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 "vcard.h"
  12 #include "webserver.h"
  13
  14
  15 /**
  16  * \brief       Strip surrounding single or double quotes from a string.
  17  *
  18  * \param s     String to be stripped.
  19  */
  20 void stripquotes(char *s)
  21 {
  22         int len;
  23
  24         if (!s) return;
  25
  26         len = strlen(s);
  27         if (len < 2) return;
  28
  29         if ( ( (s[0] == '\"') && (s[len-1] == '\"') ) || ( (s[0] == '\'') && (s[len-1] == '\'') ) ) {
  30                 s[len-1] = 0;
  31                 strcpy(s, &s[1]);
  32         }
  33 }
  34
  35
  36 /**
  37  * \brief Check to see if a META tag has overridden the declared MIME character set.
  38  *
  39  * \param charset               Character set name (left unchanged if we don't do anything)
  40  * \param meta_http_equiv       Content of the "http-equiv" portion of the META tag
  41  * \param meta_content          Content of the "content" portion of the META tag
  42  */
  43 void extract_charset_from_meta(char *charset, char *meta_http_equiv, char *meta_content)
  44 {
  45         char *ptr;
  46         char buf[64];
  47
  48         if (!charset) return;
  49         if (!meta_http_equiv) return;
  50         if (!meta_content) return;
  51
  52
  53         if (strcasecmp(meta_http_equiv, "Content-type")) return;
  54
  55         ptr = strchr(meta_content, ';');
  56         if (!ptr) return;
  57
  58         safestrncpy(buf, ++ptr, sizeof buf);
  59         striplt(buf);
  60         if (!strncasecmp(buf, "charset=", 8)) {
  61                 strcpy(charset, &buf[8]);
  62         }
  63 }
  64
  65
  66
  67 /**
  68  * \brief Sanitize and enhance an HTML message for display.
  69  *        Also convert weird character sets to UTF-8 if necessary.
  70  *
  71  * \param supplied_charset the input charset as declared in the MIME headers
  72  */
  73 void output_html(char *supplied_charset, int treat_as_wiki) {
  74         char buf[SIZ];
  75         char *msg;
  76         char *ptr;
  77         char *msgstart;
  78         char *msgend;
  79         char *converted_msg;
  80         size_t converted_alloc = 0;
  81         int buffer_length = 1;
  82         int line_length = 0;
  83         int content_length = 0;
  84         int output_length = 0;
  85         char new_window[SIZ];
  86         int brak = 0;
  87         int alevel = 0;
  88         int i;
  89         int linklen;
  90         char charset[128];
  91 #ifdef HAVE_ICONV
  92         iconv_t ic = (iconv_t)(-1) ;
  93         char *ibuf;                   /**< Buffer of characters to be converted */
  94         char *obuf;                   /**< Buffer for converted characters      */
  95         size_t ibuflen;               /**< Length of input buffer               */
  96         size_t obuflen;               /**< Length of output buffer              */
  97         char *osav;                   /**< Saved pointer to output buffer       */
  98 #endif
  99
 100         safestrncpy(charset, supplied_charset, sizeof charset);
 101         msg = strdup("");
 102         sprintf(new_window, "<a target=\"%s\" href=", TARGET);
 103
 104         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 105                 line_length = strlen(buf);
 106                 buffer_length = content_length + line_length + 2;
 107                 ptr = realloc(msg, buffer_length);
 108                 if (ptr == NULL) {
 109                         wprintf("<b>");
 110                         wprintf(_("realloc() error! couldn't get %d bytes: %s"),
 111                                 buffer_length + 1,
 112                                 strerror(errno));
 113                         wprintf("</b><br /><br />\n");
 114                         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
 115                                 /** flush */
 116                         }
 117                         free(msg);
 118                         return;
 119                 }
 120                 msg = ptr;
 121                 strcpy(&msg[content_length], buf);
 122                 content_length += line_length;
 123                 strcpy(&msg[content_length], "\n");
 124                 content_length += 1;
 125         }
 126
 127         /** Do a first pass to isolate the message body */
 128         ptr = msg;
 129         msgstart = msg;
 130         msgend = &msg[content_length];
 131
 132         while (ptr < msgend) {
 133
 134                 /** Advance to next tag */
 135                 ptr = strchr(ptr, '<');
 136                 if ((ptr == NULL) || (ptr >= msgend)) break;
 137                 ++ptr;
 138                 if ((ptr == NULL) || (ptr >= msgend)) break;
 139
 140                 /**
 141                  *  Look for META tags.  Some messages (particularly in
 142                  *  Asian locales) illegally declare a message's character
 143                  *  set in the HTML instead of in the MIME headers.  This
 144                  *  is wrong but we have to work around it anyway.
 145                  */
 146                 if (!strncasecmp(ptr, "META", 4)) {
 147
 148                         char *meta_start;
 149                         char *meta_end;
 150                         int meta_length;
 151                         char *meta;
 152                         char *meta_http_equiv;
 153                         char *meta_content;
 154                         char *spaceptr;
 155
 156                         meta_start = &ptr[4];
 157                         meta_end = strchr(ptr, '>');
 158                         if ((meta_end != NULL) && (meta_end <= msgend)) {
 159                                 meta_length = meta_end - meta_start + 1;
 160                                 meta = malloc(meta_length + 1);
 161                                 safestrncpy(meta, meta_start, meta_length);
 162                                 meta[meta_length] = 0;
 163                                 striplt(meta);
 164                                 if (!strncasecmp(meta, "HTTP-EQUIV=", 11)) {
 165                                         meta_http_equiv = strdup(&meta[11]);
 166                                         spaceptr = strchr(meta_http_equiv, ' ');
 167                                         if (spaceptr != NULL) {
 168                                                 *spaceptr = 0;
 169                                                 meta_content = strdup(++spaceptr);
 170                                                 if (!strncasecmp(meta_content, "content=", 8)) {
 171                                                         strcpy(meta_content, &meta_content[8]);
 172                                                         stripquotes(meta_http_equiv);
 173                                                         stripquotes(meta_content);
 174                                                         extract_charset_from_meta(charset,
 175                                                                 meta_http_equiv, meta_content);
 176                                                 }
 177                                                 free(meta_content);
 178                                         }
 179                                         free(meta_http_equiv);
 180                                 }
 181                                 free(meta);
 182                         }
 183                 }
 184
 185                 /**
 186                  * Any of these tags cause everything up to and including
 187                  * the tag to be removed.
 188                  */
 189                 if ( (!strncasecmp(ptr, "HTML", 4))
 190                    ||(!strncasecmp(ptr, "HEAD", 4))
 191                    ||(!strncasecmp(ptr, "/HEAD", 5))
 192                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
 193                         ptr = strchr(ptr, '>');
 194                         if ((ptr == NULL) || (ptr >= msgend)) break;
 195                         ++ptr;
 196                         if ((ptr == NULL) || (ptr >= msgend)) break;
 197                         msgstart = ptr;
 198                 }
 199
 200                 /**
 201                  * Any of these tags cause everything including and following
 202                  * the tag to be removed.
 203                  */
 204                 if ( (!strncasecmp(ptr, "/HTML", 5))
 205                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 206                         --ptr;
 207                         msgend = ptr;
 208                         strcpy(ptr, "");
 209
 210                 }
 211
 212                 ++ptr;
 213         }
 214         if (msgstart > msg) {
 215                 strcpy(msg, msgstart);
 216         }
 217
 218         /** Convert foreign character sets to UTF-8 if necessary. */
 219 #ifdef HAVE_ICONV
 220         if ( (strcasecmp(charset, "us-ascii"))
 221            && (strcasecmp(charset, "UTF-8"))
 222            && (strcasecmp(charset, ""))
 223         ) {
 224                 lprintf(9, "Converting %s to UTF-8\n", charset);
 225                 ic = ctdl_iconv_open("UTF-8", charset);
 226                 if (ic == (iconv_t)(-1) ) {
 227                         lprintf(5, "%s:%d iconv_open() failed: %s\n",
 228                                 __FILE__, __LINE__, strerror(errno));
 229                 }
 230         }
 231         if (ic != (iconv_t)(-1) ) {
 232                 ibuf = msg;
 233                 ibuflen = content_length;
 234                 obuflen = content_length + (content_length / 2) ;
 235                 obuf = (char *) malloc(obuflen);
 236                 osav = obuf;
 237                 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
 238                 content_length = content_length + (content_length / 2) - obuflen;
 239                 osav[content_length] = 0;
 240                 free(msg);
 241                 msg = osav;
 242                 iconv_close(ic);
 243         }
 244 #endif
 245
 246         /**
 247          *      At this point, the message has been stripped down to
 248          *      only the content inside the <BODY></BODY> tags, and has
 249          *      been converted to UTF-8 if it was originally in a foreign
 250          *      character set.  The text is also guaranteed to be null
 251          *      terminated now.
 252          */
 253
 254         /** Now go through the message, parsing tags as necessary. */
 255         converted_alloc = content_length + 8192;
 256         converted_msg = malloc(converted_alloc);
 257         if (converted_msg == NULL) {
 258                 wprintf("Error %d: %s<br />%s:%s", errno, strerror(errno), __FILE__, __LINE__);
 259                 goto BAIL;
 260         }
 261
 262         strcpy(converted_msg, "");
 263         ptr = msg;
 264         msgend = strchr(msg, 0);
 265         while (ptr < msgend) {
 266
 267                 /**
 268                  * Change mailto: links to WebCit mail, by replacing the
 269                  * link with one that points back to our mail room.  Due to
 270                  * the way we parse URL's, it'll even handle mailto: links
 271                  * that have "?subject=" in them.
 272                  */
 273                 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
 274                         content_length += 64;
 275                         if (content_length >= converted_alloc) {
 276                                 converted_alloc += 8192;
 277                                 converted_msg = realloc(converted_msg, converted_alloc);
 278                                 if (converted_msg == NULL) {
 279                                         abort();
 280                                 }
 281                         }
 282                         sprintf(&converted_msg[output_length],
 283                                 "<a href=\"display_enter"
 284                                 "?force_room=_MAIL_&recp=");
 285                         output_length += 47;
 286                         ptr = &ptr[16];
 287                         ++alevel;
 288                 }
 289                 /** Make external links open in a separate window */
 290                 else if (!strncasecmp(ptr, "<a href=\"", 9)) {
 291                         ++alevel;
 292                         if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
 293                              &&  ((strchr(ptr, '/') < strchr(ptr, '>')))
 294                              ) {
 295                                 /* open external links to new window */
 296                                 content_length += 64;
 297                                 if (content_length >= converted_alloc) {
 298                                         converted_alloc += 8192;
 299                                         converted_msg = realloc(converted_msg, converted_alloc);
 300                                         if (converted_msg == NULL) {
 301                                                 abort();
 302                                         }
 303                                 }
 304                                 sprintf(&converted_msg[output_length], new_window);
 305                                 output_length += strlen(new_window);
 306                                 ptr = &ptr[8];
 307                         }
 308                         else if ( (treat_as_wiki) && (strncasecmp(ptr, "<a href=\"wiki?", 14)) ) {
 309                                 content_length += 64;
 310                                 if (content_length >= converted_alloc) {
 311                                         converted_alloc += 8192;
 312                                         converted_msg = realloc(converted_msg, converted_alloc);
 313                                         if (converted_msg == NULL) {
 314                                                 abort();
 315                                         }
 316                                 }
 317                                 sprintf(&converted_msg[output_length], "<a href=\"wiki?page=");
 318                                 output_length += 19;
 319                                 ptr = &ptr[9];
 320                         }
 321                         else {
 322                                 sprintf(&converted_msg[output_length], "<a href=\"");
 323                                 output_length += 9;
 324                                 ptr = &ptr[9];
 325                         }
 326                 }
 327                 /**
 328                  * Turn anything that looks like a URL into a real link, as long
 329                  * as it's not inside a tag already
 330                  */
 331                 else if ( (brak == 0) && (alevel == 0)
 332                      && (!strncasecmp(ptr, "http://", 7))) {
 333                                 linklen = 0;
 334                                 /** Find the end of the link */
 335                                 for (i=0; i<=strlen(ptr); ++i) {
 336                                         if ((ptr[i]==0)
 337                                            ||(isspace(ptr[i]))
 338                                            ||(ptr[i]==10)
 339                                            ||(ptr[i]==13)
 340                                            ||(ptr[i]=='(')
 341                                            ||(ptr[i]==')')
 342                                            ||(ptr[i]=='<')
 343                                            ||(ptr[i]=='>')
 344                                            ||(ptr[i]=='[')
 345                                            ||(ptr[i]==']')
 346                                         ) linklen = i;
 347                                         if (linklen > 0) break;
 348                                 }
 349                                 if (linklen > 0) {
 350                                         content_length += (32 + linklen);
 351                                         if (content_length >= converted_alloc) {
 352                                                 converted_alloc += 8192;
 353                                                 converted_msg = realloc(converted_msg, converted_alloc);
 354                                                 if (converted_msg == NULL) {
 355                                                         abort();
 356                                                 }
 357                                         }
 358                                         sprintf(&converted_msg[output_length], new_window);
 359                                         output_length += strlen(new_window);
 360                                         converted_msg[output_length] = '\"';
 361                                         converted_msg[++output_length] = 0;
 362                                         for (i=0; i<linklen; ++i) {
 363                                                 converted_msg[output_length] = ptr[i];
 364                                                 converted_msg[++output_length] = 0;
 365                                         }
 366                                         sprintf(&converted_msg[output_length], "\">");
 367                                         output_length += 2;
 368                                         for (i=0; i<linklen; ++i) {
 369                                                 converted_msg[output_length] = *ptr++;
 370                                                 converted_msg[++output_length] = 0;
 371                                         }
 372                                         sprintf(&converted_msg[output_length], "</A>");
 373                                         output_length += 4;
 374                                 }
 375                 }
 376                 else {
 377                         /**
 378                          * We need to know when we're inside a tag,
 379                          * so we don't turn things that look like URL's into
 380                          * links, when they're already links - or image sources.
 381                          */
 382                         if (*ptr == '<') ++brak;
 383                         if (*ptr == '>') --brak;
 384                         if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 385                         converted_msg[output_length] = *ptr++;
 386                         converted_msg[++output_length] = 0;
 387                 }
 388         }
 389
 390         /**     uncomment these two lines to override conversion        */
 391         /**     memcpy(converted_msg, msg, content_length);             */
 392         /**     output_length = content_length;                         */
 393
 394         /** Output our big pile of markup */
 395         client_write(converted_msg, output_length);
 396
 397 BAIL:   /** A little trailing vertical whitespace... */
 398         wprintf("<br /><br />\n");
 399
 400         /** Now give back the memory */
 401         if (converted_msg != NULL) free(converted_msg);
 402         if (msg != NULL) free(msg);
 403 }
 404
 405 /*@}*/