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 Sanitize and enhance an HTML message for display.
  17  * Also convert weird character sets to UTF-8 if necessary.
  18  * \param charset the input charset
  19  */
  20 void output_html(char *charset, int treat_as_wiki) {
  21         char buf[SIZ];
  22         char *msg;
  23         char *ptr;
  24         char *msgstart;
  25         char *msgend;
  26         char *converted_msg;
  27         int buffer_length = 1;
  28         int line_length = 0;
  29         int content_length = 0;
  30         int output_length = 0;
  31         char new_window[SIZ];
  32         int brak = 0;
  33         int alevel = 0;
  34         int i;
  35         int linklen;
  36 #ifdef HAVE_ICONV
  37         iconv_t ic = (iconv_t)(-1) ;
  38         char *ibuf;                   /**< Buffer of characters to be converted */
  39         char *obuf;                   /**< Buffer for converted characters      */
  40         size_t ibuflen;               /**< Length of input buffer               */
  41         size_t obuflen;               /**< Length of output buffer              */
  42         char *osav;                   /**< Saved pointer to output buffer       */
  43 #endif
  44
  45         msg = strdup("");
  46         sprintf(new_window, "<a target=\"%s\" href=", TARGET);
  47
  48         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
  49                 line_length = strlen(buf);
  50                 buffer_length = content_length + line_length + 2;
  51                 msg = realloc(msg, buffer_length);
  52                 if (msg == NULL) {
  53                         wprintf("<b>");
  54                         wprintf(_("realloc() error! couldn't get %d bytes: %s"),
  55                                 buffer_length + 1,
  56                                 strerror(errno));
  57                         wprintf("</b><br /><br />\n");
  58                         return;
  59                 }
  60                 strcpy(&msg[content_length], buf);
  61                 content_length += line_length;
  62                 strcpy(&msg[content_length], "\n");
  63                 content_length += 1;
  64         }
  65
  66 #ifdef HAVE_ICONV
  67         if ( (strcasecmp(charset, "us-ascii"))
  68            && (strcasecmp(charset, "UTF-8"))
  69            && (strcasecmp(charset, ""))
  70         ) {
  71                 ic = iconv_open("UTF-8", charset);
  72                 if (ic == (iconv_t)(-1) ) {
  73                         lprintf(5, "%s:%d iconv_open() failed: %s\n",
  74                                 __FILE__, __LINE__, strerror(errno));
  75                 }
  76         }
  77         if (ic != (iconv_t)(-1) ) {
  78                 ibuf = msg;
  79                 ibuflen = content_length;
  80                 obuflen = content_length + (content_length / 2) ;
  81                 obuf = (char *) malloc(obuflen);
  82                 osav = obuf;
  83                 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
  84                 content_length = content_length + (content_length / 2) - obuflen;
  85                 osav[content_length] = 0;
  86                 free(msg);
  87                 msg = osav;
  88                 iconv_close(ic);
  89         }
  90 #endif
  91
  92         ptr = msg;
  93         msgstart = msg;
  94         msgend = &msg[content_length];
  95
  96         while (ptr < msgend) {
  97
  98                 /** Advance to next tag */
  99                 ptr = strchr(ptr, '<');
 100                 if ((ptr == NULL) || (ptr >= msgend)) break;
 101                 ++ptr;
 102                 if ((ptr == NULL) || (ptr >= msgend)) break;
 103
 104                 /**
 105                  * Any of these tags cause everything up to and including
 106                  * the tag to be removed.
 107                  */
 108                 if ( (!strncasecmp(ptr, "HTML", 4))
 109                    ||(!strncasecmp(ptr, "HEAD", 4))
 110                    ||(!strncasecmp(ptr, "/HEAD", 5))
 111                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
 112                         ptr = strchr(ptr, '>');
 113                         if ((ptr == NULL) || (ptr >= msgend)) break;
 114                         ++ptr;
 115                         if ((ptr == NULL) || (ptr >= msgend)) break;
 116                         msgstart = ptr;
 117                 }
 118
 119                 /**
 120                  * Any of these tags cause everything including and following
 121                  * the tag to be removed.
 122                  */
 123                 if ( (!strncasecmp(ptr, "/HTML", 5))
 124                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 125                         --ptr;
 126                         msgend = ptr;
 127                         strcpy(ptr, "");
 128
 129                 }
 130
 131                 ++ptr;
 132         }
 133
 134         converted_msg = malloc(content_length);
 135         strcpy(converted_msg, "");
 136         ptr = msgstart;
 137         while (ptr < msgend) {
 138                 /**
 139                  * Change mailto: links to WebCit mail, by replacing the
 140                  * link with one that points back to our mail room.  Due to
 141                  * the way we parse URL's, it'll even handle mailto: links
 142                  * that have "?subject=" in them.
 143                  */
 144                 if (!strncasecmp(ptr, "<a href=\"mailto:", 16)) {
 145                         content_length += 64;
 146                         converted_msg = realloc(converted_msg, content_length);
 147                         sprintf(&converted_msg[output_length],
 148                                 "<a href=\"display_enter"
 149                                 "?force_room=_MAIL_&recp=");
 150                         output_length += 47;
 151                         ptr = &ptr[16];
 152                         ++alevel;
 153                 }
 154                 /** Make external links open in a separate window */
 155                 else if (!strncasecmp(ptr, "<a href=", 8)) {
 156                         ++alevel;
 157                         if ( ((strchr(ptr, ':') < strchr(ptr, '/')))
 158                              &&  ((strchr(ptr, '/') < strchr(ptr, '>')))
 159                              ) {
 160                                 /* open external links to new window */
 161                                 content_length += 64;
 162                                 converted_msg = realloc(converted_msg, content_length);
 163                                 sprintf(&converted_msg[output_length], new_window);
 164                                 output_length += strlen(new_window);
 165                                 ptr = &ptr[8];
 166                         }
 167                         else if ( (treat_as_wiki) && (strncasecmp(ptr, "<a href=\"wiki?", 14)) ) {
 168                                 lprintf(9, "converting wiki link\n");
 169                                 content_length += 64;
 170                                 converted_msg = realloc(converted_msg, content_length);
 171                                 sprintf(&converted_msg[output_length], "<a href=\"wiki?page=");
 172                                 output_length += 19;
 173                                 ptr = &ptr[9];
 174                         }
 175                         else {
 176                                 sprintf(&converted_msg[output_length], "<a href=");
 177                                 output_length += 8;
 178                                 ptr = &ptr[8];
 179                         }
 180                 }
 181                 /**
 182                  * Turn anything that looks like a URL into a real link, as long
 183                  * as it's not inside a tag already
 184                  */
 185                 else if ( (brak == 0) && (alevel == 0)
 186                      && (!strncasecmp(ptr, "http://", 7))) {
 187                                 linklen = 0;
 188                                 /** Find the end of the link */
 189                                 for (i=0; i<=strlen(ptr); ++i) {
 190                                         if ((ptr[i]==0)
 191                                            ||(isspace(ptr[i]))
 192                                            ||(ptr[i]==10)
 193                                            ||(ptr[i]==13)
 194                                            ||(ptr[i]=='(')
 195                                            ||(ptr[i]==')')
 196                                            ||(ptr[i]=='<')
 197                                            ||(ptr[i]=='>')
 198                                            ||(ptr[i]=='[')
 199                                            ||(ptr[i]==']')
 200                                         ) linklen = i;
 201                                         if (linklen > 0) break;
 202                                 }
 203                                 if (linklen > 0) {
 204                                         content_length += (32 + linklen);
 205                                         converted_msg = realloc(converted_msg, content_length);
 206                                         sprintf(&converted_msg[output_length], new_window);
 207                                         output_length += strlen(new_window);
 208                                         converted_msg[output_length] = '\"';
 209                                         converted_msg[++output_length] = 0;
 210                                         for (i=0; i<linklen; ++i) {
 211                                                 converted_msg[output_length] = ptr[i];
 212                                                 converted_msg[++output_length] = 0;
 213                                         }
 214                                         sprintf(&converted_msg[output_length], "\">");
 215                                         output_length += 2;
 216                                         for (i=0; i<linklen; ++i) {
 217                                                 converted_msg[output_length] = *ptr++;
 218                                                 converted_msg[++output_length] = 0;
 219                                         }
 220                                         sprintf(&converted_msg[output_length], "</A>");
 221                                         output_length += 4;
 222                                 }
 223                 }
 224                 else {
 225                         /**
 226                          * We need to know when we're inside a tag,
 227                          * so we don't turn things that look like URL's into
 228                          * links, when they're already links - or image sources.
 229                          */
 230                         if (*ptr == '<') ++brak;
 231                         if (*ptr == '>') --brak;
 232                         if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 233                         converted_msg[output_length] = *ptr++;
 234                         converted_msg[++output_length] = 0;
 235                 }
 236         }
 237
 238         /** Output our big pile of markup */
 239         client_write(converted_msg, output_length);
 240
 241         /** A little trailing vertical whitespace... */
 242         wprintf("<br /><br />\n");
 243
 244         /** Now give back the memory */
 245         free(converted_msg);
 246         free(msg);
 247 }
 248
 249 /*@}*/