test_skip "configure did not detect multibyte support"
fi
-LC_ALL=en_US.UTF-8; export LC_ALL
+require_locale en_US.UTF-8 en_US.UTF8 en_US.utf-8 en_US.utf8
#
# Create a test message with RFC 2047 headers we can scan
expected="$MH_TEST_DIR/$$.expected"
actual="$MH_TEST_DIR/$$.actual"
+
+start_test 'RFC 2047 headers'
if test "$width" -eq 3; then
cat > "$expected" <<EOF
- 11 03/02 David st Hubbins Spın̈al Tap → Tap into America!<<Things are look
+ 11 03/02 David st Hubbins Spın̈al Tap → Tap into America!<<Things are looki
EOF
elif test "$width" -eq 2; then
cat > "$expected" <<EOF
- 11 03/02 David st Hubbins Spın̈al Tap → Tap into America!<<Things are looki
+ 11 03/02 David st Hubbins Spın̈al Tap → Tap into America!<<Things are lookin
EOF
else
echo "Unsupported width for UTF-8 test string: $width"
run_prog scan -width 80 +inbox 11 > $actual || exit 1
check "$expected" "$actual"
+
#
# Check decoding with an invalid multibyte sequence. We skip this test
# if we don't have iconv support, since it requires converting from one
# it's required for the test right after it.
#
+start_test 'invalid multibyte sequence'
cat >`mhpath new` <<EOF
From: Test12 <test12@example.com>
To: Some User <user@example.com>
Subject: =?UTF-8?B?MjAxMyBOZXcgWWVhcuKAmXMgRGVhbHMhIFN0YXJ0IHRoZSB5ZWFy?=
=?UTF-8?B?IHJpZ2h0IHdpdGggYmlnIHNhdmluZ3M=?=
-This message has an encoded Subject with an invalid character for the
-ISO-8859-1 character set, but it (U+2019) is valid UTF-8.
+This message has an encoded Subject with an invalid character for
+single-byte character sets, but it (U+2019) is valid UTF-8.
EOF
-cat >"$expected" <<EOF
+if test "$ICONV_ENABLED" -eq 1; then
+ cat >"$expected" <<EOF
12 12/31 Test12 2013 New Year?s Deals! Start the year right
EOF
-if test "$ICONV_ENABLED" -eq 1; then
- # Don't use run_prog here because it loses those environment settings.
- LC_CTYPE=ISO-8859-1 MM_CHARSET=ISO-8859-1 scan -width 75 last >"$actual"
+ # Don't use run_prog here because it loses the environment setting.
+ LC_ALL=C scan -width 74 last >"$actual"
check "$expected" "$actual"
fi
fi
# check scan width with a valid multibyte sequence
+start_test 'scan width with a valid multibyte sequence'
if test "$width" -eq 1; then
cat >"$expected" <<EOF
12 12/31 Test12 2013 New Year’s Deals! Start the year right
echo "Unsupported width for U+2019: $width"
fi
-LC_CTYPE=en_US.UTF-8 MM_CHARSET=UTF-8 run_prog scan -width 75 last >"$actual"
+run_prog scan -width 74 last >"$actual"
+check "$expected" "$actual"
+
+
+if test "$ICONV_ENABLED" -eq 1; then
+ start_test 'encoded single quote'
+ cat >"$expected" <<EOF
+ 13 01/13 sender@example.co <<The Subject: is an encoded single quote, 0x92.
+EOF
+
+ cat >"${MH_TEST_DIR}/Mail/inbox/13" <<EOF
+From: <sender@example.com>
+Subject: =?iso-8859-1?B?kgo=?=
+Date: Mon, 13 Jan 2014 14:18:33 -0600
+
+The Subject: is an encoded single quote, 0x92. cpstripped() didn't
+properly count it when decoding, which could be seen with:
+
+ scan -format '%(decode{subject})%{body}'
+
+The scan listing was two characters too long.
+EOF
+
+ run_prog scan -width 80 last >"$actual"
+ check "$expected" "$actual"
+fi
+
+
+start_test 'insufficient room for multicolumn character'
+#### This multibyte character requires 2 columns for display, but
+#### only 1 is availble. cpstripped() used to get this wrong.
+
+cat >"$expected" <<EOF
+
+EOF
+
+cat >`mhpath new` <<EOF
+Mime-Version: 1.0
+
+在 Should not see any of this text
+EOF
+
+run_prog scan -format '%{body}' -width 1 last >"$actual"
check "$expected" "$actual"
+finish_test
exit $failed