1#!/usr/bin/perl
2
3use strict;
4use warnings;
5
6use WgetFeature qw(iri);
7use HTTPTest;
8
9# cf. http://en.wikipedia.org/wiki/Latin1
10#     http://en.wikipedia.org/wiki/ISO-8859-15
11
12###############################################################################
13#
14# mime : charset found in Content-Type HTTP MIME header
15# meta : charset found in Content-Type meta tag
16#
17# index.html                  mime + file = iso-8859-15
18# p1_fran��ais.html            meta + file = iso-8859-1, mime = utf-8
19# p2_����n.html                 meta + file = utf-8, mime =iso-8859-1
20# p3_���������.html                 meta + file = utf-8, mime = iso-8859-1
21# p4_m����r.html                mime + file = utf-8
22#
23
24my $ccedilla_l15 = "\xE7";
25my $ccedilla_u8 = "\xC3\xA7";
26my $eacute_l1 = "\xE9";
27my $eacute_u8 = "\xC3\xA9";
28my $eurosign_l15 = "\xA4";
29my $eurosign_u8 = "\xE2\x82\xAC";
30
31my $pageindex = <<EOF;
32<html>
33<head>
34  <title>Main Page</title>
35</head>
36<body>
37  <p>
38    Link to page 1 <a href="http://localhost:{{port}}/p1_fran${ccedilla_l15}ais.html">La seule page en fran&ccedil;ais</a>.
39    Link to page 3 <a href="http://localhost:{{port}}/p3_${eurosign_l15}${eurosign_l15}${eurosign_l15}.html">My tailor is rich</a>.
40  </p>
41</body>
42</html>
43EOF
44
45my $pagefrancais = <<EOF;
46<html>
47<head>
48  <title>La seule page en fran??ais</title>
49  <meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1"/>
50</head>
51<body>
52  <p>
53    Link to page 2 <a href="http://localhost:{{port}}/p2_${eacute_l1}${eacute_l1}n.html">Die enkele nerderlangstalige pagina</a>.
54  </p>
55</body>
56</html>
57EOF
58
59my $pageeen = <<EOF;
60<html>
61<head>
62  <title>Die enkele nederlandstalige pagina</title>
63  <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
64</head>
65<body>
66  <p>
67    &Eacute;&eacute;n is niet veel maar toch meer dan nul.<br/>
68    Nerdelands is een mooie taal... dit zin stuckje spreekt vanzelf, of niet :)<br/>
69    <a href="http://localhost:{{port}}/p4_m${eacute_u8}${eacute_u8}r.html">M&eacute&eacute;r</a>
70  </p>
71</body>
72</html>
73EOF
74
75my $pageeuro = <<EOF;
76<html>
77<head>
78  <title>Euro page</title>
79  <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
80</head>
81<body>
82  <p>
83    My tailor isn't rich anymore.
84  </p>
85</body>
86</html>
87EOF
88
89my $pagemeer = <<EOF;
90<html>
91<head>
92  <title>Bekende supermarkt</title>
93</head>
94<body>
95  <p>
96    Ik ben toch niet gek !
97  </p>
98</body>
99</html>
100EOF
101
102my $page404 = <<EOF;
103<html>
104<head>
105  <title>404</title>
106</head>
107<body>
108  <p>
109    Nop nop nop...
110  </p>
111</body>
112</html>
113EOF
114
115# code, msg, headers, content
116my %urls = (
117    '/index.html' => {
118        code => "200",
119        msg => "Ok",
120        headers => {
121            "Content-type" => "text/html; charset=ISO-8859-15",
122        },
123        content => $pageindex,
124    },
125    '/robots.txt' => {
126        code => "200",
127        msg => "Ok",
128        headers => {
129            "Content-type" => "text/plain",
130        },
131        content => "",
132    },
133    '/p1_fran%C3%A7ais.html' => {	# UTF-8 encoded
134        code => "404",
135        msg => "File not found",
136        headers => {
137            "Content-type" => "text/html; charset=UTF-8",
138        },
139        content => $page404,
140    },
141    '/p1_fran%E7ais.html' => {
142        code => "200",
143        msg => "Ok",
144        headers => {
145            "Content-type" => "text/html; charset=UTF-8",
146        },
147        content => $pagefrancais,
148    },
149    '/p2_%C3%A9%C3%A9n.html' => {	# UTF-8 encoded
150        code => "200",
151        msg => "Ok",
152        request_headers => {
153            "Referer" => qr|http://localhost:[0-9]+/p1_fran%E7ais.html|,
154        },
155        headers => {
156            "Content-type" => "text/html; charset=ISO-8859-1",
157        },
158        content => $pageeen,
159    },
160    '/p3_%E2%82%AC%E2%82%AC%E2%82%AC.html' => {	# UTF-8 encoded
161        code => "200",
162        msg => "Ok",
163        headers => {
164            "Content-type" => "text/plain; charset=ISO-8859-1",
165        },
166        content => $pageeuro,
167    },
168    '/p3_%A4%A4%A4.html' => {
169        code => "200",
170        msg => "Ok",
171        headers => {
172            "Content-type" => "text/plain; charset=ISO-8859-1",
173        },
174        content => $pageeuro,
175    },
176    '/p4_m%C3%A9%C3%A9r.html' => {
177        code => "200",
178        msg => "Ok",
179        request_headers => {
180            "Referer" => qr|http://localhost:[0-9]+/p2_%C3%A9%C3%A9n.html|,
181        },
182        headers => {
183            "Content-type" => "text/plain; charset=UTF-8",
184        },
185        content => $pagemeer,
186    },
187);
188
189my $cmdline = $WgetTest::WGETPATH . " -d --iri --restrict-file-names=nocontrol -nH -r http://localhost:{{port}}/";
190
191my $expected_error_code = 0;
192
193my %expected_downloaded_files = (
194    'index.html' => {
195        content => $pageindex,
196    },
197    'robots.txt' => {
198        content => "",
199    },
200    "p1_fran${ccedilla_l15}ais.html" => {
201        content => $pagefrancais,
202    },
203    "p2_${eacute_u8}${eacute_u8}n.html" => {
204        content => $pageeen,
205    },
206    "p3_${eurosign_u8}${eurosign_u8}${eurosign_u8}.html" => {
207        content => $pageeuro,
208    },
209    "p4_m${eacute_u8}${eacute_u8}r.html" => {
210        content => $pagemeer,
211    },
212);
213
214###############################################################################
215
216my $the_test = HTTPTest->new (name => "Test-iri",
217                              input => \%urls,
218                              cmdline => $cmdline,
219                              errcode => $expected_error_code,
220                              output => \%expected_downloaded_files);
221exit $the_test->run();
222
223# vim: et ts=4 sw=4
224
225