7 # cf. http://en.wikipedia.org/wiki/Latin1
8 # http://en.wikipedia.org/wiki/ISO-8859-15
10 ###############################################################################
12 # mime : charset found in Content-Type HTTP MIME header
13 # meta : charset found in Content-Type meta tag
15 # index.html mime + file = iso-8859-15
16 # p1_français.html meta + file = iso-8859-1, mime = utf-8
17 # p2_één.html meta + file = utf-8, mime =iso-8859-1
18 # p3_€€€.html meta + file = utf-8, mime = iso-8859-1
19 # p4_méér.html mime + file = utf-8
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";
29 my $pageindex = <<EOF;
32 <title>Main Page</title>
36 Link to page 1 <a href="http://localhost:{{port}}/p1_fran${ccedilla_l15}ais.html">La seule page en français</a>.
37 Link to page 3 <a href="http://localhost:{{port}}/p3_${eurosign_l15}${eurosign_l15}${eurosign_l15}.html">My tailor is rich</a>.
43 my $pagefrancais = <<EOF;
46 <title>La seule page en français</title>
47 <meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1"/>
51 Link to page 2 <a href="http://localhost:{{port}}/p2_${eacute_l1}${eacute_l1}n.html">Die enkele nerderlangstalige pagina</a>.
60 <title>Die enkele nederlandstalige pagina</title>
61 <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
65 Één is niet veel maar toch meer dan nul.<br/>
66 Nerdelands is een mooie taal... dit zin stuckje spreekt vanzelf, of niet :)<br/>
67 <a href="http://localhost:{{port}}/p4_m${eacute_u8}${eacute_u8}r.html">Méér</a>
76 <title>Euro page</title>
77 <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
81 My tailor isn't rich anymore.
90 <title>Bekende supermarkt</title>
94 Ik ben toch niet gek !
113 # code, msg, headers, content
119 "Content-type" => "text/html; charset=ISO-8859-15",
121 content => $pageindex,
127 "Content-type" => "text/plain",
131 '/p1_fran%C3%A7ais.html' => { # UTF-8 encoded
133 msg => "File not found",
135 "Content-type" => "text/html; charset=UTF-8",
139 '/p1_fran%E7ais.html' => {
143 "Content-type" => "text/html; charset=UTF-8",
145 content => $pagefrancais,
147 '/p2_%C3%A9%C3%A9n.html' => { # UTF-8 encoded
151 "Content-type" => "text/html; charset=ISO-8859-1",
155 '/p2_%E9%E9n.html' => {
159 "Content-type" => "text/html; charset=ISO-8859-1",
163 '/p3_%E2%82%AC%E2%82%AC%E2%82%AC.html' => { # UTF-8 encoded
167 "Content-type" => "text/plain; charset=ISO-8859-1",
169 content => $pageeuro,
171 '/p3_%A4%A4%A4.html' => {
175 "Content-type" => "text/plain; charset=ISO-8859-1",
177 content => $pageeuro,
179 '/p4_m%C3%A9%C3%A9r.html' => {
183 "Content-type" => "text/plain; charset=UTF-8",
185 content => $pagemeer,
189 my $cmdline = $WgetTest::WGETPATH . " --iri --restrict-file-names=nocontrol -nH -r http://localhost:{{port}}/";
191 my $expected_error_code = 0;
193 my %expected_downloaded_files = (
195 content => $pageindex,
200 "p1_fran${ccedilla_l15}ais.html" => {
201 content => $pagefrancais,
203 "p2_${eacute_u8}${eacute_u8}n.html" => {
206 "p3_${eurosign_u8}${eurosign_u8}${eurosign_u8}.html" => {
207 content => $pageeuro,
209 "p4_m${eacute_u8}${eacute_u8}r.html" => {
210 content => $pagemeer,
214 ###############################################################################
216 my $the_test = HTTPTest->new (name => "Test-iri",
219 errcode => $expected_error_code,
220 output => \%expected_downloaded_files);
221 exit $the_test->run();