File mailx-12.5-mime.dif of Package mailx

--- def.h
+++ def.h	2011-05-13 11:04:51.779926337 +0000
@@ -142,7 +142,8 @@ enum mimeclean {
 	MIME_LONGLINES	= 002,		/* has lines too long for RFC 2822 */
 	MIME_CTRLCHAR	= 004,		/* contains control characters */
 	MIME_HASNUL	= 010,		/* contains \0 characters */
-	MIME_NOTERMNL	= 020		/* lacks a terminating newline */
+	MIME_NOTERMNL	= 020,		/* lacks a terminating newline */
+	MIME_UTF8	= 040		/* UTF-8 high bit characters */
 };
 
 enum tdflags {
--- mime.c
+++ mime.c	2011-05-13 11:04:51.779926337 +0000
@@ -258,8 +258,11 @@ getcharset(int isclean)
 	if (isclean & (MIME_CTRLCHAR|MIME_HASNUL))
 		charset = NULL;
 	else if (isclean & MIME_HIGHBIT) {
-		charset = (wantcharset && wantcharset != (char *)-1) ?
-			wantcharset : value("charset");
+		if (isclean & MIME_UTF8)
+			charset = defcharset;
+		if (charset == NULL)
+			charset = (wantcharset && wantcharset != (char *)-1) ?
+				wantcharset : value("charset");
 		if (charset == NULL) {
 			char *t = value("ttycharset");
 			if (t == NULL || (ascncasecmp("ANSI_X3.4", t, 9) == 0))
@@ -737,6 +740,7 @@ mime_isclean(FILE *f)
 		lastc = c;
 		c = getc(f);
 		curlen++;
+	check:
 		if (c == '\n' || c == EOF) {
 			/*
 			 * RFC 821 imposes a maximum line length of 1000
@@ -749,10 +753,38 @@ mime_isclean(FILE *f)
 			curlen = 1;
 		} else if (c & 0200) {
 			isclean |= MIME_HIGHBIT;
+			if (c & 0100) {
+				int n, follow;
+
+				if      ((c & 040) == 0)	/* 110xxxxx */
+					follow = 1;
+				else if ((c & 020) == 0)	/* 1110xxxx */
+					follow = 2;
+				else if ((c & 010) == 0)	/* 11110xxx */
+					follow = 3;
+				else if ((c & 004) == 0)	/* 111110xx */
+					follow = 4;
+				else if ((c & 002) == 0)	/* 1111110x */
+					follow = 5;
+				else
+					continue;
+
+				for (n = 0; n < follow; n++) {
+					lastc = c;
+					c = getc(f);
+
+					if ((c & 0200) == 0 || (c & 0100) ||
+					    (c == '\0') || (c == '\n') || (c == EOF)) {
+						curlen += n;
+						goto check;
+					}
+				}
+				isclean |= MIME_UTF8;
+			}
 		} else if (c == '\0') {
 			isclean |= MIME_HASNUL;
 			break;
-		} else if ((c < 040 && (c != '\t' && c != '\f')) || c == 0177) {
+		} else if ((c < 040 && (c != '\t' && c != '\f' && c != '\b')) || c == 0177) {
 			isclean |= MIME_CTRLCHAR;
 		}
 	} while (c != EOF);
@@ -826,12 +858,15 @@ get_mime_convert(FILE *fp, char **conten
 			 * ^I or ^L in text/plain bodies. However, some
 			 * obscure character sets actually contain these
 			 * characters, so the content type can be set.
+			 * Beside ^I or ^L from RFC 2046 we accept also
+			 * backspace ^H often used in enhanced text.
 			 */
 			if ((*contenttype = value("contenttype-cntrl")) == NULL)
 				*contenttype = "application/octet-stream";
 		} else if (*contenttype == NULL)
 			*contenttype = "text/plain";
-	}
+	} else if (ascncasecmp(*contenttype, "text/", 5) == 0)
+		*charset = getcharset(*isclean);
 	return convert;
 }
 
--- sendout.c
+++ sendout.c	2011-05-13 11:07:31.623926237 +0000
@@ -230,6 +230,11 @@ attach_file1(struct attachment *ap, FILE
 		"\n--%s\n"
 		"Content-Type: %s",
 		send_boundary, contenttype);
+	tcs = gettcharset();
+#ifdef HAVE_ICONV
+	if (wantcharset && ascncasecmp(wantcharset, "ANSI_X3.4", 9))
+		charset = wantcharset;
+#endif
 	if (charset == NULL)
 		putc('\n', fo);
 	else
@@ -259,7 +264,7 @@ attach_file1(struct attachment *ap, FILE
 	if ((isclean & (MIME_HASNUL|MIME_CTRLCHAR)) == 0 &&
 			ascncasecmp(contenttype, "text/", 5) == 0 &&
 			isclean & MIME_HIGHBIT &&
-			charset != NULL) {
+			charset != NULL && tcs != NULL && asccasecmp(charset, tcs)) {
 		if ((iconvd = iconv_open_ft(charset, tcs)) == (iconv_t)-1 &&
 				errno != 0) {
 			if (errno == EINVAL)