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            && (strcasecmp(charset, ""))
  68         ) {
  69                 ic = iconv_open("UTF-8", charset);
  70                 if (ic == (iconv_t)(-1) ) {
  71                         lprintf(5, "%s:%d iconv_open() failed: %s\n", __FILE__, __LINE__, strerror(errno));
  72                 }
  73         }
  74         if (ic != (iconv_t)(-1) ) {
  75                 ibuf = msg;
  76                 ibuflen = content_length;
  77                 obuflen = content_length + (content_length / 2) ;
  78                 obuf = (char *) malloc(obuflen);
  79                 osav = obuf;
  80                 iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
  81                 content_length = content_length + (content_length / 2) - obuflen;
  82                 osav[content_length] = 0;
  83                 free(msg);
  84                 msg = osav;
  85                 iconv_close(ic);
  86         }
  87 #endif
  88
  89         ptr = msg;
  90         msgstart = msg;
  91         msgend = &msg[content_length];
  92
  93         while (ptr < msgend) {
  94
  95                 /* Advance to next tag */
  96                 ptr = strchr(ptr, '<');
  97                 if ((ptr == NULL) || (ptr >= msgend)) break;
  98                 ++ptr;
  99                 if ((ptr == NULL) || (ptr >= msgend)) break;
 100
 101                 /* Any of these tags cause everything up to and including
 102                  * the tag to be removed.
 103                  */
 104                 if ( (!strncasecmp(ptr, "HTML", 4))
 105                    ||(!strncasecmp(ptr, "HEAD", 4))
 106                    ||(!strncasecmp(ptr, "/HEAD", 5))
 107                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
 108                         ptr = strchr(ptr, '>');
 109                         if ((ptr == NULL) || (ptr >= msgend)) break;
 110                         ++ptr;
 111                         if ((ptr == NULL) || (ptr >= msgend)) break;
 112                         msgstart = ptr;
 113                 }
 114
 115                 /* Any of these tags cause everything including and following
 116                  * the tag to be removed.
 117                  */
 118                 if ( (!strncasecmp(ptr, "/HTML", 5))
 119                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 120                         --ptr;
 121                         msgend = ptr;
 122                         strcpy(ptr, "");
 123
 124                 }
 125
 126                 ++ptr;
 127         }
 128
 129         converted_msg = malloc(content_length);
 130         strcpy(converted_msg, "");
 131         ptr = msgstart;
 132         while (ptr < msgend) {
 133                 /* Change mailto: links to WebCit mail, by replacing the
 134                  * link with one that points back to our mail room.  Due to
 135                  * the way we parse URL's, it'll even handle mailto: links
 136                  * that have "?subject=" in them.
 137                  */
 138                 if (!strncasecmp(ptr, "<A HREF=\"mailto:", 16)) {
 139                         content_length += 64;
 140                         converted_msg = realloc(converted_msg, content_length);
 141                         sprintf(&converted_msg[output_length],
 142                                 "<A HREF=\"/display_enter"
 143                                 "?force_room=_MAIL_&recp=");
 144                         output_length += 47;
 145                         ptr = &ptr[16];
 146                         ++alevel;
 147                 }
 148                 /* Make links open in a separate window */
 149                 else if (!strncasecmp(ptr, "<A HREF=", 8)) {
 150                         content_length += 64;
 151                         converted_msg = realloc(converted_msg, content_length);
 152                         sprintf(&converted_msg[output_length], new_window);
 153                         output_length += strlen(new_window);
 154                         ptr = &ptr[8];
 155                         ++alevel;
 156                 }
 157                 /* Turn anything that looks like a URL into a real link, as long
 158                  * as it's not inside a tag already
 159                  */
 160                 else if ( (brak == 0) && (alevel == 0)
 161                      && (!strncasecmp(ptr, "http://", 7))) {
 162                                 linklen = 0;
 163                                 /* Find the end of the link */
 164                                 for (i=0; i<=strlen(ptr); ++i) {
 165                                         if ((ptr[i]==0)
 166                                            ||(isspace(ptr[i]))
 167                                            ||(ptr[i]==10)
 168                                            ||(ptr[i]==13)
 169                                            ||(ptr[i]=='(')
 170                                            ||(ptr[i]==')')
 171                                            ||(ptr[i]=='<')
 172                                            ||(ptr[i]=='>')
 173                                            ||(ptr[i]=='[')
 174                                            ||(ptr[i]==']')
 175                                         ) linklen = i;
 176                                         if (linklen > 0) break;
 177                                 }
 178                                 if (linklen > 0) {
 179                                         content_length += (32 + linklen);
 180                                         converted_msg = realloc(converted_msg, content_length);
 181                                         sprintf(&converted_msg[output_length], new_window);
 182                                         output_length += strlen(new_window);
 183                                         converted_msg[output_length] = '\"';
 184                                         converted_msg[++output_length] = 0;
 185                                         for (i=0; i<linklen; ++i) {
 186                                                 converted_msg[output_length] = ptr[i];
 187                                                 converted_msg[++output_length] = 0;
 188                                         }
 189                                         sprintf(&converted_msg[output_length], "\">");
 190                                         output_length += 2;
 191                                         for (i=0; i<linklen; ++i) {
 192                                                 converted_msg[output_length] = *ptr++;
 193                                                 converted_msg[++output_length] = 0;
 194                                         }
 195                                         sprintf(&converted_msg[output_length], "</A>");
 196                                         output_length += 4;
 197                                 }
 198                 }
 199                 else {
 200                         /*
 201                          * We need to know when we're inside a tag,
 202                          * so we don't turn things that look like URL's into
 203                          * links, when they're already links - or image sources.
 204                          */
 205                         if (*ptr == '<') ++brak;
 206                         if (*ptr == '>') --brak;
 207                         if (!strncasecmp(ptr, "</A>", 3)) --alevel;
 208                         converted_msg[output_length] = *ptr++;
 209                         converted_msg[++output_length] = 0;
 210                 }
 211         }
 212
 213         /* Output our big pile of markup */
 214         client_write(converted_msg, output_length);
 215
 216         /* A little trailing vertical whitespace... */
 217         wprintf("<br /><br />\n");
 218
 219         /* Now give back the memory */
 220         free(converted_msg);
 221         free(msg);
 222 }
 223