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  *
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU General Public License for more details.
  16  *
  17  *
  18  *
  19  */
  20
  21 #include "webcit.h"
  22 #include "webserver.h"
  23
  24
  25 /*
  26  * Strip surrounding single or double quotes from a string.
  27  */
  28 void stripquotes(char *s)
  29 {
  30         int len;
  31
  32         if (!s) return;
  33
  34         len = strlen(s);
  35         if (len < 2) return;
  36
  37         if ( ( (s[0] == '\"') && (s[len-1] == '\"') ) || ( (s[0] == '\'') && (s[len-1] == '\'') ) ) {
  38                 s[len-1] = 0;
  39                 strcpy(s, &s[1]);
  40         }
  41 }
  42
  43
  44 /*
  45  * Check to see if a META tag has overridden the declared MIME character set.
  46  *
  47  * charset              Character set name (left unchanged if we don't do anything)
  48  * meta_http_equiv      Content of the "http-equiv" portion of the META tag
  49  * meta_content         Content of the "content" portion of the META tag
  50  */
  51 void extract_charset_from_meta(char *charset, char *meta_http_equiv, char *meta_content)
  52 {
  53         char *ptr;
  54         char buf[64];
  55
  56         if (!charset) return;
  57         if (!meta_http_equiv) return;
  58         if (!meta_content) return;
  59
  60
  61         if (strcasecmp(meta_http_equiv, "Content-type")) return;
  62
  63         ptr = strchr(meta_content, ';');
  64         if (!ptr) return;
  65
  66         safestrncpy(buf, ++ptr, sizeof buf);
  67         striplt(buf);
  68         if (!strncasecmp(buf, "charset=", 8)) {
  69                 strcpy(charset, &buf[8]);
  70
  71                 /*
  72                  * The brain-damaged webmail program in Microsoft Exchange declares
  73                  * a charset of "unicode" when they really mean "UTF-8".  GNU iconv
  74                  * treats "unicode" as an alias for "UTF-16" so we have to manually
  75                  * fix this here, otherwise messages generated in Exchange webmail
  76                  * show up as a big pile of weird characters.
  77                  */
  78                 if (!strcasecmp(charset, "unicode")) {
  79                         strcpy(charset, "UTF-8");
  80                 }
  81
  82                 /* Remove wandering punctuation */
  83                 if ((ptr=strchr(charset, '\"'))) *ptr = 0;
  84                 striplt(charset);
  85         }
  86 }
  87
  88
  89
  90 /*
  91  * Sanitize and enhance an HTML message for display.
  92  * Also convert weird character sets to UTF-8 if necessary.
  93  * Also fixup img src="cid:..." type inline images to fetch the image
  94  *
  95  */
  96 void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, StrBuf *Source, StrBuf *Target) {
  97         char buf[SIZ];
  98         char *msg;
  99         char *ptr;
 100         char *msgstart;
 101         char *msgend;
 102         StrBuf *converted_msg;
 103         int buffer_length = 1;
 104         int line_length = 0;
 105         int content_length = 0;
 106         char new_window[SIZ];
 107         int brak = 0;
 108         int alevel = 0;
 109         int scriptlevel = 0;
 110         int script_start_pos = (-1);
 111         int i;
 112         int linklen;
 113         char charset[128];
 114         StrBuf *BodyArea = NULL;
 115 #ifdef HAVE_ICONV
 116         iconv_t ic = (iconv_t)(-1) ;
 117         char *ibuf;                   /* Buffer of characters to be converted */
 118         char *obuf;                   /* Buffer for converted characters      */
 119         size_t ibuflen;               /* Length of input buffer               */
 120         size_t obuflen;               /* Length of output buffer              */
 121         char *osav;                   /* Saved pointer to output buffer       */
 122 #endif
 123         if (Target == NULL)
 124                 Target = WC->WBuf;
 125
 126         safestrncpy(charset, supplied_charset, sizeof charset);
 127         msg = strdup("");
 128         sprintf(new_window, "<a target=\"%s\" href=", TARGET);
 129
 130         if (Source == NULL) while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 131                 line_length = strlen(buf);
 132                 buffer_length = content_length + line_length + 2;
 133                 ptr = realloc(msg, buffer_length);
 134                 if (ptr == NULL) {
 135                         StrBufAppendPrintf(Target, "<b>");
 136                         StrBufAppendPrintf(Target, _("realloc() error! couldn't get %d bytes: %s"),
 137                                         buffer_length + 1,
 138                                         strerror(errno));
 139                         StrBufAppendPrintf(Target, "</b><br><br>\n");
 140                         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 141                                 /** flush */
 142                         }
 143                         free(msg);
 144                         return;
 145                 }
 146                 msg = ptr;
 147                 strcpy(&msg[content_length], buf);
 148                 content_length += line_length;
 149                 strcpy(&msg[content_length], "\n");
 150                 content_length += 1;
 151         }
 152         else {
 153                 content_length = StrLength(Source);
 154                 free(msg);
 155                 msg = (char*) ChrPtr(Source);/* TODO: remove cast */
 156                 buffer_length = content_length;
 157         }
 158
 159         /** Do a first pass to isolate the message body */
 160         ptr = msg + 1;
 161         msgstart = msg;
 162         msgend = &msg[content_length];
 163
 164         while (ptr < msgend) {
 165
 166                 /** Advance to next tag */
 167                 ptr = strchr(ptr, '<');
 168                 if ((ptr == NULL) || (ptr >= msgend)) break;
 169                 ++ptr;
 170                 if ((ptr == NULL) || (ptr >= msgend)) break;
 171
 172                 /*
 173                  *  Look for META tags.  Some messages (particularly in
 174                  *  Asian locales) illegally declare a message's character
 175                  *  set in the HTML instead of in the MIME headers.  This
 176                  *  is wrong but we have to work around it anyway.
 177                  */
 178                 if (!strncasecmp(ptr, "META", 4)) {
 179
 180                         char *meta_start;
 181                         char *meta_end;
 182                         int meta_length;
 183                         char *meta;
 184                         char *meta_http_equiv;
 185                         char *meta_content;
 186                         char *spaceptr;
 187
 188                         meta_start = &ptr[4];
 189                         meta_end = strchr(ptr, '>');
 190                         if ((meta_end != NULL) && (meta_end <= msgend)) {
 191                                 meta_length = meta_end - meta_start + 1;
 192                                 meta = malloc(meta_length + 1);
 193                                 safestrncpy(meta, meta_start, meta_length);
 194                                 meta[meta_length] = 0;
 195                                 striplt(meta);
 196                                 if (!strncasecmp(meta, "HTTP-EQUIV=", 11)) {
 197                                         meta_http_equiv = strdup(&meta[11]);
 198                                         spaceptr = strchr(meta_http_equiv, ' ');
 199                                         if (spaceptr != NULL) {
 200                                                 *spaceptr = 0;
 201                                                 meta_content = strdup(++spaceptr);
 202                                                 if (!strncasecmp(meta_content, "content=", 8)) {
 203                                                         strcpy(meta_content, &meta_content[8]);
 204                                                         stripquotes(meta_http_equiv);
 205                                                         stripquotes(meta_content);
 206                                                         extract_charset_from_meta(charset,
 207                                                                         meta_http_equiv, meta_content);
 208                                                 }
 209                                                 free(meta_content);
 210                                         }
 211                                         free(meta_http_equiv);
 212                                 }
 213                                 free(meta);
 214                         }
 215                 }
 216
 217                 /*
 218                  * Any of these tags cause everything up to and including
 219                  * the tag to be removed.
 220                  */
 221                 if ( (!strncasecmp(ptr, "HTML", 4))
 222                                 ||(!strncasecmp(ptr, "HEAD", 4))
 223                                 ||(!strncasecmp(ptr, "/HEAD", 5))
 224                                 ||(!strncasecmp(ptr, "BODY", 4)) ) {
 225                         char *pBody = NULL;
 226
 227                         if (!strncasecmp(ptr, "BODY", 4)) {
 228                                 pBody = ptr;
 229                         }
 230                         ptr = strchr(ptr, '>');
 231                         if ((ptr == NULL) || (ptr >= msgend)) break;
 232                         if ((pBody != NULL) && (ptr - pBody > 4)) {
 233                                 char* src;
 234                                 char *cid_start, *cid_end;
 235
 236                                 *ptr = '\0';
 237                                 pBody += 4;
 238                                 while ((isspace(*pBody)) && (pBody < ptr))
 239                                         pBody ++;
 240                                 BodyArea = NewStrBufPlain(NULL,  ptr - pBody);
 241
 242                                 if (pBody < ptr) {
 243                                         src = strstr(pBody, "cid:");
 244                                         if (src) {
 245                                                 cid_start = src + 4;
 246                                                 cid_end = cid_start;
 247                                                 while ((*cid_end != '"') &&
 248                                                                 !isspace(*cid_end) &&
 249                                                                 (cid_end < ptr))
 250                                                         cid_end ++;
 251
 252                                                 /* copy tag and attributes up to src="cid: */
 253                                                 StrBufAppendBufPlain(BodyArea, pBody, src - pBody, 0);
 254
 255                                                 /* add in /webcit/mimepart/<msgno>/CID/
 256                                                    trailing / stops dumb URL filters getting excited */
 257                                                 StrBufAppendPrintf(BodyArea,
 258                                                                 "/webcit/mimepart/%d/",msgnum);
 259                                                 StrBufAppendBufPlain(BodyArea, cid_start, cid_end - cid_start, 0);
 260
 261                                                 if (ptr - cid_end > 0)
 262                                                         StrBufAppendBufPlain(BodyArea,
 263                                                                         cid_end + 1,
 264                                                                         ptr - cid_end, 0);
 265                                         }
 266                                         else
 267                                                 StrBufAppendBufPlain(BodyArea, pBody, ptr - pBody, 0);
 268                                 }
 269                                 *ptr = '>';
 270                         }
 271                         ++ptr;
 272                         if ((ptr == NULL) || (ptr >= msgend)) break;
 273                         msgstart = ptr;
 274                 }
 275
 276                 /*
 277                  * Any of these tags cause everything including and following
 278                  * the tag to be removed.
 279                  */
 280                 if ( (!strncasecmp(ptr, "/HTML", 5))
 281                                 ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 282                         --ptr;
 283                         msgend = ptr;
 284                         strcpy(ptr, "");
 285
 286                 }
 287
 288                 ++ptr;
 289         }
 290         if (msgstart > msg) {
 291                 strcpy(msg, msgstart);
 292         }
 293
 294         /* Now go through the message, parsing tags as necessary. */
 295         converted_msg = NewStrBufPlain(NULL, content_length + 8192);
 296
 297
 298         /** Convert foreign character sets to UTF-8 if necessary. */
 299 #ifdef HAVE_ICONV
 300         if ( (strcasecmp(charset, "us-ascii"))
 301                         && (strcasecmp(charset, "UTF-8"))
 302                         && (strcasecmp(charset, ""))
 303            ) {
 304                 syslog(9, "Converting %s to UTF-8\n", charset);
 305                 ctdl_iconv_open("UTF-8", charset, &ic);
 306                 if (ic == (iconv_t)(-1) ) {
 307                         syslog(5, "%s:%d iconv_open() failed: %s\n",
 308                                         __FILE__, __LINE__, strerror(errno));
 309                 }
 310         }
 311         if  (Source == NULL) {
 312                 if (ic != (iconv_t)(-1) ) {
 313                         ibuf = msg;
 314                         ibuflen = content_length;
 315                         obuflen = content_length + (content_length / 2) ;
 316                         obuf = (char *) malloc(obuflen);
 317                         osav = obuf;
 318                         iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
 319                         content_length = content_length + (content_length / 2) - obuflen;
 320                         osav[content_length] = 0;
 321                         free(msg);
 322                         msg = osav;
 323                         iconv_close(ic);
 324                 }
 325         }
 326         else {
 327                 if (ic != (iconv_t)(-1) ) {
 328                         StrBuf *Buf = NewStrBufPlain(NULL, StrLength(Source) + 8096);;
 329                         StrBufConvert(Source, Buf, &ic);
 330                         FreeStrBuf(&Buf);
 331                         iconv_close(ic);
 332                         msg = (char*)ChrPtr(Source); /* TODO: get rid of this. */
 333                 }
 334         }
 335
 336 #endif
 337
 338         /*
 339          *      At this point, the message has been stripped down to
 340          *      only the content inside the <BODY></BODY> tags, and has
 341          *      been converted to UTF-8 if it was originally in a foreign
 342          *      character set.  The text is also guaranteed to be null
 343          *      terminated now.
 344          */
 345
 346         if (converted_msg == NULL) {
 347                 StrBufAppendPrintf(Target, "Error %d: %s<br>%s:%d", errno, strerror(errno), __FILE__, __LINE__);
 348                 goto BAIL;
 349         }
 350
 351         if (BodyArea != NULL) {
 352                 StrBufAppendBufPlain(converted_msg, HKEY("<table "), 0);
 353                 StrBufAppendBuf(converted_msg, BodyArea, 0);
 354                 StrBufAppendBufPlain(converted_msg, HKEY(" width=\"100%\"><tr><td>"), 0);
 355         }
 356         ptr = msg;
 357         msgend = strchr(msg, 0);
 358         while (ptr < msgend) {
 359
 360                 /** Try to sanitize the html of any rogue scripts */
 361                 if (!strncasecmp(ptr, "<script", 7)) {
 362                         if (scriptlevel == 0) {
 363                                 script_start_pos = StrLength(converted_msg);
 364                         }
 365                         ++scriptlevel;
 366                 }
 367                 if (!strncasecmp(ptr, "</script", 8)) {
 368                         --scriptlevel;
 369                 }
 370
 371                 /**
 372                  * Change mailto: links to WebCit mail, by replacing the
 373                  * link with one that points back to our mail room.  Due to
 374                  * the way we parse URL's, it'll even handle mailto: links
 375                  * that have "?subject=" in them.
 376                  */
 377                 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
 378                         content_length += 64;
 379                         StrBufAppendPrintf(converted_msg,
 380                                         "<a href=\"display_enter?force_room=_MAIL_?recp=");
 381                         ptr = &ptr[16];
 382                         ++alevel;
 383                         ++brak;
 384                 }
 385                 /** Make external links open in a separate window */
 386                 else if (!strncasecmp(ptr, "<a href=\"", 9)) {
 387                         ++alevel;
 388                         ++brak;
 389                         if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
 390                                         &&  ((strchr(ptr, '/') < strchr(ptr, '>')))
 391                            ) {
 392                                 /* open external links to new window */
 393                                 StrBufAppendPrintf(converted_msg, new_window);
 394                                 ptr = &ptr[8];
 395                         }
 396                         else if (
 397                                 (treat_as_wiki)
 398                                 && (strncasecmp(ptr, "<a href=\"wiki?", 14))
 399                                 && (strncasecmp(ptr, "<a href=\"dotgoto?", 17))
 400                                 && (strncasecmp(ptr, "<a href=\"knrooms?", 17))
 401                         ) {
 402                                 content_length += 64;
 403                                 StrBufAppendPrintf(converted_msg, "<a href=\"wiki?go=");
 404                                 StrBufUrlescAppend(converted_msg, WC->CurRoom.name, NULL);
 405                                 StrBufAppendPrintf(converted_msg, "?page=");
 406                                 ptr = &ptr[9];
 407                         }
 408                         else {
 409                                 StrBufAppendPrintf(converted_msg, "<a href=\"");
 410                                 ptr = &ptr[9];
 411                         }
 412                 }
 413                 /** Fixup <img src="cid:... ...> to fetch the mime part */
 414                 else if (!strncasecmp(ptr, "<img ", 5)) {
 415                         char *cid_start, *cid_end;
 416                         char* tag_end=strchr(ptr,'>');
 417                         char* src;
 418                         /* FIXME - handle this situation (maybe someone opened an <img cid...
 419                          * and then ended the message)
 420                          */
 421                         if (!tag_end) {
 422                                 syslog(9, "tag_end is null and ptr is:\n");
 423                                 syslog(9, "%s\n", ptr);
 424                                 syslog(9, "Theoretical bytes remaining: %d\n", (int)(msgend - ptr));
 425                         }
 426
 427                         src=strstr(ptr, "src=\"cid:");
 428                         ++brak;
 429
 430                         if (src
 431                             && isspace(*(src-1))
 432                                 && tag_end
 433                                 && (cid_start=strchr(src,':'))
 434                                 && (cid_end=strchr(cid_start,'"'))
 435                                 && (cid_end < tag_end)
 436                         ) {
 437                                 /* copy tag and attributes up to src="cid: */
 438                                 StrBufAppendBufPlain(converted_msg, ptr, src - ptr, 0);
 439                                 cid_start++;
 440
 441                                 /* add in /webcit/mimepart/<msgno>/CID/
 442                                    trailing / stops dumb URL filters getting excited */
 443                                 StrBufAppendPrintf(converted_msg,
 444                                                 " src=\"/webcit/mimepart/%d/",msgnum);
 445                                 StrBufAppendBufPlain(converted_msg, cid_start, cid_end - cid_start, 0);
 446                                 StrBufAppendBufPlain(converted_msg, "/\"", -1, 0);
 447
 448                                 ptr = cid_end+1;
 449                         }
 450                         StrBufAppendBufPlain(converted_msg, ptr, tag_end - ptr, 0);
 451                         ptr = tag_end;
 452                 }
 453
 454                 /**
 455                  * Turn anything that looks like a URL into a real link, as long
 456                  * as it's not inside a tag already
 457                  */
 458                 else if ( (brak == 0) && (alevel == 0)
 459                      && (!strncasecmp(ptr, "http://", 7))) {
 460                                 /** Find the end of the link */
 461                                 int strlenptr;
 462                                 linklen = 0;
 463
 464                                 strlenptr = strlen(ptr);
 465                                 for (i=0; i<=strlenptr; ++i) {
 466                                         if ((ptr[i]==0)
 467                                            ||(isspace(ptr[i]))
 468                                            ||(ptr[i]==10)
 469                                            ||(ptr[i]==13)
 470                                            ||(ptr[i]=='(')
 471                                            ||(ptr[i]==')')
 472                                            ||(ptr[i]=='<')
 473                                            ||(ptr[i]=='>')
 474                                            ||(ptr[i]=='[')
 475                                            ||(ptr[i]==']')
 476                                            ||(ptr[i]=='"')
 477                                            ||(ptr[i]=='\'')
 478                                         ) linklen = i;
 479                                         /* did s.b. send us an entity? */
 480                                         if (ptr[i] == '&') {
 481                                                 if ((ptr[i+2] ==';') ||
 482                                                     (ptr[i+3] ==';') ||
 483                                                     (ptr[i+5] ==';') ||
 484                                                     (ptr[i+6] ==';') ||
 485                                                     (ptr[i+7] ==';'))
 486                                                         linklen = i;
 487                                         }
 488                                         if (linklen > 0) break;
 489                                 }
 490                                 if (linklen > 0) {
 491                                         char *ltreviewptr;
 492                                         char *nbspreviewptr;
 493                                         char linkedchar;
 494                                         int len;
 495
 496                                         len = linklen;
 497                                         linkedchar = ptr[len];
 498                                         ptr[len] = '\0';
 499                                         /* spot for some subject strings tinymce tends to give us. */
 500                                         ltreviewptr = strchr(ptr, '<');
 501                                         if (ltreviewptr != NULL) {
 502                                                 *ltreviewptr = '\0';
 503                                                 linklen = ltreviewptr - ptr;
 504                                         }
 505
 506                                         nbspreviewptr = strstr(ptr, "&nbsp;");
 507                                         if (nbspreviewptr != NULL) {
 508                                                 /* nbspreviewptr = '\0'; */
 509                                                 linklen = nbspreviewptr - ptr;
 510                                         }
 511                                         if (ltreviewptr != 0)
 512                                                 *ltreviewptr = '<';
 513
 514                                         ptr[len] = linkedchar;
 515
 516                                         content_length += (32 + linklen);
 517                                         StrBufAppendPrintf(converted_msg, "%s\"", new_window);
 518                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 519                                         StrBufAppendPrintf(converted_msg, "\">");
 520                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 521                                         ptr += linklen;
 522                                         StrBufAppendPrintf(converted_msg, "</A>");
 523                                 }
 524                 }
 525                 else {
 526                         StrBufAppendBufPlain(converted_msg, ptr, 1, 0);
 527                         ptr++;
 528                 }
 529
 530
 531                 if ((ptr >= msg) && (ptr <= msgend)) {
 532                         /*
 533                          * We need to know when we're inside a tag,
 534                          * so we don't turn things that look like URL's into
 535                          * links, when they're already links - or image sources.
 536                          */
 537                         if ((ptr > msg) && (*(ptr-1) == '<')) {
 538                                 ++brak;
 539                         }
 540                         if ((ptr > msg) && (*(ptr-1) == '>')) {
 541                                 --brak;
 542                                 if ((scriptlevel == 0) && (script_start_pos >= 0)) {
 543                                         StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
 544                                         script_start_pos = (-1);
 545                                 }
 546                         }
 547                         if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 548                 }
 549         }
 550
 551         if (BodyArea != NULL) {
 552                 StrBufAppendBufPlain(converted_msg, HKEY("</td></tr></table>"), 0);
 553                 FreeStrBuf(&BodyArea);
 554         }
 555
 556         /**     uncomment these two lines to override conversion        */
 557         /**     memcpy(converted_msg, msg, content_length);             */
 558         /**     output_length = content_length;                         */
 559
 560         /** Output our big pile of markup */
 561         StrBufAppendBuf(Target, converted_msg, 0);
 562
 563 BAIL:   /** A little trailing vertical whitespace... */
 564         StrBufAppendPrintf(Target, "<br><br>\n");
 565
 566         /** Now give back the memory */
 567         FreeStrBuf(&converted_msg);
 568         if ((msg != NULL) && (Source == NULL)) free(msg);
 569 }
 570
 571
 572
 573
 574
 575
 576 /*
 577  * Look for URL's embedded in a buffer and make them linkable.  We use a
 578  * target window in order to keep the Citadel session in its own window.
 579  */
 580 void UrlizeText(StrBuf* Target, StrBuf *Source, StrBuf *WrkBuf)
 581 {
 582         int len, UrlLen, Offset, TrailerLen;
 583         const char *start, *end, *pos;
 584
 585         FlushStrBuf(Target);
 586
 587         start = NULL;
 588         len = StrLength(Source);
 589         end = ChrPtr(Source) + len;
 590         for (pos = ChrPtr(Source); (pos < end) && (start == NULL); ++pos) {
 591                 if (!strncasecmp(pos, "http://", 7))
 592                         start = pos;
 593                 else if (!strncasecmp(pos, "ftp://", 6))
 594                         start = pos;
 595         }
 596
 597         if (start == NULL) {
 598                 StrBufAppendBuf(Target, Source, 0);
 599                 return;
 600         }
 601         FlushStrBuf(WrkBuf);
 602
 603         for (pos = ChrPtr(Source) + len; pos > start; --pos) {
 604                 if (  (!isprint(*pos))
 605                    || (isspace(*pos))
 606                    || (*pos == '{')
 607                    || (*pos == '}')
 608                    || (*pos == '|')
 609                    || (*pos == '\\')
 610                    || (*pos == '^')
 611                    || (*pos == '[')
 612                    || (*pos == ']')
 613                    || (*pos == '`')
 614                    || (*pos == '<')
 615                    || (*pos == '>')
 616                    || (*pos == '(')
 617                    || (*pos == ')')
 618                 ) {
 619                         end = pos;
 620                 }
 621         }
 622
 623         UrlLen = end - start;
 624         StrBufAppendBufPlain(WrkBuf, start, UrlLen, 0);
 625
 626         Offset = start - ChrPtr(Source);
 627         if (Offset != 0)
 628                 StrBufAppendBufPlain(Target, ChrPtr(Source), Offset, 0);
 629         StrBufAppendPrintf(Target, "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
 630                            LB, QU, ChrPtr(WrkBuf), QU, QU, TARGET,
 631                            QU, RB, ChrPtr(WrkBuf), LB, RB);
 632
 633         TrailerLen = StrLength(Source) - (end - ChrPtr(Source));
 634         if (TrailerLen > 0)
 635                 StrBufAppendBufPlain(Target, end, TrailerLen, 0);
 636 }
 637
 638
 639 void url(char *buf, size_t bufsize)
 640 {
 641         int len, UrlLen, Offset, TrailerLen, outpos;
 642         char *start, *end, *pos;
 643         char urlbuf[SIZ];
 644         char outbuf[SIZ];
 645
 646         start = NULL;
 647         len = strlen(buf);
 648         if (len > bufsize) {
 649                 syslog(1, "URL: content longer than buffer!");
 650                 return;
 651         }
 652         end = buf + len;
 653         for (pos = buf; (pos < end) && (start == NULL); ++pos) {
 654                 if (!strncasecmp(pos, "http://", 7))
 655                         start = pos;
 656                 if (!strncasecmp(pos, "ftp://", 6))
 657                         start = pos;
 658         }
 659
 660         if (start == NULL)
 661                 return;
 662
 663         for (pos = buf+len; pos > start; --pos) {
 664                 if (  (!isprint(*pos))
 665                    || (isspace(*pos))
 666                    || (*pos == '{')
 667                    || (*pos == '}')
 668                    || (*pos == '|')
 669                    || (*pos == '\\')
 670                    || (*pos == '^')
 671                    || (*pos == '[')
 672                    || (*pos == ']')
 673                    || (*pos == '`')
 674                    || (*pos == '<')
 675                    || (*pos == '>')
 676                    || (*pos == '(')
 677                    || (*pos == ')')
 678                 ) {
 679                         end = pos;
 680                 }
 681         }
 682
 683         UrlLen = end - start;
 684         if (UrlLen > sizeof(urlbuf)){
 685                 syslog(1, "URL: content longer than buffer!");
 686                 return;
 687         }
 688         memcpy(urlbuf, start, UrlLen);
 689         urlbuf[UrlLen] = '\0';
 690
 691         Offset = start - buf;
 692         if ((Offset != 0) && (Offset < sizeof(outbuf)))
 693                 memcpy(outbuf, buf, Offset);
 694         outpos = snprintf(&outbuf[Offset], sizeof(outbuf) - Offset,
 695                           "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
 696                           LB, QU, urlbuf, QU, QU, TARGET, QU, RB, urlbuf, LB, RB);
 697         if (outpos >= sizeof(outbuf) - Offset) {
 698                 syslog(1, "URL: content longer than buffer!");
 699                 return;
 700         }
 701
 702         TrailerLen = len - (end - start);
 703         if (TrailerLen > 0)
 704                 memcpy(outbuf + Offset + outpos, end, TrailerLen);
 705         if (Offset + outpos + TrailerLen > bufsize) {
 706                 syslog(1, "URL: content longer than buffer!");
 707                 return;
 708         }
 709         memcpy (buf, outbuf, Offset + outpos + TrailerLen);
 710         *(buf + Offset + outpos + TrailerLen) = '\0';
 711 }
 712