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(char *supplied_charset, int treat_as_wiki, int msgnum) {
  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
 112         safestrncpy(charset, supplied_charset, sizeof charset);
 113         msg = strdup("");
 114         sprintf(new_window, "<a target=\"%s\" href=", TARGET);
 115
 116         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 117                 line_length = strlen(buf);
 118                 buffer_length = content_length + line_length + 2;
 119                 ptr = realloc(msg, buffer_length);
 120                 if (ptr == NULL) {
 121                         wprintf("<b>");
 122                         wprintf(_("realloc() error! couldn't get %d bytes: %s"),
 123                                 buffer_length + 1,
 124                                 strerror(errno));
 125                         wprintf("</b><br /><br />\n");
 126                         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 127                                 /** flush */
 128                         }
 129                         free(msg);
 130                         return;
 131                 }
 132                 msg = ptr;
 133                 strcpy(&msg[content_length], buf);
 134                 content_length += line_length;
 135                 strcpy(&msg[content_length], "\n");
 136                 content_length += 1;
 137         }
 138
 139         /** Do a first pass to isolate the message body */
 140         ptr = msg + 1;
 141         msgstart = msg;
 142         msgend = &msg[content_length];
 143
 144         while (ptr < msgend) {
 145
 146                 /** Advance to next tag */
 147                 ptr = strchr(ptr, '<');
 148                 if ((ptr == NULL) || (ptr >= msgend)) break;
 149                 ++ptr;
 150                 if ((ptr == NULL) || (ptr >= msgend)) break;
 151
 152                 /**
 153                  *  Look for META tags.  Some messages (particularly in
 154                  *  Asian locales) illegally declare a message's character
 155                  *  set in the HTML instead of in the MIME headers.  This
 156                  *  is wrong but we have to work around it anyway.
 157                  */
 158                 if (!strncasecmp(ptr, "META", 4)) {
 159
 160                         char *meta_start;
 161                         char *meta_end;
 162                         int meta_length;
 163                         char *meta;
 164                         char *meta_http_equiv;
 165                         char *meta_content;
 166                         char *spaceptr;
 167
 168                         meta_start = &ptr[4];
 169                         meta_end = strchr(ptr, '>');
 170                         if ((meta_end != NULL) && (meta_end <= msgend)) {
 171                                 meta_length = meta_end - meta_start + 1;
 172                                 meta = malloc(meta_length + 1);
 173                                 safestrncpy(meta, meta_start, meta_length);
 174                                 meta[meta_length] = 0;
 175                                 striplt(meta);
 176                                 if (!strncasecmp(meta, "HTTP-EQUIV=", 11)) {
 177                                         meta_http_equiv = strdup(&meta[11]);
 178                                         spaceptr = strchr(meta_http_equiv, ' ');
 179                                         if (spaceptr != NULL) {
 180                                                 *spaceptr = 0;
 181                                                 meta_content = strdup(++spaceptr);
 182                                                 if (!strncasecmp(meta_content, "content=", 8)) {
 183                                                         strcpy(meta_content, &meta_content[8]);
 184                                                         stripquotes(meta_http_equiv);
 185                                                         stripquotes(meta_content);
 186                                                         extract_charset_from_meta(charset,
 187                                                                 meta_http_equiv, meta_content);
 188                                                 }
 189                                                 free(meta_content);
 190                                         }
 191                                         free(meta_http_equiv);
 192                                 }
 193                                 free(meta);
 194                         }
 195                 }
 196
 197                 /**
 198                  * Any of these tags cause everything up to and including
 199                  * the tag to be removed.
 200                  */
 201                 if ( (!strncasecmp(ptr, "HTML", 4))
 202                    ||(!strncasecmp(ptr, "HEAD", 4))
 203                    ||(!strncasecmp(ptr, "/HEAD", 5))
 204                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
 205                         ptr = strchr(ptr, '>');
 206                         if ((ptr == NULL) || (ptr >= msgend)) break;
 207                         ++ptr;
 208                         if ((ptr == NULL) || (ptr >= msgend)) break;
 209                         msgstart = ptr;
 210                 }
 211
 212                 /**
 213                  * Any of these tags cause everything including and following
 214                  * the tag to be removed.
 215                  */
 216                 if ( (!strncasecmp(ptr, "/HTML", 5))
 217                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 218                         --ptr;
 219                         msgend = ptr;
 220                         strcpy(ptr, "");
 221
 222                 }
 223
 224                 ++ptr;
 225         }
 226         if (msgstart > msg) {
 227                 strcpy(msg, msgstart);
 228         }
 229
 230         /** Convert foreign character sets to UTF-8 if necessary. */
 231 #ifdef HAVE_ICONV
 232         if ( (strcasecmp(charset, "us-ascii"))
 233            && (strcasecmp(charset, "UTF-8"))
 234            && (strcasecmp(charset, ""))
 235         ) {
 236                 lprintf(9, "Converting %s to UTF-8\n", charset);
 237                 ic = ctdl_iconv_open("UTF-8", charset);
 238                 if (ic == (iconv_t)(-1) ) {
 239                         lprintf(5, "%s:%d iconv_open() failed: %s\n",
 240                                 __FILE__, __LINE__, strerror(errno));
 241                 }
 242         }
 243         if (ic != (iconv_t)(-1) ) {
 244                 ibuf = msg;
 245                 ibuflen = content_length;
 246                 obuflen = content_length + (content_length / 2) ;
 247                 obuf = (char *) malloc(obuflen);
 248                 osav = obuf;
 249                 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
 250                 content_length = content_length + (content_length / 2) - obuflen;
 251                 osav[content_length] = 0;
 252                 free(msg);
 253                 msg = osav;
 254                 iconv_close(ic);
 255         }
 256 #endif
 257
 258         /**
 259          *      At this point, the message has been stripped down to
 260          *      only the content inside the <BODY></BODY> tags, and has
 261          *      been converted to UTF-8 if it was originally in a foreign
 262          *      character set.  The text is also guaranteed to be null
 263          *      terminated now.
 264          */
 265
 266         /** Now go through the message, parsing tags as necessary. */
 267         converted_msg = NewStrBufPlain(NULL, content_length + 8192);
 268         if (converted_msg == NULL) {
 269                 wprintf("Error %d: %s<br />%s:%d", errno, strerror(errno), __FILE__, __LINE__);
 270                 goto BAIL;
 271         }
 272
 273         ptr = msg;
 274         msgend = strchr(msg, 0);
 275         while (ptr < msgend) {
 276
 277                 /** Try to sanitize the html of any rogue scripts */
 278                 if (!strncasecmp(ptr, "<script", 7)) {
 279                         if (scriptlevel == 0) {
 280                                 script_start_pos = StrLength(converted_msg);
 281                         }
 282                         ++scriptlevel;
 283                 }
 284                 if (!strncasecmp(ptr, "</script", 8)) {
 285                         --scriptlevel;
 286                 }
 287
 288                 /**
 289                  * Change mailto: links to WebCit mail, by replacing the
 290                  * link with one that points back to our mail room.  Due to
 291                  * the way we parse URL's, it'll even handle mailto: links
 292                  * that have "?subject=" in them.
 293                  */
 294                 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
 295                         content_length += 64;
 296                         StrBufAppendPrintf(converted_msg,
 297                                 "<a href=\"display_enter?force_room=_MAIL_&recp=");
 298                         ptr = &ptr[16];
 299                         ++alevel;
 300                         ++brak;
 301                 }
 302                 /** Make external links open in a separate window */
 303                 else if (!strncasecmp(ptr, "<a href=\"", 9)) {
 304                         ++alevel;
 305                         ++brak;
 306                         if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
 307                              &&  ((strchr(ptr, '/') < strchr(ptr, '>')))
 308                              ) {
 309                                 /* open external links to new window */
 310                                 StrBufAppendPrintf(converted_msg, new_window);
 311                                 ptr = &ptr[8];
 312                         }
 313                         else if ( (treat_as_wiki) && (strncasecmp(ptr, "<a href=\"wiki?", 14)) ) {
 314                                 content_length += 64;
 315                                 StrBufAppendPrintf(converted_msg, "<a href=\"wiki?page=");
 316                                 ptr = &ptr[9];
 317                         }
 318                         else {
 319                                 StrBufAppendPrintf(converted_msg, "<a href=\"");
 320                                 ptr = &ptr[9];
 321                         }
 322                 }
 323                 /** Fixup <img src="cid:... ...> to fetch the mime part */
 324                 else if (!strncasecmp(ptr, "<img ", 5)) {
 325                         char* tag_end=strchr(ptr,'>');
 326                         char* src=strstr(ptr, " src=\"cid:");
 327                         char *cid_start, *cid_end;
 328                         ++brak;
 329
 330                         if (src &&
 331                                 (cid_start=strchr(src,':')) &&
 332                                 (cid_end=strchr(cid_start,'"')) &&
 333                                 (cid_end < tag_end)) {
 334
 335                                 /* copy tag and attributes up to src="cid: */
 336                                 StrBufAppendBufPlain(converted_msg, ptr, src - ptr, 0);
 337                                 cid_start++;
 338
 339                                 /* add in /webcit/mimepart/<msgno>/CID/
 340                                    trailing / stops dumb URL filters getting excited */
 341                                 StrBufAppendPrintf(converted_msg,
 342                                         "src=\"/webcit/mimepart/%d/",msgnum);
 343                                 StrBufAppendBufPlain(converted_msg, cid_start, cid_end - cid_start, 0);
 344                                 StrBufAppendBufPlain(converted_msg, "/\"", -1, 0);
 345
 346                                 ptr = cid_end+1;
 347                         }
 348                         StrBufAppendBufPlain(converted_msg, ptr, tag_end - ptr, 0);
 349                         ptr = tag_end;
 350                 }
 351
 352                 /**
 353                  * Turn anything that looks like a URL into a real link, as long
 354                  * as it's not inside a tag already
 355                  */
 356                 else if ( (brak == 0) && (alevel == 0)
 357                      && (!strncasecmp(ptr, "http://", 7))) {
 358                                 /** Find the end of the link */
 359                                 int strlenptr;
 360                                 linklen = 0;
 361
 362                                 strlenptr = strlen(ptr);
 363                                 for (i=0; i<=strlenptr; ++i) {
 364                                         if ((ptr[i]==0)
 365                                            ||(isspace(ptr[i]))
 366                                            ||(ptr[i]==10)
 367                                            ||(ptr[i]==13)
 368                                            ||(ptr[i]=='(')
 369                                            ||(ptr[i]==')')
 370                                            ||(ptr[i]=='<')
 371                                            ||(ptr[i]=='>')
 372                                            ||(ptr[i]=='[')
 373                                            ||(ptr[i]==']')
 374                                            ||(ptr[i]=='"')
 375                                            ||(ptr[i]=='\'')
 376                                         ) linklen = i;
 377                                         /* did s.b. send us an entity? */
 378                                         if (ptr[i] == '&') {
 379                                                 if ((ptr[i+2] ==';') ||
 380                                                     (ptr[i+3] ==';') ||
 381                                                     (ptr[i+5] ==';') ||
 382                                                     (ptr[i+6] ==';') ||
 383                                                     (ptr[i+7] ==';'))
 384                                                         linklen = i;
 385                                         }
 386                                         if (linklen > 0) break;
 387                                 }
 388                                 if (linklen > 0) {
 389                                         char *ltreviewptr;
 390                                         char *nbspreviewptr;
 391                                         char linkedchar;
 392                                         int len = linklen;
 393
 394                                         len = linklen;
 395                                         linkedchar = ptr[len];
 396                                         ptr[len] = '\0';
 397                                         /* spot for some subject strings tinymce tends to give us. */
 398                                         ltreviewptr = strchr(ptr, '<');
 399                                         if (ltreviewptr != NULL) {
 400                                                 *ltreviewptr = '\0';
 401                                                 linklen = ltreviewptr - ptr;
 402                                         }
 403
 404                                         nbspreviewptr = strstr(ptr, "&nbsp;");
 405                                         if (nbspreviewptr != NULL) {
 406                                                 ///*nbspreviewptr = '\0';
 407                                                 linklen = nbspreviewptr - ptr;
 408                                         }
 409                                         if (ltreviewptr != 0)
 410                                                 *ltreviewptr = '<';
 411
 412                                         ptr[len] = linkedchar;
 413
 414                                         content_length += (32 + linklen);
 415                                         StrBufAppendPrintf(converted_msg, "%s\"", new_window);
 416                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 417                                         StrBufAppendPrintf(converted_msg, "\">");
 418                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 419                                         ptr += linklen;
 420                                         StrBufAppendPrintf(converted_msg, "</A>");
 421                                 }
 422                 }
 423                 else {
 424                         StrBufAppendBufPlain(converted_msg, ptr, 1, 0);
 425                         ptr++;
 426                 }
 427
 428                 /**
 429                  * We need to know when we're inside a tag,
 430                  * so we don't turn things that look like URL's into
 431                  * links, when they're already links - or image sources.
 432                  */
 433                 if (*(ptr-1) == '<') {
 434                         ++brak;
 435                 }
 436                 if (*(ptr-1) == '>') {
 437                         --brak;
 438                         if ((scriptlevel == 0) && (script_start_pos >= 0)) {
 439                                 StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
 440                                 script_start_pos = (-1);
 441                         }
 442                 }
 443                 if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 444         }
 445
 446         /**     uncomment these two lines to override conversion        */
 447         /**     memcpy(converted_msg, msg, content_length);             */
 448         /**     output_length = content_length;                         */
 449
 450         /** Output our big pile of markup */
 451         StrBufAppendBuf(WC->WBuf, converted_msg, 0);
 452
 453 BAIL:   /** A little trailing vertical whitespace... */
 454         wprintf("<br /><br />\n");
 455
 456         /** Now give back the memory */
 457         if (converted_msg != NULL) free(converted_msg);
 458         if (msg != NULL) free(msg);
 459 }
 460
 461 /*@}*/