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                  * The brain-damaged webmail program in Microsoft Exchange declares
  65                  * a charset of "unicode" when they really mean "UTF-8".  GNU iconv
  66                  * treats "unicode" as an alias for "UTF-16" so we have to manually
  67                  * fix this here, otherwise messages generated in Exchange webmail
  68                  * show up as a big pile of weird characters.
  69                  */
  70                 if (!strcasecmp(charset, "unicode")) {
  71                         strcpy(charset, "UTF-8");
  72                 }
  73
  74         }
  75 }
  76
  77
  78
  79 /**
  80  * \brief Sanitize and enhance an HTML message for display.
  81  *        Also convert weird character sets to UTF-8 if necessary.
  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) {
  86         char buf[SIZ];
  87         char *msg;
  88         char *ptr;
  89         char *msgstart;
  90         char *msgend;
  91         char *converted_msg;
  92         size_t converted_alloc = 0;
  93         int buffer_length = 1;
  94         int line_length = 0;
  95         int content_length = 0;
  96         int output_length = 0;
  97         char new_window[SIZ];
  98         int brak = 0;
  99         int alevel = 0;
 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;
 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_alloc = content_length + 8192;
 268         converted_msg = malloc(converted_alloc);
 269         if (converted_msg == NULL) {
 270                 wprintf("Error %d: %s<br />%s:%s", errno, strerror(errno), __FILE__, __LINE__);
 271                 goto BAIL;
 272         }
 273
 274         strcpy(converted_msg, "");
 275         ptr = msg;
 276         msgend = strchr(msg, 0);
 277         while (ptr < msgend) {
 278
 279                 /**
 280                  * Change mailto: links to WebCit mail, by replacing the
 281                  * link with one that points back to our mail room.  Due to
 282                  * the way we parse URL's, it'll even handle mailto: links
 283                  * that have "?subject=" in them.
 284                  */
 285                 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
 286                         content_length += 64;
 287                         if (content_length >= converted_alloc) {
 288                                 converted_alloc += 8192;
 289                                 converted_msg = realloc(converted_msg, converted_alloc);
 290                                 if (converted_msg == NULL) {
 291                                         abort();
 292                                 }
 293                         }
 294                         sprintf(&converted_msg[output_length],
 295                                 "<a href=\"display_enter"
 296                                 "?force_room=_MAIL_&recp=");
 297                         output_length += 47;
 298                         ptr = &ptr[16];
 299                         ++alevel;
 300                 }
 301                 /** Make external links open in a separate window */
 302                 else if (!strncasecmp(ptr, "<a href=\"", 9)) {
 303                         ++alevel;
 304                         if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
 305                              &&  ((strchr(ptr, '/') < strchr(ptr, '>')))
 306                              ) {
 307                                 /* open external links to new window */
 308                                 content_length += 64;
 309                                 if (content_length >= converted_alloc) {
 310                                         converted_alloc += 8192;
 311                                         converted_msg = realloc(converted_msg, converted_alloc);
 312                                         if (converted_msg == NULL) {
 313                                                 abort();
 314                                         }
 315                                 }
 316                                 sprintf(&converted_msg[output_length], new_window);
 317                                 output_length += strlen(new_window);
 318                                 ptr = &ptr[8];
 319                         }
 320                         else if ( (treat_as_wiki) && (strncasecmp(ptr, "<a href=\"wiki?", 14)) ) {
 321                                 content_length += 64;
 322                                 if (content_length >= converted_alloc) {
 323                                         converted_alloc += 8192;
 324                                         converted_msg = realloc(converted_msg, converted_alloc);
 325                                         if (converted_msg == NULL) {
 326                                                 abort();
 327                                         }
 328                                 }
 329                                 sprintf(&converted_msg[output_length], "<a href=\"wiki?page=");
 330                                 output_length += 19;
 331                                 ptr = &ptr[9];
 332                         }
 333                         else {
 334                                 sprintf(&converted_msg[output_length], "<a href=\"");
 335                                 output_length += 9;
 336                                 ptr = &ptr[9];
 337                         }
 338                 }
 339                 /**
 340                  * Turn anything that looks like a URL into a real link, as long
 341                  * as it's not inside a tag already
 342                  */
 343                 else if ( (brak == 0) && (alevel == 0)
 344                      && (!strncasecmp(ptr, "http://", 7))) {
 345                                 linklen = 0;
 346                                 /** Find the end of the link */
 347                                 for (i=0; i<=strlen(ptr); ++i) {
 348                                         if ((ptr[i]==0)
 349                                            ||(isspace(ptr[i]))
 350                                            ||(ptr[i]==10)
 351                                            ||(ptr[i]==13)
 352                                            ||(ptr[i]=='(')
 353                                            ||(ptr[i]==')')
 354                                            ||(ptr[i]=='<')
 355                                            ||(ptr[i]=='>')
 356                                            ||(ptr[i]=='[')
 357                                            ||(ptr[i]==']')
 358                                         ) linklen = i;
 359                                         if (linklen > 0) break;
 360                                 }
 361                                 if (linklen > 0) {
 362                                         content_length += (32 + linklen);
 363                                         if (content_length >= converted_alloc) {
 364                                                 converted_alloc += 8192;
 365                                                 converted_msg = realloc(converted_msg, converted_alloc);
 366                                                 if (converted_msg == NULL) {
 367                                                         abort();
 368                                                 }
 369                                         }
 370                                         sprintf(&converted_msg[output_length], new_window);
 371                                         output_length += strlen(new_window);
 372                                         converted_msg[output_length] = '\"';
 373                                         converted_msg[++output_length] = 0;
 374                                         for (i=0; i<linklen; ++i) {
 375                                                 converted_msg[output_length] = ptr[i];
 376                                                 converted_msg[++output_length] = 0;
 377                                         }
 378                                         sprintf(&converted_msg[output_length], "\">");
 379                                         output_length += 2;
 380                                         for (i=0; i<linklen; ++i) {
 381                                                 converted_msg[output_length] = *ptr++;
 382                                                 converted_msg[++output_length] = 0;
 383                                         }
 384                                         sprintf(&converted_msg[output_length], "</A>");
 385                                         output_length += 4;
 386                                 }
 387                 }
 388                 else {
 389                         /**
 390                          * We need to know when we're inside a tag,
 391                          * so we don't turn things that look like URL's into
 392                          * links, when they're already links - or image sources.
 393                          */
 394                         if (*ptr == '<') ++brak;
 395                         if (*ptr == '>') --brak;
 396                         if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 397                         converted_msg[output_length] = *ptr++;
 398                         converted_msg[++output_length] = 0;
 399                 }
 400         }
 401
 402         /**     uncomment these two lines to override conversion        */
 403         /**     memcpy(converted_msg, msg, content_length);             */
 404         /**     output_length = content_length;                         */
 405
 406         /** Output our big pile of markup */
 407         client_write(converted_msg, output_length);
 408
 409 BAIL:   /** A little trailing vertical whitespace... */
 410         wprintf("<br /><br />\n");
 411
 412         /** Now give back the memory */
 413         if (converted_msg != NULL) free(converted_msg);
 414         if (msg != NULL) free(msg);
 415 }
 416
 417 /*@}*/