7 # cf. http://en.wikipedia.org/wiki/Latin1
8 # http://en.wikipedia.org/wiki/ISO-8859-15
10 ###############################################################################
11 # Force remote encoding to ISO-8859-1
13 # mime : charset found in Content-Type HTTP MIME header
14 # meta : charset found in Content-Type meta tag
16 # index.html mime + file = iso-8859-15
17 # p1_français.html meta + file = iso-8859-1, mime = utf-8
18 # p2_één.html mime + file = iso-8859-1
19 # p3_€€€.html meta + file = utf-8, mime = iso-8859-1
22 my $ccedilla_l15 = "\xE7";
23 my $ccedilla_u8 = "\xC3\xA7";
24 my $eacute_l1 = "\xE9";
25 my $eacute_u8 = "\xC3\xA9";
26 my $eurosign_l15 = "\xA4";
27 my $eurosign_u8 = "\xE2\x82\xAC";
28 my $currency_l1 = "\xA4";
29 my $currency_u8 = "\xC2\xA4";
31 my $pageindex = <<EOF;
34 <title>Main Page</title>
38 Link to page 1 <a href="http://localhost:{{port}}/p1_fran${ccedilla_l15}ais.html">La seule page en français</a>.
39 Link to page 3 <a href="http://localhost:{{port}}/p3_${eurosign_l15}${eurosign_l15}${eurosign_l15}.html">My tailor is rich</a>.
45 my $pagefrancais = <<EOF;
48 <title>La seule page en français</title>
49 <meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1"/>
53 Link to page 2 <a href="http://localhost:{{port}}/p2_${eacute_l1}${eacute_l1}n.html">Die enkele nerderlangstalige pagina</a>.
62 <title>Die enkele nederlandstalige pagina</title>
66 Één is niet veel maar toch meer dan nul.<br/>
67 Nerdelands is een mooie taal... dit zin stuckje spreekt vanzelf, of niet :)
76 <title>Euro page</title>
80 My tailor isn't rich anymore.
99 # code, msg, headers, content
105 "Content-type" => "text/html; charset=ISO-8859-15",
107 content => $pageindex,
113 "Content-type" => "text/plain",
117 '/p1_fran%C3%A7ais.html' => { # UTF-8 encoded
119 msg => "File not found",
121 "Content-type" => "text/html; charset=UTF-8",
125 '/p1_fran%E7ais.html' => {
129 "Content-type" => "text/html; charset=UTF-8",
131 content => $pagefrancais,
133 '/p2_%C3%A9%C3%A9n.html' => { # UTF-8 encoded
137 "Content-type" => "text/html; charset=UTF-8",
141 '/p2_%E9%E9n.html' => {
145 "Content-type" => "text/html; charset=ISO-8859-1",
149 '/p3_%E2%82%AC%E2%82%AC%E2%82%AC.html' => { # UTF-8 encoded
153 "Content-type" => "text/plain",
155 content => $pageeuro,
157 '/p3_%A4%A4%A4.html' => {
161 "Content-type" => "text/plain",
163 content => $pageeuro,
165 '/p3_%C2%A4%C2%A4%C2%A4.html' => { # UTF-8 encoded
169 "Content-type" => "text/plain",
171 content => $pageeuro,
175 my $cmdline = $WgetTest::WGETPATH . " --iri --remote-encoding=iso-8859-1 -nH -r http://localhost:{{port}}/";
177 my $expected_error_code = 0;
179 my %expected_downloaded_files = (
181 content => $pageindex,
186 "p1_fran${ccedilla_l15}ais.html" => {
187 content => $pagefrancais,
189 "p2_${eacute_u8}${eacute_u8}n.html" => {
192 "p3_${currency_u8}${currency_u8}${currency_u8}.html" => {
193 content => $pageeuro,
197 ###############################################################################
199 my $the_test = HTTPTest->new (name => "Test-iri-forced-remote",
202 errcode => $expected_error_code,
203 output => \%expected_downloaded_files);
204 exit $the_test->run();