It is usefull to add two test for IDN with UTF-8 local encoding: 1. Testing for translating hostname to punycode. 2. Testing for downloading robots.txt with UTF-8 local encoding.
Any trivial fix breaks at least the second test. Following patch creates this tests based on Test-idn-cmd.px Test-idn-robots.px tests.
=== modified file 'tests/ChangeLog' --- tests/ChangeLog 2011-04-19 10:33:46 +0000 +++ tests/ChangeLog 2011-07-07 15:25:39 +0000 @@ -1,3 +1,9 @@ +2011-06-03 Merinov Nikolay <[email protected]> + + * Test-idn-cmd-utf8.px: Added test for idn with utf-8 local encoding. + * Test-idn-robots-utf8.px: Added test for idn with utf-8 local encoding + and robots.txt file. + * Makefile.am, run-px: Add new tests. 2011-04-19 Giuseppe Scrivano <[email protected]> * Makefile.am (LIBS): Add $(LIB_CLOCK_GETTIME). === modified file 'tests/Makefile.am' --- tests/Makefile.am 2011-04-19 10:33:46 +0000 +++ tests/Makefile.am 2011-07-07 15:25:20 +0000 @@ -90,7 +90,9 @@ Test-idn-headers.px \ Test-idn-meta.px \ Test-idn-cmd.px \ + Test-idn-cmd-utf8.px \ Test-idn-robots.px \ + Test-idn-robots-utf8.px \ Test-iri.px \ Test-iri-percent.px \ Test-iri-disabled.px \ === added file 'tests/Test-idn-cmd-utf8.px' --- tests/Test-idn-cmd-utf8.px 1970-01-01 00:00:00 +0000 +++ tests/Test-idn-cmd-utf8.px 2011-07-07 15:23:08 +0000 @@ -0,0 +1,52 @@ +#!/usr/bin/env perl + +use strict; +use warnings; + +use WgetFeature qw(iri); +use HTTPTest; + +# " Kon'nichiwa <dot> Japan +my $utf8_hostname = "\344\273\212\346\227\245\343\201\257.\346\227\245\346\234\254"; +my $punycoded_hostname = 'xn--v9ju72g90p.xn--wgv71a'; + +############################################################################### + +my $result_file = <<EOF; +Found me! +EOF + +# code, msg, headers, content +my %urls = ( + "http://$punycoded_hostname/index.html"; => { + code => "200", + msg => "Yes, please", + headers => { + 'Content-Type' => 'text/plain', + }, + content => $result_file, + }, +); + +my $cmdline = $WgetTest::WGETPATH . " --iri -rH" + . " -e http_proxy=localhost:{{port}} --local-encoding=UTF-8 $utf8_hostname"; + +my $expected_error_code = 0; + +my %expected_downloaded_files = ( + "$punycoded_hostname/index.html" => { + content => $result_file, + }, +); + +############################################################################### + +my $the_test = HTTPTest->new (name => "Test-idn-cmd", + input => \%urls, + cmdline => $cmdline, + errcode => $expected_error_code, + output => \%expected_downloaded_files); +exit $the_test->run(); + +# vim: et ts=4 sw=4 + === added file 'tests/Test-idn-robots-utf8.px' --- tests/Test-idn-robots-utf8.px 1970-01-01 00:00:00 +0000 +++ tests/Test-idn-robots-utf8.px 2011-07-03 18:26:26 +0000 @@ -0,0 +1,79 @@ +#!/usr/bin/env perl + +use strict; +use warnings; + +use WgetFeature qw(iri); +use HTTPTest; + +# " Kon'nichiwa <dot> Japan +my $utf8_hostname = "\344\273\212\346\227\245\343\201\257.\346\227\245\346\234\254"; +my $punycoded_hostname = 'xn--v9ju72g90p.xn--wgv71a'; + +############################################################################### + +my $starter_file = <<EOF; +<a href="http://$utf8_hostname/foo.txt";>The link</a> +EOF + +my $result_file = <<EOF; +Found me! +EOF + +# code, msg, headers, content +my %urls = ( + "http://$punycoded_hostname/index.html"; => { + code => "200", + msg => "Yes, please", + headers => { + 'Content-Type' => 'text/html; charset=UTF-8', + }, + content => $starter_file, + }, + "http://$punycoded_hostname/foo.txt"; => { + code => "200", + msg => "Uh-huh", + headers => { + 'Content-Type' => 'text/plain', + }, + content => $result_file, + }, + "http://$punycoded_hostname/robots.txt"; => { + code => "200", + msg => "Uh-huh", + headers => { + 'Content-Type' => 'text/plain', + }, + content => '', + }, +); + +my $cmdline = $WgetTest::WGETPATH . " --iri -rH" + . " -e http_proxy=localhost:{{port}} --local-encoding=UTF-8" + . " http://$utf8_hostname/";; + +my $expected_error_code = 0; + +my %expected_downloaded_files = ( + "$punycoded_hostname/index.html" => { + content => $starter_file, + }, + "$punycoded_hostname/foo.txt" => { + content => $result_file, + }, + "$punycoded_hostname/robots.txt" => { + content => '', + }, +); + +############################################################################### + +my $the_test = HTTPTest->new (name => "Test-idn-robots", + input => \%urls, + cmdline => $cmdline, + errcode => $expected_error_code, + output => \%expected_downloaded_files); +exit $the_test->run(); + +# vim: et ts=4 sw=4 + === modified file 'tests/run-px' --- tests/run-px 2010-05-29 21:10:59 +0000 +++ tests/run-px 2011-07-07 15:25:32 +0000 @@ -43,7 +43,9 @@ 'Test-idn-headers.px', 'Test-idn-meta.px', 'Test-idn-cmd.px', + 'Test-idn-cmd-utf8.px', 'Test-idn-robots.px', + 'Test-idn-robots-utf8.px', 'Test-iri.px', 'Test-iri-percent.px', 'Test-iri-disabled.px',
