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# 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_français.html            meta + file = iso-8859-1, mime = utf-8
20# p2_één.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 $pagefrancais = <<EOF;
48<html>
49<head>
50  <title>La seule page en fran��ais</title>
51  <meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1"/>
52</head>
53<body>
54  <p>
55    Link to page 2 <a href="http://localhost:{{port}}/p2_${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 => "404",
121        msg => "File not found",
122        headers => {
123            "Content-type" => "text/html; charset=UTF-8",
124        },
125        content => $page404,
126    },
127    '/p1_fran%E7ais.html' => {
128        code => "200",
129        msg => "Ok",
130        headers => {
131            "Content-type" => "text/html; charset=UTF-8",
132        },
133        content => $pagefrancais,
134    },
135    '/p2_%C3%A9%C3%A9n.html' => {	# UTF-8 encoded
136        code => "200",
137        msg => "Ok",
138        headers => {
139            "Content-type" => "text/html; charset=UTF-8",
140        },
141        content => $pageeen,
142    },
143    '/p2_%E9%E9n.html' => {
144        code => "200",
145        msg => "Ok",
146        headers => {
147            "Content-type" => "text/html; charset=ISO-8859-1",
148        },
149        content => $pageeen,
150    },
151    '/p3_%E2%82%AC%E2%82%AC%E2%82%AC.html' => {	# UTF-8 encoded
152        code => "200",
153        msg => "Ok",
154        headers => {
155            "Content-type" => "text/plain",
156        },
157        content => $pageeuro,
158    },
159    '/p3_%A4%A4%A4.html' => {
160        code => "200",
161        msg => "Ok",
162        headers => {
163            "Content-type" => "text/plain",
164        },
165        content => $pageeuro,
166    },
167    '/p3_%C2%A4%C2%A4%C2%A4.html' => {	# UTF-8 encoded
168        code => "200",
169        msg => "Ok",
170        headers => {
171            "Content-type" => "text/plain",
172        },
173        content => $pageeuro,
174    },
175);
176
177my $cmdline = $WgetTest::WGETPATH . " --iri --remote-encoding=iso-8859-1 -nH -r http://localhost:{{port}}/";
178
179my $expected_error_code = 0;
180
181my %expected_downloaded_files = (
182    'index.html' => {
183        content => $pageindex,
184    },
185    'robots.txt' => {
186        content => "",
187    },
188    "p1_fran${ccedilla_l15}ais.html" => {
189        content => $pagefrancais,
190    },
191    "p2_${eacute_u8}${eacute_u8}n.html" => {
192        content => $pageeen,
193    },
194    "p3_${currency_u8}${currency_u8}${currency_u8}.html" => {
195        content => $pageeuro,
196    },
197);
198
199###############################################################################
200
201my $the_test = HTTPTest->new (name => "Test-iri-forced-remote",
202                              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
210