wget/tests/Test-iri-percent.px

89 lines
2.0 KiB
Plaintext
Raw Normal View History

2022-05-14 03:18:11 +08:00
#!/usr/bin/env -S perl -I .
use strict;
use warnings;
use WgetFeature qw(iri);
use HTTPTest;
# Just a sanity check to verify that %-encoded values are always left
# untouched.
my $ccedilla_l15 = "\xE7";
my $ccedilla_l15_pct = "%E7";
my $ccedilla_u8 = "\xC3\xA7";
my $ccedilla_u8_pct = "%C3%A7";
my $eacute_l1 = "\xE9";
my $eacute_u8 = "\xC3\xA9";
my $eacute_u8_pct = "%C3%A9";
my $pageindex = <<EOF;
<html>
<head>
<title>Main Page</title>
</head>
<body>
<p>
Link to page 1 <a
href="http://localhost:{{port}}/hello_${ccedilla_l15_pct}${eacute_l1}.html">La seule page en fran&ccedil;ais</a>.
</p>
</body>
</html>
EOF
my $pagefrancais = <<EOF;
<html>
<head>
<title>La seule page en français</title>
<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1"/>
</head>
<body>
<p>
</p>
</body>
</html>
EOF
# code, msg, headers, content
my %urls = (
'/index.html' => {
code => "200",
msg => "Ok",
headers => {
"Content-type" => "text/html; charset=ISO-8859-15",
},
content => $pageindex,
},
"/hello_${ccedilla_u8_pct}${eacute_u8_pct}.html" => {
code => "200",
msg => "Ok",
headers => {
"Content-type" => "text/html; charset=UTF-8",
},
content => $pagefrancais,
},
);
my $cmdline = $WgetTest::WGETPATH . " --iri -e robots=off --restrict-file-names=nocontrol -nH -r http://localhost:{{port}}/";
my $expected_error_code = 0;
my %expected_downloaded_files = (
'index.html' => {
content => $pageindex,
},
"hello_${ccedilla_u8}${eacute_u8}.html" => {
content => $pagefrancais,
},
);
###############################################################################
my $the_test = HTTPTest->new (input => \%urls,
cmdline => $cmdline,
errcode => $expected_error_code,
output => \%expected_downloaded_files);
exit $the_test->run();
# vim: et ts=4 sw=4