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         if (Target == NULL)
 112                 Target = WC->WBuf;
 113
 114         safestrncpy(charset, supplied_charset, sizeof charset);
 115         msg = strdup("");
 116         sprintf(new_window, "<a target=\"%s\" href=", TARGET);
 117
 118         if (Source == NULL) while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 119                 line_length = strlen(buf);
 120                 buffer_length = content_length + line_length + 2;
 121                 ptr = realloc(msg, buffer_length);
 122                 if (ptr == NULL) {
 123                         StrBufAppendPrintf(Target, "<b>");
 124                         StrBufAppendPrintf(Target, _("realloc() error! couldn't get %d bytes: %s"),
 125                                 buffer_length + 1,
 126                                 strerror(errno));
 127                         StrBufAppendPrintf(Target, "</b><br /><br />\n");
 128                         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 129                                 /** flush */
 130                         }
 131                         free(msg);
 132                         return;
 133                 }
 134                 msg = ptr;
 135                 strcpy(&msg[content_length], buf);
 136                 content_length += line_length;
 137                 strcpy(&msg[content_length], "\n");
 138                 content_length += 1;
 139         }
 140         else {
 141                 content_length = StrLength(Source);
 142                 msg = (char*) ChrPtr(Source);/* TODO: remove cast */
 143                 buffer_length = content_length;
 144         }
 145
 146         /** Do a first pass to isolate the message body */
 147         ptr = msg + 1;
 148         msgstart = msg;
 149         msgend = &msg[content_length];
 150
 151         while (ptr < msgend) {
 152
 153                 /** Advance to next tag */
 154                 ptr = strchr(ptr, '<');
 155                 if ((ptr == NULL) || (ptr >= msgend)) break;
 156                 ++ptr;
 157                 if ((ptr == NULL) || (ptr >= msgend)) break;
 158
 159                 /**
 160                  *  Look for META tags.  Some messages (particularly in
 161                  *  Asian locales) illegally declare a message's character
 162                  *  set in the HTML instead of in the MIME headers.  This
 163                  *  is wrong but we have to work around it anyway.
 164                  */
 165                 if (!strncasecmp(ptr, "META", 4)) {
 166
 167                         char *meta_start;
 168                         char *meta_end;
 169                         int meta_length;
 170                         char *meta;
 171                         char *meta_http_equiv;
 172                         char *meta_content;
 173                         char *spaceptr;
 174
 175                         meta_start = &ptr[4];
 176                         meta_end = strchr(ptr, '>');
 177                         if ((meta_end != NULL) && (meta_end <= msgend)) {
 178                                 meta_length = meta_end - meta_start + 1;
 179                                 meta = malloc(meta_length + 1);
 180                                 safestrncpy(meta, meta_start, meta_length);
 181                                 meta[meta_length] = 0;
 182                                 striplt(meta);
 183                                 if (!strncasecmp(meta, "HTTP-EQUIV=", 11)) {
 184                                         meta_http_equiv = strdup(&meta[11]);
 185                                         spaceptr = strchr(meta_http_equiv, ' ');
 186                                         if (spaceptr != NULL) {
 187                                                 *spaceptr = 0;
 188                                                 meta_content = strdup(++spaceptr);
 189                                                 if (!strncasecmp(meta_content, "content=", 8)) {
 190                                                         strcpy(meta_content, &meta_content[8]);
 191                                                         stripquotes(meta_http_equiv);
 192                                                         stripquotes(meta_content);
 193                                                         extract_charset_from_meta(charset,
 194                                                                 meta_http_equiv, meta_content);
 195                                                 }
 196                                                 free(meta_content);
 197                                         }
 198                                         free(meta_http_equiv);
 199                                 }
 200                                 free(meta);
 201                         }
 202                 }
 203
 204                 /**
 205                  * Any of these tags cause everything up to and including
 206                  * the tag to be removed.
 207                  */
 208                 if ( (!strncasecmp(ptr, "HTML", 4))
 209                    ||(!strncasecmp(ptr, "HEAD", 4))
 210                    ||(!strncasecmp(ptr, "/HEAD", 5))
 211                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
 212                         ptr = strchr(ptr, '>');
 213                         if ((ptr == NULL) || (ptr >= msgend)) break;
 214                         ++ptr;
 215                         if ((ptr == NULL) || (ptr >= msgend)) break;
 216                         msgstart = ptr;
 217                 }
 218
 219                 /**
 220                  * Any of these tags cause everything including and following
 221                  * the tag to be removed.
 222                  */
 223                 if ( (!strncasecmp(ptr, "/HTML", 5))
 224                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 225                         --ptr;
 226                         msgend = ptr;
 227                         strcpy(ptr, "");
 228
 229                 }
 230
 231                 ++ptr;
 232         }
 233         if (msgstart > msg) {
 234                 strcpy(msg, msgstart);
 235         }
 236
 237         /** Convert foreign character sets to UTF-8 if necessary. */
 238 #ifdef HAVE_ICONV
 239         if ( (strcasecmp(charset, "us-ascii"))
 240            && (strcasecmp(charset, "UTF-8"))
 241            && (strcasecmp(charset, ""))
 242         ) {
 243                 lprintf(9, "Converting %s to UTF-8\n", charset);
 244                 ic = ctdl_iconv_open("UTF-8", charset);
 245                 if (ic == (iconv_t)(-1) ) {
 246                         lprintf(5, "%s:%d iconv_open() failed: %s\n",
 247                                 __FILE__, __LINE__, strerror(errno));
 248                 }
 249         }
 250         if (ic != (iconv_t)(-1) ) {
 251                 ibuf = msg;
 252                 ibuflen = content_length;
 253                 obuflen = content_length + (content_length / 2) ;
 254                 obuf = (char *) malloc(obuflen);
 255                 osav = obuf;
 256                 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
 257                 content_length = content_length + (content_length / 2) - obuflen;
 258                 osav[content_length] = 0;
 259                 free(msg);
 260                 msg = osav;
 261                 iconv_close(ic);
 262         }
 263 #endif
 264
 265         /**
 266          *      At this point, the message has been stripped down to
 267          *      only the content inside the <BODY></BODY> tags, and has
 268          *      been converted to UTF-8 if it was originally in a foreign
 269          *      character set.  The text is also guaranteed to be null
 270          *      terminated now.
 271          */
 272
 273         /** Now go through the message, parsing tags as necessary. */
 274         converted_msg = NewStrBufPlain(NULL, content_length + 8192);
 275         if (converted_msg == NULL) {
 276                 StrBufAppendPrintf(Target, "Error %d: %s<br />%s:%d", errno, strerror(errno), __FILE__, __LINE__);
 277                 goto BAIL;
 278         }
 279
 280         ptr = msg;
 281         msgend = strchr(msg, 0);
 282         while (ptr < msgend) {
 283
 284                 /** Try to sanitize the html of any rogue scripts */
 285                 if (!strncasecmp(ptr, "<script", 7)) {
 286                         if (scriptlevel == 0) {
 287                                 script_start_pos = StrLength(converted_msg);
 288                         }
 289                         ++scriptlevel;
 290                 }
 291                 if (!strncasecmp(ptr, "</script", 8)) {
 292                         --scriptlevel;
 293                 }
 294
 295                 /**
 296                  * Change mailto: links to WebCit mail, by replacing the
 297                  * link with one that points back to our mail room.  Due to
 298                  * the way we parse URL's, it'll even handle mailto: links
 299                  * that have "?subject=" in them.
 300                  */
 301                 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
 302                         content_length += 64;
 303                         StrBufAppendPrintf(converted_msg,
 304                                 "<a href=\"display_enter?force_room=_MAIL_&recp=");
 305                         ptr = &ptr[16];
 306                         ++alevel;
 307                         ++brak;
 308                 }
 309                 /** Make external links open in a separate window */
 310                 else if (!strncasecmp(ptr, "<a href=\"", 9)) {
 311                         ++alevel;
 312                         ++brak;
 313                         if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
 314                              &&  ((strchr(ptr, '/') < strchr(ptr, '>')))
 315                              ) {
 316                                 /* open external links to new window */
 317                                 StrBufAppendPrintf(converted_msg, new_window);
 318                                 ptr = &ptr[8];
 319                         }
 320                         else if ( (treat_as_wiki) && (strncasecmp(ptr, "<a href=\"wiki?", 14)) ) {
 321                                 content_length += 64;
 322                                 StrBufAppendPrintf(converted_msg, "<a href=\"wiki?page=");
 323                                 ptr = &ptr[9];
 324                         }
 325                         else {
 326                                 StrBufAppendPrintf(converted_msg, "<a href=\"");
 327                                 ptr = &ptr[9];
 328                         }
 329                 }
 330                 /** Fixup <img src="cid:... ...> to fetch the mime part */
 331                 else if (!strncasecmp(ptr, "<img ", 5)) {
 332                         char* tag_end=strchr(ptr,'>');
 333                         char* src=strstr(ptr, " src=\"cid:");
 334                         char *cid_start, *cid_end;
 335                         ++brak;
 336
 337                         if (src &&
 338                                 (cid_start=strchr(src,':')) &&
 339                                 (cid_end=strchr(cid_start,'"')) &&
 340                                 (cid_end < tag_end)) {
 341
 342                                 /* copy tag and attributes up to src="cid: */
 343                                 StrBufAppendBufPlain(converted_msg, ptr, src - ptr, 0);
 344                                 cid_start++;
 345
 346                                 /* add in /webcit/mimepart/<msgno>/CID/
 347                                    trailing / stops dumb URL filters getting excited */
 348                                 StrBufAppendPrintf(converted_msg,
 349                                         "src=\"/webcit/mimepart/%d/",msgnum);
 350                                 StrBufAppendBufPlain(converted_msg, cid_start, cid_end - cid_start, 0);
 351                                 StrBufAppendBufPlain(converted_msg, "/\"", -1, 0);
 352
 353                                 ptr = cid_end+1;
 354                         }
 355                         StrBufAppendBufPlain(converted_msg, ptr, tag_end - ptr, 0);
 356                         ptr = tag_end;
 357                 }
 358
 359                 /**
 360                  * Turn anything that looks like a URL into a real link, as long
 361                  * as it's not inside a tag already
 362                  */
 363                 else if ( (brak == 0) && (alevel == 0)
 364                      && (!strncasecmp(ptr, "http://", 7))) {
 365                                 /** Find the end of the link */
 366                                 int strlenptr;
 367                                 linklen = 0;
 368
 369                                 strlenptr = strlen(ptr);
 370                                 for (i=0; i<=strlenptr; ++i) {
 371                                         if ((ptr[i]==0)
 372                                            ||(isspace(ptr[i]))
 373                                            ||(ptr[i]==10)
 374                                            ||(ptr[i]==13)
 375                                            ||(ptr[i]=='(')
 376                                            ||(ptr[i]==')')
 377                                            ||(ptr[i]=='<')
 378                                            ||(ptr[i]=='>')
 379                                            ||(ptr[i]=='[')
 380                                            ||(ptr[i]==']')
 381                                            ||(ptr[i]=='"')
 382                                            ||(ptr[i]=='\'')
 383                                         ) linklen = i;
 384                                         /* did s.b. send us an entity? */
 385                                         if (ptr[i] == '&') {
 386                                                 if ((ptr[i+2] ==';') ||
 387                                                     (ptr[i+3] ==';') ||
 388                                                     (ptr[i+5] ==';') ||
 389                                                     (ptr[i+6] ==';') ||
 390                                                     (ptr[i+7] ==';'))
 391                                                         linklen = i;
 392                                         }
 393                                         if (linklen > 0) break;
 394                                 }
 395                                 if (linklen > 0) {
 396                                         char *ltreviewptr;
 397                                         char *nbspreviewptr;
 398                                         char linkedchar;
 399                                         int len = linklen;
 400
 401                                         len = linklen;
 402                                         linkedchar = ptr[len];
 403                                         ptr[len] = '\0';
 404                                         /* spot for some subject strings tinymce tends to give us. */
 405                                         ltreviewptr = strchr(ptr, '<');
 406                                         if (ltreviewptr != NULL) {
 407                                                 *ltreviewptr = '\0';
 408                                                 linklen = ltreviewptr - ptr;
 409                                         }
 410
 411                                         nbspreviewptr = strstr(ptr, "&nbsp;");
 412                                         if (nbspreviewptr != NULL) {
 413                                                 ///*nbspreviewptr = '\0';
 414                                                 linklen = nbspreviewptr - ptr;
 415                                         }
 416                                         if (ltreviewptr != 0)
 417                                                 *ltreviewptr = '<';
 418
 419                                         ptr[len] = linkedchar;
 420
 421                                         content_length += (32 + linklen);
 422                                         StrBufAppendPrintf(converted_msg, "%s\"", new_window);
 423                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 424                                         StrBufAppendPrintf(converted_msg, "\">");
 425                                         StrBufAppendBufPlain(converted_msg, ptr, linklen, 0);
 426                                         ptr += linklen;
 427                                         StrBufAppendPrintf(converted_msg, "</A>");
 428                                 }
 429                 }
 430                 else {
 431                         StrBufAppendBufPlain(converted_msg, ptr, 1, 0);
 432                         ptr++;
 433                 }
 434
 435                 /**
 436                  * We need to know when we're inside a tag,
 437                  * so we don't turn things that look like URL's into
 438                  * links, when they're already links - or image sources.
 439                  */
 440                 if (*(ptr-1) == '<') {
 441                         ++brak;
 442                 }
 443                 if (*(ptr-1) == '>') {
 444                         --brak;
 445                         if ((scriptlevel == 0) && (script_start_pos >= 0)) {
 446                                 StrBufCutRight(converted_msg, StrLength(converted_msg) - script_start_pos);
 447                                 script_start_pos = (-1);
 448                         }
 449                 }
 450                 if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 451         }
 452
 453         /**     uncomment these two lines to override conversion        */
 454         /**     memcpy(converted_msg, msg, content_length);             */
 455         /**     output_length = content_length;                         */
 456
 457         /** Output our big pile of markup */
 458         StrBufAppendBuf(Target, converted_msg, 0);
 459
 460 BAIL:   /** A little trailing vertical whitespace... */
 461         StrBufAppendPrintf(Target, "<br /><br />\n");
 462
 463         /** Now give back the memory */
 464         FreeStrBuf(&converted_msg);
 465         if (msg != NULL) free(msg);
 466 }
 467
 468 /*@}*/