diplodocus.org Git - nmh/blob - test/mhshow/test-charset

   1 #!/bin/sh
   2 ######################################################
   3 #
   4 # Test charset conversion functionality
   5 #
   6 ######################################################
   7
   8 set -e
   9
  10 if test -z "${MH_OBJ_DIR}"; then
  11     srcdir=`dirname "$0"`/../..
  12     MH_OBJ_DIR=`cd "$srcdir" && pwd`; export MH_OBJ_DIR
  13 fi
  14
  15 . "$MH_OBJ_DIR/test/common.sh"
  16
  17 setup_test
  18
  19 check_exit '-eq 1' mhshow -
  20
  21 require_locale en_US.UTF-8 en_US.UTF8 en_US.utf-8 en_US.utf8
  22
  23 iconv_elides_question_marks=0
  24 if test "$ICONV_ENABLED" -eq 0; then
  25     text_size=10
  26 else
  27     text_size=11
  28     #### The GNU iconv library normalises charset names by eliding '?', along
  29     #### with some other characters.  The iconv libraries used on FreeBSD/NetBSD
  30     #### and Mac OS X don't.
  31     printf x | iconv -f '?UTF-8' -t UTF-8 >/dev/null 2>&1  &&
  32         iconv_elides_question_marks=1
  33 fi
  34
  35 expected=$MH_TEST_DIR/$$.expected
  36 actual=$MH_TEST_DIR/$$.actual
  37
  38 echo 'mhshow-charset-windows-1252: echo "This is a test"' >> "$MH"
  39
  40 # Write message with bogus quoted-printable data.
  41 start_test 'Write message with bogus quoted-printable data'
  42 msgfile=`mhpath new`
  43 msgnum=`basename $msgfile`
  44 cat > $msgfile <<EOF
  45 From: foo@example.edu
  46 To: bar@example.edu
  47 Subject: test
  48 MIME-Version: 1.0
  49 Content-Type: text/plain; charset=windows-1252
  50 Content-Transfer-Encoding: quoted-printable
  51 Content-Description: bogus quoted-printable data
  52 Date: Sun, 18 Dec 2005 00:52:39 +0100
  53
  54 4 =F7 2 =3D 2
  55 EOF
  56
  57 cat > $expected <<EOF
  58 [ Message inbox:11 ]
  59 Date:    Sun, 18 Dec 2005 00:52:39 +0100
  60 To:      bar@example.edu
  61 From:    foo@example.edu
  62 Subject: test
  63
  64 MIME-Version: 1.0
  65 Content-Description: bogus quoted-printable data
  66
  67 [ part  - text/plain - bogus quoted-printable data  ${text_size}B  ]
  68 This is a test
  69 EOF
  70
  71 run_prog mhshow $msgnum > $actual 2>&1
  72 check "$expected" "$actual" 'keep first' : bogus quoted-printable data
  73
  74
  75 start_test 'locale profile component'
  76 #### Used LC_ALL above, now use locale profile component instead.
  77 printf "locale: $LC_ALL\n" >> "${MH}"
  78 unset LC_ALL
  79
  80 run_prog mhshow $msgnum > $actual 2>&1
  81 check "$expected" "$actual" : locale profile component
  82
  83
  84 #### All of the remaining tests rely on iconv.a
  85 if test "$ICONV_ENABLED" -eq 0; then
  86   finish_test
  87   exit $failed
  88 fi
  89
  90
  91 #### RFC 2231 defines the extended value syntax:
  92 #### extended-initial-value := [charset] "'" [language] "'"
  93 ####                           extended-other-values
  94 #### used below, where language is blank.
  95
  96 start_test 'Encoded parameter value'
  97 msgfile=`mhpath new`
  98 msgnum=`basename $msgfile`
  99 #### Use a charset to exercise the iconv code, if it's available,
 100 #### in get_param_value().  The Content-Type charset is UTF-8, encoded
 101 #### to exercise iconv.
 102 cat > $msgfile <<EOF
 103 From: foo@example.edu
 104 To: bar@example.edu
 105 Subject: test
 106 MIME-Version: 1.0
 107 Content-Type: text/plain; charset*=ISO-2022-KR''%1b%24%29%43%55%54%46%2d%38
 108 Content-Transfer-Encoding: quoted-printable
 109 Content-Description: bogus quoted-printable data
 110 Date: Sun, 18 Dec 2005 00:52:39 +0100
 111
 112 4 =C3=B7 2 =3D 2
 113 EOF
 114
 115 cat > $expected <<EOF
 116 [ Message inbox:12 ]
 117 Date:    Sun, 18 Dec 2005 00:52:39 +0100
 118 To:      bar@example.edu
 119 From:    foo@example.edu
 120 Subject: test
 121
 122 MIME-Version: 1.0
 123 Content-Description: bogus quoted-printable data
 124
 125 [ part  - text/plain - bogus quoted-printable data  11B  ]
 126 4 ÷ 2 = 2
 127 EOF
 128
 129 run_prog mhshow $msgnum > $actual 2>&1
 130 check "$expected" "$actual" : encoded parameter value
 131
 132
 133 #### mhshow replaces the invalid character in the charset parameter value with
 134 #### a '?'.  So, a charset name of "?us-ascii" is fed to iconv.  If
 135 #### $iconv_elides_question_marks, iconv will normalize that to "us-ascii".
 136 #### Otherwise, we expect iconv to fail.
 137 start_test 'replacement character in parameter value'
 138 cat > $msgfile <<'EOF'
 139 Subject: invalid parameter value charset
 140 MIME-Version: 1.0
 141 Content-Type: text/plain; charset*=invalid''%0Dus-ascii
 142 EOF
 143
 144 cat > $expected <<EOF
 145 [ Message inbox:12 ]
 146 Subject: invalid parameter value charset
 147
 148 MIME-Version: 1.0
 149
 150 [ part  - text/plain -   0B  ]
 151 EOF
 152
 153 run_prog mhshow $msgnum > $actual 2>&1
 154 if [ $iconv_elides_question_marks -eq 1 ]; then
 155     check "$expected" "$actual"
 156 else
 157     if grep "Can't convert .*us-ascii to UTF-8" "$actual" >/dev/null; then
 158         rm -f "$expected" "$actual"
 159     else
 160         failed=1
 161     fi
 162 fi
 163
 164
 165 finish_test
 166 exit $failed