xref: /openbsd-src/gnu/usr.bin/perl/cpan/Encode/t/at-cn.t (revision 5486feefcc8cb79b19e014ab332cc5dfd05b3b33) 
1BEGIN {
2    if ($ENV{'PERL_CORE'}){
3        chdir 't';
4    unshift @INC, '../lib';
5    }
6    require Config; Config->import();
7    if ($Config{'extensions'} !~ /\bEncode\b/) {
8      print "1..0 # Skip: Encode was not built\n";
9      exit 0;
10    }
11    if (ord("A") == 193) {
12    print "1..0 # Skip: EBCDIC\n";
13    exit 0;
14    }
15    $| = 1;
16}
17
18use strict;
19use Test::More tests => 29;
20use Encode;
21
22no utf8; # we have raw Chinese encodings here
23
24BEGIN {
25    use_ok('Encode::CN');
26}
27
28# Since JP.t already tests basic file IO, we will just focus on
29# internal encode / decode test here. Unfortunately, to test
30# against all the UniHan characters will take a huge disk space,
31# not to mention the time it will take, and the fact that Perl
32# did not bundle UniHan.txt anyway.
33
34# So, here we just test a typical snippet spanning multiple Unicode
35# blocks, and hope it can point out obvious errors.
36
37run_tests('Simplified Chinese only', {
38    'utf'	=> (
3912298.26131.32463.12299.31532.19968.21350.
4024406.26352.65306.
4122823.21705.20094.20803.65292.19975.29289.36164.22987.65292.
4220035.32479.22825.12290.
4320113.34892.38632.26045.65292.21697.29289.27969.24418.12290.
4422823.26126.22987.32456.65292.20845.20301.26102.25104.65292.
4526102.20056.20845.40857.20197.24481.22825.12290.
4620094.36947.21464.21270.65292.21508.27491.24615.21629.65292.
4720445.21512.22823.21644.65292.20035.21033.36126.12290.
4839318.20986.24246.29289.65292.19975.22269.21688.23425.12290
49    ),
50
51    'euc-cn'	=> join('',
52'���׾�����һ��',
53'��Ի��',
54'����ǬԪ��������ʼ��',
55'��ͳ�졣',
56'������ʩ��Ʒ�����Ρ�',
57'����ʼ�գ���λʱ�ɣ�',
58'ʱ�����������졣',
59'Ǭ���仯������������',
60'���ϴ�ͣ������ꡣ',
61'�׳�������������',
62    ),
63
64    'gb2312-raw'	=> join('',
65'!6RW>-!75ZR;XT',
66'ehT;#:',
67'4sTUG,T*#,MrNoWJJ<#,',
68'DKM3Ll!#',
69'TFPPSjJ)#,F7NoAwPN!#',
70'4sCwJ<VU#,AyN;J13I#,',
71'J13KAyAzRTSyLl!#',
72'G,5@1d;/#,8wU}PTC|#,',
73'1#:O4s:M#,DK@{Uj!#',
74'JW3vJ|No#,Mr9zOLD~!#'
75    ),
76
77    'iso-ir-165'=> join('',
78'!6RW>-!75ZR;XT',
79'ehT;#:',
80'4sTUG,T*#,MrNoWJJ<#,',
81'DKM3Ll!#',
82'TFPPSjJ)#,F7NoAwPN!#',
83'4sCwJ<VU#,AyN;J13I#,',
84'J13KAyAzRTSyLl!#',
85'G,5@1d;/#,8wU}PTC|#,',
86'1#:O4s:M#,DK@{Uj!#',
87'JW3vJ|No#,Mr9zOLD~!#'
88    ),
89});
90
91run_tests('Simplified Chinese + ASCII', {
92    'utf'	=> (
9335937.26352.65306.10.
9422825.34892.20581.65292.21531.23376.20197.33258.24378.19981.24687.12290.10.
9528508.40857.21247.29992.65292.38451.22312.19979.20063.12290.32.
9635265.40857.22312.30000.65292.24503.26045.26222.20063.12290.32.
9732456.26085.20094.20094.65292.21453.22797.36947.20063.12290.10.
9825110.36291.22312.28170.65292.36827.26080.21646.20063.12290.39134.
9940857.22312.22825.65292.22823.20154.36896.20063.12290.32.
10020130.40857.26377.24724.65292.30408.19981.21487.20037.20063.12290.10.
10129992.20061.65292.22825.24503.19981.21487.20026.39318.20063.12290
102    ),
103
104    'cp936'	=> join(chr(10),
105'��Ի��',
106'���н�����������ǿ��Ϣ��',
107'DZ�����ã�������Ҳ�� ���������ʩ��Ҳ�� ����ǬǬ��������Ҳ��',
108'��Ծ��Ԩ�����޾�Ҳ���������죬������Ҳ�� �����лڣ�ӯ���ɾ�Ҳ��',
109'�þţ���²���Ϊ��Ҳ��',
110    ),
111
112    'hz'	=> join(chr(10),
113'~{OsT;#:~}',
114'~{LlPP=!#,>}WSRTWTG?2;O"!#~}',
115'~{G1AzNpSC#,QtTZOBR2!#~} ~{<{AzTZLo#,5BJ)FUR2!#~} ~{VUHUG,G,#,74845@R2!#~}',
116'~{;rT>TZT(#,=xN^>LR2!#7IAzTZLl#,4sHKTlR2!#~} ~{?:AzSP;Z#,S/2;?I>CR2!#~}',
117'~{SC>E#,Ll5B2;?IN*JWR2!#~}',
118    ),
119});
120
121run_tests('Traditional Chinese', {
122    'utf',	=> 20094.65306.20803.12289.20136.12289.21033.12289.35998,
123    'gb12345-raw'	=> 'G,#:T*!":`!"@{!"Uj',
124    'gbk'	=> 'Ǭ��Ԫ���ࡢ����ؑ',
125});
126
127sub run_tests {
128    my ($title, $tests) = @_;
129    my $utf = delete $tests->{'utf'};
130
131    # $enc = encoding, $str = content
132    foreach my $enc (sort keys %{$tests}) {
133    my $str = $tests->{$enc};
134
135    is(Encode::decode($enc, $str), $utf, "[$enc] decode - $title");
136    is(Encode::encode($enc, $utf), $str, "[$enc] encode - $title");
137
138    my $str2 = $str;
139    my $utf8 = Encode::encode('utf-8', $utf);
140
141    Encode::from_to($str2, $enc, 'utf-8');
142    is($str2, $utf8, "[$enc] from_to => utf8 - $title");
143
144    Encode::from_to($utf8, 'utf-8', $enc); # convert $utf8 as $enc
145    is($utf8, $str,  "[$enc] utf8 => from_to - $title");
146    }
147}
148