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 in_link = 0;
  48         int i;
  49         int linklen;
  50
  51         msg = strdup("");
  52         sprintf(new_window, "<A TARGET=\"%s\" HREF=", TARGET);
  53
  54         while (serv_gets(buf), strcmp(buf, "000")) {
  55                 line_length = strlen(buf);
  56                 buffer_length = content_length + line_length + 2;
  57                 msg = realloc(msg, buffer_length);
  58                 if (msg == NULL) {
  59                         wprintf("<B>realloc() error!  "
  60                                 "couldn't get %d bytes: %s</B><BR><BR>\n",
  61                                 buffer_length + 1,
  62                                 strerror(errno));
  63                         return;
  64                 }
  65                 strcpy(&msg[content_length], buf);
  66                 content_length += line_length;
  67                 strcpy(&msg[content_length], "\n");
  68                 content_length += 1;
  69         }
  70
  71         ptr = msg;
  72         msgstart = msg;
  73         msgend = &msg[content_length];
  74
  75         while (ptr < msgend) {
  76
  77                 /* Advance to next tag */
  78                 ptr = strchr(ptr, '<');
  79                 if ((ptr == NULL) || (ptr >= msgend)) break;
  80                 ++ptr;
  81                 if ((ptr == NULL) || (ptr >= msgend)) break;
  82
  83                 /* Any of these tags cause everything up to and including
  84                  * the tag to be removed.
  85                  */
  86                 if ( (!strncasecmp(ptr, "HTML", 4))
  87                    ||(!strncasecmp(ptr, "HEAD", 4))
  88                    ||(!strncasecmp(ptr, "/HEAD", 5))
  89                    ||(!strncasecmp(ptr, "BODY", 4)) ) {
  90                         ptr = strchr(ptr, '>');
  91                         if ((ptr == NULL) || (ptr >= msgend)) break;
  92                         ++ptr;
  93                         if ((ptr == NULL) || (ptr >= msgend)) break;
  94                         msgstart = ptr;
  95                 }
  96
  97                 /* Any of these tags cause everything including and following
  98                  * the tag to be removed.
  99                  */
 100                 if ( (!strncasecmp(ptr, "/HTML", 5))
 101                    ||(!strncasecmp(ptr, "/BODY", 5)) ) {
 102                         --ptr;
 103                         msgend = ptr;
 104                         strcpy(ptr, "");
 105
 106                 }
 107
 108                 ++ptr;
 109         }
 110
 111         converted_msg = malloc(content_length);
 112         strcpy(converted_msg, "");
 113         ptr = msgstart;
 114         while (ptr < msgend) {
 115                 /* Change mailto: links to WebCit mail, by replacing the
 116                  * link with one that points back to our mail room.  Due to
 117                  * the way we parse URL's, it'll even handle mailto: links
 118                  * that have "?subject=" in them.
 119                  */
 120                 if (!strncasecmp(ptr, "<A HREF=\"mailto:", 16)) {
 121                         content_length += 64;
 122                         converted_msg = realloc(converted_msg, content_length);
 123                         sprintf(&converted_msg[output_length],
 124                                 "<A HREF=\"/display_enter"
 125                                 "?force_room=_MAIL_&recp=");
 126                         output_length += 47;
 127                         ptr = &ptr[16];
 128                         ++brak;
 129                 }
 130                 /* Make links open in a separate window */
 131                 else if (!strncasecmp(ptr, "<A HREF=", 8)) {
 132                         content_length += 64;
 133                         converted_msg = realloc(converted_msg, content_length);
 134                         sprintf(&converted_msg[output_length], new_window);
 135                         output_length += strlen(new_window);
 136                         ptr = &ptr[8];
 137                         ++brak;
 138                 }
 139                 /* Turn anything that looks like a URL into a real link */
 140                 else if ( (in_link == 0)
 141                      && (!strncasecmp(ptr, "http://", 7))) {
 142                                 linklen = 0;
 143                                 /* Find the end of the link */
 144                                 for (i=0; i<=strlen(ptr); ++i) {
 145                                         if ((ptr[i]==0)
 146                                            ||(isspace(ptr[i]))
 147                                            ||(ptr[i]==10)
 148                                            ||(ptr[i]==13)
 149                                            ||(ptr[i]==')')
 150                                            ||(ptr[i]=='>')
 151                                            ||(ptr[i]==']')
 152                                         ) linklen = i;
 153                                         if (linklen > 0) break;
 154                                 }
 155                                 if (linklen > 0) {
 156                                         content_length += (32 + linklen);
 157                                         converted_msg = realloc(converted_msg, content_length);
 158                                         sprintf(&converted_msg[output_length], new_window);
 159                                         output_length += strlen(new_window);
 160                                         converted_msg[output_length] = '\"';
 161                                         converted_msg[++output_length] = 0;
 162                                         for (i=0; i<linklen; ++i) {
 163                                                 converted_msg[output_length] = ptr[i];
 164                                                 converted_msg[++output_length] = 0;
 165                                         }
 166                                         sprintf(&converted_msg[output_length], "\">");
 167                                         output_length += 2;
 168                                         for (i=0; i<linklen; ++i) {
 169                                                 converted_msg[output_length] = *ptr++;
 170                                                 converted_msg[++output_length] = 0;
 171                                         }
 172                                         sprintf(&converted_msg[output_length], "</A>");
 173                                         output_length += 4;
 174                                 }
 175                 }
 176                 else {
 177                         /*
 178                          * We need to know when we're inside a pair of <A>...</A>
 179                          * tags, so we don't turn things that look like URL's into
 180                          * links, when they're already links.
 181                          */
 182                         if (!strncasecmp(ptr, "<A ", 3)) {
 183                                 ++in_link;
 184                         }
 185                         if (!strncasecmp(ptr, "</A", 3)) {
 186                                 --in_link;
 187                         }
 188                         if (*ptr == '<') ++brak;
 189                         if (*ptr == '>') --brak;
 190                         converted_msg[output_length] = *ptr++;
 191                         converted_msg[++output_length] = 0;
 192                 }
 193         }
 194
 195         /* Output our big pile of markup */
 196         client_write(converted_msg, output_length);
 197
 198         /* A little trailing vertical whitespace... */
 199         wprintf("<BR><BR>\n");
 200
 201         /* Now give back the memory */
 202         free(converted_msg);
 203         free(msg);
 204 }
 205