From: Ken Hornstein <kenh@pobox.com>
Date: Sun, 8 Jun 2014 03:14:12 +0000 (-0400)
Subject: Return an appropriate error if the input characters are 8-bit, but the
X-Git-Url: https://diplodocus.org/git/nmh/commitdiff_plain/c3bb80a82ee1c5f74fbd38e870af591227ac2de3?ds=sidebyside;hp=6b1962e92b9a3bfc7570aecb844fe67141fb87d7

Return an appropriate error if the input characters are 8-bit, but the
locale character set is US-ASCII.
---

diff --git a/test/mhbuild/test-ext-params b/test/mhbuild/test-ext-params
index adde9d95..0b5e66da 100755
--- a/test/mhbuild/test-ext-params
+++ b/test/mhbuild/test-ext-params
@@ -157,7 +157,6 @@ This is a test message
 EOF
 
 run_prog mhbuild "$draft"
-rm -f "$MH_TEST_DIR/tïny.jpg"
 
 cat > "$expected" <<EOF
 To: Mr Test <mrtest@example.com>
@@ -183,6 +182,31 @@ EOF
 
 check "$draft" "$expected"
 
+#
+# Test to make sure things fail if we try to put 8-bit characters when the
+# locale is US-ASCII
+#
+
+cat > "$draft" <<EOF
+To: Mr Test <mrtest@example.com>
+cc:
+Fcc: +outbox
+------
+This is a test message
+#image/jpeg {attachment} $MH_TEST_DIR/tïny.jpg
+EOF
+
+old_locale="$LC_ALL"
+LC_ALL=C; export LC_ALL
+
+set +e
+run_test 'eval mhbuild "$draft" 2>&1' 'mhbuild: 8-bit characters in parameter "filename", but local character set is US-ASCII'
+set -e
+
+LC_ALL="$old_locale"; export LC_ALL
+
+rm -f "$MH_TEST_DIR/tïny.jpg"
+
 #
 # Test out message/external-body decoding
 #
diff --git a/test/mhbuild/test-utf8-body b/test/mhbuild/test-utf8-body
index 650133c9..13629706 100755
--- a/test/mhbuild/test-utf8-body
+++ b/test/mhbuild/test-utf8-body
@@ -16,6 +16,8 @@ fi
 
 setup_test
 testname="${MH_TEST_DIR}/$$"
+expected="$MH_TEST_DIR/$$.expected"
+actual="$MH_TEST_DIR/$$.actual"
 
 LC_ALL=en_US.UTF-8; export LC_ALL
 
@@ -124,6 +126,31 @@ EOF
 
 test_attachment "${testname}.expected"
 
+# Repeat the previous test, but make sure that the locale is set to C, which 
+# should cause mhbuild to fail
+
+cat > "${MH_TEST_DIR}/Mail/draft" <<EOF
+From: Mr Nobody <nobody@example.com>
+To: Somebody <somebody@example.com>
+Subject: Test
+Attach: $MH_TEST_DIR/attachment.txt
+
+¡Ay, caramba!
+EOF
+
+LC_ALL="C"; export LC_ALL
+
+set +e
+run_prog send -draft -mts sendmail/pipe > "$actual" 2>&1
+set -e
+
+cat > "$expected" <<EOF
+mhbuild: Text content contains 8 bit characters, but character set is US-ASCII
+${MH_INST_DIR}${bindir}/mhbuild: exit 1
+EOF
+
+check "$expected" "$actual"
+
 rm -f ${MHMTSCONF} "${MH_TEST_DIR}/attachment.txt"
 
 exit ${failed:-0}
diff --git a/uip/mhbuildsbr.c b/uip/mhbuildsbr.c
index 19d31316..77763337 100644
--- a/uip/mhbuildsbr.c
+++ b/uip/mhbuildsbr.c
@@ -1476,9 +1476,16 @@ scan_content (CT ct, size_t maxunencoded)
 	t = (struct text *) ct->c_ctparams;
 	if (t->tx_charset == CHARSET_UNSPECIFIED) {
 	    CI ci = &ct->c_ctinfo;
+	    char *eightbitcharset = write_charset_8bit();
+
+	    if (contains8bit && strcasecmp(eightbitcharset, "US-ASCII") == 0) {
+		adios(NULL, "Text content contains 8 bit characters, but "
+		      "character set is US-ASCII");
+	    }
 
 	    add_param(&ci->ci_first_pm, &ci->ci_last_pm, "charset",
-			contains8bit ? write_charset_8bit() : "us-ascii", 0);
+			contains8bit ? eightbitcharset : "us-ascii", 0);
+
 	    t->tx_charset = CHARSET_SPECIFIED;
 	}
     }
diff --git a/uip/mhparse.c b/uip/mhparse.c
index 6e16a5aa..75414667 100644
--- a/uip/mhparse.c
+++ b/uip/mhparse.c
@@ -3851,8 +3851,12 @@ param_len(PM pm, int index, size_t valueoff, int *encode, int *cont,
 	 * add them now.
 	 */
 
-	if (! pm->pm_charset)
+	if (! pm->pm_charset) {
 	    pm->pm_charset = getcpy(write_charset_8bit());
+	    if (strcasecmp(pm->pm_charset, "US-ASCII") == 0)
+		adios(NULL, "8-bit characters in parameter \"%s\", but "
+		      "local character set is US-ASCII", pm->pm_name);
+	}
 	if (! pm->pm_lang)
 	    pm->pm_lang = getcpy(NULL);	/* Default to a blank lang tag */