset_from_meta(char *charset, char *meta_http_equiv, char *meta_content)
  47 {
  48         char *ptr;
  49         char buf[64];
  50
  51         if (!charset) return;
  52         if (!meta_http_equiv) return;
  53         if (!meta_content) return;
  54
  55
  56         if (strcasecmp(meta_http_equiv, "Content-type")) return;
  57
  58         ptr = strchr(meta_content, ';');
  59         if (!ptr) return;
  60
  61         safestrncpy(buf, ++ptr, sizeof buf);
  62         striplt(buf);
  63         if (!strncasecmp(buf, "charset=", 8)) {
  64                 strcpy(charset, &buf[8]);
  65
  66                 /*
  67                  * The brain-damaged webmail program in Microsoft Exchange declares
  68                  * a charset of "unicode" when they really mean "UTF-8".  GNU iconv
  69                  * treats "unicode" as an alias for "UTF-16" so we have to manually
  70                  * fix this here, otherwise messages generated in Exchange webmail
  71                  * show up as a big pile of weird characters.
  72                  */
  73                 if (!strcasecmp(charset, "unicode")) {
  74                         strcpy(charset, "UTF-8");
  75                 }
  76
  77                 /* Remove wandering punctuation */
  78                 if ((ptr=strchr(charset, '\"'))) *ptr = 0;
  79                 striplt(charset);
  80         }
  81 }
  82
  83
  84
  85 /*
  86  * Sanitize and enhance an HTML message for display.
  87  * Also convert weird character sets to UTF-8 if necessary.
  88  * Also fixup img src="cid:..." type inline images to fetch the image
  89  *
  90  */
  91 void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, StrBuf *Source, StrBuf *Target) {
  92         char buf[SIZ];
  93         char *msg;
  94         char *ptr;
  95         char *msgstart;
  96         char *msgend;
  97         StrBuf *converted_msg;
  98         int buffer_length = 1;
  99         int line_length = 0;
 100         int content_length = 0;
 101         char new_window[SIZ];
 102         int brak = 0;
 103         int alevel = 0;
 104         int scriptlevel = 0;
 105         int script_start_pos = (-1);
 106         int i;
 107         int linklen;
 108         char charset[128];
 109         StrBuf *BodyArea = NULL;
 110 #ifdef HAVE_ICONV
 111         iconv_t ic = (iconv_t)(-1) ;
 112         char *ibuf;                   /* Buffer of characters to be converted */
 113         char *obuf;                   /* Buffer for converted characters      */
 114         size_t ibuflen;               /* Length of input buffer               */
 115         size_t obuflen;               /* Length of output buffer              */
 116         char *osav;                   /* Saved pointer to output buffer       */
 117 #endif
 118         if (Target == NULL)
 119                 Target = WC->WBuf;
 120
 121         safestrncpy(charset, supplied_charset, sizeof charset);
 122         msg = strdup("");
 123         sprintf(new_window, "<a target=\"%s\" href=", TARGET);
 124
 125         if (Source == NULL) while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 126                 line_length = strlen(buf);
 127                 buffer_length = content_length + line_length + 2;
 128                 ptr = realloc(msg, buffer_length);
 129                 if (ptr == NULL) {
 130                         StrBufAppendPrintf(Target, "<b>");
 131                         StrBufAppendPrintf(Target, _("realloc() error! couldn't get %d bytes: %s"),
 132                                         buffer_length + 1,
 133                                         strerror(errno));
 134                         StrBufAppendPrintf(Target, "</b><br><br>\n");
 135                         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 136                                 /** flush */
 137                         }
 138                         free(msg);
 139                         return;
 140                 }
 141                 msg = ptr;
 142                 strcpy(&msg[content_length], buf);
 143                 content_length += line_length;
 144                 strcpy(&msg[content_length], "\n");
 145                 content_length += 1;
 146         }
 147         else {
 148                 content_length = StrLength(Source);
 149                 free(msg);
 150                 msg = (char*) ChrPtr(Source);/* TODO: remove cast */
 151                 buffer_length = content_length;
 152         }
 153
 154         /** Do a first pass to isolate the message body */
 155         ptr = msg + 1;
 156         msgstart = msg;
 157         msgend = &msg[content_length];
 158
 159         while (ptr < msgend) {
 160
 161                 /** Advance to next tag */
 162                 ptr = strchr(ptr, '<');
 163                 if ((ptr == NULL) || (ptr >= msgend)) break;
 164                 ++ptr;
 165                 if ((ptr == NULL) || (ptr >= msgend)) break;
 166
 167                 /*
 168                  *  Look for META tags.  Some messages (particularly in
 169                  *  Asian locales) illegally declare a message's character
 170                  *  set in the HTML instead of in the MIME headers.  This
 171                  *  is wrong but we have to work around it anyway.
 172                  */
 173                 if (!strncasecmp(ptr, "META", 4)) {
 174
 175                         char *meta_start;
 176                         char *meta_end;
 177                         int meta_length;
 178                         char *meta;
 179                         char *meta_http_equiv;
 180                         char *meta_content;
 181                         char *spaceptr;
 182
 183                         meta_start = &ptr[4];
 184                         meta_end = strchr(ptr, '>');
 185                         if ((meta_end != NULL) && (meta_end <= msgend)) {
 186                                 meta_length = meta_end - meta_start + 1;
 187                                 meta = malloc(meta_length + 1);
 188                                 safestrncpy(meta, meta_start, meta_length);
 189                                 meta[meta_length] = 0;
 190                                 striplt(meta);
 191                                 if (!strncasecmp(meta, "HTTP-EQUIV=", 11)) {
 192                                         meta_http_equiv = strdup(&meta[11]);
 193                                         spaceptr = strchr(meta_http_equiv, ' ');
 194                                         if (spaceptr != NULL) {
 195                                                 *spaceptr = 0;
 196                                                 meta_content = strdup(++spaceptr);
 197                                                 if (!strncasecmp(meta_content, "content=", 8)) {
 198                                                         strcpy(meta_content, &meta_content[8]);
 199                                                         stripquotes(meta_http_equiv);
 200                                                         stripquotes(meta_content);
 201                                                         extract_charset_from_meta(charset,
 202                                                                         meta_http_equiv, meta_content);
 203                                                 }
 204                                                 free(meta_content);
 205                                         }
 206                                         free(meta_http_equiv);
 207                                 }
 208                                 free(meta);
 209                         }
 210                 }
 211
 212                 /*
 213                  * Any of these tags cause everything up to and including
 214                  * the tag to be removed.
 215                  */
 216                 if ( (!strncasecmp(ptr, "HTML", 4))
 217                                 ||(!strncasecmp(ptr, "HEAD", 4))
 218                                 ||(!strncasecmp(ptr, "/HEAD", 5))
 219                                 ||(!strncasecmp(ptr, "BODY", 4)) ) {
 220                         char *pBody = NULL;
 221
 222                         if (!strncasecmp(ptr, "BODY", 4)) {
 223                                 pBody = ptr;
 224                         }
 225                         ptr = strchr(ptr, '>');
 226                         if ((ptr == NULL) || (ptr >= msgend)) break;
 227                         if ((pBody != NULL) && (ptr - pBody > 4)) {
 228                                 char* src;
 229                                 char *cid_start, *cid_end;
 230
 231                                 *ptr = '\0';
 232                                 pBody += 4;
 233                                 while ((isspace(*pBody)) && (pBody < ptr))
 234                                         pBody ++;
 235                                 BodyArea = NewStrBufPlain(NULL,  ptr - pBody);
 236
 237                                 if (pBody < ptr) {
 238                                         src = strstr(pBody, "cid:");
 239                                         if (src) {
 240                                                 cid_start = src + 4;
 241                                                 cid_end = cid_start;
 242                                                 while ((*cid_end != '"') &&
 243                                                                 !isspace(*cid_end) &&
 244                                                                 (cid_end < ptr))
 245                                                         cid_end ++;
 246
 247                                                 /* copy tag and attributes up to src="cid: */
 248                                                 StrBufAppendBufPlain(BodyArea, pBody, src - pBody, 0);
 249
 250                                                 /* add in /webcit/mimepart/<msgno>/CID/
 251                                                    trailing / stops dumb URL filters getting excited */
 252                                                 StrBufAppendPrintf(BodyArea,
 253                                                                 "/webcit/mimepart/%d/",msgnum);
 254                                                 StrBufAppendBufPlain(BodyArea, cid_start, cid_end - cid_start, 0);
 255
 256                                                 if (ptr - cid_end > 0)
 257                                                         StrBufAppendBufPlain(BodyArea,
 258                                                                         cid_end + 1,
 259                                                                         ptr - cid_end, 0);
 260                                         }
 261                                         else
 262                                                 StrBufAppendBufPlain(BodyArea, pBody, ptr - pBody, 0);
 263                                 }
 264                                 *ptr = '>';
 265                         }
 266                         ++ptr;
 267                         if ((ptr == NULL) || (ptr >= msgend)) break;
 268                         msgstart = ptr;
 269                 }
 270
 271                 /*
 272                  * Any of these tags cause everything including and following
 273                  * the tag to be removed.
 274                  */
 275                 if ( (!strncasecmp(ptr, "/HTML", 5))
 276                                 ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 277                         --ptr;
 278                         msgend = ptr;
 279                         strcpy(ptr, "");
 280
 281                 }
 282
 283                 ++ptr;
 284         }
 285         if (msgstart > msg) {
 286                 strcpy(msg, msgstart);
 287         }
 288
 289         /* Now go through the message, parsing tags as necessary. */
 290         converted_msg = NewStrBufPlain(NULL, content_length + 8192);
 291
 292
 293         /** Convert foreign character sets to UTF-8 if necessary. */
 294 #ifdef HAVE_ICONV
 295         if ( (strcasecmp(charset, "us-ascii"))
 296                         && (strcasecmp(charset, "UTF-8"))
 297                         && (strcasecmp(charset, ""))
 298            ) {
 299                 syslog(LOG_DEBUG, "Converting %s to UTF-8\n", charset);
 300                 ctdl_iconv_open("UTF-8", charset, &ic);
 301                 if (ic == (iconv_t)(-1) ) {
 302                         syslog(LOG_WARNING, "%s:%d iconv_open() failed: %s\n",
 303                                         __FILE__, __LINE__, strerror(errno));
 304                 }
 305         }
 306         if  (Source == NULL) {
 307                 if (ic != (iconv_t)(-1) ) {
 308                         ibuf = msg;
 309                         ibuflen = content_length;
 310                         obuflen = content_length + (content_length / 2) ;
 311                         obuf = (char *) malloc(obuflen);
 312                         osav = obuf;
 313                         iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
 314                         content_length = content_length + (content_length / 2) - obuflen;
 315                         osav[content_length] = 0;
 316                         free(msg);
 317                         msg = osav;
 318                         iconv_close(ic);
 319                 }
 320         }
 321         else {
 322                 if (ic != (iconv_t)(-1) ) {
 323                         StrBuf *Buf = NewStrBufPlain(NULL, StrLength(Source) + 8096);;
 324                         StrBufConvert(Source, Buf, &ic);
 325                         FreeStrBuf(&Buf);
 326                         iconv_close(ic);
 327                         msg = (char*)ChrPtr(Source); /* TODO: get rid of this. */
 328                 }
 329         }
 330
 331 #endif
 332
 333         /*
 334          *      At this point, the message has been stripped down to
 335          *      only the content inside the <BODY></BODY> tags, and has
 336          *      been converted to UTF-8 if it was originally in a foreign
 337          *      character set.  The text is also guaranteed to be null
 338          *      terminated now.
 339          */
 340
 341         if (converted_msg == NULL) {
 342                 StrBufAppendPrintf(Target, "Error %d: %s<br>%s:%d", errno, strerror(errno), __FILE__, __LINE__);
 343                 goto BAIL;
 344         }
 345
 346         if (BodyArea != NULL) {
 347                 StrBufAppendBufPlain(converted_msg, HKEY("<table "), 0);
 348                 StrBufAppendBuf(converted_msg, BodyArea, 0);
 349                 StrBufAppendBufPlain(converted_msg, HKEY(" width=\"100%\"><tr><td>"), 0);
 350         }
 351         ptr = msg;
 352         msgend = strchr(msg, 0);
 353         while (ptr < msgend) {
 354
 355                 /** Try to sanitize the html of any rogue scripts */
 356                 if (!strncasecmp(ptr, "<script", 7)) {
 357                         if (scriptlevel == 0) {
 358                                 script_start_pos = StrLength(converted_msg);
 359                         }
 360                         ++scriptlevel;
 361                 }
 362                 if (!strncasecmp(ptr, "</script", 8)) {
 363                         --scriptlevel;
 364                 }
 365
 366                 /**
 367                  * Change mailto: links to WebCit mail, by replacing the
 368                  * link with one that points back to our mail room.  Due to
 369                  * the way we parse URL's, it'll even handle mailto: links
 370                  * that have "?subject=" in them.
 371                  */
 372                 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
 373                         content_length += 64;
 374                         StrBufAppendPrintf(converted_msg,
 375                                         "<a href=\"display_enter?force_room=_MAIL_?recp=");
 376                         ptr = &ptr[16];
 377                         ++alevel;
 378                         ++brak;
 379                 }
 380                 /** Make external links open in a separate window */
 381                 else if (!strncasecmp(ptr, "<a href=\"", 9)) {
 382                         ++alevel;
 383                         ++brak;
 384                         if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
 385                                         &&  ((strchr(ptr, '/') < strchr(ptr, '>')))
 386                            ) {
 387                                 /* open external links to new window */
 388                                 StrBufAppendPrintf(converted_msg, new_window);
 389                                 ptr = &ptr[8];
 390                         }
 391                         else if (
 392                                 (treat_as_wiki)
 393                                 && (strncasecmp(ptr, "<a href=\"wiki?", 14))
 394                                 && (strncasecmp(ptr, "<a href=\"dotgoto?", 17))
 395                                 && (strncasecmp(ptr, "<a href=\"knrooms?", 17))
 396                         ) {
 397                                 content_length += 64;
 398                                 StrBufAppendPrintf(converted_msg, "<a href=\"wiki?go=");
 399                                 StrBufUrlescAppend(converted_msg, WC->CurRoom.name, NULL);
 400                                 StrBufAppendPrintf(converted_msg, "?page=");
 401                                 ptr = &ptr[9];
 402                         }
 403                         else {
 404                                 StrBufAppendPrintf(converted_msg, "<a href=\"");
 405                                 ptr = &ptr[9];
 406                         }
 407                 }
 408                 /** Fixup <img src="cid:... ...> to fetch the mime part */
 409                 else if (!strncasecmp(ptr, "<img ", 5)) {
 410                         char *cid_start, *cid_end;
 411                         char* tag_end=strchr(ptr,'>');
 412                         char* src;
 413                         /* FIXME - handle this situation (maybe someone opened an <img cid...
 414                          * and then ended the message)
 415                          */
 416                         if (!tag_end) {
 417                                 syslog(LOG_DEBUG, "tag_end is null and ptr is:\n");
 418                                 syslog(LOG_DEBUG, "%s\n", ptr);
 419                                 syslog(LOG_DEBUG, "Theoretical bytes remaining: %d\n", (int)(msgend - ptr));
 420                         }
 421
 422                         src=strstr(ptr, "src=\"cid:");
 423                         ++brak;
 424
 425                         if (src
 426                             && isspace(*(src-1))
 427                                 && tag_end
 428                                 && (cid_start=strchr(src,':'))
 429                                 && (cid_end=strchr(cid_start,'"'))
 430                                 && (cid_end < tag_end)
 431                         ) {
 432                                 /* copy tag and attributes up to src="cid: */
 433                                 StrBufAppendBufPlain(converted_msg, ptr, src - ptr, 0);
 434                                 cid_start++;
 435
 436                                 /* add in /webcit/mimepart/<msgno>/CID/
 437                                    trailing / stops dumb URL filters getting excited */
 438                                 StrBufAppendPrintf(converted_msg,
 439                                                 " src=\"/webcit/mimepart/%d/",msgnum);
 440                                 StrBufAppendBufPlain(converted_msg, cid_start, cid_end - cid_start, 0);
 441                                 StrBufAppendBufPlain(converted_msg, "/\"", -1, 0);
 442
 443                                 ptr = cid_end+1;
 444                         }
 445                         StrBufAppendBufPlain(converted_msg, ptr, tag_end - ptr, 0);
 446                         ptr = tag_end;
 447                 }
 448
 449                 /**
 450                  * Turn anything that looks like a URL into a real link, as long
 451                  * as it's not inside a tag already
 452                  */
 453                 else if ( (brak == 0) && (alevel == 0) &&
 454                           ( (!strncasecmp(ptr, "http://", 7)) ||
 455                             (!strncasecmp(ptr, "https://", 8)))) {
 456                         /** Find the end of the link */
 457                         int strlenptr;
 458                         linklen = 0;
 459
 460                         strlenptr = strlen(ptr);
 461                         for (i=0; i<=strlenptr; ++i) {
 462                                 if ((ptr[i]==0)
 463                                     ||(isspace(ptr[i]))
 464                                     ||(ptr[i]==10)
 465                                     ||(ptr[i]==13)
 466                                     ||(ptr[i]=='(')
 467                                     ||(ptr[i]==')')
 468                                     ||(ptr[i]=='<')
 469                                     ||(ptr[i]=='>')
 470                                     ||(ptr[i]=='[')
 471                                     ||(ptr[i]==']')
 472                                     ||(ptr[i]=='"')
 473                                     ||(ptr[i]=='\'')
 474                                         ) linklen = i;
 475                                 /* did s.b. send us an entity? */
 476                                 if (ptr[i] == '&') {
 477                                         if ((ptr[i+2] ==';') ||
 478                                             (ptr[i+3] ==';') ||
 479                                             (ptr[i+5] ==';') ||
 480                                             (ptr[i+6] ==';') ||
 481                                             (ptr[i+7] ==';'))
 482                                                 linklen = i;
 483                                 }
 484                                 if (linklen > 0) break;
 485                         }
 486                         if (linklen > 0) {
 487                                 char *ltreviewptr;
 488                                 char *nbspreviewptr;
 489                                 char linkedchar;
 490                                 int len;
 491
 492                                 len = linklen;
 493                                 linkedchar = ptr[len];
 494                                 ptr[len] = '\0';
 495                                 /* spot for some subject strings tinymce tends to give us. */
 496                                 ltreviewptr = strchr(ptr, '<');
 497                                 if (ltreviewptr != NULL) {
 498                                         *ltreviewptr = '\0';
 499                                         linklen = ltreviewptr - ptr;
 500                                 }
 501
 502                                 nbspreviewptr = strstr(ptr, "&nbsp;");
 503                                 if (nbspreviewptr != NULL) {
 504                                         /* nbspreviewptr = '\0'; */
 505                                         linklen = nbspreviewptr - ptr;
 506                                 }
 507                                 if (ltreviewptr != 0)
 508                                         *ltreviewptr = '<';
 509
 510                                 ptr[len] = linkedchar;
 511
 512                                 content_length += (32 + linklen);
 513                                 StrBufAppendPrintf(converted_msg, "%s\"", new_window);
 514                                 StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 515                                 StrBufAppendPrintf(converted_msg, "\">");
 516                                 StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 517                                 ptr += linklen;
 518                                 StrBufAppendPrintf(converted_msg, "</A>");
 519                         }
 520                 }
 521                 else {
 522                         StrBufAppendBufPlain(converted_msg, ptr, 1, 0);
 523                         ptr++;
 524                 }
 525
 526
 527                 if ((ptr >= msg) && (ptr <= msgend)) {
 528                         /*
 529                          * We need to know when we're inside a tag,
 530                          * so we don't turn things that look like URL's into
 531                          * links, when they're already links - or image sources.
 532                          */
 533                         if ((ptr > msg) && (*(ptr-1) == '<')) {
 534                                 ++brak;
 535                         }
 536                         if ((ptr > msg) && (*(ptr-1) == '>')) {
 537                                 --brak;
 538                                 if ((scriptlevel == 0) && (script_start_pos >= 0)) {
 539                                         StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
 540                                         script_start_pos = (-1);
 541                                 }
 542                         }
 543                         if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 544                 }
 545         }
 546
 547         if (BodyArea != NULL) {
 548                 StrBufAppendBufPlain(converted_msg, HKEY("</td></tr></table>"), 0);
 549                 FreeStrBuf(&BodyArea);
 550         }
 551
 552         /**     uncomment these two lines to override conversion        */
 553         /**     memcpy(converted_msg, msg, content_length);             */
 554         /**     output_length = content_length;                         */
 555
 556         /** Output our big pile of markup */
 557         StrBufAppendBuf(Target, converted_msg, 0);
 558
 559 BAIL:   /** A little trailing vertical whitespace... */
 560         StrBufAppendPrintf(Target, "<br><br>\n");
 561
 562         /** Now give back the memory */
 563         FreeStrBuf(&converted_msg);
 564         if ((msg != NULL) && (Source == NULL)) free(msg);
 565 }
 566
 567
 568
 569
 570
 571
 572 /*
 573  * Look for URL's embedded in a buffer and make them linkable.  We use a
 574  * target window in order to keep the Citadel session in its own window.
 575  */
 576 void UrlizeText(StrBuf* Target, StrBuf *Source, StrBuf *WrkBuf)
 577 {
 578         int len, UrlLen, Offset, TrailerLen;
 579         const char *start, *end, *pos;
 580
 581         FlushStrBuf(Target);
 582
 583         start = NULL;
 584         len = StrLength(Source);
 585         end = ChrPtr(Source) + len;
 586         for (pos = ChrPtr(Source); (pos < end) && (start == NULL); ++pos) {
 587                 if (!strncasecmp(pos, "http://", 7))
 588                         start = pos;
 589                 else if (!strncasecmp(pos, "ftp://", 6))
 590                         start = pos;
 591         }
 592
 593         if (start == NULL) {
 594                 StrBufAppendBuf(Target, Source, 0);
 595                 return;
 596         }
 597         FlushStrBuf(WrkBuf);
 598
 599         for (pos = ChrPtr(Source) + len; pos > start; --pos) {
 600                 if (  (!isprint(*pos))
 601                    || (isspace(*pos))
 602                    || (*pos == '{')
 603                    || (*pos == '}')
 604                    || (*pos == '|')
 605                    || (*pos == '\\')
 606                    || (*pos == '^')
 607                    || (*pos == '[')
 608                    || (*pos == ']')
 609                    || (*pos == '`')
 610                    || (*pos == '<')
 611                    || (*pos == '>')
 612                    || (*pos == '(')
 613                    || (*pos == ')')
 614                 ) {
 615                         end = pos;
 616                 }
 617         }
 618
 619         UrlLen = end - start;
 620         StrBufAppendBufPlain(WrkBuf, start, UrlLen, 0);
 621
 622         Offset = start - ChrPtr(Source);
 623         if (Offset != 0)
 624                 StrBufAppendBufPlain(Target, ChrPtr(Source), Offset, 0);
 625         StrBufAppendPrintf(Target, "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
 626                            LB, QU, ChrPtr(WrkBuf), QU, QU, TARGET,
 627                            QU, RB, ChrPtr(WrkBuf), LB, RB);
 628
 629         TrailerLen = StrLength(Source) - (end - ChrPtr(Source));
 630         if (TrailerLen > 0)
 631                 StrBufAppendBufPlain(Target, end, TrailerLen, 0);
 632 }
 633
 634
 635 void url(char *buf, size_t bufsize)
 636 {
 637         int len, UrlLen, Offset, TrailerLen, outpos;
 638         char *start, *end, *pos;
 639         char urlbuf[SIZ];
 640         char outbuf[SIZ];
 641
 642         start = NULL;
 643         len = strlen(buf);
 644         if (len > bufsize) {
 645                 syslog(LOG_WARNING, "URL: content longer than buffer!");
 646                 return;
 647         }
 648         end = buf + len;
 649         for (pos = buf; (pos < end) && (start == NULL); ++pos) {
 650                 if (!strncasecmp(pos, "http://", 7))
 651                         start = pos;
 652                 if (!strncasecmp(pos, "ftp://", 6))
 653                         start = pos;
 654         }
 655
 656         if (start == NULL)
 657                 return;
 658
 659         for (pos = buf+len; pos > start; --pos) {
 660                 if (  (!isprint(*pos))
 661                    || (isspace(*pos))
 662                    || (*pos == '{')
 663                    || (*pos == '}')
 664                    || (*pos == '|')
 665                    || (*pos == '\\')
 666                    || (*pos == '^')
 667                    || (*pos == '[')
 668                    || (*pos == ']')
 669                    || (*pos == '`')
 670                    || (*pos == '<')
 671                    || (*pos == '>')
 672                    || (*pos == '(')
 673                    || (*pos == ')')
 674                 ) {
 675                         end = pos;
 676                 }
 677         }
 678
 679         UrlLen = end - start;
 680         if (UrlLen > sizeof(urlbuf)){
 681                 syslog(LOG_WARNING, "URL: content longer than buffer!");
 682                 return;
 683         }
 684         memcpy(urlbuf, start, UrlLen);
 685         urlbuf[UrlLen] = '\0';
 686
 687         Offset = start - buf;
 688         if ((Offset != 0) && (Offset < sizeof(outbuf)))
 689                 memcpy(outbuf, buf, Offset);
 690         outpos = snprintf(&outbuf[Offset], sizeof(outbuf) - Offset,
 691                           "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
 692                           LB, QU, urlbuf, QU, QU, TARGET, QU, RB, urlbuf, LB, RB);
 693         if (outpos >= sizeof(outbuf) - Offset) {
 694                 syslog(LOG_WARNING, "URL: content longer than buffer!");
 695                 return;
 696         }
 697
 698         TrailerLen = len - (end - start);
 699         if (TrailerLen > 0)
 700                 memcpy(outbuf + Offset + outpos, end, TrailerLen);
 701         if (Offset + outpos + TrailerLen > bufsize) {
 702                 syslog(LOG_WARNING, "URL: content longer than buffer!");
 703                 return;
 704         }
 705         memcpy (buf, outbuf, Offset + outpos + TrailerLen);
 706         *(buf + Offset + outpos + TrailerLen) = '\0';
 707 }
 708