[p5sagit/p5-mst-13.2.git] / t / pragma / utf8.t

#!./perl 

BEGIN {
    chdir 't' if -d 't';
    unshift @INC, '../lib';
    $ENV{PERL5LIB} = '../lib';
    if ( ord("\t") != 9 ) { # skip on ebcdic platforms
        print "1..0 # Skip utf8 tests on ebcdic platform.\n";
        exit;
    }
}

print "1..65\n";

my $test = 1;

sub ok {
    my ($got,$expect) = @_;
    print "# expected [$expect], got [$got]\nnot " if $got ne $expect;
    print "ok $test\n";
}

sub ok_bytes {
    use bytes;
    my ($got,$expect) = @_;
    print "# expected [$expect], got [$got]\nnot " if $got ne $expect;
    print "ok $test\n";
}


{
    use utf8;
    $_ = ">\x{263A}<"; 
    s/([\x{80}-\x{10ffff}])/"&#".ord($1).";"/eg; 
    ok $_, '>&#9786;<';
    $test++;				# 1

    $_ = ">\x{263A}<"; 
    my $rx = "\x{80}-\x{10ffff}";
    s/([$rx])/"&#".ord($1).";"/eg; 
    ok $_, '>&#9786;<';
    $test++;				# 2

    $_ = ">\x{263A}<"; 
    my $rx = "\\x{80}-\\x{10ffff}";
    s/([$rx])/"&#".ord($1).";"/eg; 
    ok $_, '>&#9786;<';
    $test++;				# 3

    $_ = "alpha,numeric"; 
    m/([[:alpha:]]+)/; 
    ok $1, 'alpha';
    $test++;				# 4

    $_ = "alphaNUMERICstring";
    m/([[:^lower:]]+)/; 
    ok $1, 'NUMERIC';
    $test++;				# 5

    $_ = "alphaNUMERICstring";
    m/(\p{Ll}+)/; 
    ok $1, 'alpha';
    $test++;				# 6

    $_ = "alphaNUMERICstring"; 
    m/(\p{Lu}+)/; 
    ok $1, 'NUMERIC';
    $test++;				# 7

    $_ = "alpha,numeric"; 
    m/([\p{IsAlpha}]+)/; 
    ok $1, 'alpha';
    $test++;				# 8

    $_ = "alphaNUMERICstring";
    m/([^\p{IsLower}]+)/; 
    ok $1, 'NUMERIC';
    $test++;				# 9

    $_ = "alpha123numeric456"; 
    m/([\p{IsDigit}]+)/; 
    ok $1, '123';
    $test++;				# 10

    $_ = "alpha123numeric456"; 
    m/([^\p{IsDigit}]+)/; 
    ok $1, 'alpha';
    $test++;				# 11

    $_ = ",123alpha,456numeric"; 
    m/([\p{IsAlnum}]+)/; 
    ok $1, '123alpha';
    $test++;				# 12
}
{
    use utf8;

    $_ = "\x{263A}>\x{263A}\x{263A}"; 

    ok length, 4;
    $test++;				# 13

    ok length((m/>(.)/)[0]), 1;
    $test++;				# 14

    ok length($&), 2;
    $test++;				# 15

    ok length($'), 1;
    $test++;				# 16

    ok length($`), 1;
    $test++;				# 17

    ok length($1), 1;
    $test++;				# 18

    ok length($tmp=$&), 2;
    $test++;				# 19

    ok length($tmp=$'), 1;
    $test++;				# 20

    ok length($tmp=$`), 1;
    $test++;				# 21

    ok length($tmp=$1), 1;
    $test++;				# 22

    {
	use bytes;

	my $tmp = $&;
	ok $tmp, pack("C*", ord(">"), 0342, 0230, 0272);
	$test++;				# 23

	$tmp = $';
	ok $tmp, pack("C*", 0342, 0230, 0272);
	$test++;				# 24

	$tmp = $`;
	ok $tmp, pack("C*", 0342, 0230, 0272);
	$test++;				# 25

	$tmp = $1;
	ok $tmp, pack("C*", 0342, 0230, 0272);
	$test++;				# 26
    }

    ok_bytes $&, pack("C*", ord(">"), 0342, 0230, 0272);
    $test++;				# 27

    ok_bytes $', pack("C*", 0342, 0230, 0272);
    $test++;				# 28

    ok_bytes $`, pack("C*", 0342, 0230, 0272);
    $test++;				# 29

    ok_bytes $1, pack("C*", 0342, 0230, 0272);
    $test++;				# 30

    {
	use bytes;
	no utf8;

	ok length, 10;
	$test++;				# 31

    	ok length((m/>(.)/)[0]), 1;
    	$test++;				# 32

    	ok length($&), 2;
    	$test++;				# 33

    	ok length($'), 5;
    	$test++;				# 34

    	ok length($`), 3;
    	$test++;				# 35

    	ok length($1), 1;
    	$test++;				# 36

	ok $&, pack("C*", ord(">"), 0342);
	$test++;				# 37

	ok $', pack("C*", 0230, 0272, 0342, 0230, 0272);
	$test++;				# 38

	ok $`, pack("C*", 0342, 0230, 0272);
	$test++;				# 39

	ok $1, pack("C*", 0342);
	$test++;				# 40

    }


    {
	no utf8;
	$_="\342\230\272>\342\230\272\342\230\272";
    }

    ok length, 10;
    $test++;				# 41

    ok length((m/>(.)/)[0]), 1;
    $test++;				# 42

    ok length($&), 2;
    $test++;				# 43

    ok length($'), 1;
    $test++;				# 44

    ok length($`), 1;
    $test++;				# 45

    ok length($1), 1;
    $test++;				# 46

    ok length($tmp=$&), 2;
    $test++;				# 47

    ok length($tmp=$'), 1;
    $test++;				# 48

    ok length($tmp=$`), 1;
    $test++;				# 49

    ok length($tmp=$1), 1;
    $test++;				# 50

    {
	use bytes;

        my $tmp = $&;
	ok $tmp, pack("C*", ord(">"), 0342, 0230, 0272);
	$test++;				# 51

        $tmp = $';
	ok $tmp, pack("C*", 0342, 0230, 0272);
	$test++;				# 52

        $tmp = $`;
	ok $tmp, pack("C*", 0342, 0230, 0272);
	$test++;				# 53

        $tmp = $1;
	ok $tmp, pack("C*", 0342, 0230, 0272);
	$test++;				# 54
    }
    {
	use bytes;
	no utf8;

	ok length, 10;
	$test++;				# 55

    	ok length((m/>(.)/)[0]), 1;
    	$test++;				# 56

    	ok length($&), 2;
    	$test++;				# 57

    	ok length($'), 5;
    	$test++;				# 58

    	ok length($`), 3;
    	$test++;				# 59

    	ok length($1), 1;
    	$test++;				# 60

	ok $&, pack("C*", ord(">"), 0342);
	$test++;				# 61

	ok $', pack("C*", 0230, 0272, 0342, 0230, 0272);
	$test++;				# 62

	ok $`, pack("C*", 0342, 0230, 0272);
	$test++;				# 63

	ok $1, pack("C*", 0342);
	$test++;				# 64

    }

    ok "\x{ab}" =~ /^\x{ab}$/, 1;
    $test++;					# 65
}
Commit	Line	Data
f96ec2a2	1	#!./perl
	2
	3	BEGIN {
	4	chdir 't' if -d 't';
	5	unshift @INC, '../lib';
	6	$ENV{PERL5LIB} = '../lib';
146174a9	7	if ( ord("\t") != 9 ) { # skip on ebcdic platforms
	8	print "1..0 # Skip utf8 tests on ebcdic platform.\n";
	9	exit;
	10	}
f96ec2a2	11	}
f96ec2a2	12
ee8c7f54	13	print "1..65\n";
f96ec2a2	14
	15	my $test = 1;
	16
	17	sub ok {
	18	my ($got,$expect) = @_;
	19	print "# expected [$expect], got [$got]\nnot " if $got ne $expect;
	20	print "ok $test\n";
	21	}
	22
ee8c7f54	23	sub ok_bytes {
	24	use bytes;
	25	my ($got,$expect) = @_;
	26	print "# expected [$expect], got [$got]\nnot " if $got ne $expect;
	27	print "ok $test\n";
	28	}
	29
	30
f96ec2a2	31	{
	32	use utf8;
	33	$_ = ">\x{263A}<";
	34	s/([\x{80}-\x{10ffff}])/"&#".ord($1).";"/eg;
	35	ok $_, '>☺<';
ee8c7f54	36	$test++; # 1
f96ec2a2	37
	38	$_ = ">\x{263A}<";
	39	my $rx = "\x{80}-\x{10ffff}";
	40	s/([$rx])/"&#".ord($1).";"/eg;
	41	ok $_, '>☺<';
ee8c7f54	42	$test++; # 2
f96ec2a2	43
	44	$_ = ">\x{263A}<";
	45	my $rx = "\\x{80}-\\x{10ffff}";
	46	s/([$rx])/"&#".ord($1).";"/eg;
	47	ok $_, '>☺<';
ee8c7f54	48	$test++; # 3
b8c5462f	49
	50	$_ = "alpha,numeric";
	51	m/([[:alpha:]]+)/;
	52	ok $1, 'alpha';
ee8c7f54	53	$test++; # 4
b8c5462f	54
	55	$_ = "alphaNUMERICstring";
	56	m/([[:^lower:]]+)/;
	57	ok $1, 'NUMERIC';
ee8c7f54	58	$test++; # 5
b8c5462f	59
	60	$_ = "alphaNUMERICstring";
	61	m/(\p{Ll}+)/;
	62	ok $1, 'alpha';
ee8c7f54	63	$test++; # 6
b8c5462f	64
	65	$_ = "alphaNUMERICstring";
	66	m/(\p{Lu}+)/;
	67	ok $1, 'NUMERIC';
ee8c7f54	68	$test++; # 7
b8c5462f	69
	70	$_ = "alpha,numeric";
	71	m/([\p{IsAlpha}]+)/;
	72	ok $1, 'alpha';
ee8c7f54	73	$test++; # 8
b8c5462f	74
	75	$_ = "alphaNUMERICstring";
	76	m/([^\p{IsLower}]+)/;
	77	ok $1, 'NUMERIC';
ee8c7f54	78	$test++; # 9
b8c5462f	79
0f4b6630	80	$_ = "alpha123numeric456";
	81	m/([\p{IsDigit}]+)/;
	82	ok $1, '123';
ee8c7f54	83	$test++; # 10
b8c5462f	84
0f4b6630	85	$_ = "alpha123numeric456";
	86	m/([^\p{IsDigit}]+)/;
	87	ok $1, 'alpha';
ee8c7f54	88	$test++; # 11
b8c5462f	89
0f4b6630	90	$_ = ",123alpha,456numeric";
	91	m/([\p{IsAlnum}]+)/;
	92	ok $1, '123alpha';
ee8c7f54	93	$test++; # 12
	94	}
	95	{
	96	use utf8;
	97
	98	$_ = "\x{263A}>\x{263A}\x{263A}";
	99
	100	ok length, 4;
	101	$test++; # 13
	102
	103	ok length((m/>(.)/)[0]), 1;
	104	$test++; # 14
	105
	106	ok length($&), 2;
	107	$test++; # 15
	108
	109	ok length($'), 1;
	110	$test++; # 16
	111
	112	ok length($`), 1;
	113	$test++; # 17
	114
	115	ok length($1), 1;
	116	$test++; # 18
	117
	118	ok length($tmp=$&), 2;
	119	$test++; # 19
	120
	121	ok length($tmp=$'), 1;
	122	$test++; # 20
	123
	124	ok length($tmp=$`), 1;
	125	$test++; # 21
	126
	127	ok length($tmp=$1), 1;
	128	$test++; # 22
	129
	130	{
	131	use bytes;
	132
	133	my $tmp = $&;
	134	ok $tmp, pack("C*", ord(">"), 0342, 0230, 0272);
	135	$test++; # 23
	136
	137	$tmp = $';
	138	ok $tmp, pack("C*", 0342, 0230, 0272);
	139	$test++; # 24
	140
	141	$tmp = $`;
	142	ok $tmp, pack("C*", 0342, 0230, 0272);
	143	$test++; # 25
	144
	145	$tmp = $1;
	146	ok $tmp, pack("C*", 0342, 0230, 0272);
	147	$test++; # 26
	148	}
	149
	150	ok_bytes $&, pack("C*", ord(">"), 0342, 0230, 0272);
	151	$test++; # 27
	152
	153	ok_bytes $', pack("C*", 0342, 0230, 0272);
	154	$test++; # 28
	155
	156	ok_bytes $`, pack("C*", 0342, 0230, 0272);
157	$test++; # 29
158
159	ok_bytes $1, pack("C*", 0342, 0230, 0272);
160	$test++; # 30
161
162	{
163	use bytes;
164	no utf8;
165
166	ok length, 10;
167	$test++; # 31
168
169	ok length((m/>(.)/)[0]), 1;
170	$test++; # 32
171
172	ok length($&), 2;
173	$test++; # 33
174
175	ok length($'), 5;
176	$test++; # 34
177
178	ok length($`), 3;
179	$test++; # 35
180
181	ok length($1), 1;
182	$test++; # 36
183
184	ok $&, pack("C*", ord(">"), 0342);
185	$test++; # 37
186
187	ok $', pack("C*", 0230, 0272, 0342, 0230, 0272);
188	$test++; # 38
189
190	ok $`, pack("C*", 0342, 0230, 0272);
191	$test++; # 39
192
193	ok $1, pack("C*", 0342);
194	$test++; # 40
195
196	}
197
198
199	{
200	no utf8;
201	$_="\342\230\272>\342\230\272\342\230\272";
202	}
203
204	ok length, 10;
205	$test++; # 41
206
207	ok length((m/>(.)/)[0]), 1;
208	$test++; # 42
209
210	ok length($&), 2;
211	$test++; # 43
212
213	ok length($'), 1;
214	$test++; # 44
215
216	ok length($`), 1;
217	$test++; # 45
218
219	ok length($1), 1;
220	$test++; # 46
221
222	ok length($tmp=$&), 2;
223	$test++; # 47
224
225	ok length($tmp=$'), 1;
226	$test++; # 48
227
228	ok length($tmp=$`), 1;
229	$test++; # 49
230
231	ok length($tmp=$1), 1;
232	$test++; # 50
233
234	{
235	use bytes;
236
237	my $tmp = $&;
238	ok $tmp, pack("C*", ord(">"), 0342, 0230, 0272);
239	$test++; # 51
240
241	$tmp = $';
242	ok $tmp, pack("C*", 0342, 0230, 0272);
243	$test++; # 52
244
245	$tmp = $`;
246	ok $tmp, pack("C*", 0342, 0230, 0272);
247	$test++; # 53
248
249	$tmp = $1;
250	ok $tmp, pack("C*", 0342, 0230, 0272);
251	$test++; # 54
252	}
253	{
254	use bytes;
255	no utf8;
256
257	ok length, 10;
258	$test++; # 55
259
260	ok length((m/>(.)/)[0]), 1;
261	$test++; # 56
262
263	ok length($&), 2;
264	$test++; # 57
265
266	ok length($'), 5;
267	$test++; # 58
268
269	ok length($`), 3;
270	$test++; # 59
271
272	ok length($1), 1;
273	$test++; # 60
274
275	ok $&, pack("C*", ord(">"), 0342);
276	$test++; # 61
277
278	ok $', pack("C*", 0230, 0272, 0342, 0230, 0272);
279	$test++; # 62
280
281	ok $`, pack("C*", 0342, 0230, 0272);
282	$test++; # 63
283
284	ok $1, pack("C*", 0342);
285	$test++; # 64
286
287	}
288
289	ok "\x{ab}" =~ /^\x{ab}$/, 1;
290	$test++; # 65
0f4b6630	291	}