webcit/html2html.c

   1
   2 /*
   3  * Output an HTML message, modifying it slightly to make sure it plays nice
   4  * with the rest of our web framework.
   5  *
   6  * Copyright (c) 2005-2012 by the citadel.org team
   7  *
   8  * This program is open source software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License, version 3.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  * GNU General Public License for more details.
  15  */
  16
  17 #include "webcit.h"
  18 #include "webserver.h"
  19
  20
  21 /*
  22  * Strip surrounding single or double quotes from a string.
  23  */
  24 void stripquotes(char *s) {
  25         int len;
  26
  27         if (!s)
  28                 return;
  29
  30         len = strlen(s);
  31         if (len < 2)
  32                 return;
  33
  34         if (((s[0] == '\"') && (s[len - 1] == '\"')) || ((s[0] == '\'') && (s[len - 1] == '\''))) {
  35                 s[len - 1] = 0;
  36                 strcpy(s, &s[1]);
  37         }
  38 }
  39
  40
  41 /*
  42  * Check to see if a META tag has overridden the declared MIME character set.
  43  *
  44  * charset              Character set name (left unchanged if we don't do anything)
  45  * meta_http_equiv      Content of the "http-equiv" portion of the META tag
  46  * meta_content         Content of the "content" portion of the META tag
  47  */
  48 void extract_charset_from_meta(char *charset, char *meta_http_equiv, char *meta_content) {
  49         char *ptr;
  50         char buf[64];
  51
  52         if (!charset)
  53                 return;
  54         if (!meta_http_equiv)
  55                 return;
  56         if (!meta_content)
  57                 return;
  58
  59
  60         if (strcasecmp(meta_http_equiv, "Content-type"))
  61                 return;
  62
  63         ptr = strchr(meta_content, ';');
  64         if (!ptr)
  65                 return;
  66
  67         safestrncpy(buf, ++ptr, sizeof buf);
  68         striplt(buf);
  69         if (!strncasecmp(buf, "charset=", 8)) {
  70                 strcpy(charset, &buf[8]);
  71
  72                 /*
  73                  * The brain-damaged webmail program in Microsoft Exchange declares
  74                  * a charset of "unicode" when they really mean "UTF-8".  GNU iconv
  75                  * treats "unicode" as an alias for "UTF-16" so we have to manually
  76                  * fix this here, otherwise messages generated in Exchange webmail
  77                  * show up as a big pile of weird characters.
  78                  */
  79                 if (!strcasecmp(charset, "unicode")) {
  80                         strcpy(charset, "UTF-8");
  81                 }
  82
  83                 /* Remove wandering punctuation */
  84                 if ((ptr = strchr(charset, '\"')))
  85                         *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)
 133                 while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 134                         line_length = strlen(buf);
 135                         buffer_length = content_length + line_length + 2;
 136                         ptr = realloc(msg, buffer_length);
 137                         if (ptr == NULL) {
 138                                 StrBufAppendPrintf(Target, "<b>");
 139                                 StrBufAppendPrintf(Target, _("realloc() error! couldn't get %d bytes: %s"),
 140                                                    buffer_length + 1, strerror(errno));
 141                                 StrBufAppendPrintf(Target, "</b><br><br>\n");
 142                                 while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 143
 144                                 /** flush */
 145                                 }
 146                                 free(msg);
 147                                 return;
 148                         }
 149                         msg = ptr;
 150                         strcpy(&msg[content_length], buf);
 151                         content_length += line_length;
 152                         strcpy(&msg[content_length], "\n");
 153                         content_length += 1;
 154                 }
 155         else {
 156                 content_length = StrLength(Source);
 157                 free(msg);
 158                 msg = (char *) ChrPtr(Source);  /* TODO: remove cast */
 159                 buffer_length = content_length;
 160         }
 161
 162         /** Do a first pass to isolate the message body */
 163         ptr = msg + 1;
 164         msgstart = msg;
 165         msgend = &msg[content_length];
 166
 167         while (ptr < msgend) {
 168
 169                 /** Advance to next tag */
 170                 ptr = strchr(ptr, '<');
 171                 if ((ptr == NULL) || (ptr >= msgend))
 172                         break;
 173                 ++ptr;
 174                 if ((ptr == NULL) || (ptr >= msgend))
 175                         break;
 176
 177                 /*
 178                  *  Look for META tags.  Some messages (particularly in
 179                  *  Asian locales) illegally declare a message's character
 180                  *  set in the HTML instead of in the MIME headers.  This
 181                  *  is wrong but we have to work around it anyway.
 182                  */
 183                 if (!strncasecmp(ptr, "META", 4)) {
 184
 185                         char *meta_start;
 186                         char *meta_end;
 187                         int meta_length;
 188                         char *meta;
 189                         char *meta_http_equiv;
 190                         char *meta_content;
 191                         char *spaceptr;
 192
 193                         meta_start = &ptr[4];
 194                         meta_end = strchr(ptr, '>');
 195                         if ((meta_end != NULL) && (meta_end <= msgend)) {
 196                                 meta_length = meta_end - meta_start + 1;
 197                                 meta = malloc(meta_length + 1);
 198                                 safestrncpy(meta, meta_start, meta_length);
 199                                 meta[meta_length] = 0;
 200                                 striplt(meta);
 201                                 if (!strncasecmp(meta, "HTTP-EQUIV=", 11)) {
 202                                         meta_http_equiv = strdup(&meta[11]);
 203                                         spaceptr = strchr(meta_http_equiv, ' ');
 204                                         if (spaceptr != NULL) {
 205                                                 *spaceptr = 0;
 206                                                 meta_content = strdup(++spaceptr);
 207                                                 if (!strncasecmp(meta_content, "content=", 8)) {
 208                                                         strcpy(meta_content, &meta_content[8]);
 209                                                         stripquotes(meta_http_equiv);
 210                                                         stripquotes(meta_content);
 211                                                         extract_charset_from_meta(charset, meta_http_equiv, meta_content);
 212                                                 }
 213                                                 free(meta_content);
 214                                         }
 215                                         free(meta_http_equiv);
 216                                 }
 217                                 free(meta);
 218                         }
 219                 }
 220
 221                 /*
 222                  * Any of these tags cause everything up to and including
 223                  * the tag to be removed.
 224                  */
 225                 if ((!strncasecmp(ptr, "HTML", 4))
 226                     || (!strncasecmp(ptr, "HEAD", 4))
 227                     || (!strncasecmp(ptr, "/HEAD", 5))
 228                     || (!strncasecmp(ptr, "BODY", 4))) {
 229                         char *pBody = NULL;
 230
 231                         if (!strncasecmp(ptr, "BODY", 4)) {
 232                                 pBody = ptr;
 233                         }
 234                         ptr = strchr(ptr, '>');
 235                         if ((ptr == NULL) || (ptr >= msgend))
 236                                 break;
 237                         if ((pBody != NULL) && (ptr - pBody > 4)) {
 238                                 char *src;
 239                                 char *cid_start, *cid_end;
 240
 241                                 *ptr = '\0';
 242                                 pBody += 4;
 243                                 while ((isspace(*pBody)) && (pBody < ptr))
 244                                         pBody++;
 245                                 BodyArea = NewStrBufPlain(NULL, ptr - pBody);
 246
 247                                 if (pBody < ptr) {
 248                                         src = strstr(pBody, "cid:");
 249                                         if (src) {
 250                                                 cid_start = src + 4;
 251                                                 cid_end = cid_start;
 252                                                 while ((*cid_end != '"') && !isspace(*cid_end) && (cid_end < ptr))
 253                                                         cid_end++;
 254
 255                                                 /* copy tag and attributes up to src="cid: */
 256                                                 StrBufAppendBufPlain(BodyArea, pBody, src - pBody, 0);
 257
 258                                                 /* add in /webcit/mimepart/<msgno>/CID/
 259                                                    trailing / stops dumb URL filters getting excited */
 260                                                 StrBufAppendPrintf(BodyArea, "/webcit/mimepart/%d/", msgnum);
 261                                                 StrBufAppendBufPlain(BodyArea, cid_start, cid_end - cid_start, 0);
 262
 263                                                 if (ptr - cid_end > 0)
 264                                                         StrBufAppendBufPlain(BodyArea, cid_end + 1, 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))
 273                                 break;
 274                         msgstart = ptr;
 275                 }
 276
 277                 /*
 278                  * Any of these tags cause everything including and following
 279                  * the tag to be removed.
 280                  */
 281                 if ((!strncasecmp(ptr, "/HTML", 5))
 282                     || (!strncasecmp(ptr, "/BODY", 5))) {
 283                         --ptr;
 284                         msgend = ptr;
 285                         strcpy(ptr, "");
 286
 287                 }
 288
 289                 ++ptr;
 290         }
 291         if (msgstart > msg) {
 292                 strcpy(msg, msgstart);
 293         }
 294
 295         /* Now go through the message, parsing tags as necessary. */
 296         converted_msg = NewStrBufPlain(NULL, content_length + 8192);
 297
 298
 299         /** Convert foreign character sets to UTF-8 if necessary. */
 300 #ifdef HAVE_ICONV
 301         if ((strcasecmp(charset, "us-ascii"))
 302             && (strcasecmp(charset, "UTF-8"))
 303             && (strcasecmp(charset, ""))
 304             ) {
 305                 syslog(LOG_DEBUG, "Converting %s to UTF-8\n", charset);
 306                 ctdl_iconv_open("UTF-8", charset, &ic);
 307                 if (ic == (iconv_t) (-1)) {
 308                         syslog(LOG_WARNING, "%s:%d iconv_open() failed: %s\n", __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, "<a href=\"display_enter?force_room=_MAIL_?recp=");
 380                         ptr = &ptr[16];
 381                         ++alevel;
 382                         ++brak;
 383                 }
 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 ((treat_as_wiki)
 397                                  && (strncasecmp(ptr, "<a href=\"wiki?", 14))
 398                                  && (strncasecmp(ptr, "<a href=\"dotgoto?", 17))
 399                                  && (strncasecmp(ptr, "<a href=\"knrooms?", 17))
 400                             ) {
 401                                 content_length += 64;
 402                                 StrBufAppendPrintf(converted_msg, "<a href=\"wiki?go=");
 403                                 StrBufUrlescAppend(converted_msg, WC->CurRoom.name, NULL);
 404                                 StrBufAppendPrintf(converted_msg, "?page=");
 405                                 ptr = &ptr[9];
 406                         }
 407                         else {
 408                                 StrBufAppendPrintf(converted_msg, "<a href=\"");
 409                                 ptr = &ptr[9];
 410                         }
 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(LOG_DEBUG, "tag_end is null and ptr is:\n");
 423                                 syslog(LOG_DEBUG, "%s\n", ptr);
 424                                 syslog(LOG_DEBUG, "Theoretical bytes remaining: %d\n", (int) (msgend - ptr));
 425                         }
 426
 427                         src = strstr(ptr, "src=\"cid:");
 428                         ++brak;
 429
 430                         if (src && isspace(*(src - 1))
 431                             && tag_end && (cid_start = strchr(src, ':'))
 432                             && (cid_end = strchr(cid_start, '"'))
 433                             && (cid_end < tag_end)
 434                             ) {
 435                                 /* copy tag and attributes up to src="cid: */
 436                                 StrBufAppendBufPlain(converted_msg, ptr, src - ptr, 0);
 437                                 cid_start++;
 438
 439                                 /* add in /webcit/mimepart/<msgno>/CID/
 440                                    trailing / stops dumb URL filters getting excited */
 441                                 StrBufAppendPrintf(converted_msg, " src=\"/webcit/mimepart/%d/", msgnum);
 442                                 StrBufAppendBufPlain(converted_msg, cid_start, cid_end - cid_start, 0);
 443                                 StrBufAppendBufPlain(converted_msg, "/\"", -1, 0);
 444
 445                                 ptr = cid_end + 1;
 446                         }
 447                         StrBufAppendBufPlain(converted_msg, ptr, tag_end - ptr, 0);
 448                         ptr = tag_end;
 449                 }
 450
 451                 /**
 452                  * Turn anything that looks like a URL into a real link, as long
 453                  * as it's not inside a tag already
 454                  */
 455                 else if ((brak == 0) && (alevel == 0) && ((!strncasecmp(ptr, "http://", 7)) || (!strncasecmp(ptr, "https://", 8)))) {
 456
 457                         /** Find the end of the link */
 458                         int strlenptr;
 459                         linklen = 0;
 460
 461                         strlenptr = strlen(ptr);
 462                         for (i = 0; i <= strlenptr; ++i) {
 463                                 if ((ptr[i] == 0)
 464                                     || (isspace(ptr[i]))
 465                                     || (ptr[i] == 10)
 466                                     || (ptr[i] == 13)
 467                                     || (ptr[i] == '(')
 468                                     || (ptr[i] == ')')
 469                                     || (ptr[i] == '<')
 470                                     || (ptr[i] == '>')
 471                                     || (ptr[i] == '[')
 472                                     || (ptr[i] == ']')
 473                                     || (ptr[i] == '"')
 474                                     || (ptr[i] == '\'')
 475                                     )
 476                                         linklen = i;
 477                                 /* did s.b. send us an entity? */
 478                                 if (ptr[i] == '&') {
 479                                         if ((ptr[i + 2] == ';') ||
 480                                             (ptr[i + 3] == ';') ||
 481                                             (ptr[i + 5] == ';') || (ptr[i + 6] == ';') || (ptr[i + 7] == ';'))
 482                                                 linklen = i;
 483                                 }
 484                                 if (linklen > 0)
 485                                         break;
 486                         }
 487                         if (linklen > 0) {
 488                                 char *ltreviewptr;
 489                                 char *nbspreviewptr;
 490                                 char linkedchar;
 491                                 int len;
 492
 493                                 len = linklen;
 494                                 linkedchar = ptr[len];
 495                                 ptr[len] = '\0';
 496                                 /* spot for some subject strings tinymce tends to give us. */
 497                                 ltreviewptr = strchr(ptr, '<');
 498                                 if (ltreviewptr != NULL) {
 499                                         *ltreviewptr = '\0';
 500                                         linklen = ltreviewptr - ptr;
 501                                 }
 502
 503                                 nbspreviewptr = strstr(ptr, "&nbsp;");
 504                                 if (nbspreviewptr != NULL) {
 505                                         /* nbspreviewptr = '\0'; */
 506                                         linklen = nbspreviewptr - ptr;
 507                                 }
 508                                 if (ltreviewptr != 0)
 509                                         *ltreviewptr = '<';
 510
 511                                 ptr[len] = linkedchar;
 512
 513                                 content_length += (32 + linklen);
 514                                 StrBufAppendPrintf(converted_msg, "%s\"", new_window);
 515                                 StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 516                                 StrBufAppendPrintf(converted_msg, "\">");
 517                                 StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 518                                 ptr += linklen;
 519                                 StrBufAppendPrintf(converted_msg, "</A>");
 520                         }
 521                 }
 522                 else {
 523                         StrBufAppendBufPlain(converted_msg, ptr, 1, 0);
 524                         ptr++;
 525                 }
 526
 527
 528                 if ((ptr >= msg) && (ptr <= msgend)) {
 529                         /*
 530                          * We need to know when we're inside a tag,
 531                          * so we don't turn things that look like URL's into
 532                          * links, when they're already links - or image sources.
 533                          */
 534                         if ((ptr > msg) && (*(ptr - 1) == '<')) {
 535                                 ++brak;
 536                         }
 537                         if ((ptr > msg) && (*(ptr - 1) == '>')) {
 538                                 --brak;
 539                                 if ((scriptlevel == 0) && (script_start_pos >= 0)) {
 540                                         StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
 541                                         script_start_pos = (-1);
 542                                 }
 543                         }
 544                         if (!strncasecmp(ptr, "</A>", 3))
 545                                 --alevel;
 546                 }
 547         }
 548
 549         if (BodyArea != NULL) {
 550                 StrBufAppendBufPlain(converted_msg, HKEY("</td></tr></table>"), 0);
 551                 FreeStrBuf(&BodyArea);
 552         }
 553
 554         /**     uncomment these two lines to override conversion        */
 555
 556         /**     memcpy(converted_msg, msg, content_length);             */
 557
 558         /**     output_length = content_length;                         */
 559
 560         /** Output our big pile of markup */
 561         StrBufAppendBuf(Target, converted_msg, 0);
 562
 563       BAIL:
 564         /** A little trailing vertical whitespace... */
 565         StrBufAppendPrintf(Target, "<br><br>\n");
 566
 567         /** Now give back the memory */
 568         FreeStrBuf(&converted_msg);
 569         if ((msg != NULL) && (Source == NULL))
 570                 free(msg);
 571 }
 572
 573
 574
 575
 576
 577
 578 /*
 579  * Look for URL's embedded in a buffer and make them linkable.  We use a
 580  * target window in order to keep the Citadel session in its own window.
 581  */
 582 void UrlizeText(StrBuf * Target, StrBuf * Source, StrBuf * WrkBuf) {
 583         int len, UrlLen, Offset, TrailerLen;
 584         const char *start, *end, *pos;
 585
 586         FlushStrBuf(Target);
 587
 588         start = NULL;
 589         len = StrLength(Source);
 590         end = ChrPtr(Source) + len;
 591         for (pos = ChrPtr(Source); (pos < end) && (start == NULL); ++pos) {
 592                 if (!strncasecmp(pos, "http://", 7))
 593                         start = pos;
 594                 else if (!strncasecmp(pos, "ftp://", 6))
 595                         start = pos;
 596         }
 597
 598         if (start == NULL) {
 599                 StrBufAppendBuf(Target, Source, 0);
 600                 return;
 601         }
 602         FlushStrBuf(WrkBuf);
 603
 604         for (pos = ChrPtr(Source) + len; pos > start; --pos) {
 605                 if ((!isprint(*pos))
 606                     || (isspace(*pos))
 607                     || (*pos == '{')
 608                     || (*pos == '}')
 609                     || (*pos == '|')
 610                     || (*pos == '\\')
 611                     || (*pos == '^')
 612                     || (*pos == '[')
 613                     || (*pos == ']')
 614                     || (*pos == '`')
 615                     || (*pos == '<')
 616                     || (*pos == '>')
 617                     || (*pos == '(')
 618                     || (*pos == ')')
 619                     ) {
 620                         end = pos;
 621                 }
 622         }
 623
 624         UrlLen = end - start;
 625         StrBufAppendBufPlain(WrkBuf, start, UrlLen, 0);
 626
 627         Offset = start - ChrPtr(Source);
 628         if (Offset != 0)
 629                 StrBufAppendBufPlain(Target, ChrPtr(Source), Offset, 0);
 630         StrBufAppendPrintf(Target, "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
 631                            LB, QU, ChrPtr(WrkBuf), QU, QU, TARGET, 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         int len, UrlLen, Offset, TrailerLen, outpos;
 641         char *start, *end, *pos;
 642         char urlbuf[SIZ];
 643         char outbuf[SIZ];
 644
 645         start = NULL;
 646         len = strlen(buf);
 647         if (len > bufsize) {
 648                 syslog(LOG_WARNING, "URL: content longer than buffer!");
 649                 return;
 650         }
 651         end = buf + len;
 652         for (pos = buf; (pos < end) && (start == NULL); ++pos) {
 653                 if (!strncasecmp(pos, "http://", 7))
 654                         start = pos;
 655                 if (!strncasecmp(pos, "ftp://", 6))
 656                         start = pos;
 657         }
 658
 659         if (start == NULL)
 660                 return;
 661
 662         for (pos = buf + len; pos > start; --pos) {
 663                 if ((!isprint(*pos))
 664                     || (isspace(*pos))
 665                     || (*pos == '{')
 666                     || (*pos == '}')
 667                     || (*pos == '|')
 668                     || (*pos == '\\')
 669                     || (*pos == '^')
 670                     || (*pos == '[')
 671                     || (*pos == ']')
 672                     || (*pos == '`')
 673                     || (*pos == '<')
 674                     || (*pos == '>')
 675                     || (*pos == '(')
 676                     || (*pos == ')')
 677                     ) {
 678                         end = pos;
 679                 }
 680         }
 681
 682         UrlLen = end - start;
 683         if (UrlLen > sizeof(urlbuf)) {
 684                 syslog(LOG_WARNING, "URL: content longer than buffer!");
 685                 return;
 686         }
 687         memcpy(urlbuf, start, UrlLen);
 688         urlbuf[UrlLen] = '\0';
 689
 690         Offset = start - buf;
 691         if ((Offset != 0) && (Offset < sizeof(outbuf)))
 692                 memcpy(outbuf, buf, Offset);
 693         outpos = snprintf(&outbuf[Offset], sizeof(outbuf) - Offset,
 694                           "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c", LB, QU, urlbuf, QU, QU, TARGET, QU, RB, urlbuf, LB, RB);
 695         if (outpos >= sizeof(outbuf) - Offset) {
 696                 syslog(LOG_WARNING, "URL: content longer than buffer!");
 697                 return;
 698         }
 699
 700         TrailerLen = len - (end - start);
 701         if (TrailerLen > 0)
 702                 memcpy(outbuf + Offset + outpos, end, TrailerLen);
 703         if (Offset + outpos + TrailerLen > bufsize) {
 704                 syslog(LOG_WARNING, "URL: content longer than buffer!");
 705                 return;
 706         }
 707         memcpy(buf, outbuf, Offset + outpos + TrailerLen);
 708         *(buf + Offset + outpos + TrailerLen) = '\0';
 709 }