webcit/html2html.c

   1 /*
   2  * $Id$
   3  *
   4  * Output an HTML message, modifying it slightly to make sure it plays nice
   5  * with the rest of our web framework.
   6  *
   7  * Copyright (c) 2005-2010 by the citadel.org team
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 3 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License
  20  * along with this program; if not, write to the Free Software
  21  */
  22
  23 #include "webcit.h"
  24 #include "webserver.h"
  25
  26
  27 /*
  28  * Strip surrounding single or double quotes from a string.
  29  */
  30 void stripquotes(char *s)
  31 {
  32         int len;
  33
  34         if (!s) return;
  35
  36         len = strlen(s);
  37         if (len < 2) return;
  38
  39         if ( ( (s[0] == '\"') && (s[len-1] == '\"') ) || ( (s[0] == '\'') && (s[len-1] == '\'') ) ) {
  40                 s[len-1] = 0;
  41                 strcpy(s, &s[1]);
  42         }
  43 }
  44
  45
  46 /*
  47  * Check to see if a META tag has overridden the declared MIME character set.
  48  *
  49  * charset              Character set name (left unchanged if we don't do anything)
  50  * meta_http_equiv      Content of the "http-equiv" portion of the META tag
  51  * meta_content         Content of the "content" portion of the META tag
  52  */
  53 void extract_charset_from_meta(char *charset, char *meta_http_equiv, char *meta_content)
  54 {
  55         char *ptr;
  56         char buf[64];
  57
  58         if (!charset) return;
  59         if (!meta_http_equiv) return;
  60         if (!meta_content) return;
  61
  62
  63         if (strcasecmp(meta_http_equiv, "Content-type")) return;
  64
  65         ptr = strchr(meta_content, ';');
  66         if (!ptr) return;
  67
  68         safestrncpy(buf, ++ptr, sizeof buf);
  69         striplt(buf);
  70         if (!strncasecmp(buf, "charset=", 8)) {
  71                 strcpy(charset, &buf[8]);
  72
  73                 /*
  74                  * The brain-damaged webmail program in Microsoft Exchange declares
  75                  * a charset of "unicode" when they really mean "UTF-8".  GNU iconv
  76                  * treats "unicode" as an alias for "UTF-16" so we have to manually
  77                  * fix this here, otherwise messages generated in Exchange webmail
  78                  * show up as a big pile of weird characters.
  79                  */
  80                 if (!strcasecmp(charset, "unicode")) {
  81                         strcpy(charset, "UTF-8");
  82                 }
  83
  84                 /* Remove wandering punctuation */
  85                 if ((ptr=strchr(charset, '\"'))) *ptr = 0;
  86                 striplt(charset);
  87         }
  88 }
  89
  90
  91
  92 /*
  93  * Sanitize and enhance an HTML message for display.
  94  * Also convert weird character sets to UTF-8 if necessary.
  95  * Also fixup img src="cid:..." type inline images to fetch the image
  96  *
  97  */
  98 void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, StrBuf *Source, StrBuf *Target) {
  99         char buf[SIZ];
 100         char *msg;
 101         char *ptr;
 102         char *msgstart;
 103         char *msgend;
 104         StrBuf *converted_msg;
 105         int buffer_length = 1;
 106         int line_length = 0;
 107         int content_length = 0;
 108         char new_window[SIZ];
 109         int brak = 0;
 110         int alevel = 0;
 111         int scriptlevel = 0;
 112         int script_start_pos = (-1);
 113         int i;
 114         int linklen;
 115         char charset[128];
 116         StrBuf *BodyArea = NULL;
 117 #ifdef HAVE_ICONV
 118         iconv_t ic = (iconv_t)(-1) ;
 119         char *ibuf;                   /* Buffer of characters to be converted */
 120         char *obuf;                   /* Buffer for converted characters      */
 121         size_t ibuflen;               /* Length of input buffer               */
 122         size_t obuflen;               /* Length of output buffer              */
 123         char *osav;                   /* Saved pointer to output buffer       */
 124 #endif
 125         if (Target == NULL)
 126                 Target = WC->WBuf;
 127
 128         safestrncpy(charset, supplied_charset, sizeof charset);
 129         msg = strdup("");
 130         sprintf(new_window, "<a target=\"%s\" href=", TARGET);
 131
 132         if (Source == NULL) while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 133                 line_length = strlen(buf);
 134                 buffer_length = content_length + line_length + 2;
 135                 ptr = realloc(msg, buffer_length);
 136                 if (ptr == NULL) {
 137                         StrBufAppendPrintf(Target, "<b>");
 138                         StrBufAppendPrintf(Target, _("realloc() error! couldn't get %d bytes: %s"),
 139                                         buffer_length + 1,
 140                                         strerror(errno));
 141                         StrBufAppendPrintf(Target, "</b><br /><br />\n");
 142                         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 143                                 /** flush */
 144                         }
 145                         free(msg);
 146                         return;
 147                 }
 148                 msg = ptr;
 149                 strcpy(&msg[content_length], buf);
 150                 content_length += line_length;
 151                 strcpy(&msg[content_length], "\n");
 152                 content_length += 1;
 153         }
 154         else {
 155                 content_length = StrLength(Source);
 156                 free(msg);
 157                 msg = (char*) ChrPtr(Source);/* TODO: remove cast */
 158                 buffer_length = content_length;
 159         }
 160
 161         /** Do a first pass to isolate the message body */
 162         ptr = msg + 1;
 163         msgstart = msg;
 164         msgend = &msg[content_length];
 165
 166         while (ptr < msgend) {
 167
 168                 /** Advance to next tag */
 169                 ptr = strchr(ptr, '<');
 170                 if ((ptr == NULL) || (ptr >= msgend)) break;
 171                 ++ptr;
 172                 if ((ptr == NULL) || (ptr >= msgend)) break;
 173
 174                 /*
 175                  *  Look for META tags.  Some messages (particularly in
 176                  *  Asian locales) illegally declare a message's character
 177                  *  set in the HTML instead of in the MIME headers.  This
 178                  *  is wrong but we have to work around it anyway.
 179                  */
 180                 if (!strncasecmp(ptr, "META", 4)) {
 181
 182                         char *meta_start;
 183                         char *meta_end;
 184                         int meta_length;
 185                         char *meta;
 186                         char *meta_http_equiv;
 187                         char *meta_content;
 188                         char *spaceptr;
 189
 190                         meta_start = &ptr[4];
 191                         meta_end = strchr(ptr, '>');
 192                         if ((meta_end != NULL) && (meta_end <= msgend)) {
 193                                 meta_length = meta_end - meta_start + 1;
 194                                 meta = malloc(meta_length + 1);
 195                                 safestrncpy(meta, meta_start, meta_length);
 196                                 meta[meta_length] = 0;
 197                                 striplt(meta);
 198                                 if (!strncasecmp(meta, "HTTP-EQUIV=", 11)) {
 199                                         meta_http_equiv = strdup(&meta[11]);
 200                                         spaceptr = strchr(meta_http_equiv, ' ');
 201                                         if (spaceptr != NULL) {
 202                                                 *spaceptr = 0;
 203                                                 meta_content = strdup(++spaceptr);
 204                                                 if (!strncasecmp(meta_content, "content=", 8)) {
 205                                                         strcpy(meta_content, &meta_content[8]);
 206                                                         stripquotes(meta_http_equiv);
 207                                                         stripquotes(meta_content);
 208                                                         extract_charset_from_meta(charset,
 209                                                                         meta_http_equiv, meta_content);
 210                                                 }
 211                                                 free(meta_content);
 212                                         }
 213                                         free(meta_http_equiv);
 214                                 }
 215                                 free(meta);
 216                         }
 217                 }
 218
 219                 /*
 220                  * Any of these tags cause everything up to and including
 221                  * the tag to be removed.
 222                  */
 223                 if ( (!strncasecmp(ptr, "HTML", 4))
 224                                 ||(!strncasecmp(ptr, "HEAD", 4))
 225                                 ||(!strncasecmp(ptr, "/HEAD", 5))
 226                                 ||(!strncasecmp(ptr, "BODY", 4)) ) {
 227                         char *pBody = NULL;
 228
 229                         if (!strncasecmp(ptr, "BODY", 4)) {
 230                                 pBody = ptr;
 231                         }
 232                         ptr = strchr(ptr, '>');
 233                         if ((ptr == NULL) || (ptr >= msgend)) break;
 234                         if ((pBody != NULL) && (ptr - pBody > 4)) {
 235                                 char* src;
 236                                 char *cid_start, *cid_end;
 237
 238                                 *ptr = '\0';
 239                                 pBody += 4;
 240                                 while ((isspace(*pBody)) && (pBody < ptr))
 241                                         pBody ++;
 242                                 BodyArea = NewStrBufPlain(NULL,  ptr - pBody);
 243
 244                                 if (pBody < ptr) {
 245                                         src = strstr(pBody, "cid:");
 246                                         if (src) {
 247                                                 cid_start = src + 4;
 248                                                 cid_end = cid_start;
 249                                                 while ((*cid_end != '"') &&
 250                                                                 !isspace(*cid_end) &&
 251                                                                 (cid_end < ptr))
 252                                                         cid_end ++;
 253
 254                                                 /* copy tag and attributes up to src="cid: */
 255                                                 StrBufAppendBufPlain(BodyArea, pBody, src - pBody, 0);
 256
 257                                                 /* add in /webcit/mimepart/<msgno>/CID/
 258                                                    trailing / stops dumb URL filters getting excited */
 259                                                 StrBufAppendPrintf(BodyArea,
 260                                                                 "/webcit/mimepart/%d/",msgnum);
 261                                                 StrBufAppendBufPlain(BodyArea, cid_start, cid_end - cid_start, 0);
 262
 263                                                 if (ptr - cid_end > 0)
 264                                                         StrBufAppendBufPlain(BodyArea,
 265                                                                         cid_end + 1,
 266                                                                         ptr - cid_end, 0);
 267                                         }
 268                                         else
 269                                                 StrBufAppendBufPlain(BodyArea, pBody, ptr - pBody, 0);
 270                                 }
 271                                 *ptr = '>';
 272                         }
 273                         ++ptr;
 274                         if ((ptr == NULL) || (ptr >= msgend)) break;
 275                         msgstart = ptr;
 276                 }
 277
 278                 /*
 279                  * Any of these tags cause everything including and following
 280                  * the tag to be removed.
 281                  */
 282                 if ( (!strncasecmp(ptr, "/HTML", 5))
 283                                 ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 284                         --ptr;
 285                         msgend = ptr;
 286                         strcpy(ptr, "");
 287
 288                 }
 289
 290                 ++ptr;
 291         }
 292         if (msgstart > msg) {
 293                 strcpy(msg, msgstart);
 294         }
 295
 296         /* Now go through the message, parsing tags as necessary. */
 297         converted_msg = NewStrBufPlain(NULL, content_length + 8192);
 298
 299
 300         /** Convert foreign character sets to UTF-8 if necessary. */
 301 #ifdef HAVE_ICONV
 302         if ( (strcasecmp(charset, "us-ascii"))
 303                         && (strcasecmp(charset, "UTF-8"))
 304                         && (strcasecmp(charset, ""))
 305            ) {
 306                 lprintf(9, "Converting %s to UTF-8\n", charset);
 307                 ctdl_iconv_open("UTF-8", charset, &ic);
 308                 if (ic == (iconv_t)(-1) ) {
 309                         lprintf(5, "%s:%d iconv_open() failed: %s\n",
 310                                         __FILE__, __LINE__, strerror(errno));
 311                 }
 312         }
 313         if  (Source == NULL) {
 314                 if (ic != (iconv_t)(-1) ) {
 315                         ibuf = msg;
 316                         ibuflen = content_length;
 317                         obuflen = content_length + (content_length / 2) ;
 318                         obuf = (char *) malloc(obuflen);
 319                         osav = obuf;
 320                         iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
 321                         content_length = content_length + (content_length / 2) - obuflen;
 322                         osav[content_length] = 0;
 323                         free(msg);
 324                         msg = osav;
 325                         iconv_close(ic);
 326                 }
 327         }
 328         else {
 329                 if (ic != (iconv_t)(-1) ) {
 330                         StrBuf *Buf = NewStrBufPlain(NULL, StrLength(Source) + 8096);;
 331                         StrBufConvert(Source, Buf, &ic);
 332                         FreeStrBuf(&Buf);
 333                         iconv_close(ic);
 334                         msg = (char*)ChrPtr(Source); /* TODO: get rid of this. */
 335                 }
 336         }
 337
 338 #endif
 339
 340         /*
 341          *      At this point, the message has been stripped down to
 342          *      only the content inside the <BODY></BODY> tags, and has
 343          *      been converted to UTF-8 if it was originally in a foreign
 344          *      character set.  The text is also guaranteed to be null
 345          *      terminated now.
 346          */
 347
 348         if (converted_msg == NULL) {
 349                 StrBufAppendPrintf(Target, "Error %d: %s<br />%s:%d", errno, strerror(errno), __FILE__, __LINE__);
 350                 goto BAIL;
 351         }
 352
 353         if (BodyArea != NULL) {
 354                 StrBufAppendBufPlain(converted_msg, HKEY("<table "), 0);
 355                 StrBufAppendBuf(converted_msg, BodyArea, 0);
 356                 StrBufAppendBufPlain(converted_msg, HKEY(" width=\"100%\"><tr><td>"), 0);
 357         }
 358         ptr = msg;
 359         msgend = strchr(msg, 0);
 360         while (ptr < msgend) {
 361
 362                 /** Try to sanitize the html of any rogue scripts */
 363                 if (!strncasecmp(ptr, "<script", 7)) {
 364                         if (scriptlevel == 0) {
 365                                 script_start_pos = StrLength(converted_msg);
 366                         }
 367                         ++scriptlevel;
 368                 }
 369                 if (!strncasecmp(ptr, "</script", 8)) {
 370                         --scriptlevel;
 371                 }
 372
 373                 /**
 374                  * Change mailto: links to WebCit mail, by replacing the
 375                  * link with one that points back to our mail room.  Due to
 376                  * the way we parse URL's, it'll even handle mailto: links
 377                  * that have "?subject=" in them.
 378                  */
 379                 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
 380                         content_length += 64;
 381                         StrBufAppendPrintf(converted_msg,
 382                                         "<a href=\"display_enter?force_room=_MAIL_?recp=");
 383                         ptr = &ptr[16];
 384                         ++alevel;
 385                         ++brak;
 386                 }
 387                 /** Make external links open in a separate window */
 388                 else if (!strncasecmp(ptr, "<a href=\"", 9)) {
 389                         ++alevel;
 390                         ++brak;
 391                         if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
 392                                         &&  ((strchr(ptr, '/') < strchr(ptr, '>')))
 393                            ) {
 394                                 /* open external links to new window */
 395                                 StrBufAppendPrintf(converted_msg, new_window);
 396                                 ptr = &ptr[8];
 397                         }
 398                         else if ( (treat_as_wiki) && (strncasecmp(ptr, "<a href=\"wiki?", 14)) ) {
 399                                 content_length += 64;
 400                                 StrBufAppendPrintf(converted_msg, "<a href=\"wiki?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                                 lprintf(9, "tag_end is null and ptr is:\n");
 418                                 lprintf(9, "%s\n", ptr);
 419                                 lprintf(9, "Theoretical bytes remaining: %d\n", 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                                 /** Find the end of the link */
 456                                 int strlenptr;
 457                                 linklen = 0;
 458
 459                                 strlenptr = strlen(ptr);
 460                                 for (i=0; i<=strlenptr; ++i) {
 461                                         if ((ptr[i]==0)
 462                                            ||(isspace(ptr[i]))
 463                                            ||(ptr[i]==10)
 464                                            ||(ptr[i]==13)
 465                                            ||(ptr[i]=='(')
 466                                            ||(ptr[i]==')')
 467                                            ||(ptr[i]=='<')
 468                                            ||(ptr[i]=='>')
 469                                            ||(ptr[i]=='[')
 470                                            ||(ptr[i]==']')
 471                                            ||(ptr[i]=='"')
 472                                            ||(ptr[i]=='\'')
 473                                         ) linklen = i;
 474                                         /* did s.b. send us an entity? */
 475                                         if (ptr[i] == '&') {
 476                                                 if ((ptr[i+2] ==';') ||
 477                                                     (ptr[i+3] ==';') ||
 478                                                     (ptr[i+5] ==';') ||
 479                                                     (ptr[i+6] ==';') ||
 480                                                     (ptr[i+7] ==';'))
 481                                                         linklen = i;
 482                                         }
 483                                         if (linklen > 0) break;
 484                                 }
 485                                 if (linklen > 0) {
 486                                         char *ltreviewptr;
 487                                         char *nbspreviewptr;
 488                                         char linkedchar;
 489                                         int len;
 490
 491                                         len = linklen;
 492                                         linkedchar = ptr[len];
 493                                         ptr[len] = '\0';
 494                                         /* spot for some subject strings tinymce tends to give us. */
 495                                         ltreviewptr = strchr(ptr, '<');
 496                                         if (ltreviewptr != NULL) {
 497                                                 *ltreviewptr = '\0';
 498                                                 linklen = ltreviewptr - ptr;
 499                                         }
 500
 501                                         nbspreviewptr = strstr(ptr, "&nbsp;");
 502                                         if (nbspreviewptr != NULL) {
 503                                                 /* nbspreviewptr = '\0'; */
 504                                                 linklen = nbspreviewptr - ptr;
 505                                         }
 506                                         if (ltreviewptr != 0)
 507                                                 *ltreviewptr = '<';
 508
 509                                         ptr[len] = linkedchar;
 510
 511                                         content_length += (32 + linklen);
 512                                         StrBufAppendPrintf(converted_msg, "%s\"", new_window);
 513                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 514                                         StrBufAppendPrintf(converted_msg, "\">");
 515                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 516                                         ptr += linklen;
 517                                         StrBufAppendPrintf(converted_msg, "</A>");
 518                                 }
 519                 }
 520                 else {
 521                         StrBufAppendBufPlain(converted_msg, ptr, 1, 0);
 522                         ptr++;
 523                 }
 524
 525
 526                 if ((ptr >= msg) && (ptr <= msgend)) {
 527                         /*
 528                          * We need to know when we're inside a tag,
 529                          * so we don't turn things that look like URL's into
 530                          * links, when they're already links - or image sources.
 531                          */
 532                         if ((ptr > msg) && (*(ptr-1) == '<')) {
 533                                 ++brak;
 534                         }
 535                         if ((ptr > msg) && (*(ptr-1) == '>')) {
 536                                 --brak;
 537                                 if ((scriptlevel == 0) && (script_start_pos >= 0)) {
 538                                         StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
 539                                         script_start_pos = (-1);
 540                                 }
 541                         }
 542                         if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 543                 }
 544         }
 545
 546         if (BodyArea != NULL) {
 547                 StrBufAppendBufPlain(converted_msg, HKEY("</td></tr></table>"), 0);
 548                 FreeStrBuf(&BodyArea);
 549         }
 550
 551         /**     uncomment these two lines to override conversion        */
 552         /**     memcpy(converted_msg, msg, content_length);             */
 553         /**     output_length = content_length;                         */
 554
 555         /** Output our big pile of markup */
 556         StrBufAppendBuf(Target, converted_msg, 0);
 557
 558 BAIL:   /** A little trailing vertical whitespace... */
 559         StrBufAppendPrintf(Target, "<br /><br />\n");
 560
 561         /** Now give back the memory */
 562         FreeStrBuf(&converted_msg);
 563         if ((msg != NULL) && (Source == NULL)) free(msg);
 564 }
 565
 566
 567
 568
 569
 570
 571 /*
 572  * Look for URL's embedded in a buffer and make them linkable.  We use a
 573  * target window in order to keep the Citadel session in its own window.
 574  */
 575 void UrlizeText(StrBuf* Target, StrBuf *Source, StrBuf *WrkBuf)
 576 {
 577         int len, UrlLen, Offset, TrailerLen;
 578         const char *start, *end, *pos;
 579
 580         FlushStrBuf(Target);
 581
 582         start = NULL;
 583         len = StrLength(Source);
 584         end = ChrPtr(Source) + len;
 585         for (pos = ChrPtr(Source); (pos < end) && (start == NULL); ++pos) {
 586                 if (!strncasecmp(pos, "http://", 7))
 587                         start = pos;
 588                 else if (!strncasecmp(pos, "ftp://", 6))
 589                         start = pos;
 590         }
 591
 592         if (start == NULL) {
 593                 StrBufAppendBuf(Target, Source, 0);
 594                 return;
 595         }
 596         FlushStrBuf(WrkBuf);
 597
 598         for (pos = ChrPtr(Source) + len; pos > start; --pos) {
 599                 if (  (!isprint(*pos))
 600                    || (isspace(*pos))
 601                    || (*pos == '{')
 602                    || (*pos == '}')
 603                    || (*pos == '|')
 604                    || (*pos == '\\')
 605                    || (*pos == '^')
 606                    || (*pos == '[')
 607                    || (*pos == ']')
 608                    || (*pos == '`')
 609                    || (*pos == '<')
 610                    || (*pos == '>')
 611                    || (*pos == '(')
 612                    || (*pos == ')')
 613                 ) {
 614                         end = pos;
 615                 }
 616         }
 617
 618         UrlLen = end - start;
 619         StrBufAppendBufPlain(WrkBuf, start, UrlLen, 0);
 620
 621         Offset = start - ChrPtr(Source);
 622         if (Offset != 0)
 623                 StrBufAppendBufPlain(Target, ChrPtr(Source), Offset, 0);
 624         StrBufAppendPrintf(Target, "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
 625                            LB, QU, ChrPtr(WrkBuf), QU, QU, TARGET,
 626                            QU, RB, ChrPtr(WrkBuf), LB, RB);
 627
 628         TrailerLen = StrLength(Source) - (end - ChrPtr(Source));
 629         if (TrailerLen > 0)
 630                 StrBufAppendBufPlain(Target, end, TrailerLen, 0);
 631 }
 632
 633
 634 void url(char *buf, size_t bufsize)
 635 {
 636         int len, UrlLen, Offset, TrailerLen, outpos;
 637         char *start, *end, *pos;
 638         char urlbuf[SIZ];
 639         char outbuf[SIZ];
 640
 641         start = NULL;
 642         len = strlen(buf);
 643         if (len > bufsize) {
 644                 lprintf(1, "URL: content longer than buffer!");
 645                 return;
 646         }
 647         end = buf + len;
 648         for (pos = buf; (pos < end) && (start == NULL); ++pos) {
 649                 if (!strncasecmp(pos, "http://", 7))
 650                         start = pos;
 651                 if (!strncasecmp(pos, "ftp://", 6))
 652                         start = pos;
 653         }
 654
 655         if (start == NULL)
 656                 return;
 657
 658         for (pos = buf+len; pos > start; --pos) {
 659                 if (  (!isprint(*pos))
 660                    || (isspace(*pos))
 661                    || (*pos == '{')
 662                    || (*pos == '}')
 663                    || (*pos == '|')
 664                    || (*pos == '\\')
 665                    || (*pos == '^')
 666                    || (*pos == '[')
 667                    || (*pos == ']')
 668                    || (*pos == '`')
 669                    || (*pos == '<')
 670                    || (*pos == '>')
 671                    || (*pos == '(')
 672                    || (*pos == ')')
 673                 ) {
 674                         end = pos;
 675                 }
 676         }
 677
 678         UrlLen = end - start;
 679         if (UrlLen > sizeof(urlbuf)){
 680                 lprintf(1, "URL: content longer than buffer!");
 681                 return;
 682         }
 683         memcpy(urlbuf, start, UrlLen);
 684         urlbuf[UrlLen] = '\0';
 685
 686         Offset = start - buf;
 687         if ((Offset != 0) && (Offset < sizeof(outbuf)))
 688                 memcpy(outbuf, buf, Offset);
 689         outpos = snprintf(&outbuf[Offset], sizeof(outbuf) - Offset,
 690                           "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
 691                           LB, QU, urlbuf, QU, QU, TARGET, QU, RB, urlbuf, LB, RB);
 692         if (outpos >= sizeof(outbuf) - Offset) {
 693                 lprintf(1, "URL: content longer than buffer!");
 694                 return;
 695         }
 696
 697         TrailerLen = len - (end - start);
 698         if (TrailerLen > 0)
 699                 memcpy(outbuf + Offset + outpos, end, TrailerLen);
 700         if (Offset + outpos + TrailerLen > bufsize) {
 701                 lprintf(1, "URL: content longer than buffer!");
 702                 return;
 703         }
 704         memcpy (buf, outbuf, Offset + outpos + TrailerLen);
 705         *(buf + Offset + outpos + TrailerLen) = '\0';
 706 }
 707