diplodocus.org Git - nmh/blob - test/scan/test-scan-multibyte

   1 #!/bin/sh
   2 ############################################################
   3 #
   4 # Test scan to see if multibyte support (UTF-8 locale) works
   5 #
   6 # Other tests will get the normal ASCII case, so all we care
   7 # about here is UTF-8 encoded headers (RFC 2047).
   8 #
   9 # Note that this file should be edited via a UTF-8 aware
  10 # editor, since UTF-8 characters are in it.
  11 #
  12 ############################################################
  13
  14 set -e
  15
  16 if test -z "${MH_OBJ_DIR}"; then
  17     srcdir=`dirname "$0"`/../..
  18     MH_OBJ_DIR=`cd "$srcdir" && pwd`; export MH_OBJ_DIR
  19 fi
  20
  21 . "$MH_OBJ_DIR/test/common.sh"
  22
  23 setup_test
  24
  25 if test "${MULTIBYTE_ENABLED}" -ne 1; then
  26     test_skip "configure did not detect multibyte support"
  27 fi
  28
  29 LC_ALL=en_US.UTF-8; export LC_ALL
  30
  31 #
  32 # Create a test message with RFC 2047 headers we can scan
  33 #
  34
  35 cat > "${MH_TEST_DIR}/Mail/inbox/11" <<EOF
  36 From: David =?utf-8?q?=EF=AC=86?= Hubbins <hubbins@example.com>
  37 To: Sir Denis =?utf-8?q?Eton=E2=80=93Hogg? <sirdenis@example.com>
  38 Date: Friday, 2 Mar 1984 00:00:00
  39 Subject: =?utf-8?q?Spin=CC=88al_Tap_=E2=86=92_Tap_into_America!?=
  40
  41 Things are looking great!
  42 EOF
  43
  44 expected="$MH_TEST_DIR/$$.expected"
  45 actual="$MH_TEST_DIR/$$.actual"
  46
  47 cat > "$expected" <<EOF
  48   11  03/02 David ﬆ Hubbins    Spin̈al Tap → Tap into America!<<Things are
  49 EOF
  50
  51 scan -width 80 +inbox 11 > $actual || exit 1
  52 check "$expected" "$actual"
  53
  54 # check decoding with an invalid multibyte sequence
  55 cat >`mhpath new` <<EOF
  56 From: Test12 <test12@example.com>
  57 To: Some User <user@example.com>
  58 Date: Mon, 31 Dec 2012 00:00:00
  59 Message-Id: 12@test.nmh
  60 Subject: =?UTF-8?B?MjAxMyBOZXcgWWVhcuKAmXMgRGVhbHMhIFN0YXJ0IHRoZSB5ZWFy?=
  61         =?UTF-8?B?IHJpZ2h0IHdpdGggYmlnIHNhdmluZ3M=?=
  62
  63 This message has an encoded Subject with an invalid character for the
  64 ISO-8859-1 character set, but it (U+2019) is valid UTF-8.
  65 EOF
  66
  67 cat >"$expected" <<EOF
  68   12  12/31 Test12             2013 New Year?s Deals! Start the year right
  69 EOF
  70
  71 LC_CTYPE=ISO-8859-1 MM_CHARSET=ISO-8859-1 scan -width 75 last >"$actual"
  72 check "$expected" "$actual"
  73
  74 # check scan width with a valid multibyte sequence
  75 cat >"$expected" <<EOF
  76   12  12/31 Test12             2013 New Year’s Deals! Start the year right
  77 EOF
  78
  79 LC_CTYPE=en_US.UTF-8 MM_CHARSET=UTF-8 scan -width 75 last >"$actual"
  80 check "$expected" "$actual"
  81
  82
  83 exit $failed