When reading file, identify encoding (ASCII, UTF-8, EXTENDED, MIXED).

file.c

@@ -48,6 +48,13 @@ static const char *eolname[] = {
 	"MIXED"
 } ;
 
+static const char *codename[] = {
+	"ASCII",
+	"UTF-8",
+	"EXTENDED",
+	"MIXED"
+} ;
+
 boolean restflag = FALSE ;	/* restricted use?              */
 
 boolean resterr( void) {
@@ -290,7 +297,7 @@ int readin(const char *fname, boolean lockfl)
     eoltype found_eol ;
     int nbytes;
     int nline;
-    char mesg[NSTRING];
+	char *errmsg ;
 
 #if (FILOCK && BSD) || SVR4
     if (lockfl && lockchk(fname) == ABORT)
@@ -370,28 +377,29 @@ int readin(const char *fname, boolean lockfl)
     	break ;
     default:
     	found_eol = EOL_MIXED ;
-		curbp->b_mode |= MDVIEW ;	/* add view mode as we have lost
-									** information */
+		curbp->b_mode |= MDVIEW ;	/* force view mode as we have lost
+									** EOL information */
 	}
 
-    ffclose();      /* Ignore errors.       */
-    strcpy(mesg, "(");
-    if (s == FIOERR) {
-        strcat(mesg, "I/O ERROR, ");
-        curbp->b_flag |= BFTRUNC;
-    }
-    if (s == FIOMEM) {
-        strcat(mesg, "OUT OF MEMORY, ");
-        curbp->b_flag |= BFTRUNC;
-    }
-    sprintf(&mesg[strlen(mesg)], "Read %d line", nline);
-    if (nline != 1)
-        strcat(mesg, "s");
+	if( fcode == FCODE_UTF_8)
+		curbp->b_mode |= MDUTF8 ;
 
-    strcat( mesg, ", eol = ") ;
-    strcat( mesg, eolname[ found_eol]) ;
-    strcat(mesg, ")");
-    mloutstr( mesg) ;
+    if( s == FIOERR) {
+		errmsg = "I/O ERROR, " ;
+        curbp->b_flag |= BFTRUNC ;
+    } else if( s == FIOMEM) {
+        errmsg = "OUT OF MEMORY, " ;
+        curbp->b_flag |= BFTRUNC ;
+    } else
+    	errmsg = "" ;
+
+	mloutfmt( "(%sRead %d line%s, code/eol: %s/%s)",
+		errmsg,
+		nline,
+		(nline != 1) ? "s" : "",
+		codename[ fcode & (FCODE_MASK -1)],
+		eolname[ found_eol]) ;
+    ffclose();      /* Ignore errors.       */
 
       out:
     for (wp = wheadp; wp != NULL; wp = wp->w_wndp) {

fileio.c

@@ -21,6 +21,7 @@
 
 #include "defines.h"
 #include "retcode.h"
+#include "utf8.h"
 
 #if CRYPT
 boolean	is_crypted ;		/* currently encrypting?   */
@@ -29,6 +30,7 @@ boolean	is_crypted ;		/* currently encrypting?   */
 char	*fline = NULL ;		/* dynamic return line     */
 int		flen = 0 ;			/* current allocated length of fline */
 int		ftype ;
+int		fcode ;				/* encoding type FCODE_xxxxx */
 int		fpayload ;			/* actual length of fline content */
 
 
@@ -45,6 +47,7 @@ fio_code ffropen( const char *fn)
         return FIOFNF;
     eofflag = FALSE;
     ftype = FTYPE_NONE ;
+    fcode = FCODE_ASCII ;
     return FIOSUC;
 }
 
@@ -79,6 +82,7 @@ fio_code ffclose(void)
     }
     eofflag = FALSE;
     ftype = FTYPE_NONE ;
+    fcode = FCODE_ASCII ;
 
 #if MSDOS & CTRLZ
     fputc(26, ffp);     /* add a ^Z at the end of the file */
@@ -136,6 +140,7 @@ fio_code ffgetline(void)
 {
     int c;      /* current character read */
     int i;      /* current index into fline */
+    int lcode = FCODE_ASCII ;	/* line encoding, defaults to ASCII */
 
     /* if we are at the end...return it */
     if (eofflag)
@@ -156,6 +161,7 @@ fio_code ffgetline(void)
     i = 0;
     while ((c = fgetc(ffp)) != EOF && c != '\r' && c != '\n') {
 		fline[i++] = c;
+		lcode |= c ;
         /* if it's longer, get more room */
         if (i >= flen) {
 		    char *tmpline;  /* temp storage for expanding line */
@@ -173,6 +179,24 @@ fio_code ffgetline(void)
     }
 
 	fpayload = i ;
+	lcode &= FCODE_MASK ;
+	if( lcode && (fcode != FCODE_MIXED)) {	/* line contains extended chars */
+	/* Check if consistent UTF-8 encoding */
+		int bytes ;
+		int pos = 0 ;
+		unicode_t uc ;
+
+		while( (pos < i) && (lcode != FCODE_MIXED)) {
+			bytes = utf8_to_unicode( fline, pos, i, &uc) ;
+			pos += bytes ;
+			if( bytes > 1)		/* Multi byte UTF-8 sequence */
+				lcode |= FCODE_UTF_8 ;
+			else if( uc > 127)	/* Extended ASCII */
+				lcode |= FCODE_EXTND ;
+		}
+
+		fcode |= lcode ;
+	}
 
     /* test for any errors that may have occured */
     if (c == EOF) {

fileio.h

@@ -18,6 +18,12 @@ typedef enum {
 #define FTYPE_MAC   4
 /*      FTYPE_MIXED [ 3, 5, 6, 7] */
 
+#define FCODE_ASCII 0
+#define FCODE_MASK  0x80
+#define FCODE_UTF_8 0x81
+#define FCODE_EXTND 0x82
+#define FCODE_MIXED 0x83
+
 #if CRYPT
 #include "retcode.h"
 
@@ -27,6 +33,7 @@ extern boolean	is_crypted ;	/* currently encrypting?   */
 extern char		*fline ;		/* dynamic return line     */
 extern int 		flen ;			/* current allocated length of fline */
 extern int 		ftype ;
+extern int		fcode ;			/* encoding type */
 extern int		fpayload ;		/* actual length of fline content */
 
 fio_code ffclose( void) ;