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# Force remote encoding to ISO-8859-1
14#
15# mime : charset found in Content-Type HTTP MIME header
16# meta : charset found in Content-Type meta tag
17#
18# index.html                  mime + file = iso-8859-15
19# p1_fran��ais.html            meta + file = iso-8859-1, mime = utf-8
20# p2_����n.html                 mime + file = iso-8859-1
21# p3_���������.html                 meta + file = utf-8, mime = iso-8859-1
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";
30my $currency_l1 = "\xA4";
31my $currency_u8 = "\xC2\xA4";
32
33my $pageindex = <<EOF;
34<html>
35<head>
36  <title>Main Page</title>
37</head>
38<body>
39  <p>
40    Link to page 1 <a href="http://localhost:{{port}}/p1_fran${ccedilla_l15}ais.html">La seule page en fran&ccedil;ais</a>.
41    Link to page 3 <a href="http://localhost:{{port}}/p3_${eurosign_l15}${eurosign_l15}${eurosign_l15}.html">My tailor is rich</a>.
42  </p>
43</body>
44</html>
45EOF
46
47my $pagefrancais = <<EOF;
48<html>
49<head>
50  <title>La seule page en fran??ais</title>
51  <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
52</head>
53<body>
54  <p>
55    Link to page 2 <a href="http://localhost:{{port}}/p2_${eacute_l1}${eacute_l1}n.html">Die enkele nerderlangstalige pagina</a>.
56  </p>
57</body>
58</html>
59EOF
60
61my $pageeen = <<EOF;
62<html>
63<head>
64  <title>Die enkele nederlandstalige pagina</title>
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 :)
70  </p>
71</body>
72</html>
73EOF
74
75my $pageeuro = <<EOF;
76<html>
77<head>
78  <title>Euro page</title>
79</head>
80<body>
81  <p>
82    My tailor isn't rich anymore.
83  </p>
84</body>
85</html>
86EOF
87
88my $page404 = <<EOF;
89<html>
90<head>
91  <title>404</title>
92</head>
93<body>
94  <p>
95    Nop nop nop...
96  </p>
97</body>
98</html>
99EOF
100
101# code, msg, headers, content
102my %urls = (
103    '/index.html' => {
104        code => "200",
105        msg => "Ok",
106        headers => {
107            "Content-type" => "text/html; charset=ISO-8859-15",
108        },
109        content => $pageindex,
110    },
111    '/robots.txt' => {
112        code => "200",
113        msg => "Ok",
114        headers => {
115            "Content-type" => "text/plain",
116        },
117        content => "",
118    },
119    '/p1_fran%C3%A7ais.html' => {	# UTF-8 encoded
120        code => "200",
121        msg => "Ok",
122        headers => {
123            # wrong charset here, overridden by --remote-encoding=iso-8859-1
124            "Content-type" => "text/html; charset=UTF-8",
125        },
126        content => $pagefrancais,
127    },
128    '/p2_%C3%A9%C3%A9n.html' => {	# UTF-8 encoded
129        code => "200",
130        msg => "Ok",
131        headers => {
132            "Content-type" => "text/html; charset=UTF-8",
133        },
134        content => $pageeen,
135    },
136    '/p3_%E2%82%AC%E2%82%AC%E2%82%AC.html' => {	# UTF-8 encoded
137        code => "200",
138        msg => "Ok",
139        headers => {
140            "Content-type" => "text/plain",
141        },
142        content => $pageeuro,
143    },
144    '/p3_%C2%A4%C2%A4%C2%A4.html' => {	# UTF-8 encoded
145        code => "200",
146        msg => "Ok",
147        headers => {
148            "Content-type" => "text/plain",
149        },
150        content => $pageeuro,
151    },
152);
153
154my $cmdline = $WgetTest::WGETPATH . " --iri -e robots=on --trust-server-names --remote-encoding=iso-8859-1 -nH -r http://localhost:{{port}}/";
155
156my $expected_error_code = 0;
157
158my %expected_downloaded_files = (
159    'index.html' => {
160        content => $pageindex,
161    },
162    'robots.txt' => {
163        content => "",
164    },
165    "p1_fran${ccedilla_u8}ais.html" => {
166        content => $pagefrancais,
167    },
168    "p2_${eacute_u8}${eacute_u8}n.html" => {
169        content => $pageeen,
170    },
171    "p3_${currency_u8}${currency_u8}${currency_u8}.html" => {
172        content => $pageeuro,
173    },
174);
175
176###############################################################################
177
178my $the_test = HTTPTest->new (input => \%urls,
179                              cmdline => $cmdline,
180                              errcode => $expected_error_code,
181                              output => \%expected_downloaded_files);
182exit $the_test->run();
183
184# vim: et ts=4 sw=4
185