Review inc.man.

[nmh] / test / scan / test-scan-multibyte

diff --git a/test/scan/test-scan-multibyte b/test/scan/test-scan-multibyte
index f093a82496ae31c48fc3b2d735cab733a01a8ae5..dc2a284c76e55d8bc31f935a814dd9560a7d40a1 100755 (executable)
--- a/test/scan/test-scan-multibyte
+++ b/test/scan/test-scan-multibyte
@@ -26,7 +26,7 @@ if test "${MULTIBYTE_ENABLED}" -ne 1; then
     test_skip "configure did not detect multibyte support"
 fi
 
     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
 
 #
 # Create a test message with RFC 2047 headers we can scan


@@ -63,13 +63,15 @@ fi
 expected="$MH_TEST_DIR/$$.expected"
 actual="$MH_TEST_DIR/$$.actual"
 
 expected="$MH_TEST_DIR/$$.expected"
 actual="$MH_TEST_DIR/$$.actual"
 

+
+start_test 'RFC 2047 headers'

 if test "$width" -eq 3; then
 cat > "$expected" <<EOF
 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
 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"
 EOF
 else
     echo "Unsupported width for UTF-8 test string: $width"


@@ -79,6 +81,7 @@ fi
 run_prog scan -width 80 +inbox 11 > $actual || exit 1
 check "$expected" "$actual"
 
 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
 #
 # Check decoding with an invalid multibyte sequence.  We skip this test
 # if we don't have iconv support, since it requires converting from one


@@ -86,6 +89,7 @@ check "$expected" "$actual"
 # it's required for the test right after it.
 #
 
 # 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>
 cat >`mhpath new` <<EOF
 From: Test12 <test12@example.com>
 To: Some User <user@example.com>


@@ -94,8 +98,8 @@ Message-Id: 12@test.nmh
 Subject: =?UTF-8?B?MjAxMyBOZXcgWWVhcuKAmXMgRGVhbHMhIFN0YXJ0IHRoZSB5ZWFy?=
        =?UTF-8?B?IHJpZ2h0IHdpdGggYmlnIHNhdmluZ3M=?=
 
 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
 
 if test "$ICONV_ENABLED" -eq 1; then
 EOF
 
 if test "$ICONV_ENABLED" -eq 1; then


@@ -103,8 +107,8 @@ if test "$ICONV_ENABLED" -eq 1; then
   12  12/31 Test12             2013 New Year?s Deals! Start the year right
 EOF
 
   12  12/31 Test12             2013 New Year?s Deals! Start the year right
 EOF
 

-    # 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
 
     check "$expected" "$actual"
 fi
 


@@ -120,6 +124,7 @@ if test $? -ne 0; then
 fi
 
 # check scan width with a valid multibyte sequence
 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
 if test "$width" -eq 1; then
     cat >"$expected" <<EOF
   12  12/31 Test12             2013 New Year’s Deals! Start the year right


@@ -132,18 +137,17 @@ else
     echo "Unsupported width for U+2019: $width"
 fi
 
     echo "Unsupported width for U+2019: $width"
 fi
 

-LC_CTYPE=en_US.UTF-8 MM_CHARSET=UTF-8 scan -width 75 last >"$actual"
+run_prog scan -width 74 last >"$actual"

 check "$expected" "$actual"
 
 
 check "$expected" "$actual"
 
 

-#### Skip this test if we don't have iconv support because it
-#### requires converting from ISO-8859-1 to UTF-8.

 if test "$ICONV_ENABLED" -eq 1; then
 if test "$ICONV_ENABLED" -eq 1; then

-    cat >"$expected" <<EOF
-  13  01/13 sender@example.co  <<The Subject: is an encoded single quote, 0x92.
+  start_test 'encoded single quote'
+  cat >"$expected" <<EOF
+  13  01/13 sender@example.co  <<The Subject: is an encoded single quote, 0x92. 

 EOF
 
 EOF
 

-    cat >"${MH_TEST_DIR}/Mail/inbox/13" <<EOF
+  cat >"${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
 From: <sender@example.com>
 Subject: =?iso-8859-1?B?kgo=?=
 Date: Mon, 13 Jan 2014 14:18:33 -0600


@@ -156,9 +160,28 @@ properly count it when decoding, which could be seen with:
 The scan listing was two characters too long.
 EOF
 
 The scan listing was two characters too long.
 EOF
 

-    run_prog scan -width 80 last >"$actual"
-    check "$expected" "$actual"
+  run_prog scan -width 80 last >"$actual"
+  check "$expected" "$actual"

 fi
 
 
 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
 exit $failed