Experimental character set translation message reading

ChangeLog

@@ -35,6 +35,13 @@ v0.51.1		21-Feb-2004
 		Added login check to see if user has a valid character set.
 		Added Control-U 9 to display file to show the users character
 		set in the menus screens.
+		Added experimental characterset translation to message read.
+		This will translate the message character set used to the users
+		character set. When translation is impossible a ? is printed.
+		Doesn't work well with Russion, but seems to work with German
+		text. 
+		In delete file in home directory the Y/N keys were read from
+		the wrong language prompt.
 
 	mbmsg:
 		Writes the ^aCHRS: kludge again in new created messages.

lib/charset.c

@@ -66,3 +66,35 @@ char *getchrsdesc(int val)
 }
 
 
+
+char *get_iconv_name(char *name)
+{
+    if (!strncasecmp(name, "CP437", 5))
+	return (char *)"IBM437";
+    if (!strncasecmp(name, "CP850", 5))
+	return (char *)"IBM850";
+    if (!strncasecmp(name, "CP865", 5))
+	return (char *)"IBM865";
+    if (!strncasecmp(name, "CP866", 5))
+	return (char *)"IBM866";
+    if (!strncasecmp(name, "LATIN-1", 7))
+	return (char *)"ISO_8859-1";
+    if (!strncasecmp(name, "LATIN-2", 7))
+	return (char *)"ISO_8859-2";
+    if (!strncasecmp(name, "LATIN-5", 7))
+	return (char *)"ISO_8859-5";
+    if (!strncasecmp(name, "PC-8", 4))
+	return (char *)"IBM437";
+    if (!strncasecmp(name, "IBMPC", 5))
+	return (char *)"IBM437";
+    if (!strncasecmp(name, "+7_FIDO", 7))
+	return (char *)"IBM866";
+    if (!strncasecmp(name, "CP1125", 6))
+	return (char *)"CP1125";
+
+    Syslog('+', "get_iconv_name(%s): no usable character set name found", name);
+    return NULL;
+}
+
+
+

lib/mbselib.h

@@ -2134,6 +2134,7 @@ int	attach(faddr, char *, int, char);
  */
 char	*getchrs(int);			/* Return characterset name	    */
 char	*getchrsdesc(int);		/* Return characterset description  */
+char	*get_iconv_name(char *);	/* Return usable name for iconv	    */
 
 
 

mbsebbs/file.c

@@ -1481,7 +1481,7 @@ void Delete_Home()
 			fflush(stdout);
 			i = toupper(Getone());
 
-			if (i == Keystroke(368, 0) || i == 13) {
+			if (i == Keystroke(369, 0) || i == 13) {
 				i = unlink(temp);
 
 				if (i == -1) {

mbsebbs/mail.c

@@ -1136,9 +1136,15 @@ int Export_a_Msg(unsigned long Num)
  */
 int Read_a_Msg(unsigned long Num, int UpdateLR)
 {
-	char		*p = NULL, *fn;
+    char	*p = NULL, *fn, *charset = NULL, *charsin = NULL, *charsout = NULL;
-	int		ShowMsg = TRUE;
+    int		ShowMsg = TRUE, UseIconv = FALSE;
     lastread	LR;
+    iconv_t	iconvstr = (iconv_t)(0);
+#ifdef HAVE_ICONV_H
+    size_t	cnt, inleft, outleft;
+    char	inbuf[256], outbuf[1024];
+    char	*inb, *outb;
+#endif
 
     LastNum = Num;
     iLineCount = 7;
@@ -1168,6 +1174,7 @@ int Read_a_Msg(unsigned long Num, int UpdateLR)
 	sleep(3);
 	return FALSE;
     }
+    
     if (Msg.Private) {
 	ShowMsg = FALSE;
 	if ((strcasecmp(exitinfo.sUserName, Msg.From) == 0) || (strcasecmp(exitinfo.sUserName, Msg.To) == 0))
@@ -1195,6 +1202,15 @@ int Read_a_Msg(unsigned long Num, int UpdateLR)
 	    if ((p = (char *)MsgText_First()) != NULL)
 		do {
 		    if ((p[0] == '\001') || (!strncmp(p, "SEEN-BY:", 8)) || (!strncmp(p, "AREA:", 5))) {
+			/*
+			 * While we are here, check for the ^aCHRS: kludge and set the used charset.
+			 */
+			if (strncmp(p, "\001CHRS: ", 7) == 0) {
+			    charset = xstrcpy(p + 7);
+			}
+			if (strncmp(p, "\001CHARSET: ", 10) == 0) {
+			    charset = xstrcpy(p + 10);
+			}
 			if (Kludges) {
 			    if (p[0] == '\001') {
 				p[0] = 'a';
@@ -1212,6 +1228,32 @@ int Read_a_Msg(unsigned long Num, int UpdateLR)
     }
     free(fn);
 
+    if ((charset == NULL) && (msgs.Charset != FTNC_NONE)) {
+	/*
+	 * No charset marked in the message, use the area charset
+	 */
+	charset = xstrcpy(getchrs(msgs.Charset));
+    }
+    charsin = xstrcpy(get_iconv_name(charset));
+    charsout = xstrcpy(get_iconv_name(getchrs(exitinfo.Charset)));
+    Syslog('b', "Stage 3: charset %s, translate %s to %s", MBSE_SS(charset), MBSE_SS(charsin), MBSE_SS(charsout));
+
+#ifdef HAVE_ICONV_H
+    /*
+     * Try to setup iconv if the charactersets are different.
+     */
+    if (charsin && charsout && strcmp(charsout, charsin)) {
+
+	charsout = xstrcat(charsout, (char *)"//TRANSLIT");
+	iconvstr = iconv_open(charsout, charsin);
+	if (iconvstr == (iconv_t)(-1)) {
+	    Syslog('!', "$open_iconv(%s, %s)", charsin, charsout);
+	} else {
+	    UseIconv = TRUE;
+	}
+    }
+#endif
+
     /*
      * Show message text
      */
@@ -1231,7 +1273,42 @@ int Read_a_Msg(unsigned long Num, int UpdateLR)
 		    if (strchr(p, '>') != NULL)
 			if ((strlen(p) - strlen(strchr(p, '>'))) < 10)
 			    colour(CFG.HiliteF, CFG.HiliteB);
+#ifdef HAVE_ICONV_H
+		    if (UseIconv) {
+			/*
+			 * Try to translate character sets with iconv
+			 */
+			inleft = outleft = strlen(p);
+			memset(&inbuf, 0, sizeof(inbuf));
+			memset(&outbuf, 0, sizeof(inbuf));
+			sprintf(inbuf, "%s", p);
+			inb = inbuf;
+			outb = outbuf;
+			cnt = iconv(iconvstr, &inb, &inleft, &outb, &outleft);
+			if (cnt == (size_t)(-1)) {
+			    /*
+			     * Failed, log and show original line.
+			     */
+			    Syslog('b', "iconv cnt=%d inleft=%d outleft=%d", cnt, inleft, outleft);
+			    Syslog('b', "%s", printable(p, 0));
+			    Syslog('b', "$iconv");
 			    printf("%s\n", p);
+			} else {
+			    if (strcmp(inbuf, outbuf)) {
+				/*
+				 * Success, translated and log
+				 */
+				Syslog('b', "< %s", MBSE_SS(inbuf));
+				Syslog('b', "> %s", MBSE_SS(outbuf));
+			    }
+			    printf("%s\n", outbuf);
+			}
+		    } else {
+			printf("%s\n", p);
+		    }
+#else
+		    printf("%s\n", p);
+#endif
 		    if (CheckLine(CFG.TextColourF, CFG.TextColourB, FALSE))
 			break;
 		}
@@ -1239,6 +1316,19 @@ int Read_a_Msg(unsigned long Num, int UpdateLR)
 	}
     }
 
+    if (charset)
+	free(charset);
+    if (charsout)
+	free(charsout);
+    if (charsin)
+	free(charsin);
+    
+#ifdef HAVE_ICONV_H
+    if (UseIconv && (iconv_close(iconvstr) == -1)) {
+	Syslog('!', "$iconv_close()");
+    }
+#endif
+    
     /*
      * Set the Received status on this message if it's for the user.
      */

mbsebbs/mbsebbs.c

@@ -43,6 +43,9 @@
 #include "timeout.h"
 #include "funcs.h"
 
+#include <locale.h>
+#include <langinfo.h>
+
 
 extern	int	do_quiet;	/* Logging quiet flag */
 time_t		t_start;
@@ -205,6 +208,12 @@ int main(int argc, char **argv)
 	}
     }
 
+    /*
+     * Some debugging for me
+     */
+    Syslog('b', "setlocale() returns \"%s\"", printable(setlocale(LC_ALL, NULL), 0));
+    Syslog('b', "nl_langinfo(LC_CTYPE) returns \"%s\"", printable(nl_langinfo(LC_CTYPE), 0));
+
     sprintf(sMailbox, "mailbox");
     colour(LIGHTGRAY, BLACK);
     user();