X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/d5e78944632398a4e31bca0c43cb4df3b30b1640..66cfb7956488139a507375b9b8964cf78a221962:/test/scan/test-scan-multibyte?ds=sidebyside

diff --git a/test/scan/test-scan-multibyte b/test/scan/test-scan-multibyte
index 819c34c0..98dd9a26 100755
--- a/test/scan/test-scan-multibyte
+++ b/test/scan/test-scan-multibyte
@@ -47,7 +47,7 @@ Subject: =?utf-8?q?Sp=C4=B1n=CC=88al_Tap_=E2=86=92_Tap_into_America!?=
 Things are looking great!
 EOF
 
-width=`${MH_OBJ_DIR}/test/getcwidth`
+width=`${MH_OBJ_DIR}/test/getcwidth "→n̈"`
 if test $? -ne 0; then
     echo "getcwidth failed to run"
     exit 1
@@ -56,16 +56,16 @@ fi
 expected="$MH_TEST_DIR/$$.expected"
 actual="$MH_TEST_DIR/$$.actual"
 
-if test "$width" -eq 1; then
+if test "$width" -eq 3; then
 cat > "$expected" <<EOF
   11  03/02 David st Hubbins    Spın̈al Tap → Tap into America!<<Things are look
 EOF
-elif test "$width" -eq 0; then
+elif test "$width" -eq 2; then
 cat > "$expected" <<EOF
   11  03/02 David st Hubbins    Spın̈al Tap → Tap into America!<<Things are looki
 EOF
 else
-    echo "Unsupported width for combining diaeresis: $width"
+    echo "Unsupported width for UTF-8 test string: $width"
     exit 1
 fi
 
@@ -75,11 +75,11 @@ 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
-# character set to another
+# character set to another.  Be sure we created the test file, though, because
+# it's required for the test right after it.
 #
 
-if test "$ICONV_ENABLED" -eq 1; then
-    cat >`mhpath new` <<EOF
+cat >`mhpath new` <<EOF
 From: Test12 <test12@example.com>
 To: Some User <user@example.com>
 Date: Mon, 31 Dec 2012 00:00:00
@@ -91,18 +91,38 @@ 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.
 EOF
 
-    cat >"$expected" <<EOF
+cat >"$expected" <<EOF
   12  12/31 Test12             2013 New Year?s Deals! Start the year right
 EOF
 
+if test "$ICONV_ENABLED" -eq 1; then
     LC_CTYPE=ISO-8859-1 MM_CHARSET=ISO-8859-1 scan -width 75 last >"$actual"
     check "$expected" "$actual"
 fi
 
+#
+# Find out the width of our Unicode apostrophe (U+2019).  Some implementations
+# say it has a width of 2, but that seems totally bizarre to me.
+#
+
+width=`${MH_OBJ_DIR}/test/getcwidth U+2019`
+if test $? -ne 0; then
+    echo "getcwidth failed to run"
+    exit 1
+fi
+
 # check scan width with a valid multibyte sequence
-cat >"$expected" <<EOF
+if test "$width" -eq 1; then
+    cat >"$expected" <<EOF
   12  12/31 Test12             2013 New Year’s Deals! Start the year right
 EOF
+elif test "$width" -eq 2; then
+    cat >"$expected" <<EOF
+  12  12/31 Test12             2013 New Year’s Deals! Start the year righ
+EOF
+else
+    echo "Unsupported width for U+2019: $width"
+fi
 
 LC_CTYPE=en_US.UTF-8 MM_CHARSET=UTF-8 scan -width 75 last >"$actual"
 check "$expected" "$actual"