webcit/html2html.c

   1 /*
   2  * $Id$
   3  *
   4  * Output an HTML message, modifying it slightly to make sure it plays nice
   5  * with the rest of our web framework.
   6  *
   7  */
   8
   9 #include "webcit.h"
  10 #include "vcard.h"
  11 #include "webserver.h"
  12
  13
  14 /*
  15  * Sanitize and enhance an HTML message for display.
  16  * Also convert weird character sets to UTF-8 if necessary.
  17  */
  18 void output_html(char *charset) {
  19         char buf[SIZ];
  20         char *msg;
  21         char *ptr;
  22         char *msgstart;
  23         char *msgend;
  24         char *converted_msg;
  25         int buffer_length = 1;
  26         int line_length = 0;
  27         int content_length = 0;
  28         int output_length = 0;
  29         char new_window[SIZ];
  30         int brak = 0;
  31         int alevel = 0;
  32         int i;
  33         int linklen;
  34 #ifdef HAVE_ICONV
  35         iconv_t ic = (iconv_t)(-1) ;
  36         char *ibuf;                   /* Buffer of characters to be converted */
  37         char *obuf;                   /* Buffer for converted characters      */
  38         size_t ibuflen;               /* Length of input buffer               */
  39         size_t obuflen;               /* Length of output buffer              */
  40         char *osav;                   /* Saved pointer to output buffer       */
  41 #endif
  42
  43         msg = strdup("");
  44         sprintf(new_window, "<A TARGET=\"%s\" HREF=", TARGET);
  45
  46         while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
  47                 line_length = strlen(buf);
  48                 buffer_length = content_length + line_length + 2;
  49                 msg = realloc(msg, buffer_length);
  50                 if (msg == NULL) {
  51                         wprintf("<B>realloc() error!  "
  52                                 "couldn't get %d bytes: %s</B><br /><br />\n",
  53                                 buffer_length + 1,
  54                                 strerror(errno));
  55                         return;
  56                 }
  57                 strcpy(&msg[content_length], buf);
  58                 content_length += line_length;
  59                 strcpy(&msg[content_length], "\n");
  60                 content_length += 1;
  61         }
  62
  63 #ifdef HAVE_ICONV
  64         if ( (strcasecmp(charset, "us-ascii"))
  65            && (strcasecmp(charset, "UTF-8")) ) {
  66                 ic = iconv_open("UTF-8", charset);
  67                 if (ic == (iconv_t)(-1) ) {
  68                         lprintf(5, "iconv_open() failed: %s\n", strerror(errno));
  69                 }
  70         }
  71         if (ic != (iconv_t)(-1) ) {
  72                 ibuf = msg;
  73                 ibuflen = content_length;
  74                 obuflen = content_length + (content_length / 2) ;
  75                 obuf = (char *) malloc(obuflen);
  76                 osav = obuf;
  77                 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
  78                 content_length = content_length + (content_length / 2) - obuflen;
  79                 osav[content_length] = 0;
  80                 free(msg);
  81                 msg = osav;
  82                 iconv_close(ic);
  83         }
  84 #endif
  85
  86         ptr = msg;
  87         msgstart = msg;
  88         msgend = &msg[content_length];
  89
  90         while (ptr < msgend) {
  91
  92                 /* Advance to next tag */
  93                 ptr = strchr(ptr, '<');
  94                 if ((ptr == NULL) || (ptr >= msgend)) break;
  95                 ++ptr;
  96                 if ((ptr == NULL) || (ptr >= msgend)) break;
  97
  98                 /* Any of these tags cause everything up to and including
  99                  * the tag to be removed.
 100                  */
 101                 if ( (!strncasecmp(ptr, "HTML", 4))
 102                    ||(!strncasecmp(ptr, "HEAD", 4))
 103                    ||(!strncasecmp(ptr, "/HEAD", 5))
 104                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
 105                         ptr = strchr(ptr, '>');
 106                         if ((ptr == NULL) || (ptr >= msgend)) break;
 107                         ++ptr;
 108                         if ((ptr == NULL) || (ptr >= msgend)) break;
 109                         msgstart = ptr;
 110                 }
 111
 112                 /* Any of these tags cause everything including and following
 113                  * the tag to be removed.
 114                  */
 115                 if ( (!strncasecmp(ptr, "/HTML", 5))
 116                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 117                         --ptr;
 118                         msgend = ptr;
 119                         strcpy(ptr, "");
 120
 121                 }
 122
 123                 ++ptr;
 124         }
 125
 126         converted_msg = malloc(content_length);
 127         strcpy(converted_msg, "");
 128         ptr = msgstart;
 129         while (ptr < msgend) {
 130                 /* Change mailto: links to WebCit mail, by replacing the
 131                  * link with one that points back to our mail room.  Due to
 132                  * the way we parse URL's, it'll even handle mailto: links
 133                  * that have "?subject=" in them.
 134                  */
 135                 if (!strncasecmp(ptr, "<A HREF=\"mailto:", 16)) {
 136                         content_length += 64;
 137                         converted_msg = realloc(converted_msg, content_length);
 138                         sprintf(&converted_msg[output_length],
 139                                 "<A HREF=\"/display_enter"
 140                                 "?force_room=_MAIL_&recp=");
 141                         output_length += 47;
 142                         ptr = &ptr[16];
 143                         ++alevel;
 144                 }
 145                 /* Make links open in a separate window */
 146                 else if (!strncasecmp(ptr, "<A HREF=", 8)) {
 147                         content_length += 64;
 148                         converted_msg = realloc(converted_msg, content_length);
 149                         sprintf(&converted_msg[output_length], new_window);
 150                         output_length += strlen(new_window);
 151                         ptr = &ptr[8];
 152                         ++alevel;
 153                 }
 154                 /* Turn anything that looks like a URL into a real link, as long
 155                  * as it's not inside a tag already
 156                  */
 157                 else if ( (brak == 0) && (alevel == 0)
 158                      && (!strncasecmp(ptr, "http://", 7))) {
 159                                 linklen = 0;
 160                                 /* Find the end of the link */
 161                                 for (i=0; i<=strlen(ptr); ++i) {
 162                                         if ((ptr[i]==0)
 163                                            ||(isspace(ptr[i]))
 164                                            ||(ptr[i]==10)
 165                                            ||(ptr[i]==13)
 166                                            ||(ptr[i]=='(')
 167                                            ||(ptr[i]==')')
 168                                            ||(ptr[i]=='<')
 169                                            ||(ptr[i]=='>')
 170                                            ||(ptr[i]=='[')
 171                                            ||(ptr[i]==']')
 172                                         ) linklen = i;
 173                                         if (linklen > 0) break;
 174                                 }
 175                                 if (linklen > 0) {
 176                                         content_length += (32 + linklen);
 177                                         converted_msg = realloc(converted_msg, content_length);
 178                                         sprintf(&converted_msg[output_length], new_window);
 179                                         output_length += strlen(new_window);
 180                                         converted_msg[output_length] = '\"';
 181                                         converted_msg[++output_length] = 0;
 182                                         for (i=0; i<linklen; ++i) {
 183                                                 converted_msg[output_length] = ptr[i];
 184                                                 converted_msg[++output_length] = 0;
 185                                         }
 186                                         sprintf(&converted_msg[output_length], "\">");
 187                                         output_length += 2;
 188                                         for (i=0; i<linklen; ++i) {
 189                                                 converted_msg[output_length] = *ptr++;
 190                                                 converted_msg[++output_length] = 0;
 191                                         }
 192                                         sprintf(&converted_msg[output_length], "</A>");
 193                                         output_length += 4;
 194                                 }
 195                 }
 196                 else {
 197                         /*
 198                          * We need to know when we're inside a tag,
 199                          * so we don't turn things that look like URL's into
 200                          * links, when they're already links - or image sources.
 201                          */
 202                         if (*ptr == '<') ++brak;
 203                         if (*ptr == '>') --brak;
 204                         if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 205                         converted_msg[output_length] = *ptr++;
 206                         converted_msg[++output_length] = 0;
 207                 }
 208         }
 209
 210         /* Output our big pile of markup */
 211         client_write(converted_msg, output_length);
 212
 213         /* A little trailing vertical whitespace... */
 214         wprintf("<br /><br />\n");
 215
 216         /* Now give back the memory */
 217         free(converted_msg);
 218         free(msg);
 219 }
 220