X-Git-Url: https://diplodocus.org/git/nmh/blobdiff_plain/2e768e3cc0ce768373d3320eca4cdd08b8cf646a..ec4b22fc47186bd8c46495c0bee5592b225c9abf:/test/scan/test-scan-multibyte

diff --git a/test/scan/test-scan-multibyte b/test/scan/test-scan-multibyte
index 98dd9a26..fb9747d6 100755
--- a/test/scan/test-scan-multibyte
+++ b/test/scan/test-scan-multibyte
@@ -37,10 +37,17 @@ LC_ALL=en_US.UTF-8; export LC_ALL
 # We use a test program to determine what the output width of U+0308 is
 # and adjust our test output appropriately.
 #
+# True Spın̈al Tap fans will note that David st Hubbins was born in Squatney,
+# London, England, and thus having his name language-tagged with "cy" is almost
+# certainly incorrect.  But in his own words: "Here lies David st Hubbins,
+# and why not?".
+#
+# The second "* in the To line is just to exercise the parser a bit.
+#
 
 cat > "${MH_TEST_DIR}/Mail/inbox/11" <<EOF
-From: David =?utf-8?q?=EF=AC=86?= Hubbins <hubbins@example.com>
-To: Sir Denis =?utf-8?q?Eton=E2=80=93Hogg? <sirdenis@example.com>
+From: David =?utf-8*cy?q?=EF=AC=86?= Hubbins <hubbins@example.com>
+To: Sir Denis =?utf-8*?q?Eton=E2=80=93Hogg? <sirdenis@example.com>
 Date: Friday, 2 Mar 1984 00:00:00
 Subject: =?utf-8?q?Sp=C4=B1n=CC=88al_Tap_=E2=86=92_Tap_into_America!?=
 
@@ -69,7 +76,7 @@ else
     exit 1
 fi
 
-scan -width 80 +inbox 11 > $actual || exit 1
+run_prog scan -width 80 +inbox 11 > $actual || exit 1
 check "$expected" "$actual"
 
 #
@@ -87,16 +94,17 @@ Message-Id: 12@test.nmh
 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 "$ICONV_ENABLED" -eq 1; then
-    LC_CTYPE=ISO-8859-1 MM_CHARSET=ISO-8859-1 scan -width 75 last >"$actual"
+    # Don't use run_prog here because it loses the environment setting.
+    LC_ALL=C scan -width 75 last >"$actual"
     check "$expected" "$actual"
 fi
 
@@ -124,8 +132,31 @@ else
     echo "Unsupported width for U+2019: $width"
 fi
 
-LC_CTYPE=en_US.UTF-8 MM_CHARSET=UTF-8 scan -width 75 last >"$actual"
+run_prog scan -width 75 last >"$actual"
 check "$expected" "$actual"
 
 
+if test "$ICONV_ENABLED" -eq 1; then
+  cat >"$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
+
+
 exit $failed