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>");
  52                         wprintf(_("realloc() error! couldn't get %d bytes: %s"),
  53                                 buffer_length + 1,
  54                                 strerror(errno));
  55                         wprintf("</B><br /><br />\n");
  56                         return;
  57                 }
  58                 strcpy(&msg[content_length], buf);
  59                 content_length += line_length;
  60                 strcpy(&msg[content_length], "\n");
  61                 content_length += 1;
  62         }
  63
  64 #ifdef HAVE_ICONV
  65         if ( (strcasecmp(charset, "us-ascii"))
  66            && (strcasecmp(charset, "UTF-8")) ) {
  67                 ic = iconv_open("UTF-8", charset);
  68                 if (ic == (iconv_t)(-1) ) {
  69                         lprintf(5, "iconv_open() failed: %s\n", strerror(errno));
  70                 }
  71         }
  72         if (ic != (iconv_t)(-1) ) {
  73                 ibuf = msg;
  74                 ibuflen = content_length;
  75                 obuflen = content_length + (content_length / 2) ;
  76                 obuf = (char *) malloc(obuflen);
  77                 osav = obuf;
  78                 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
  79                 content_length = content_length + (content_length / 2) - obuflen;
  80                 osav[content_length] = 0;
  81                 free(msg);
  82                 msg = osav;
  83                 iconv_close(ic);
  84         }
  85 #endif
  86
  87         ptr = msg;
  88         msgstart = msg;
  89         msgend = &msg[content_length];
  90
  91         while (ptr < msgend) {
  92
  93                 /* Advance to next tag */
  94                 ptr = strchr(ptr, '<');
  95                 if ((ptr == NULL) || (ptr >= msgend)) break;
  96                 ++ptr;
  97                 if ((ptr == NULL) || (ptr >= msgend)) break;
  98
  99                 /* Any of these tags cause everything up to and including
 100                  * the tag to be removed.
 101                  */
 102                 if ( (!strncasecmp(ptr, "HTML", 4))
 103                    ||(!strncasecmp(ptr, "HEAD", 4))
 104                    ||(!strncasecmp(ptr, "/HEAD", 5))
 105                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
 106                         ptr = strchr(ptr, '>');
 107                         if ((ptr == NULL) || (ptr >= msgend)) break;
 108                         ++ptr;
 109                         if ((ptr == NULL) || (ptr >= msgend)) break;
 110                         msgstart = ptr;
 111                 }
 112
 113                 /* Any of these tags cause everything including and following
 114                  * the tag to be removed.
 115                  */
 116                 if ( (!strncasecmp(ptr, "/HTML", 5))
 117                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 118                         --ptr;
 119                         msgend = ptr;
 120                         strcpy(ptr, "");
 121
 122                 }
 123
 124                 ++ptr;
 125         }
 126
 127         converted_msg = malloc(content_length);
 128         strcpy(converted_msg, "");
 129         ptr = msgstart;
 130         while (ptr < msgend) {
 131                 /* Change mailto: links to WebCit mail, by replacing the
 132                  * link with one that points back to our mail room.  Due to
 133                  * the way we parse URL's, it'll even handle mailto: links
 134                  * that have "?subject=" in them.
 135                  */
 136                 if (!strncasecmp(ptr, "<A HREF=\"mailto:", 16)) {
 137                         content_length += 64;
 138                         converted_msg = realloc(converted_msg, content_length);
 139                         sprintf(&converted_msg[output_length],
 140                                 "<A HREF=\"/display_enter"
 141                                 "?force_room=_MAIL_&recp=");
 142                         output_length += 47;
 143                         ptr = &ptr[16];
 144                         ++alevel;
 145                 }
 146                 /* Make links open in a separate window */
 147                 else if (!strncasecmp(ptr, "<A HREF=", 8)) {
 148                         content_length += 64;
 149                         converted_msg = realloc(converted_msg, content_length);
 150                         sprintf(&converted_msg[output_length], new_window);
 151                         output_length += strlen(new_window);
 152                         ptr = &ptr[8];
 153                         ++alevel;
 154                 }
 155                 /* Turn anything that looks like a URL into a real link, as long
 156                  * as it's not inside a tag already
 157                  */
 158                 else if ( (brak == 0) && (alevel == 0)
 159                      && (!strncasecmp(ptr, "http://", 7))) {
 160                                 linklen = 0;
 161                                 /* Find the end of the link */
 162                                 for (i=0; i<=strlen(ptr); ++i) {
 163                                         if ((ptr[i]==0)
 164                                            ||(isspace(ptr[i]))
 165                                            ||(ptr[i]==10)
 166                                            ||(ptr[i]==13)
 167                                            ||(ptr[i]=='(')
 168                                            ||(ptr[i]==')')
 169                                            ||(ptr[i]=='<')
 170                                            ||(ptr[i]=='>')
 171                                            ||(ptr[i]=='[')
 172                                            ||(ptr[i]==']')
 173                                         ) linklen = i;
 174                                         if (linklen > 0) break;
 175                                 }
 176                                 if (linklen > 0) {
 177                                         content_length += (32 + linklen);
 178                                         converted_msg = realloc(converted_msg, content_length);
 179                                         sprintf(&converted_msg[output_length], new_window);
 180                                         output_length += strlen(new_window);
 181                                         converted_msg[output_length] = '\"';
 182                                         converted_msg[++output_length] = 0;
 183                                         for (i=0; i<linklen; ++i) {
 184                                                 converted_msg[output_length] = ptr[i];
 185                                                 converted_msg[++output_length] = 0;
 186                                         }
 187                                         sprintf(&converted_msg[output_length], "\">");
 188                                         output_length += 2;
 189                                         for (i=0; i<linklen; ++i) {
 190                                                 converted_msg[output_length] = *ptr++;
 191                                                 converted_msg[++output_length] = 0;
 192                                         }
 193                                         sprintf(&converted_msg[output_length], "</A>");
 194                                         output_length += 4;
 195                                 }
 196                 }
 197                 else {
 198                         /*
 199                          * We need to know when we're inside a tag,
 200                          * so we don't turn things that look like URL's into
 201                          * links, when they're already links - or image sources.
 202                          */
 203                         if (*ptr == '<') ++brak;
 204                         if (*ptr == '>') --brak;
 205                         if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 206                         converted_msg[output_length] = *ptr++;
 207                         converted_msg[++output_length] = 0;
 208                 }
 209         }
 210
 211         /* Output our big pile of markup */
 212         client_write(converted_msg, output_length);
 213
 214         /* A little trailing vertical whitespace... */
 215         wprintf("<br /><br />\n");
 216
 217         /* Now give back the memory */
 218         free(converted_msg);
 219         free(msg);
 220 }
 221