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 require_locale en_US.UTF-8 en_US.UTF8 en_US.utf-8 en_US.utf8
  30
  31 #
  32 # Create a test message with RFC 2047 headers we can scan
  33 #
  34 # In this Subject header in this message is a "n" with a Combining Diaeresis
  35 # (U+0308).  There is different interpretation of this character with respect
  36 # to wcwidth() (which is supposed to return the column width of a character).
  37 # We use a test program to determine what the output width of U+0308 is
  38 # and adjust our test output appropriately.
  39 #
  40 # True Spın̈al Tap fans will note that David ﬆ Hubbins was born in Squatney,
  41 # London, England, and thus having his name language-tagged with "cy" is almost
  42 # certainly incorrect.  But in his own words: "Here lies David ﬆ Hubbins,
  43 # and why not?".
  44 #
  45 # The second "* in the To line is just to exercise the parser a bit.
  46 #
  47
  48 cat > "${MH_TEST_DIR}/Mail/inbox/11" <<EOF
  49 From: David =?utf-8*cy?q?=EF=AC=86?= Hubbins <hubbins@example.com>
  50 To: Sir Denis =?utf-8*?q?Eton=E2=80=93Hogg? <sirdenis@example.com>
  51 Date: Friday, 2 Mar 1984 00:00:00
  52 Subject: =?utf-8?q?Sp=C4=B1n=CC=88al_Tap_=E2=86=92_Tap_into_America!?=
  53
  54 Things are looking great!
  55 EOF
  56
  57 width=`${MH_OBJ_DIR}/test/getcwidth "→n̈"`
  58 if test $? -ne 0; then
  59     echo "getcwidth failed to run"
  60     exit 1
  61 fi
  62
  63 expected="$MH_TEST_DIR/$$.expected"
  64 actual="$MH_TEST_DIR/$$.actual"
  65
  66 if test "$width" -eq 3; then
  67 cat > "$expected" <<EOF
  68   11  03/02 David ﬆ Hubbins    Spın̈al Tap → Tap into America!<<Things are looki
  69 EOF
  70 elif test "$width" -eq 2; then
  71 cat > "$expected" <<EOF
  72   11  03/02 David ﬆ Hubbins    Spın̈al Tap → Tap into America!<<Things are lookin
  73 EOF
  74 else
  75     echo "Unsupported width for UTF-8 test string: $width"
  76     exit 1
  77 fi
  78
  79 run_prog scan -width 80 +inbox 11 > $actual || exit 1
  80 check "$expected" "$actual"
  81
  82 #
  83 # Check decoding with an invalid multibyte sequence.  We skip this test
  84 # if we don't have iconv support, since it requires converting from one
  85 # character set to another.  Be sure we created the test file, though, because
  86 # it's required for the test right after it.
  87 #
  88
  89 cat >`mhpath new` <<EOF
  90 From: Test12 <test12@example.com>
  91 To: Some User <user@example.com>
  92 Date: Mon, 31 Dec 2012 00:00:00
  93 Message-Id: 12@test.nmh
  94 Subject: =?UTF-8?B?MjAxMyBOZXcgWWVhcuKAmXMgRGVhbHMhIFN0YXJ0IHRoZSB5ZWFy?=
  95         =?UTF-8?B?IHJpZ2h0IHdpdGggYmlnIHNhdmluZ3M=?=
  96
  97 This message has an encoded Subject with an invalid character for
  98 single-byte character sets, but it (U+2019) is valid UTF-8.
  99 EOF
 100
 101 if test "$ICONV_ENABLED" -eq 1; then
 102     cat >"$expected" <<EOF
 103   12  12/31 Test12             2013 New Year?s Deals! Start the year right
 104 EOF
 105
 106     # Don't use run_prog here because it loses the environment setting.
 107     LC_ALL=C scan -width 74 last >"$actual"
 108     check "$expected" "$actual"
 109 fi
 110
 111 #
 112 # Find out the width of our Unicode apostrophe (U+2019).  Some implementations
 113 # say it has a width of 2, but that seems totally bizarre to me.
 114 #
 115
 116 width=`${MH_OBJ_DIR}/test/getcwidth U+2019`
 117 if test $? -ne 0; then
 118     echo "getcwidth failed to run"
 119     exit 1
 120 fi
 121
 122 # check scan width with a valid multibyte sequence
 123 if test "$width" -eq 1; then
 124     cat >"$expected" <<EOF
 125   12  12/31 Test12             2013 New Year’s Deals! Start the year right
 126 EOF
 127 elif test "$width" -eq 2; then
 128     cat >"$expected" <<EOF
 129   12  12/31 Test12             2013 New Year’s Deals! Start the year righ
 130 EOF
 131 else
 132     echo "Unsupported width for U+2019: $width"
 133 fi
 134
 135 run_prog scan -width 74 last >"$actual"
 136 check "$expected" "$actual"
 137
 138
 139 if test "$ICONV_ENABLED" -eq 1; then
 140   cat >"$expected" <<EOF
 141   13  01/13 sender@example.co  <<The Subject: is an encoded single quote, 0x92.
 142 EOF
 143
 144   cat >"${MH_TEST_DIR}/Mail/inbox/13" <<EOF
 145 From: <sender@example.com>
 146 Subject: =?iso-8859-1?B?kgo=?=
 147 Date: Mon, 13 Jan 2014 14:18:33 -0600
 148
 149 The Subject: is an encoded single quote, 0x92.  cpstripped() didn't
 150 properly count it when decoding, which could be seen with:
 151
 152     scan -format '%(decode{subject})%{body}'
 153
 154 The scan listing was two characters too long.
 155 EOF
 156
 157   run_prog scan -width 80 last >"$actual"
 158   check "$expected" "$actual"
 159 fi
 160
 161
 162 exit $failed