1#!/usr/bin/perl
2
3use strict;
4use warnings;
5
6use HTTPTest;
7
8# cf. http://en.wikipedia.org/wiki/Latin1
9#     http://en.wikipedia.org/wiki/ISO-8859-15
10
11###############################################################################
12#
13# mime : charset found in Content-Type HTTP MIME header
14# meta : charset found in Content-Type meta tag
15#
16# index.html                  mime + file = iso-8859-15
17# p1_fran��ais.html            meta + file = iso-8859-1, mime = utf-8
18# p2_����n.html                 mime + file = iso-8859-1
19# p3_���������.html                 meta + file = utf-8, mime = iso-8859-1
20#
21
22my $ccedilla_l15 = "\xE7";
23my $ccedilla_u8 = "\xC3\xA7";
24my $eacute_l1 = "\xE9";
25my $eacute_u8 = "\xC3\xA9";
26my $eurosign_l15 = "\xA4";
27my $eurosign_u8 = "\xE2\x82\xAC";
28
29my $pageindex = <<EOF;
30<html>
31<head>
32  <title>Main Page</title>
33</head>
34<body>
35  <p>
36    Link to page 1 <a href="http://localhost:{{port}}/p1_fran${ccedilla_l15}ais.html">La seule page en fran&ccedil;ais</a>.
37    Link to page 3 <a href="http://localhost:{{port}}/p3_${eurosign_l15}${eurosign_l15}${eurosign_l15}.html">My tailor is rich</a>.
38  </p>
39</body>
40</html>
41EOF
42
43my $pagefrancais = <<EOF;
44<html>
45<head>
46  <title>La seule page en fran��ais</title>
47  <meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1"/>
48</head>
49<body>
50  <p>
51    Link to page 2 <a href="http://localhost:{{port}}/p2_${eacute_l1}${eacute_l1}n.html">Die enkele nerderlangstalige pagina</a>.
52  </p>
53</body>
54</html>
55EOF
56
57my $pageeen = <<EOF;
58<html>
59<head>
60  <title>Die enkele nederlandstalige pagina</title>
61</head>
62<body>
63  <p>
64    &Eacute;&eacute;n is niet veel maar toch meer dan nul.<br/>
65    Nerdelands is een mooie taal... dit zin stuckje spreekt vanzelf, of niet :)
66  </p>
67</body>
68</html>
69EOF
70
71my $pageeuro = <<EOF;
72<html>
73<head>
74  <title>Euro page</title>
75</head>
76<body>
77  <p>
78    My tailor isn't rich anymore.
79  </p>
80</body>
81</html>
82EOF
83
84my $page404 = <<EOF;
85<html>
86<head>
87  <title>404</title>
88</head>
89<body>
90  <p>
91    Nop nop nop...
92  </p>
93</body>
94</html>
95EOF
96
97# code, msg, headers, content
98my %urls = (
99    '/index.html' => {
100        code => "200",
101        msg => "Ok",
102        headers => {
103            "Content-type" => "text/html; charset=ISO-8859-15",
104        },
105        content => $pageindex,
106    },
107    '/robots.txt' => {
108        code => "200",
109        msg => "Ok",
110        headers => {
111            "Content-type" => "text/plain",
112        },
113        content => "",
114    },
115    '/p1_fran%C3%A7ais.html' => {	# UTF-8 encoded
116        code => "200",
117        msg => "File not found",
118        headers => {
119            "Content-type" => "text/html; charset=UTF-8",
120        },
121        content => $pagefrancais,
122    },
123    '/p1_fran%E7ais.html' => {
124        code => "200",
125        msg => "Ok",
126        headers => {
127            "Content-type" => "text/html; charset=UTF-8",
128        },
129        content => $pagefrancais,
130    },
131    '/p2_%C3%A9%C3%A9n.html' => {	# UTF-8 encoded
132        code => "200",
133        msg => "Ok",
134        headers => {
135            "Content-type" => "text/html; charset=UTF-8",
136        },
137        content => $pageeen,
138    },
139    '/p2_%E9%E9n.html' => {
140        code => "200",
141        msg => "Ok",
142        headers => {
143            "Content-type" => "text/html; charset=ISO-8859-1",
144        },
145        content => $pageeen,
146    },
147    '/p3_%E2%82%AC%E2%82%AC%E2%82%AC.html' => {	# UTF-8 encoded
148        code => "200",
149        msg => "Ok",
150        headers => {
151            "Content-type" => "text/plain",
152        },
153        content => $pageeuro,
154    },
155    '/p3_%A4%A4%A4.html' => {
156        code => "200",
157        msg => "Ok",
158        headers => {
159            "Content-type" => "text/plain",
160        },
161        content => $pageeuro,
162    },
163);
164
165my $cmdline = $WgetTest::WGETPATH . " --no-iri -nH -r http://localhost:{{port}}/";
166
167my $expected_error_code = 0;
168
169my %expected_downloaded_files = (
170    'index.html' => {
171        content => $pageindex,
172    },
173    'robots.txt' => {
174        content => "",
175    },
176    "p1_fran${ccedilla_l15}ais.html" => {
177        content => $pagefrancais,
178    },
179    "p2_${eacute_l1}${eacute_l1}n.html" => {
180        content => $pageeen,
181    },
182    "p3_${eurosign_l15}${eurosign_l15}${eurosign_l15}.html" => {
183        content => $pageeuro,
184    },
185);
186
187###############################################################################
188
189my $the_test = HTTPTest->new (name => "Test-iri-disabled",
190                              input => \%urls,
191                              cmdline => $cmdline,
192                              errcode => $expected_error_code,
193                              output => \%expected_downloaded_files);
194exit $the_test->run();
195
196# vim: et ts=4 sw=4
197
198