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