1#!/usr/bin/env 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
45# specifying a wrong charset in http-equiv - it will be overridden by Content-Type HTTP header
46my $pagefrancais = <<EOF;
47<html>
48<head>
49  <title>La seule page en fran??ais</title>
50  <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
51</head>
52<body>
53  <p>
54    Link to page 2 <a href="http://localhost:{{port}}/p2_${eacute_l1}${eacute_l1}n.html">Die enkele nerderlangstalige pagina</a>.
55  </p>
56</body>
57</html>
58EOF
59
60my $pageeen = <<EOF;
61<html>
62<head>
63  <title>Die enkele nederlandstalige pagina</title>
64  <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
65</head>
66<body>
67  <p>
68    &Eacute;&eacute;n is niet veel maar toch meer dan nul.<br/>
69    Nerdelands is een mooie taal... dit zin stuckje spreekt vanzelf, of niet :)<br/>
70    <a href="http://localhost:{{port}}/p4_m${eacute_u8}${eacute_u8}r.html">M&eacute&eacute;r</a>
71  </p>
72</body>
73</html>
74EOF
75
76my $pageeuro = <<EOF;
77<html>
78<head>
79  <title>Euro page</title>
80  <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
81</head>
82<body>
83  <p>
84    My tailor isn't rich anymore.
85  </p>
86</body>
87</html>
88EOF
89
90my $pagemeer = <<EOF;
91<html>
92<head>
93  <title>Bekende supermarkt</title>
94</head>
95<body>
96  <p>
97    Ik ben toch niet gek !
98  </p>
99</body>
100</html>
101EOF
102
103my $page404 = <<EOF;
104<html>
105<head>
106  <title>404</title>
107</head>
108<body>
109  <p>
110    Nop nop nop...
111  </p>
112</body>
113</html>
114EOF
115
116# code, msg, headers, content
117my %urls = (
118    '/index.html' => {
119        code => "200",
120        msg => "Ok",
121        headers => {
122            "Content-type" => "text/html; charset=ISO-8859-15",
123        },
124        content => $pageindex,
125    },
126    '/robots.txt' => {
127        code => "200",
128        msg => "Ok",
129        headers => {
130            "Content-type" => "text/plain",
131        },
132        content => "",
133    },
134    '/p1_fran%C3%A7ais.html' => {	# UTF-8 encoded
135        code => "200",
136        msg => "Ok",
137        headers => {
138            # Content-Type header overrides http-equiv Content-Type
139            "Content-type" => "text/html; charset=ISO-8859-15",
140        },
141        content => $pagefrancais,
142    },
143    '/p2_%C3%A9%C3%A9n.html' => {	# UTF-8 encoded
144        code => "200",
145        msg => "Ok",
146        request_headers => {
147            "Referer" => qr|http://localhost:[0-9]+/p1_fran%C3%A7ais.html|,
148        },
149        headers => {
150            "Content-type" => "text/html; charset=UTF-8",
151        },
152        content => $pageeen,
153    },
154    '/p3_%E2%82%AC%E2%82%AC%E2%82%AC.html' => {	# UTF-8 encoded
155        code => "200",
156        msg => "Ok",
157        headers => {
158            "Content-type" => "text/plain; charset=ISO-8859-1",
159        },
160        content => $pageeuro,
161    },
162    '/p4_m%C3%A9%C3%A9r.html' => {
163        code => "200",
164        msg => "Ok",
165        request_headers => {
166            "Referer" => qr|http://localhost:[0-9]+/p2_%C3%A9%C3%A9n.html|,
167        },
168        headers => {
169            "Content-type" => "text/plain; charset=UTF-8",
170        },
171        content => $pagemeer,
172    },
173);
174
175my $cmdline = $WgetTest::WGETPATH . " --iri --trust-server-names --restrict-file-names=nocontrol -nH -r http://localhost:{{port}}/";
176
177my $expected_error_code = 0;
178
179my %expected_downloaded_files = (
180    'index.html' => {
181        content => $pageindex,
182    },
183    'robots.txt' => {
184        content => "",
185    },
186    "p1_fran${ccedilla_u8}ais.html" => {
187        content => $pagefrancais,
188    },
189    "p2_${eacute_u8}${eacute_u8}n.html" => {
190        content => $pageeen,
191    },
192    "p3_${eurosign_u8}${eurosign_u8}${eurosign_u8}.html" => {
193        content => $pageeuro,
194    },
195    "p4_m${eacute_u8}${eacute_u8}r.html" => {
196        content => $pagemeer,
197    },
198);
199
200###############################################################################
201
202my $the_test = HTTPTest->new (input => \%urls,
203                              cmdline => $cmdline,
204                              errcode => $expected_error_code,
205                              output => \%expected_downloaded_files);
206exit $the_test->run();
207
208# vim: et ts=4 sw=4
209