webcit/html2html.c

   1 /*
   2  * $Id$
   3  */
   4 /**
   5  * \defgroup HTML2HTML Output an HTML message, modifying it slightly to make sure it plays nice
   6  * with the rest of our web framework.
   7  * \ingroup WebcitHttpServer
   8  */
   9 /*@{*/
  10 #include "webcit.h"
  11 #include "webserver.h"
  12
  13
  14 /**
  15  * \brief       Strip surrounding single or double quotes from a string.
  16  *
  17  * \param s     String to be stripped.
  18  */
  19 void stripquotes(char *s)
  20 {
  21         int len;
  22
  23         if (!s) return;
  24
  25         len = strlen(s);
  26         if (len < 2) return;
  27
  28         if ( ( (s[0] == '\"') && (s[len-1] == '\"') ) || ( (s[0] == '\'') && (s[len-1] == '\'') ) ) {
  29                 s[len-1] = 0;
  30                 strcpy(s, &s[1]);
  31         }
  32 }
  33
  34
  35 /**
  36  * \brief Check to see if a META tag has overridden the declared MIME character set.
  37  *
  38  * \param charset               Character set name (left unchanged if we don't do anything)
  39  * \param meta_http_equiv       Content of the "http-equiv" portion of the META tag
  40  * \param meta_content          Content of the "content" portion of the META tag
  41  */
  42 void extract_charset_from_meta(char *charset, char *meta_http_equiv, char *meta_content)
  43 {
  44         char *ptr;
  45         char buf[64];
  46
  47         if (!charset) return;
  48         if (!meta_http_equiv) return;
  49         if (!meta_content) return;
  50
  51
  52         if (strcasecmp(meta_http_equiv, "Content-type")) return;
  53
  54         ptr = strchr(meta_content, ';');
  55         if (!ptr) return;
  56
  57         safestrncpy(buf, ++ptr, sizeof buf);
  58         striplt(buf);
  59         if (!strncasecmp(buf, "charset=", 8)) {
  60                 strcpy(charset, &buf[8]);
  61
  62                 /*
  63                  * The brain-damaged webmail program in Microsoft Exchange declares
  64                  * a charset of "unicode" when they really mean "UTF-8".  GNU iconv
  65                  * treats "unicode" as an alias for "UTF-16" so we have to manually
  66                  * fix this here, otherwise messages generated in Exchange webmail
  67                  * show up as a big pile of weird characters.
  68                  */
  69                 if (!strcasecmp(charset, "unicode")) {
  70                         strcpy(charset, "UTF-8");
  71                 }
  72
  73         }
  74 }
  75
  76
  77
  78 /**
  79  * \brief Sanitize and enhance an HTML message for display.
  80  *        Also convert weird character sets to UTF-8 if necessary.
  81  *        Also fixup img src="cid:..." type inline images to fetch the image
  82  *
  83  * \param supplied_charset the input charset as declared in the MIME headers
  84  */
  85 void output_html(const char *supplied_charset, int treat_as_wiki, int msgnum, StrBuf *Source, StrBuf *Target) {
  86         char buf[SIZ];
  87         char *msg;
  88         char *ptr;
  89         char *msgstart;
  90         char *msgend;
  91         StrBuf *converted_msg;
  92         int buffer_length = 1;
  93         int line_length = 0;
  94         int content_length = 0;
  95         char new_window[SIZ];
  96         int brak = 0;
  97         int alevel = 0;
  98         int scriptlevel = 0;
  99         int script_start_pos = (-1);
 100         int i;
 101         int linklen;
 102         char charset[128];
 103 #ifdef HAVE_ICONV
 104         iconv_t ic = (iconv_t)(-1) ;
 105         char *ibuf;                   /**< Buffer of characters to be converted */
 106         char *obuf;                   /**< Buffer for converted characters      */
 107         size_t ibuflen;               /**< Length of input buffer               */
 108         size_t obuflen;               /**< Length of output buffer              */
 109         char *osav;                   /**< Saved pointer to output buffer       */
 110 #endif
 111         lprintf(9, "output_html() called with msgnum=%ld\n", msgnum);
 112         if (Target == NULL)
 113                 Target = WC->WBuf;
 114
 115         safestrncpy(charset, supplied_charset, sizeof charset);
 116         msg = strdup("");
 117         sprintf(new_window, "<a target=\"%s\" href=", TARGET);
 118
 119         if (Source == NULL) while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 120                 line_length = strlen(buf);
 121                 buffer_length = content_length + line_length + 2;
 122                 ptr = realloc(msg, buffer_length);
 123                 if (ptr == NULL) {
 124                         StrBufAppendPrintf(Target, "<b>");
 125                         StrBufAppendPrintf(Target, _("realloc() error! couldn't get %d bytes: %s"),
 126                                 buffer_length + 1,
 127                                 strerror(errno));
 128                         StrBufAppendPrintf(Target, "</b><br /><br />\n");
 129                         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 130                                 /** flush */
 131                         }
 132                         free(msg);
 133                         return;
 134                 }
 135                 msg = ptr;
 136                 strcpy(&msg[content_length], buf);
 137                 content_length += line_length;
 138                 strcpy(&msg[content_length], "\n");
 139                 content_length += 1;
 140         }
 141         else {
 142                 content_length = StrLength(Source);
 143                 free(msg);
 144                 msg = (char*) ChrPtr(Source);/* TODO: remove cast */
 145                 buffer_length = content_length;
 146         }
 147
 148         /** Do a first pass to isolate the message body */
 149         ptr = msg + 1;
 150         msgstart = msg;
 151         msgend = &msg[content_length];
 152
 153         while (ptr < msgend) {
 154
 155                 /** Advance to next tag */
 156                 ptr = strchr(ptr, '<');
 157                 if ((ptr == NULL) || (ptr >= msgend)) break;
 158                 ++ptr;
 159                 if ((ptr == NULL) || (ptr >= msgend)) break;
 160
 161                 /**
 162                  *  Look for META tags.  Some messages (particularly in
 163                  *  Asian locales) illegally declare a message's character
 164                  *  set in the HTML instead of in the MIME headers.  This
 165                  *  is wrong but we have to work around it anyway.
 166                  */
 167                 if (!strncasecmp(ptr, "META", 4)) {
 168
 169                         char *meta_start;
 170                         char *meta_end;
 171                         int meta_length;
 172                         char *meta;
 173                         char *meta_http_equiv;
 174                         char *meta_content;
 175                         char *spaceptr;
 176
 177                         meta_start = &ptr[4];
 178                         meta_end = strchr(ptr, '>');
 179                         if ((meta_end != NULL) && (meta_end <= msgend)) {
 180                                 meta_length = meta_end - meta_start + 1;
 181                                 meta = malloc(meta_length + 1);
 182                                 safestrncpy(meta, meta_start, meta_length);
 183                                 meta[meta_length] = 0;
 184                                 striplt(meta);
 185                                 if (!strncasecmp(meta, "HTTP-EQUIV=", 11)) {
 186                                         meta_http_equiv = strdup(&meta[11]);
 187                                         spaceptr = strchr(meta_http_equiv, ' ');
 188                                         if (spaceptr != NULL) {
 189                                                 *spaceptr = 0;
 190                                                 meta_content = strdup(++spaceptr);
 191                                                 if (!strncasecmp(meta_content, "content=", 8)) {
 192                                                         strcpy(meta_content, &meta_content[8]);
 193                                                         stripquotes(meta_http_equiv);
 194                                                         stripquotes(meta_content);
 195                                                         extract_charset_from_meta(charset,
 196                                                                 meta_http_equiv, meta_content);
 197                                                 }
 198                                                 free(meta_content);
 199                                         }
 200                                         free(meta_http_equiv);
 201                                 }
 202                                 free(meta);
 203                         }
 204                 }
 205
 206                 /**
 207                  * Any of these tags cause everything up to and including
 208                  * the tag to be removed.
 209                  */
 210                 if ( (!strncasecmp(ptr, "HTML", 4))
 211                    ||(!strncasecmp(ptr, "HEAD", 4))
 212                    ||(!strncasecmp(ptr, "/HEAD", 5))
 213                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
 214                         ptr = strchr(ptr, '>');
 215                         if ((ptr == NULL) || (ptr >= msgend)) break;
 216                         ++ptr;
 217                         if ((ptr == NULL) || (ptr >= msgend)) break;
 218                         msgstart = ptr;
 219                 }
 220
 221                 /**
 222                  * Any of these tags cause everything including and following
 223                  * the tag to be removed.
 224                  */
 225                 if ( (!strncasecmp(ptr, "/HTML", 5))
 226                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 227                         --ptr;
 228                         msgend = ptr;
 229                         strcpy(ptr, "");
 230
 231                 }
 232
 233                 ++ptr;
 234         }
 235         if (msgstart > msg) {
 236                 strcpy(msg, msgstart);
 237         }
 238
 239         /** Now go through the message, parsing tags as necessary. */
 240         converted_msg = NewStrBufPlain(NULL, content_length + 8192);
 241
 242
 243         /** Convert foreign character sets to UTF-8 if necessary. */
 244 #ifdef HAVE_ICONV
 245         if ( (strcasecmp(charset, "us-ascii"))
 246            && (strcasecmp(charset, "UTF-8"))
 247            && (strcasecmp(charset, ""))
 248         ) {
 249                 lprintf(9, "Converting %s to UTF-8\n", charset);
 250                 ctdl_iconv_open("UTF-8", charset, &ic);
 251                 if (ic == (iconv_t)(-1) ) {
 252                         lprintf(5, "%s:%d iconv_open() failed: %s\n",
 253                                 __FILE__, __LINE__, strerror(errno));
 254                 }
 255         }
 256         if  (Source == NULL) {
 257                 if (ic != (iconv_t)(-1) ) {
 258                         ibuf = msg;
 259                         ibuflen = content_length;
 260                         obuflen = content_length + (content_length / 2) ;
 261                         obuf = (char *) malloc(obuflen);
 262                         osav = obuf;
 263                         iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
 264                         content_length = content_length + (content_length / 2) - obuflen;
 265                         osav[content_length] = 0;
 266                         free(msg);
 267                         msg = osav;
 268                         iconv_close(ic);
 269                 }
 270         }
 271         else {
 272                 if (ic != (iconv_t)(-1) ) {
 273                         StrBuf *Buf = NewStrBufPlain(NULL, StrLength(Source) + 8096);;
 274                         StrBufConvert(Source, Buf, &ic);
 275                         FreeStrBuf(&Buf);
 276                         iconv_close(ic);
 277                         msg = (char*)ChrPtr(Source); /* TODO: get rid of this. */
 278                 }
 279         }
 280
 281 #endif
 282
 283         /**
 284          *      At this point, the message has been stripped down to
 285          *      only the content inside the <BODY></BODY> tags, and has
 286          *      been converted to UTF-8 if it was originally in a foreign
 287          *      character set.  The text is also guaranteed to be null
 288          *      terminated now.
 289          */
 290
 291         if (converted_msg == NULL) {
 292                 StrBufAppendPrintf(Target, "Error %d: %s<br />%s:%d", errno, strerror(errno), __FILE__, __LINE__);
 293                 goto BAIL;
 294         }
 295
 296         ptr = msg;
 297         msgend = strchr(msg, 0);
 298         while (ptr < msgend) {
 299
 300                 /** Try to sanitize the html of any rogue scripts */
 301                 if (!strncasecmp(ptr, "<script", 7)) {
 302                         if (scriptlevel == 0) {
 303                                 script_start_pos = StrLength(converted_msg);
 304                         }
 305                         ++scriptlevel;
 306                 }
 307                 if (!strncasecmp(ptr, "</script", 8)) {
 308                         --scriptlevel;
 309                 }
 310
 311                 /**
 312                  * Change mailto: links to WebCit mail, by replacing the
 313                  * link with one that points back to our mail room.  Due to
 314                  * the way we parse URL's, it'll even handle mailto: links
 315                  * that have "?subject=" in them.
 316                  */
 317                 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
 318                         content_length += 64;
 319                         StrBufAppendPrintf(converted_msg,
 320                                 "<a href=\"display_enter?force_room=_MAIL_&recp=");
 321                         ptr = &ptr[16];
 322                         ++alevel;
 323                         ++brak;
 324                 }
 325                 /** Make external links open in a separate window */
 326                 else if (!strncasecmp(ptr, "<a href=\"", 9)) {
 327                         ++alevel;
 328                         ++brak;
 329                         if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
 330                              &&  ((strchr(ptr, '/') < strchr(ptr, '>')))
 331                              ) {
 332                                 /* open external links to new window */
 333                                 StrBufAppendPrintf(converted_msg, new_window);
 334                                 ptr = &ptr[8];
 335                         }
 336                         else if ( (treat_as_wiki) && (strncasecmp(ptr, "<a href=\"wiki?", 14)) ) {
 337                                 content_length += 64;
 338                                 StrBufAppendPrintf(converted_msg, "<a href=\"wiki?page=");
 339                                 ptr = &ptr[9];
 340                         }
 341                         else {
 342                                 StrBufAppendPrintf(converted_msg, "<a href=\"");
 343                                 ptr = &ptr[9];
 344                         }
 345                 }
 346                 /** Fixup <img src="cid:... ...> to fetch the mime part */
 347                 else if (!strncasecmp(ptr, "<img ", 5)) {
 348                         char* tag_end=strchr(ptr,'>');
 349                         char* src=strstr(ptr, " src=\"cid:");
 350                         char *cid_start, *cid_end;
 351                         ++brak;
 352
 353                         if (src &&
 354                                 (cid_start=strchr(src,':')) &&
 355                                 (cid_end=strchr(cid_start,'"')) &&
 356                                 (cid_end < tag_end)) {
 357
 358                                 /* copy tag and attributes up to src="cid: */
 359                                 StrBufAppendBufPlain(converted_msg, ptr, src - ptr, 0);
 360                                 cid_start++;
 361
 362                                 /* add in /webcit/mimepart/<msgno>/CID/
 363                                    trailing / stops dumb URL filters getting excited */
 364                                 StrBufAppendPrintf(converted_msg,
 365                                         "src=\"/webcit/mimepart/%d/",msgnum);
 366                                 StrBufAppendBufPlain(converted_msg, cid_start, cid_end - cid_start, 0);
 367                                 StrBufAppendBufPlain(converted_msg, "/\"", -1, 0);
 368
 369                                 ptr = cid_end+1;
 370                         }
 371                         StrBufAppendBufPlain(converted_msg, ptr, tag_end - ptr, 0);
 372                         ptr = tag_end;
 373                 }
 374
 375                 /**
 376                  * Turn anything that looks like a URL into a real link, as long
 377                  * as it's not inside a tag already
 378                  */
 379                 else if ( (brak == 0) && (alevel == 0)
 380                      && (!strncasecmp(ptr, "http://", 7))) {
 381                                 /** Find the end of the link */
 382                                 int strlenptr;
 383                                 linklen = 0;
 384
 385                                 strlenptr = strlen(ptr);
 386                                 for (i=0; i<=strlenptr; ++i) {
 387                                         if ((ptr[i]==0)
 388                                            ||(isspace(ptr[i]))
 389                                            ||(ptr[i]==10)
 390                                            ||(ptr[i]==13)
 391                                            ||(ptr[i]=='(')
 392                                            ||(ptr[i]==')')
 393                                            ||(ptr[i]=='<')
 394                                            ||(ptr[i]=='>')
 395                                            ||(ptr[i]=='[')
 396                                            ||(ptr[i]==']')
 397                                            ||(ptr[i]=='"')
 398                                            ||(ptr[i]=='\'')
 399                                         ) linklen = i;
 400                                         /* did s.b. send us an entity? */
 401                                         if (ptr[i] == '&') {
 402                                                 if ((ptr[i+2] ==';') ||
 403                                                     (ptr[i+3] ==';') ||
 404                                                     (ptr[i+5] ==';') ||
 405                                                     (ptr[i+6] ==';') ||
 406                                                     (ptr[i+7] ==';'))
 407                                                         linklen = i;
 408                                         }
 409                                         if (linklen > 0) break;
 410                                 }
 411                                 if (linklen > 0) {
 412                                         char *ltreviewptr;
 413                                         char *nbspreviewptr;
 414                                         char linkedchar;
 415                                         int len = linklen;
 416
 417                                         len = linklen;
 418                                         linkedchar = ptr[len];
 419                                         ptr[len] = '\0';
 420                                         /* spot for some subject strings tinymce tends to give us. */
 421                                         ltreviewptr = strchr(ptr, '<');
 422                                         if (ltreviewptr != NULL) {
 423                                                 *ltreviewptr = '\0';
 424                                                 linklen = ltreviewptr - ptr;
 425                                         }
 426
 427                                         nbspreviewptr = strstr(ptr, "&nbsp;");
 428                                         if (nbspreviewptr != NULL) {
 429                                                 /* nbspreviewptr = '\0'; */
 430                                                 linklen = nbspreviewptr - ptr;
 431                                         }
 432                                         if (ltreviewptr != 0)
 433                                                 *ltreviewptr = '<';
 434
 435                                         ptr[len] = linkedchar;
 436
 437                                         content_length += (32 + linklen);
 438                                         StrBufAppendPrintf(converted_msg, "%s\"", new_window);
 439                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 440                                         StrBufAppendPrintf(converted_msg, "\">");
 441                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 442                                         ptr += linklen;
 443                                         StrBufAppendPrintf(converted_msg, "</A>");
 444                                 }
 445                 }
 446                 else {
 447                         StrBufAppendBufPlain(converted_msg, ptr, 1, 0);
 448                         ptr++;
 449                 }
 450
 451                 /**
 452                  * We need to know when we're inside a tag,
 453                  * so we don't turn things that look like URL's into
 454                  * links, when they're already links - or image sources.
 455                  */
 456                 if (*(ptr-1) == '<') {
 457                         ++brak;
 458                 }
 459                 if (*(ptr-1) == '>') {
 460                         --brak;
 461                         if ((scriptlevel == 0) && (script_start_pos >= 0)) {
 462                                 StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
 463                                 script_start_pos = (-1);
 464                         }
 465                 }
 466                 if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 467         }
 468
 469         /**     uncomment these two lines to override conversion        */
 470         /**     memcpy(converted_msg, msg, content_length);             */
 471         /**     output_length = content_length;                         */
 472
 473         /** Output our big pile of markup */
 474         StrBufAppendBuf(Target, converted_msg, 0);
 475
 476 BAIL:   /** A little trailing vertical whitespace... */
 477         StrBufAppendPrintf(Target, "<br /><br />\n");
 478
 479         /** Now give back the memory */
 480         FreeStrBuf(&converted_msg);
 481         if ((msg != NULL) && (Source == NULL)) free(msg);
 482 }
 483
 484
 485
 486
 487
 488
 489 /*
 490  * Look for URL's embedded in a buffer and make them linkable.  We use a
 491  * target window in order to keep the Citadel session in its own window.
 492  */
 493 void UrlizeText(StrBuf* Target, StrBuf *Source, StrBuf *WrkBuf)
 494 {
 495         int len, UrlLen, Offset, TrailerLen;
 496         const char *start, *end, *pos;
 497
 498         FlushStrBuf(Target);
 499
 500         start = NULL;
 501         len = StrLength(Source);
 502         end = ChrPtr(Source) + len;
 503         for (pos = ChrPtr(Source); (pos < end) && (start == NULL); ++pos) {
 504                 if (!strncasecmp(pos, "http://", 7))
 505                         start = pos;
 506                 else if (!strncasecmp(pos, "ftp://", 6))
 507                         start = pos;
 508         }
 509
 510         if (start == NULL) {
 511                 StrBufAppendBuf(Target, Source, 0);
 512                 return;
 513         }
 514         FlushStrBuf(WrkBuf);
 515
 516         for (pos = ChrPtr(Source) + len; pos > start; --pos) {
 517                 if (  (!isprint(*pos))
 518                    || (isspace(*pos))
 519                    || (*pos == '{')
 520                    || (*pos == '}')
 521                    || (*pos == '|')
 522                    || (*pos == '\\')
 523                    || (*pos == '^')
 524                    || (*pos == '[')
 525                    || (*pos == ']')
 526                    || (*pos == '`')
 527                    || (*pos == '<')
 528                    || (*pos == '>')
 529                    || (*pos == '(')
 530                    || (*pos == ')')
 531                 ) {
 532                         end = pos;
 533                 }
 534         }
 535
 536         UrlLen = end - start;
 537         StrBufAppendBufPlain(WrkBuf, start, UrlLen, 0);
 538
 539         Offset = start - ChrPtr(Source);
 540         if (Offset != 0)
 541                 StrBufAppendBufPlain(Target, ChrPtr(Source), Offset, 0);
 542         StrBufAppendPrintf(Target, "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
 543                            LB, QU, ChrPtr(WrkBuf), QU, QU, TARGET,
 544                            QU, RB, ChrPtr(WrkBuf), LB, RB);
 545
 546         TrailerLen = StrLength(Source) - (end - ChrPtr(Source));
 547         if (TrailerLen > 0)
 548                 StrBufAppendBufPlain(Target, end, TrailerLen, 0);
 549 }
 550 void url(char *buf, size_t bufsize)
 551 {
 552         int len, UrlLen, Offset, TrailerLen, outpos;
 553         char *start, *end, *pos;
 554         char urlbuf[SIZ];
 555         char outbuf[SIZ];
 556
 557         start = NULL;
 558         len = strlen(buf);
 559         if (len > bufsize) {
 560                 lprintf(1, "URL: content longer than buffer!");
 561                 return;
 562         }
 563         end = buf + len;
 564         for (pos = buf; (pos < end) && (start == NULL); ++pos) {
 565                 if (!strncasecmp(pos, "http://", 7))
 566                         start = pos;
 567                 if (!strncasecmp(pos, "ftp://", 6))
 568                         start = pos;
 569         }
 570
 571         if (start == NULL)
 572                 return;
 573
 574         for (pos = buf+len; pos > start; --pos) {
 575                 if (  (!isprint(*pos))
 576                    || (isspace(*pos))
 577                    || (*pos == '{')
 578                    || (*pos == '}')
 579                    || (*pos == '|')
 580                    || (*pos == '\\')
 581                    || (*pos == '^')
 582                    || (*pos == '[')
 583                    || (*pos == ']')
 584                    || (*pos == '`')
 585                    || (*pos == '<')
 586                    || (*pos == '>')
 587                    || (*pos == '(')
 588                    || (*pos == ')')
 589                 ) {
 590                         end = pos;
 591                 }
 592         }
 593
 594         UrlLen = end - start;
 595         if (UrlLen > sizeof(urlbuf)){
 596                 lprintf(1, "URL: content longer than buffer!");
 597                 return;
 598         }
 599         memcpy(urlbuf, start, UrlLen);
 600         urlbuf[UrlLen] = '\0';
 601
 602         Offset = start - buf;
 603         if ((Offset != 0) && (Offset < sizeof(outbuf)))
 604                 memcpy(outbuf, buf, Offset);
 605         outpos = snprintf(&outbuf[Offset], sizeof(outbuf) - Offset,
 606                           "%ca href=%c%s%c TARGET=%c%s%c%c%s%c/A%c",
 607                           LB, QU, urlbuf, QU, QU, TARGET, QU, RB, urlbuf, LB, RB);
 608         if (outpos >= sizeof(outbuf) - Offset) {
 609                 lprintf(1, "URL: content longer than buffer!");
 610                 return;
 611         }
 612
 613         TrailerLen = len - (end - start);
 614         if (TrailerLen > 0)
 615                 memcpy(outbuf + Offset + outpos, end, TrailerLen);
 616         if (Offset + outpos + TrailerLen > bufsize) {
 617                 lprintf(1, "URL: content longer than buffer!");
 618                 return;
 619         }
 620         memcpy (buf, outbuf, Offset + outpos + TrailerLen);
 621         *(buf + Offset + outpos + TrailerLen) = '\0';
 622 }
 623
 624
 625
 626
 627 /*@}*/