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 <ctype.h>
  10 #include <stdlib.h>
  11 #include <unistd.h>
  12 #include <stdio.h>
  13 #include <fcntl.h>
  14 #include <signal.h>
  15 #include <sys/types.h>
  16 #include <sys/wait.h>
  17 #include <sys/socket.h>
  18 #include <limits.h>
  19 #include <netinet/in.h>
  20 #include <netdb.h>
  21 #include <string.h>
  22 #include <pwd.h>
  23 #include <errno.h>
  24 #include <stdarg.h>
  25 #include <pthread.h>
  26 #include <signal.h>
  27 #include "webcit.h"
  28 #include "vcard.h"
  29 #include "webserver.h"
  30
  31
  32 /*
  33  */
  34 void output_html(void) {
  35         char buf[SIZ];
  36         char *msg;
  37         char *ptr;
  38         char *msgstart;
  39         char *msgend;
  40         char *converted_msg;
  41         int buffer_length = 1;
  42         int line_length = 0;
  43         int content_length = 0;
  44         int output_length = 0;
  45         char new_window[SIZ];
  46         int brak = 0;
  47         int i;
  48         int linklen;
  49
  50         msg = strdup("");
  51         sprintf(new_window, "<A TARGET=\"%s\" HREF=", TARGET);
  52
  53         while (serv_gets(buf), strcmp(buf, "000")) {
  54                 line_length = strlen(buf);
  55                 buffer_length = content_length + line_length + 2;
  56                 msg = realloc(msg, buffer_length);
  57                 if (msg == NULL) {
  58                         wprintf("<B>realloc() error!  "
  59                                 "couldn't get %d bytes: %s</B><BR><BR>\n",
  60                                 buffer_length + 1,
  61                                 strerror(errno));
  62                         return;
  63                 }
  64                 strcpy(&msg[content_length], buf);
  65                 content_length += line_length;
  66                 strcpy(&msg[content_length], "\n");
  67                 content_length += 1;
  68         }
  69
  70         ptr = msg;
  71         msgstart = msg;
  72         msgend = &msg[content_length];
  73
  74         while (ptr < msgend) {
  75
  76                 /* Advance to next tag */
  77                 ptr = strchr(ptr, '<');
  78                 if ((ptr == NULL) || (ptr >= msgend)) break;
  79                 ++ptr;
  80                 if ((ptr == NULL) || (ptr >= msgend)) break;
  81
  82                 /* Any of these tags cause everything up to and including
  83                  * the tag to be removed.
  84                  */
  85                 if ( (!strncasecmp(ptr, "HTML", 4))
  86                    ||(!strncasecmp(ptr, "HEAD", 4))
  87                    ||(!strncasecmp(ptr, "/HEAD", 5))
  88                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
  89                         ptr = strchr(ptr, '>');
  90                         if ((ptr == NULL) || (ptr >= msgend)) break;
  91                         ++ptr;
  92                         if ((ptr == NULL) || (ptr >= msgend)) break;
  93                         msgstart = ptr;
  94                 }
  95
  96                 /* Any of these tags cause everything including and following
  97                  * the tag to be removed.
  98                  */
  99                 if ( (!strncasecmp(ptr, "/HTML", 5))
 100                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 101                         --ptr;
 102                         msgend = ptr;
 103                         strcpy(ptr, "");
 104
 105                 }
 106
 107                 ++ptr;
 108         }
 109
 110         converted_msg = malloc(content_length);
 111         strcpy(converted_msg, "");
 112         ptr = msgstart;
 113         while (ptr < msgend) {
 114                 /* Change mailto: links to WebCit mail, by replacing the
 115                  * link with one that points back to our mail room.  Due to
 116                  * the way we parse URL's, it'll even handle mailto: links
 117                  * that have "?subject=" in them.
 118                  */
 119                 if (!strncasecmp(ptr, "<A HREF=\"mailto:", 16)) {
 120                         content_length += 64;
 121                         converted_msg = realloc(converted_msg, content_length);
 122                         sprintf(&converted_msg[output_length],
 123                                 "<A HREF=\"/display_enter"
 124                                 "?force_room=_MAIL_&recp=");
 125                         output_length += 47;
 126                         ptr = &ptr[16];
 127                         ++brak;
 128                 }
 129                 /* Make links open in a separate window */
 130                 else if (!strncasecmp(ptr, "<A HREF=", 8)) {
 131                         content_length += 64;
 132                         converted_msg = realloc(converted_msg, content_length);
 133                         sprintf(&converted_msg[output_length], new_window);
 134                         output_length += strlen(new_window);
 135                         ptr = &ptr[8];
 136                         ++brak;
 137                 }
 138                 /* Turn anything that looks like a URL into a real link, as long
 139                  * as it's not inside a tag already
 140                  */
 141                 else if ( (brak == 0)
 142                      && (!strncasecmp(ptr, "http://", 7))) {
 143                                 linklen = 0;
 144                                 /* Find the end of the link */
 145                                 for (i=0; i<=strlen(ptr); ++i) {
 146                                         if ((ptr[i]==0)
 147                                            ||(isspace(ptr[i]))
 148                                            ||(ptr[i]==10)
 149                                            ||(ptr[i]==13)
 150                                            ||(ptr[i]==')')
 151                                            ||(ptr[i]=='>')
 152                                            ||(ptr[i]==']')
 153                                         ) linklen = i;
 154                                         if (linklen > 0) break;
 155                                 }
 156                                 if (linklen > 0) {
 157                                         content_length += (32 + linklen);
 158                                         converted_msg = realloc(converted_msg, content_length);
 159                                         sprintf(&converted_msg[output_length], new_window);
 160                                         output_length += strlen(new_window);
 161                                         converted_msg[output_length] = '\"';
 162                                         converted_msg[++output_length] = 0;
 163                                         for (i=0; i<linklen; ++i) {
 164                                                 converted_msg[output_length] = ptr[i];
 165                                                 converted_msg[++output_length] = 0;
 166                                         }
 167                                         sprintf(&converted_msg[output_length], "\">");
 168                                         output_length += 2;
 169                                         for (i=0; i<linklen; ++i) {
 170                                                 converted_msg[output_length] = *ptr++;
 171                                                 converted_msg[++output_length] = 0;
 172                                         }
 173                                         sprintf(&converted_msg[output_length], "</A>");
 174                                         output_length += 4;
 175                                 }
 176                 }
 177                 else {
 178                         /*
 179                          * We need to know when we're inside a tag,
 180                          * so we don't turn things that look like URL's into
 181                          * links, when they're already links - or image sources.
 182                          */
 183                         if (*ptr == '<') ++brak;
 184                         if (*ptr == '>') --brak;
 185                         converted_msg[output_length] = *ptr++;
 186                         converted_msg[++output_length] = 0;
 187                 }
 188         }
 189
 190         /* Output our big pile of markup */
 191         client_write(converted_msg, output_length);
 192
 193         /* A little trailing vertical whitespace... */
 194         wprintf("<BR><BR>\n");
 195
 196         /* Now give back the memory */
 197         free(converted_msg);
 198         free(msg);
 199 }
 200