* HTML messages in foreign character sets are now converted to UTF-8 for

[citadel.git] / webcit / html2html.c

diff --git a/webcit/html2html.c b/webcit/html2html.c
index 6eb490a602faf891c2a2fbb1fa5584da65157768..110d9682e31d937ffa35a47ee621b75f860431eb 100644 (file)
--- a/webcit/html2html.c
+++ b/webcit/html2html.c
@@ -24,14 +24,21 @@
 #include <stdarg.h>
 #include <pthread.h>
 #include <signal.h>
+
+#ifdef HAVE_ICONV
+#include <iconv.h>
+#endif
+
 #include "webcit.h"
 #include "vcard.h"
 #include "webserver.h"
 
 
 /*
+ * Sanitize and enhance an HTML message for display.
+ * Also convert weird character sets to UTF-8 if necessary.
  */
-void output_html(void) {
+void output_html(char *charset) {
        char buf[SIZ];
        char *msg;
        char *ptr;
@@ -47,17 +54,25 @@ void output_html(void) {
        int alevel = 0;
        int i;
        int linklen;
+#ifdef HAVE_ICONV
+       iconv_t ic = (iconv_t)(-1) ;
+       char *ibuf;                   /* Buffer of characters to be converted */
+       char *obuf;                   /* Buffer for converted characters      */
+       size_t ibuflen;               /* Length of input buffer               */
+       size_t obuflen;               /* Length of output buffer              */
+       char *osav;                   /* Saved pointer to output buffer       */
+#endif
 
        msg = strdup("");
        sprintf(new_window, "<A TARGET=\"%s\" HREF=", TARGET);
 
-       while (serv_gets(buf), strcmp(buf, "000")) {
+       while (serv_getln(buf, sizeof buf), strcmp(buf, "000")) {
                line_length = strlen(buf);
                buffer_length = content_length + line_length + 2;
                msg = realloc(msg, buffer_length);
                if (msg == NULL) {
                        wprintf("<B>realloc() error!  "
-                               "couldn't get %d bytes: %s</B><BR><BR>\n",
+                               "couldn't get %d bytes: %s</B><br /><br />\n",
                                buffer_length + 1,
                                strerror(errno));
                        return;
@@ -68,6 +83,29 @@ void output_html(void) {
                content_length += 1;
        }
 
+#ifdef HAVE_ICONV
+       if ( (strcasecmp(charset, "us-ascii"))
+          && (strcasecmp(charset, "UTF-8")) ) {
+               ic = iconv_open("UTF-8", charset);
+               if (ic == (iconv_t)(-1) ) {
+                       lprintf(5, "iconv_open() failed: %s\n", strerror(errno));
+               }
+       }
+       if (ic != (iconv_t)(-1) ) {
+               ibuf = msg;
+               ibuflen = content_length;
+               obuflen = content_length + (content_length / 2) ;
+               obuf = (char *) malloc(obuflen);
+               osav = obuf;
+               iconv(ic, &ibuf, &ibuflen, &obuf, &obuflen);
+               content_length = content_length + (content_length / 2) - obuflen;
+               osav[content_length] = 0;
+               free(msg);
+               msg = osav;
+               iconv_close(ic);
+       }
+#endif
+
        ptr = msg;
        msgstart = msg;
        msgend = &msg[content_length];
@@ -196,7 +234,7 @@ void output_html(void) {
        client_write(converted_msg, output_length);
 
        /* A little trailing vertical whitespace... */
-       wprintf("<BR><BR>\n");
+       wprintf("<br /><br />\n");
 
        /* Now give back the memory */
        free(converted_msg);