| 1 |
#!/usr/bin/perl |
| 2 |
use strict; |
| 3 |
use Test; |
| 4 |
BEGIN { plan tests => 6185 } |
| 5 |
|
| 6 |
require Whatpm::Charset::DecodeHandle; |
| 7 |
|
| 8 |
my $XML_AUTO_CHARSET = q<http://suika.fam.cx/www/2006/03/xml-entity/>; |
| 9 |
my $IANA_CHARSET = q<urn:x-suika-fam-cx:charset:>; |
| 10 |
my $PERL_CHARSET = q<http://suika.fam.cx/~wakaba/archive/2004/dis/Charset/Perl.>; |
| 11 |
my $XML_CHARSET = q<http://suika.fam.cx/~wakaba/archive/2004/dis/Charset/XML.>; |
| 12 |
|
| 13 |
## |create_decode_handle| |
| 14 |
for my $test ( |
| 15 |
['perl.utf8', $PERL_CHARSET.'utf8', 1], |
| 16 |
['xml', $XML_AUTO_CHARSET, 1], |
| 17 |
['unknown', q<http://www.unknown.test/>, 0], |
| 18 |
['iana.euc-jp', $IANA_CHARSET.'euc-jp', 1], |
| 19 |
['xml.euc-jp', $XML_CHARSET.'euc-jp', 1], |
| 20 |
['iana.shift_jis', $IANA_CHARSET.'shift_jis', 1], |
| 21 |
['xml.shift_jis', $XML_CHARSET.'shift_jis', 1], |
| 22 |
['iana.iso-2022-jp', $IANA_CHARSET.'iso-2022-jp', 1], |
| 23 |
['xml.iso-2022-jp', $XML_CHARSET.'iso-2022-jp', 1], |
| 24 |
) { |
| 25 |
open my $fh, '<', \''; |
| 26 |
my $dh = Whatpm::Charset::DecodeHandle->create_decode_handle ($test->[1], $fh); |
| 27 |
|
| 28 |
if ($test->[2]) { |
| 29 |
ok UNIVERSAL::isa ($dh, 'Whatpm::Charset::DecodeHandle::Encode') ? 1 : 0, 1, |
| 30 |
'create_decode_handle ' . $test->[0] . ' object'; |
| 31 |
ok ref $dh->onerror eq 'CODE' ? 1 : 0, 1, |
| 32 |
'create_decode_handle ' . $test->[0] . ' onerror'; |
| 33 |
} else { |
| 34 |
ok UNIVERSAL::isa ($dh, 'Whatpm::Charset::DecodeHandle::Encode') ? 1 : 0, 0, |
| 35 |
'create_decode_handle ' . $test->[0] . ' object'; |
| 36 |
|
| 37 |
Whatpm::Charset::DecodeHandle->create_decode_handle ($test->[1], $fh, sub { |
| 38 |
ok $_[1], 'charset-not-supported-error', |
| 39 |
'create_decode_handle ' . $test->[0] . ' error'; |
| 40 |
}); |
| 41 |
} |
| 42 |
} |
| 43 |
|
| 44 |
## |name_to_uri| |
| 45 |
for ( |
| 46 |
[$IANA_CHARSET.'utf-8', 'utf-8'], |
| 47 |
[$IANA_CHARSET.'x-no-such-charset', 'x-no-such-charset'], |
| 48 |
[$IANA_CHARSET.'utf-8', 'UTF-8'], |
| 49 |
[$IANA_CHARSET.'utf-8', 'uTf-8'], |
| 50 |
[$IANA_CHARSET.'utf-16be', 'utf-16be'], |
| 51 |
) { |
| 52 |
my $iname = Whatpm::Charset::DecodeHandle->name_to_uri (ietf => $_->[1]); |
| 53 |
ok $iname, $_->[0], 'ietf charset URI ' . $_->[1]; |
| 54 |
} |
| 55 |
|
| 56 |
for ( |
| 57 |
[$XML_CHARSET.'utf-8', 'utf-8'], |
| 58 |
[$XML_CHARSET.'x-no-such-charset', 'x-no-such-charset'], |
| 59 |
[$XML_CHARSET.'utf-8', 'UTF-8'], |
| 60 |
[$XML_CHARSET.'utf-8', 'uTf-8'], |
| 61 |
[$IANA_CHARSET.'utf-16be', 'utf-16be'], |
| 62 |
) { |
| 63 |
my $iname = Whatpm::Charset::DecodeHandle->name_to_uri (xml => $_->[1]); |
| 64 |
ok $iname, $_->[0], 'XML encoding URI ' . $_->[1]; |
| 65 |
} |
| 66 |
|
| 67 |
## |uri_to_name| |
| 68 |
for ( |
| 69 |
[$IANA_CHARSET.'utf-8', 'utf-8'], |
| 70 |
[$IANA_CHARSET.'x-no-such-charset', 'x-no-such-charset'], |
| 71 |
[q<http://charset.example/>, undef], |
| 72 |
) { |
| 73 |
my $uri = Whatpm::Charset::DecodeHandle->uri_to_name (ietf => $_->[0]); |
| 74 |
ok $uri, $_->[1], 'URI -> IETF charset ' . $_->[0]; |
| 75 |
} |
| 76 |
|
| 77 |
for ( |
| 78 |
[$XML_CHARSET.'utf-8', 'UTF-8'], |
| 79 |
[$XML_CHARSET.'x-no-such-charset', 'x-no-such-charset'], |
| 80 |
[q<http://charset.example/>, undef], |
| 81 |
) { |
| 82 |
my $uri = Whatpm::Charset::DecodeHandle->uri_to_name (xml => $_->[0]); |
| 83 |
ok $uri, $_->[1], 'URI -> XML encoding ' . $_->[0]; |
| 84 |
} |
| 85 |
|
| 86 |
## |getc| |
| 87 |
{ |
| 88 |
my $byte = "a\xE3\x81\x82\x81a"; |
| 89 |
open my $fh, '<', \$byte; |
| 90 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 91 |
($PERL_CHARSET.'utf8', $fh); |
| 92 |
|
| 93 |
my $error; |
| 94 |
$efh->onerror (sub { |
| 95 |
my ($efh, $type, %opt) = @_; |
| 96 |
$error = ${$opt{octets}}; |
| 97 |
}); |
| 98 |
|
| 99 |
ok $efh->getc, "a", "getc 1 [1]"; |
| 100 |
ok $error, undef, "getc 1 [1] error"; |
| 101 |
ok $efh->getc, "\x{3042}", "getc 1 [2]"; |
| 102 |
ok $error, undef, "getc 1 [2] error"; |
| 103 |
ok $efh->getc, "\x81", "getc 1 [3]"; |
| 104 |
ok $error, "\x81", "getc 1 [3] error"; |
| 105 |
undef $error; |
| 106 |
ok $efh->getc, "a", "getc 1 [4]"; |
| 107 |
ok $error, undef, "getc 1 [4] error"; |
| 108 |
ok $efh->getc, undef, "getc 1 [5]"; |
| 109 |
ok $error, undef, "getc 1 [5] error"; |
| 110 |
} |
| 111 |
|
| 112 |
{ |
| 113 |
my $byte = "a" x 256; |
| 114 |
$byte .= "b" x 256; |
| 115 |
|
| 116 |
open my $fh, '<', \$byte; |
| 117 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 118 |
($PERL_CHARSET.'utf8', $fh); |
| 119 |
|
| 120 |
my $error; |
| 121 |
$efh->onerror (sub { |
| 122 |
my ($efh, $type, %opt) = @_; |
| 123 |
$error = ${$opt{octets}}; |
| 124 |
}); |
| 125 |
|
| 126 |
for my $i (0..255) { |
| 127 |
ok $efh->getc, "a", "getc 2 [$i]"; |
| 128 |
ok $error, undef, "getc 2 [$i] error"; |
| 129 |
} |
| 130 |
|
| 131 |
for my $i (0..255) { |
| 132 |
ok $efh->getc, "b", "getc 2 [255+$i]"; |
| 133 |
ok $error, undef, "getc 2 [255+$i] error"; |
| 134 |
} |
| 135 |
|
| 136 |
ok $efh->getc, undef, "getc 2 [-1]"; |
| 137 |
ok $error, undef, "getc 2 [-1] error"; |
| 138 |
} |
| 139 |
|
| 140 |
{ |
| 141 |
my $byte = "a" x 255; |
| 142 |
$byte .= "\xE3\x81\x82"; |
| 143 |
$byte .= "b" x 256; |
| 144 |
|
| 145 |
open my $fh, '<', \$byte; |
| 146 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 147 |
($PERL_CHARSET.'utf8', $fh); |
| 148 |
|
| 149 |
my $error; |
| 150 |
$efh->onerror (sub { |
| 151 |
my ($efh, $type, %opt) = @_; |
| 152 |
$error = ${$opt{octets}}; |
| 153 |
}); |
| 154 |
|
| 155 |
for my $i (0..254) { |
| 156 |
ok $efh->getc, "a", "getc 3 [$i]"; |
| 157 |
ok $error, undef, "getc 3 [$i] error"; |
| 158 |
} |
| 159 |
|
| 160 |
ok $efh->getc, "\x{3042}", "getc 3 [255]"; |
| 161 |
ok $error, undef, "getc 3 [255] error"; |
| 162 |
|
| 163 |
for my $i (0..255) { |
| 164 |
ok $efh->getc, "b", "getc 3 [255+$i]"; |
| 165 |
ok $error, undef, "getc 3 [255+$i] error"; |
| 166 |
} |
| 167 |
|
| 168 |
ok $efh->getc, undef, "getc 3 [-1]"; |
| 169 |
ok $error, undef, "getc 3 [-1] error"; |
| 170 |
} |
| 171 |
|
| 172 |
{ |
| 173 |
my $byte = "a" x 255; |
| 174 |
$byte .= "\xE3"; |
| 175 |
|
| 176 |
open my $fh, '<', \$byte; |
| 177 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 178 |
($PERL_CHARSET.'utf8', $fh); |
| 179 |
|
| 180 |
my $error; |
| 181 |
$efh->onerror (sub { |
| 182 |
my ($efh, $type, %opt) = @_; |
| 183 |
$error = ${$opt{octets}}; |
| 184 |
}); |
| 185 |
|
| 186 |
for my $i (0..254) { |
| 187 |
ok $efh->getc, "a", "getc 4 [$i]"; |
| 188 |
ok $error, undef, "getc 4 [$i] error"; |
| 189 |
} |
| 190 |
|
| 191 |
ok $efh->getc, "\xE3", "getc 4 [255]"; |
| 192 |
ok $error, "\xE3", "getc 4 [255] error"; |
| 193 |
undef $error; |
| 194 |
|
| 195 |
ok $efh->getc, undef, "getc 4 [-1]"; |
| 196 |
ok $error, undef, "getc 4 [-1] error"; |
| 197 |
} |
| 198 |
|
| 199 |
## |ungetc| |
| 200 |
{ |
| 201 |
my $byte = "a\x{4E00}b\x{4E11}"; |
| 202 |
|
| 203 |
open my $fh, '<', \$byte; |
| 204 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 205 |
($PERL_CHARSET.'utf8', $fh); |
| 206 |
|
| 207 |
ok $efh->getc, "a", "ungetc [1]"; |
| 208 |
|
| 209 |
$efh->ungetc (ord "a"); |
| 210 |
ok $efh->getc, "a", "ungetc [2]"; |
| 211 |
|
| 212 |
ok $efh->getc, "\x{4E00}", "ungetc [3]"; |
| 213 |
|
| 214 |
$efh->ungetc (ord "\x{4E00}"); |
| 215 |
ok $efh->getc, "\x{4E00}", "ungetc [4]"; |
| 216 |
|
| 217 |
ok $efh->getc, "b", "ungetc [5]"; |
| 218 |
|
| 219 |
ok $efh->getc, "\x{4E11}", "ungetc [6]"; |
| 220 |
|
| 221 |
$efh->ungetc (ord "\x{4E11}"); |
| 222 |
ok $efh->getc, "\x{4E11}", "ungetc [7]"; |
| 223 |
} |
| 224 |
|
| 225 |
## UTF-8, UTF-16 and BOM |
| 226 |
for my $test ( |
| 227 |
["UTF-8 BOM 1", qq<\xEF\xBB\xBFabc>, $XML_CHARSET.'utf-8', |
| 228 |
["a", "b", "c", undef], 1], |
| 229 |
["UTF-8 no BOM 1", qq<abc>, $XML_CHARSET.'utf-8', |
| 230 |
["a", "b", "c", undef], 0], |
| 231 |
["UTF-8 BOM 2", qq<\xEF\xBB\xBF\xEF\xBB\xBFabc>, $XML_CHARSET.'utf-8', |
| 232 |
["\x{FEFF}", "a", "b", "c", undef], 1], |
| 233 |
["UTF-8 BOM 3", qq<\xEF\xBB\xBF>, $XML_CHARSET.'utf-8', |
| 234 |
[undef], 1], |
| 235 |
["UTF-8 no BOM 2", qq<>, $XML_CHARSET.'utf-8', |
| 236 |
[undef], 0], |
| 237 |
["UTF-8 no BOM 3", qq<ab>, $XML_CHARSET.'utf-8', |
| 238 |
[qw/a b/, undef], 0], |
| 239 |
["UTF-8 no BOM 4", qq<a>, $XML_CHARSET.'utf-8', |
| 240 |
[qw/a/, undef], 0], |
| 241 |
["UTF-16BE BOM 1", qq<\xFE\xFF\x4E\x00\x00a>, $XML_CHARSET.'utf-16', |
| 242 |
["\x{4E00}", "a", undef], 1], |
| 243 |
["UTF-16LE BOM 1", qq<\xFF\xFE\x00\x4Ea\x00>, $XML_CHARSET.'utf-16', |
| 244 |
["\x{4E00}", "a", undef], 1], |
| 245 |
["UTF-16BE BOM 2", qq<\xFE\xFF\x00a>, $XML_CHARSET.'utf-16', |
| 246 |
["a", undef], 1], |
| 247 |
["UTF-16LE BOM 2", qq<\xFF\xFEa\x00>, $XML_CHARSET.'utf-16', |
| 248 |
["a", undef], 1], |
| 249 |
["UTF-16BE BOM 3", qq<\xFE\xFF>, $XML_CHARSET.'utf-16', |
| 250 |
[undef], 1], |
| 251 |
["UTF-16LE BOM 3", qq<\xFF\xFE>, $XML_CHARSET.'utf-16', |
| 252 |
[undef], 1], |
| 253 |
) { |
| 254 |
my $error; |
| 255 |
|
| 256 |
open my $fh, '<', \($test->[1]); |
| 257 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 258 |
($test->[2], $fh, sub { $error = 1 }); |
| 259 |
|
| 260 |
for my $i (0..$#{$test->[3]}) { |
| 261 |
ok $efh->getc, $test->[3]->[$i], $test->[0] . " $i"; |
| 262 |
} |
| 263 |
ok $error, undef, $test->[0] . " error"; |
| 264 |
ok $efh->has_bom ? 1 : 0, $test->[4], $test->[0] . " has_bom"; |
| 265 |
} |
| 266 |
|
| 267 |
{ |
| 268 |
my $byte = qq<\xFE\xFFa>; |
| 269 |
|
| 270 |
my $error; |
| 271 |
|
| 272 |
open my $fh, '<', \$byte; |
| 273 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 274 |
($XML_CHARSET.'utf-16', $fh, sub { $error = $_[1] }); |
| 275 |
|
| 276 |
ok $error, undef, "UTF-16 [1]"; |
| 277 |
ok $efh->getc, "a", "UTF-16 [2]"; |
| 278 |
ok $error, 'illegal-octets-error', "UTF-16 [3]"; |
| 279 |
undef $error; |
| 280 |
ok $efh->getc, undef, "UTF-16 [4]"; |
| 281 |
ok $error, undef, "UTF-16 [5]"; |
| 282 |
ok $efh->has_bom ? 1 : 0, 1, "UTF-16 [6]"; |
| 283 |
} |
| 284 |
{ |
| 285 |
my $byte = qq<\xFF\xFEa>; |
| 286 |
|
| 287 |
my $error; |
| 288 |
|
| 289 |
open my $fh, '<', \$byte; |
| 290 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 291 |
($XML_CHARSET.'utf-16', $fh, sub { $error = $_[1] }); |
| 292 |
|
| 293 |
ok $error, undef, "UTF-16 [7]"; |
| 294 |
ok $efh->getc, "a", "UTF-16 [8]"; |
| 295 |
ok $error, 'illegal-octets-error', "UTF-16 [9]"; |
| 296 |
undef $error; |
| 297 |
ok $efh->getc, undef, "UTF-16 [10]"; |
| 298 |
ok $error, undef, "UTF-16 [11]"; |
| 299 |
ok $efh->has_bom ? 1 : 0, 1, "UTF-16 [12]"; |
| 300 |
} |
| 301 |
|
| 302 |
{ |
| 303 |
my $byte = qq<\xFD\xFF>; |
| 304 |
|
| 305 |
my $error; |
| 306 |
|
| 307 |
open my $fh, '<', \$byte; |
| 308 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 309 |
($XML_CHARSET.'utf-16', $fh, sub { $error = $_[1] }); |
| 310 |
|
| 311 |
ok $error, 'no-bom-error', "UTF-16 [13]"; |
| 312 |
undef $error; |
| 313 |
|
| 314 |
ok $efh->getc, "\x{FDFF}", "UTF-16 [14]"; |
| 315 |
ok $error, undef, "UTF-16 [15]"; |
| 316 |
ok $efh->getc, undef, "UTF-16 [16]"; |
| 317 |
ok $error, undef, "UTF-16 [17]"; |
| 318 |
ok $efh->has_bom ? 1 : 0, 0, "UTF-16 [18]"; |
| 319 |
} |
| 320 |
|
| 321 |
{ |
| 322 |
my $byte = qq<\xFD>; |
| 323 |
|
| 324 |
my $error; |
| 325 |
|
| 326 |
open my $fh, '<', \$byte; |
| 327 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 328 |
($XML_CHARSET.'utf-16', $fh, sub { $error = $_[1] }); |
| 329 |
|
| 330 |
ok $error, 'no-bom-error', "UTF-16 [19]"; |
| 331 |
undef $error; |
| 332 |
|
| 333 |
ok $efh->getc, "\xFD", "UTF-16 [20]"; |
| 334 |
ok $error, 'illegal-octets-error', "UTF-16 [21]"; |
| 335 |
undef $error; |
| 336 |
|
| 337 |
ok $efh->getc, undef, "UTF-16 [22]"; |
| 338 |
ok $error, undef, "UTF-16 [23]"; |
| 339 |
ok $efh->has_bom ? 1 : 0, 0, "UTF-16 [24]"; |
| 340 |
} |
| 341 |
|
| 342 |
{ |
| 343 |
my $byte = qq<>; |
| 344 |
|
| 345 |
my $error; |
| 346 |
|
| 347 |
open my $fh, '<', \$byte; |
| 348 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 349 |
($XML_CHARSET.'utf-16', $fh, sub { $error = $_[1] }); |
| 350 |
|
| 351 |
ok $error, 'no-bom-error', "UTF-16 [25]"; |
| 352 |
undef $error; |
| 353 |
|
| 354 |
ok $efh->getc, undef, "UTF-16 [26]"; |
| 355 |
ok $error, undef, "UTF-16 [27]"; |
| 356 |
ok $efh->has_bom ? 1 : 0, 0, "UTF-16 [28]"; |
| 357 |
} |
| 358 |
|
| 359 |
sub check_charset ($$$) { |
| 360 |
my $test_name = $_[0]; |
| 361 |
my $charset_uri = $_[1]; |
| 362 |
for my $testdata (@{$_[2]}) { |
| 363 |
my $byte = $testdata->{in}; |
| 364 |
my $error; |
| 365 |
my $i = 0; |
| 366 |
|
| 367 |
open my $fh, '<', \$byte; |
| 368 |
my $efh = Whatpm::Charset::DecodeHandle->create_decode_handle |
| 369 |
($charset_uri, $fh, sub { |
| 370 |
my (undef, $etype, %opt) = @_; |
| 371 |
$error = [$etype, \%opt]; |
| 372 |
}); |
| 373 |
|
| 374 |
ok defined $efh ? 1 : 0, 1, "$test_name $testdata->{id} return"; |
| 375 |
next unless defined $efh; |
| 376 |
ok $efh->has_bom ? 1 : 0, $testdata->{bom} || 0, |
| 377 |
"$test_name $testdata->{id} BOM"; |
| 378 |
ok $efh->input_encoding, $testdata->{name}, "$test_name $testdata->{id} ie"; |
| 379 |
|
| 380 |
while (@{$testdata->{out}}) { |
| 381 |
if ($i != 0) { |
| 382 |
my $c = shift @{$testdata->{out}}; |
| 383 |
ok $efh->getc, $c, "$test_name $testdata->{id} $i"; |
| 384 |
} |
| 385 |
|
| 386 |
my $v = shift @{$testdata->{out}}; |
| 387 |
if (defined $v) { |
| 388 |
ok defined $error ? 1 : 0, 1, "$test_name $testdata->{id} $i error"; |
| 389 |
ok $error->[0], $v->[0], "$test_name $testdata->{id} $i error 0"; |
| 390 |
} else { |
| 391 |
ok defined $error ? 1 : 0, 0, "$test_name $testdata->{id} $i error"; |
| 392 |
} |
| 393 |
undef $error; |
| 394 |
$i++; |
| 395 |
} |
| 396 |
|
| 397 |
ok $efh->getc, undef, "$test_name $testdata->{id} EOF"; |
| 398 |
if ($testdata->{eof_error}) { |
| 399 |
ok defined $error ? 1 : 0, 1, "$test_name $testdata->{id} EOF error"; |
| 400 |
ok $error->[0], $testdata->{eof_error}->[0], |
| 401 |
"$test_name $testdata->{id} EOF error 0"; |
| 402 |
} else { |
| 403 |
ok $error, undef, "$test_name $testdata->{id} EOF error"; |
| 404 |
} |
| 405 |
} # testdata |
| 406 |
} # check_charset |
| 407 |
|
| 408 |
## XML Character Encoding Autodetection |
| 409 |
{ |
| 410 |
my @testdata = ( |
| 411 |
{ |
| 412 |
id => q<l=0>, |
| 413 |
in => q<>, |
| 414 |
out => [undef], |
| 415 |
name => 'UTF-8', bom => 0, |
| 416 |
}, |
| 417 |
{ |
| 418 |
id => q<l=1>, |
| 419 |
in => "a", |
| 420 |
out => [undef, "a", undef], |
| 421 |
name => 'UTF-8', bom => 0, |
| 422 |
}, |
| 423 |
{ |
| 424 |
id => q<bom8.l=0>, |
| 425 |
in => "\xEF\xBB\xBF", |
| 426 |
out => [undef], |
| 427 |
name => 'UTF-8', bom => 1, |
| 428 |
}, |
| 429 |
{ |
| 430 |
id => q<bom8.l=1>, |
| 431 |
in => "\xEF\xBB\xBFa", |
| 432 |
out => [undef, "a", undef], |
| 433 |
name => 'UTF-8', bom => 1, |
| 434 |
}, |
| 435 |
{ |
| 436 |
id => q<bom8.zwnbsp>, |
| 437 |
in => "\xEF\xBB\xBF\xEF\xBB\xBF", |
| 438 |
out => [undef, "\x{FEFF}", undef], |
| 439 |
name => 'UTF-8', bom => 1, |
| 440 |
}, |
| 441 |
{ |
| 442 |
id => q<bom16be.l=0>, |
| 443 |
in => "\xFE\xFF", |
| 444 |
out => [undef], |
| 445 |
name => 'UTF-16', bom => 1, |
| 446 |
}, |
| 447 |
{ |
| 448 |
id => q<bom16le.l=0>, |
| 449 |
in => "\xFF\xFE", |
| 450 |
out => [undef], |
| 451 |
name => 'UTF-16', bom => 1, |
| 452 |
}, |
| 453 |
{ |
| 454 |
id => q<bom16be.l=1>, |
| 455 |
in => "\xFE\xFFa", |
| 456 |
out => [undef, "a", [q<illegal-octets-error>]], |
| 457 |
name => 'UTF-16', bom => 1, |
| 458 |
}, |
| 459 |
{ |
| 460 |
id => q<bom16le.l=1>, |
| 461 |
in => "\xFF\xFEa", |
| 462 |
out => [undef, "a", [q<illegal-octets-error>]], |
| 463 |
name => 'UTF-16', bom => 1, |
| 464 |
}, |
| 465 |
{ |
| 466 |
id => q<bom16be.l=2>, |
| 467 |
in => "\xFE\xFF\x4E\x00", |
| 468 |
out => [undef, "\x{4E00}", undef], |
| 469 |
name => 'UTF-16', bom => 1, |
| 470 |
}, |
| 471 |
{ |
| 472 |
id => q<bom16le.l=2>, |
| 473 |
in => "\xFF\xFE\x00\x4E", |
| 474 |
out => [undef, "\x{4E00}", undef], |
| 475 |
name => 'UTF-16', bom => 1, |
| 476 |
}, |
| 477 |
{ |
| 478 |
id => q<bom16be.l=2lt>, |
| 479 |
in => "\xFE\xFF\x00<", |
| 480 |
out => [undef, "<", undef], |
| 481 |
name => 'UTF-16', bom => 1, |
| 482 |
}, |
| 483 |
{ |
| 484 |
id => q<bom16le.l=2lt>, |
| 485 |
in => "\xFF\xFE<\x00", |
| 486 |
out => [undef, "<", undef], |
| 487 |
name => 'UTF-16', bom => 1, |
| 488 |
}, |
| 489 |
{ |
| 490 |
id => q<bom16be.zwnbsp>, |
| 491 |
in => "\xFE\xFF\xFE\xFF", |
| 492 |
out => [undef, "\x{FEFF}", undef], |
| 493 |
name => 'UTF-16', bom => 1, |
| 494 |
}, |
| 495 |
{ |
| 496 |
id => q<bom16le.zwnbsp>, |
| 497 |
in => "\xFF\xFE\xFF\xFE", |
| 498 |
out => [undef, "\x{FEFF}", undef], |
| 499 |
name => 'UTF-16', bom => 1, |
| 500 |
}, |
| 501 |
{ |
| 502 |
id => q<bom32e3412.l=0>, |
| 503 |
in => "\xFE\xFF\x00\x00", |
| 504 |
out => [undef, "\x00", undef], |
| 505 |
name => 'UTF-16', bom => 1, |
| 506 |
}, |
| 507 |
{ |
| 508 |
id => q<bom32e4321.l=0>, |
| 509 |
in => "\xFF\xFE\x00\x00", |
| 510 |
out => [undef, "\x00", undef], |
| 511 |
name => 'UTF-16', bom => 1, |
| 512 |
}, |
| 513 |
{ |
| 514 |
id => q<bom16be.l=4ltq>, |
| 515 |
in => "\xFE\xFF\x00<\x00?", |
| 516 |
out => [undef, "<", undef, "?", undef], |
| 517 |
name => 'UTF-16', bom => 1, |
| 518 |
}, |
| 519 |
{ |
| 520 |
id => q<bom16le.l=4ltq>, |
| 521 |
in => "\xFF\xFE<\x00?\x00", |
| 522 |
out => [undef, "<", undef, "?", undef], |
| 523 |
name => 'UTF-16', bom => 1, |
| 524 |
}, |
| 525 |
{ |
| 526 |
id => q<bom16be.decl.1>, |
| 527 |
in => qq[\xFE\xFF\x00<\x00?\x00x\x00m\x00l\x00 \x00v\x00e\x00r]. |
| 528 |
qq[\x00s\x00i\x00o\x00n\x00=\x00"\x001\x00.\x000\x00"]. |
| 529 |
qq[\x00 \x00e\x00n\x00c\x00o\x00d\x00i\x00n\x00g\x00=]. |
| 530 |
qq[\x00"\x00u\x00t\x00f\x00-\x001\x006\x00"\x00?\x00>], |
| 531 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 532 |
" ", undef, "v", undef, "e", undef, "r", undef, "s", undef, |
| 533 |
"i", undef, "o", undef, "n", undef, "=", undef, '"', undef, |
| 534 |
"1", undef, ".", undef, "0", undef, '"', undef, " ", undef, |
| 535 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 536 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 537 |
"u", undef, "t", undef, "f", undef, "-", undef, "1", undef, |
| 538 |
"6", undef, '"', undef, "?", undef, ">", undef], |
| 539 |
name => 'utf-16', bom => 1, |
| 540 |
}, |
| 541 |
{ |
| 542 |
id => q<bom16le.decl.1>, |
| 543 |
in => qq[\xFF\xFE<\x00?\x00x\x00m\x00l\x00 \x00v\x00e\x00r]. |
| 544 |
qq[\x00s\x00i\x00o\x00n\x00=\x00"\x001\x00.\x000\x00"]. |
| 545 |
qq[\x00 \x00e\x00n\x00c\x00o\x00d\x00i\x00n\x00g\x00=]. |
| 546 |
qq[\x00"\x00u\x00t\x00f\x00-\x001\x006\x00"\x00?\x00>\x00], |
| 547 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 548 |
" ", undef, "v", undef, "e", undef, "r", undef, "s", undef, |
| 549 |
"i", undef, "o", undef, "n", undef, "=", undef, '"', undef, |
| 550 |
"1", undef, ".", undef, "0", undef, '"', undef, " ", undef, |
| 551 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 552 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 553 |
"u", undef, "t", undef, "f", undef, "-", undef, "1", undef, |
| 554 |
"6", undef, '"', undef, "?", undef, ">", undef], |
| 555 |
name => 'utf-16', bom => 1, |
| 556 |
}, |
| 557 |
{ |
| 558 |
id => q<utf16be.decl.1>, |
| 559 |
in => qq[\x00<\x00?\x00x\x00m\x00l\x00 \x00v\x00e\x00r]. |
| 560 |
qq[\x00s\x00i\x00o\x00n\x00=\x00"\x001\x00.\x000\x00"]. |
| 561 |
qq[\x00 \x00e\x00n\x00c\x00o\x00d\x00i\x00n\x00g\x00=]. |
| 562 |
qq[\x00"\x00u\x00t\x00f\x00-\x001\x006\x00b\x00e\x00"\x00?\x00>], |
| 563 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 564 |
" ", undef, "v", undef, "e", undef, "r", undef, "s", undef, |
| 565 |
"i", undef, "o", undef, "n", undef, "=", undef, '"', undef, |
| 566 |
"1", undef, ".", undef, "0", undef, '"', undef, " ", undef, |
| 567 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 568 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 569 |
"u", undef, "t", undef, "f", undef, "-", undef, "1", undef, |
| 570 |
"6", undef, "b", undef, "e", undef, '"', undef, |
| 571 |
"?", undef, ">", undef], |
| 572 |
name => 'utf-16be', bom => 0, |
| 573 |
}, |
| 574 |
{ |
| 575 |
id => q<utf16le.decl.1>, |
| 576 |
in => qq[<\x00?\x00x\x00m\x00l\x00 \x00v\x00e\x00r]. |
| 577 |
qq[\x00s\x00i\x00o\x00n\x00=\x00"\x001\x00.\x000\x00"]. |
| 578 |
qq[\x00 \x00e\x00n\x00c\x00o\x00d\x00i\x00n\x00g\x00=]. |
| 579 |
qq[\x00"\x00u\x00t\x00f\x00-\x001\x006\x00l\x00e\x00"]. |
| 580 |
qq[\x00?\x00>\x00], |
| 581 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 582 |
" ", undef, "v", undef, "e", undef, "r", undef, "s", undef, |
| 583 |
"i", undef, "o", undef, "n", undef, "=", undef, '"', undef, |
| 584 |
"1", undef, ".", undef, "0", undef, '"', undef, " ", undef, |
| 585 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 586 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 587 |
"u", undef, "t", undef, "f", undef, "-", undef, "1", undef, |
| 588 |
"6", undef, "l", undef, "e", undef, '"', undef, "?", undef, |
| 589 |
">", undef], |
| 590 |
name => 'utf-16le', bom => 0, |
| 591 |
}, |
| 592 |
{ |
| 593 |
id => q<16be.decl.1>, |
| 594 |
in => qq[\x00<\x00?\x00x\x00m\x00l\x00 \x00v\x00e\x00r]. |
| 595 |
qq[\x00s\x00i\x00o\x00n\x00=\x00"\x001\x00.\x000\x00"]. |
| 596 |
qq[\x00 \x00e\x00n\x00c\x00o\x00d\x00i\x00n\x00g\x00=]. |
| 597 |
qq[\x00"\x00u\x00t\x00f\x00-\x001\x006\x00"\x00?\x00>], |
| 598 |
out => [[q<charset-name-mismatch-error>], |
| 599 |
"<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 600 |
" ", undef, "v", undef, "e", undef, "r", undef, "s", undef, |
| 601 |
"i", undef, "o", undef, "n", undef, "=", undef, '"', undef, |
| 602 |
"1", undef, ".", undef, "0", undef, '"', undef, " ", undef, |
| 603 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 604 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 605 |
"u", undef, "t", undef, "f", undef, "-", undef, "1", undef, |
| 606 |
"6", undef, '"', undef, "?", undef, ">", undef], |
| 607 |
name => 'utf-16', bom => 0, |
| 608 |
}, |
| 609 |
{ |
| 610 |
id => q<16le.decl.1>, |
| 611 |
in => qq[<\x00?\x00x\x00m\x00l\x00 \x00v\x00e\x00r]. |
| 612 |
qq[\x00s\x00i\x00o\x00n\x00=\x00"\x001\x00.\x000\x00"]. |
| 613 |
qq[\x00 \x00e\x00n\x00c\x00o\x00d\x00i\x00n\x00g\x00=]. |
| 614 |
qq[\x00"\x00u\x00t\x00f\x00-\x001\x006\x00"\x00?\x00>\x00], |
| 615 |
out => [[q<charset-name-mismatch-error>], |
| 616 |
"<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 617 |
" ", undef, "v", undef, "e", undef, "r", undef, "s", undef, |
| 618 |
"i", undef, "o", undef, "n", undef, "=", undef, '"', undef, |
| 619 |
"1", undef, ".", undef, "0", undef, '"', undef, " ", undef, |
| 620 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 621 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 622 |
"u", undef, "t", undef, "f", undef, "-", undef, "1", undef, |
| 623 |
"6", undef, '"', undef, "?", undef, ">", undef], |
| 624 |
name => 'utf-16', bom => 0, |
| 625 |
}, |
| 626 |
{ |
| 627 |
id => q<8.decl.1>, |
| 628 |
in => qq[<?xml version="1.0" encoding="utf-8"?>], |
| 629 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 630 |
" ", undef, "v", undef, "e", undef, "r", undef, "s", undef, |
| 631 |
"i", undef, "o", undef, "n", undef, "=", undef, '"', undef, |
| 632 |
"1", undef, ".", undef, "0", undef, '"', undef, " ", undef, |
| 633 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 634 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 635 |
"u", undef, "t", undef, "f", undef, "-", undef, "8", undef, |
| 636 |
'"', undef, "?", undef, ">", undef], |
| 637 |
name => 'utf-8', bom => 0, |
| 638 |
}, |
| 639 |
{ |
| 640 |
id => q<8.decl.2>, |
| 641 |
in => qq[<?xml encoding="utf-8"?>], |
| 642 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 643 |
" ", undef, |
| 644 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 645 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 646 |
"u", undef, "t", undef, "f", undef, "-", undef, "8", undef, |
| 647 |
'"', undef, "?", undef, ">", undef], |
| 648 |
name => 'utf-8', bom => 0, |
| 649 |
}, |
| 650 |
{ |
| 651 |
id => q<8.decl.3>, |
| 652 |
in => qq[<?xml version="1.1" encoding="utf-8"?>], |
| 653 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 654 |
" ", undef, "v", undef, "e", undef, "r", undef, "s", undef, |
| 655 |
"i", undef, "o", undef, "n", undef, "=", undef, '"', undef, |
| 656 |
"1", undef, ".", undef, "1", undef, '"', undef, " ", undef, |
| 657 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 658 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 659 |
"u", undef, "t", undef, "f", undef, "-", undef, "8", undef, |
| 660 |
'"', undef, "?", undef, ">", undef], |
| 661 |
name => 'utf-8', bom => 0, |
| 662 |
}, |
| 663 |
{ |
| 664 |
id => q<8.decl.4>, |
| 665 |
in => qq[<?xml version="1.0"?>], |
| 666 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 667 |
" ", undef, "v", undef, "e", undef, "r", undef, "s", undef, |
| 668 |
"i", undef, "o", undef, "n", undef, "=", undef, '"', undef, |
| 669 |
"1", undef, ".", undef, "0", undef, '"', undef, |
| 670 |
"?", undef, ">", undef], |
| 671 |
name => 'UTF-8', bom => 0, |
| 672 |
}, |
| 673 |
{ |
| 674 |
id => q<bom8.decl.1>, |
| 675 |
in => qq[\xEF\xBB\xBF<?xml encoding="utf-8"?>], |
| 676 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 677 |
" ", undef, |
| 678 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 679 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 680 |
"u", undef, "t", undef, "f", undef, "-", undef, "8", undef, |
| 681 |
'"', undef, "?", undef, ">", undef], |
| 682 |
name => 'utf-8', bom => 1, |
| 683 |
}, |
| 684 |
{ |
| 685 |
id => q<us-ascii.decl.1>, |
| 686 |
in => qq[<?xml encoding="us-ascii"?>], |
| 687 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 688 |
" ", undef, |
| 689 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 690 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 691 |
"u", undef, "s", undef, "-", undef, "a", undef, "s", undef, |
| 692 |
"c", undef, "i", undef, "i", undef, |
| 693 |
'"', undef, "?", undef, ">", undef], |
| 694 |
name => 'us-ascii', bom => 0, |
| 695 |
}, |
| 696 |
{ |
| 697 |
id => q<us-ascii.decl.2>, |
| 698 |
in => qq[<?xml encoding="US-ascii"?>], |
| 699 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 700 |
" ", undef, |
| 701 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 702 |
"i", undef, "n", undef, "g", undef, "=", undef, '"', undef, |
| 703 |
"U", undef, "S", undef, "-", undef, "a", undef, "s", undef, |
| 704 |
"c", undef, "i", undef, "i", undef, |
| 705 |
'"', undef, "?", undef, ">", undef], |
| 706 |
name => 'US-ascii', bom => 0, |
| 707 |
}, |
| 708 |
{ |
| 709 |
id => q<us-ascii.decl.3>, |
| 710 |
in => qq[<?xml encoding='us-ascii'?>], |
| 711 |
out => [undef, "<", undef, "?", undef, "x", undef, "m", undef, "l", undef, |
| 712 |
" ", undef, |
| 713 |
"e", undef, "n", undef, "c", undef, "o", undef, "d", undef, |
| 714 |
"i", undef, "n", undef, "g", undef, "=", undef, "'", undef, |
| 715 |
"u", undef, "s", undef, "-", undef, "a", undef, "s", undef, |
| 716 |
"c", undef, "i", undef, "i", undef, |
| 717 |
"'", undef, "?", undef, ">", undef], |
| 718 |
name => 'us-ascii', bom => 0, |
| 719 |
}, |
| 720 |
); |
| 721 |
check_charset ('XML', $XML_AUTO_CHARSET, \@testdata); |
| 722 |
} |
| 723 |
|
| 724 |
## EUC-JP |
| 725 |
{ |
| 726 |
my @testdata = ( |
| 727 |
{ |
| 728 |
id => q<l=0>, |
| 729 |
in => q<>, |
| 730 |
out => [undef], |
| 731 |
}, |
| 732 |
{ |
| 733 |
id => q<l=1.00>, |
| 734 |
in => qq<\x00>, |
| 735 |
out => [undef, "\x00", undef], |
| 736 |
}, |
| 737 |
{ |
| 738 |
id => q<l=1.0d>, |
| 739 |
in => qq<\x0D>, |
| 740 |
out => [undef, "\x0D", undef], |
| 741 |
}, |
| 742 |
{ |
| 743 |
id => q<l=1.0e>, |
| 744 |
in => qq<\x0E>, |
| 745 |
out => [undef, "\x0E", undef], |
| 746 |
}, # Error?? |
| 747 |
{ |
| 748 |
id => q<l=1.0f>, |
| 749 |
in => qq<\x0F>, |
| 750 |
out => [undef, "\x0F", undef], |
| 751 |
}, # Error?? |
| 752 |
{ |
| 753 |
id => q<l=1.1b>, |
| 754 |
in => qq<\x1B>, |
| 755 |
out => [undef, "\x1B", undef], |
| 756 |
}, # Error?? |
| 757 |
{ |
| 758 |
id => q<l=1.a>, |
| 759 |
in => q<a>, |
| 760 |
out => [undef, "a", undef], |
| 761 |
}, |
| 762 |
{ |
| 763 |
id => q<l=1.20>, |
| 764 |
in => qq<\x20>, |
| 765 |
out => [undef, "\x20", undef], |
| 766 |
}, |
| 767 |
{ |
| 768 |
id => q<5C>, |
| 769 |
in => qq<\x5C>, |
| 770 |
out => [undef, "\x5C", undef], |
| 771 |
}, |
| 772 |
{ |
| 773 |
id => q<l=1.7E>, |
| 774 |
in => qq<\x7E>, |
| 775 |
out => [undef, "\x7E", undef], |
| 776 |
}, |
| 777 |
{ |
| 778 |
id => q<l=1.7F>, |
| 779 |
in => qq<\x7F>, |
| 780 |
out => [undef, "\x7F", undef], |
| 781 |
}, |
| 782 |
{ |
| 783 |
id => q<l=1.80>, |
| 784 |
in => qq<\x80>, |
| 785 |
out => [undef, "\x80", undef], |
| 786 |
}, |
| 787 |
{ |
| 788 |
id => q<l=1.8c>, |
| 789 |
in => qq<\x8C>, |
| 790 |
out => [undef, "\x8C", undef], |
| 791 |
}, |
| 792 |
{ |
| 793 |
id => q<l=1.8e>, |
| 794 |
in => qq<\x8E>, |
| 795 |
out => [undef, "\x8E", [q<illegal-octets-error>]], |
| 796 |
}, |
| 797 |
{ |
| 798 |
id => q<l=1.8f>, |
| 799 |
in => qq<\x8F>, |
| 800 |
out => [undef, "\x8F", [q<illegal-octets-error>]], |
| 801 |
}, |
| 802 |
{ |
| 803 |
id => q<l=1.a0>, |
| 804 |
in => qq<\xA0>, |
| 805 |
out => [undef, "\xA0", [q<unassigned-code-point-error>]], |
| 806 |
}, |
| 807 |
{ |
| 808 |
id => q<l=1.a1>, |
| 809 |
in => qq<\xA1>, |
| 810 |
out => [undef, "\xA1", [q<illegal-octets-error>]], |
| 811 |
}, |
| 812 |
{ |
| 813 |
id => q<l=1.a2>, |
| 814 |
in => qq<\xA2>, |
| 815 |
out => [undef, "\xA2", [q<illegal-octets-error>]], |
| 816 |
}, |
| 817 |
{ |
| 818 |
id => q<l=1.fd>, |
| 819 |
in => qq<\xFD>, |
| 820 |
out => [undef, "\xFD", [q<illegal-octets-error>]], |
| 821 |
}, |
| 822 |
{ |
| 823 |
id => q<l=1.fe>, |
| 824 |
in => qq<\xFE>, |
| 825 |
out => [undef, "\xFE", [q<illegal-octets-error>]], |
| 826 |
}, |
| 827 |
{ |
| 828 |
id => q<l=1.ff>, |
| 829 |
in => qq<\xFF>, |
| 830 |
out => [undef, "\xFF", [q<unassigned-code-point-error>]], |
| 831 |
}, |
| 832 |
{ |
| 833 |
id => q<l=2.0000>, |
| 834 |
in => qq<\x00\x00>, |
| 835 |
out => [undef, "\x00", undef, "\x00", undef], |
| 836 |
}, |
| 837 |
{ |
| 838 |
id => q<l=2.0D0A>, |
| 839 |
in => qq<\x0D\x0A>, |
| 840 |
out => [undef, "\x0D", undef, "\x0A", undef], |
| 841 |
}, |
| 842 |
{ |
| 843 |
id => q<l=2.1B28>, |
| 844 |
in => qq<\x1B\x28>, |
| 845 |
out => [undef, "\x1B", undef, "\x28", undef], |
| 846 |
},# Error?? |
| 847 |
{ |
| 848 |
id => q<l=2.2020>, |
| 849 |
in => qq<\x20\x20>, |
| 850 |
out => [undef, "\x20", undef, "\x20", undef], |
| 851 |
}, |
| 852 |
{ |
| 853 |
id => q<l=2.ab>, |
| 854 |
in => qq<ab>, |
| 855 |
out => [undef, "a", undef, "b", undef], |
| 856 |
}, |
| 857 |
{ |
| 858 |
id => q<l=2.a0a1>, |
| 859 |
in => qq<\xA0\xA1>, |
| 860 |
out => [undef, "\xA0", [q<unassigned-code-point-error>], |
| 861 |
"\xA1", [q<illegal-octets-error>]], |
| 862 |
}, |
| 863 |
{ |
| 864 |
id => q<l=2.a1a1>, |
| 865 |
in => qq<\xA1\xA1>, |
| 866 |
out => [undef, "\x{3000}", undef], |
| 867 |
}, |
| 868 |
{ |
| 869 |
id => q<l=2.a1a2>, |
| 870 |
in => qq<\xA1\xA2>, |
| 871 |
out => [undef, "\x{3001}", undef], |
| 872 |
}, |
| 873 |
{ |
| 874 |
id => q<l=2.a1a4>, |
| 875 |
in => qq<\xA1\xA4>, |
| 876 |
out => [undef, "\x{FF0C}", undef], # FULLWIDTH COMMA |
| 877 |
}, |
| 878 |
{ |
| 879 |
id => q<a1a6>, |
| 880 |
in => qq<\xA1\xA6>, |
| 881 |
out => [undef, "\x{30FB}", undef], # KATAKABA MIDDLE DOT |
| 882 |
}, |
| 883 |
{ |
| 884 |
id => q<a1a7>, |
| 885 |
in => qq<\xA1\xA7>, |
| 886 |
out => [undef, "\x{FF1A}", undef], # FULLWIDTH COLON |
| 887 |
}, |
| 888 |
{ |
| 889 |
id => q<a1b1>, |
| 890 |
in => qq<\xA1\xB1>, |
| 891 |
out => [undef, "\x{203E}", undef], # OVERLINE |
| 892 |
}, |
| 893 |
{ |
| 894 |
id => q<a1bd>, |
| 895 |
in => qq<\xA1\xBD>, |
| 896 |
out => [undef, "\x{2014}", undef], # EM DASH |
| 897 |
}, |
| 898 |
{ |
| 899 |
id => q<a1c0>, |
| 900 |
in => qq<\xA1\xC0>, |
| 901 |
out => [undef, "\x{FF3C}", undef], # FULLWIDTH REVERSE SOLIDUS |
| 902 |
}, |
| 903 |
{ |
| 904 |
id => q<a1c1>, |
| 905 |
in => qq<\xA1\xC1>, |
| 906 |
out => [undef, "\x{301C}", undef], # WAVE DASH |
| 907 |
}, |
| 908 |
{ |
| 909 |
id => q<a1c2>, |
| 910 |
in => qq<\xA1\xC2>, |
| 911 |
out => [undef, "\x{2016}", undef], # DOUBLE VERTICAL LINE |
| 912 |
}, |
| 913 |
{ |
| 914 |
id => q<a1c4>, |
| 915 |
in => qq<\xA1\xC4>, |
| 916 |
out => [undef, "\x{2026}", undef], # HORIZONTAL ELLIPSIS |
| 917 |
}, |
| 918 |
{ |
| 919 |
id => q<a1dd>, |
| 920 |
in => qq<\xA1\xDD>, |
| 921 |
out => [undef, "\x{2212}", undef], # MINUS SIGN |
| 922 |
}, |
| 923 |
{ |
| 924 |
id => q<a1ef>, |
| 925 |
in => qq<\xA1\xEF>, |
| 926 |
out => [undef, "\x{00A5}", undef], # YEN SIGN |
| 927 |
}, |
| 928 |
{ |
| 929 |
id => q<a1f1>, |
| 930 |
in => qq<\xA1\xF1>, |
| 931 |
out => [undef, "\x{00A2}", undef], # CENT SIGN |
| 932 |
}, |
| 933 |
{ |
| 934 |
id => q<a1f2>, |
| 935 |
in => qq<\xA1\xF2>, |
| 936 |
out => [undef, "\x{00A3}", undef], # POUND SIGN |
| 937 |
}, |
| 938 |
{ |
| 939 |
id => q<a1f2>, |
| 940 |
in => qq<\xA1\xFF>, |
| 941 |
out => [undef, "\xA1", [q<illegal-octets-error>], |
| 942 |
"\xFF", [q<unassigned-code-point-error>]], |
| 943 |
}, |
| 944 |
{ |
| 945 |
id => q<a2ae>, |
| 946 |
in => qq<\xA2\xAE>, |
| 947 |
out => [undef, "\x{3013}", undef], # GETA MARK |
| 948 |
}, |
| 949 |
{ |
| 950 |
id => q<a2af>, |
| 951 |
in => qq<\xA2\xAF>, |
| 952 |
out => [undef, "\xA2\xAF", [q<unassigned-code-point-error>]], |
| 953 |
}, |
| 954 |
{ |
| 955 |
id => q<a2ba>, |
| 956 |
in => qq<\xA2\xBA>, |
| 957 |
out => [undef, "\x{2208}", undef], # ELEMENT OF |
| 958 |
}, |
| 959 |
{ |
| 960 |
id => q<a2fe>, |
| 961 |
in => qq<\xA2\xFE>, |
| 962 |
out => [undef, "\x{25EF}", undef], # LARGE CIRCLE |
| 963 |
}, |
| 964 |
{ |
| 965 |
id => q<adce>, |
| 966 |
in => qq<\xAD\xCE>, |
| 967 |
out => [undef, "\xAD\xCE", [q<unassigned-code-point-error>]], |
| 968 |
}, |
| 969 |
{ |
| 970 |
id => q<b0a6>, |
| 971 |
in => qq<\xB0\xA6>, |
| 972 |
out => [undef, "\x{611B}", undef], # han |
| 973 |
}, |
| 974 |
{ |
| 975 |
id => q<f4a6>, |
| 976 |
in => qq<\xF4\xA6>, |
| 977 |
out => [undef, "\x{7199}", undef], # han |
| 978 |
}, |
| 979 |
{ |
| 980 |
id => q<8ea1>, |
| 981 |
in => qq<\x8E\xA1>, |
| 982 |
out => [undef, "\x{FF61}", undef], |
| 983 |
}, |
| 984 |
{ |
| 985 |
id => q<8efe>, |
| 986 |
in => qq<\x8E\xFE>, |
| 987 |
out => [undef, "\x8E\xFE", [q<unassigned-code-point-error>]], |
| 988 |
}, |
| 989 |
{ |
| 990 |
id => q<8ffe>, |
| 991 |
in => qq<\x8F\xFE>, |
| 992 |
out => [undef, "\x8F\xFE", [q<illegal-octets-error>]], |
| 993 |
}, |
| 994 |
{ |
| 995 |
id => q<l=2.a1a2a3>, |
| 996 |
in => qq<\xA1\xA2\xA3>, |
| 997 |
out => [undef, "\x{3001}", undef, |
| 998 |
"\xA3", [q<illegal-octets-error>]], |
| 999 |
}, |
| 1000 |
{ |
| 1001 |
id => q<8ea1a1>, |
| 1002 |
in => qq<\x8E\xA1\xA1>, |
| 1003 |
out => [undef, "\x{FF61}", undef, |
| 1004 |
"\xA1", [q<illegal-octets-error>]], |
| 1005 |
}, |
| 1006 |
{ |
| 1007 |
id => q<8fa1a1>, |
| 1008 |
in => qq<\x8F\xA1\xA1>, |
| 1009 |
out => [undef, "\x8F\xA1\xA1", [q<unassigned-code-point-error>]], |
| 1010 |
}, |
| 1011 |
{ |
| 1012 |
id => q<8fa2af>, |
| 1013 |
in => qq<\x8F\xA2\xAF>, |
| 1014 |
out => [undef, "\x{02D8}", undef], |
| 1015 |
}, |
| 1016 |
{ |
| 1017 |
id => q<8fa2b7>, |
| 1018 |
in => qq<\x8F\xA2\xB7>, |
| 1019 |
out => [undef, "\x{FF5E}", undef], # FULLWIDTH TILDE |
| 1020 |
}, |
| 1021 |
{ |
| 1022 |
id => q<a1a2a1a3>, |
| 1023 |
in => qq<\xA1\xA2\xA1\xA3>, |
| 1024 |
out => [undef, "\x{3001}", undef, "\x{3002}", undef], |
| 1025 |
}, |
| 1026 |
{ |
| 1027 |
id => q<8fa2af>, |
| 1028 |
in => qq<\x8F\xA2\xAF\xAF>, |
| 1029 |
out => [undef, "\x{02D8}", undef, |
| 1030 |
"\xAF", [q<illegal-octets-error>]], |
| 1031 |
}, |
| 1032 |
{ |
| 1033 |
id => q<8fa2afafa1>, |
| 1034 |
in => qq<\x8F\xA2\xAF\xAF\xA1>, |
| 1035 |
out => [undef, "\x{02D8}", undef, |
| 1036 |
"\xAF\xA1", [q<unassigned-code-point-error>]], |
| 1037 |
}, |
| 1038 |
); |
| 1039 |
check_charset ('XML-EUC-JP', $XML_CHARSET.'euc-jp', |
| 1040 |
[map {$_->{name} = 'EUC-JP'; $_} @testdata]); |
| 1041 |
} |
| 1042 |
|
| 1043 |
## Shift_JIS |
| 1044 |
{ |
| 1045 |
my @testdata = ( |
| 1046 |
{ |
| 1047 |
id => q<l=0>, |
| 1048 |
in => q<>, |
| 1049 |
out => [undef], |
| 1050 |
}, |
| 1051 |
{ |
| 1052 |
id => q<l=1.00>, |
| 1053 |
in => qq<\x00>, |
| 1054 |
out => [undef, "\x00", undef], |
| 1055 |
}, |
| 1056 |
{ |
| 1057 |
id => q<l=1.0d>, |
| 1058 |
in => qq<\x0D>, |
| 1059 |
out => [undef, "\x0D", undef], |
| 1060 |
}, |
| 1061 |
{ |
| 1062 |
id => q<l=1.0e>, |
| 1063 |
in => qq<\x0E>, |
| 1064 |
out => [undef, "\x0E", undef], |
| 1065 |
}, # Error?? |
| 1066 |
{ |
| 1067 |
id => q<l=1.0f>, |
| 1068 |
in => qq<\x0F>, |
| 1069 |
out => [undef, "\x0F", undef], |
| 1070 |
}, # Error?? |
| 1071 |
{ |
| 1072 |
id => q<l=1.1b>, |
| 1073 |
in => qq<\x1B>, |
| 1074 |
out => [undef, "\x1B", undef], |
| 1075 |
}, # Error?? |
| 1076 |
{ |
| 1077 |
id => q<l=1.a>, |
| 1078 |
in => q<a>, |
| 1079 |
out => [undef, "a", undef], |
| 1080 |
}, |
| 1081 |
{ |
| 1082 |
id => q<l=1.20>, |
| 1083 |
in => qq<\x20>, |
| 1084 |
out => [undef, "\x20", undef], |
| 1085 |
}, |
| 1086 |
{ |
| 1087 |
id => q<l=1.5C>, |
| 1088 |
in => qq<\x5C>, |
| 1089 |
out => [undef, "\xA5", undef], # YEN SIGN |
| 1090 |
}, |
| 1091 |
{ |
| 1092 |
id => q<l=1.7E>, |
| 1093 |
in => qq<\x7E>, |
| 1094 |
out => [undef, "\x{203E}", undef], # OVERLINE |
| 1095 |
}, |
| 1096 |
{ |
| 1097 |
id => q<l=1.7F>, |
| 1098 |
in => qq<\x7F>, |
| 1099 |
out => [undef, "\x7F", undef], |
| 1100 |
}, |
| 1101 |
{ |
| 1102 |
id => q<l=1.80>, |
| 1103 |
in => qq<\x80>, |
| 1104 |
out => [undef, "\x80", [q<unassigned-code-point-error>]], |
| 1105 |
}, |
| 1106 |
{ |
| 1107 |
id => q<l=1.8c>, |
| 1108 |
in => qq<\x8C>, |
| 1109 |
out => [undef, "\x8C", [q<illegal-octets-error>]], |
| 1110 |
}, |
| 1111 |
{ |
| 1112 |
id => q<l=1.8e>, |
| 1113 |
in => qq<\x8E>, |
| 1114 |
out => [undef, "\x8E", [q<illegal-octets-error>]], |
| 1115 |
}, |
| 1116 |
{ |
| 1117 |
id => q<l=1.8f>, |
| 1118 |
in => qq<\x8F>, |
| 1119 |
out => [undef, "\x8F", [q<illegal-octets-error>]], |
| 1120 |
}, |
| 1121 |
{ |
| 1122 |
id => q<l=1.a0>, |
| 1123 |
in => qq<\xA0>, |
| 1124 |
out => [undef, "\xA0", [q<unassigned-code-point-error>]], |
| 1125 |
}, |
| 1126 |
{ |
| 1127 |
id => q<l=1.a1>, |
| 1128 |
in => qq<\xA1>, |
| 1129 |
out => [undef, "\x{FF61}", undef], |
| 1130 |
}, |
| 1131 |
{ |
| 1132 |
id => q<l=1.a2>, |
| 1133 |
in => qq<\xA2>, |
| 1134 |
out => [undef, "\x{FF62}", undef], |
| 1135 |
}, |
| 1136 |
{ |
| 1137 |
id => q<l=1.df>, |
| 1138 |
in => qq<\xdf>, |
| 1139 |
out => [undef, "\x{FF9F}", undef], |
| 1140 |
}, |
| 1141 |
{ |
| 1142 |
id => q<l=1.e0>, |
| 1143 |
in => qq<\xe0>, |
| 1144 |
out => [undef, "\xE0", [q<illegal-octets-error>]], |
| 1145 |
}, |
| 1146 |
{ |
| 1147 |
id => q<l=1.ef>, |
| 1148 |
in => qq<\xEF>, |
| 1149 |
out => [undef, "\xEF", [q<illegal-octets-error>]], |
| 1150 |
}, |
| 1151 |
{ |
| 1152 |
id => q<F0>, |
| 1153 |
in => qq<\xF0>, |
| 1154 |
out => [undef, "\xF0", [q<unassigned-code-point-error>]], |
| 1155 |
}, |
| 1156 |
{ |
| 1157 |
id => q<l=1.fc>, |
| 1158 |
in => qq<\xFC>, |
| 1159 |
out => [undef, "\xFC", [q<unassigned-code-point-error>]], |
| 1160 |
}, |
| 1161 |
{ |
| 1162 |
id => q<l=1.fd>, |
| 1163 |
in => qq<\xFD>, |
| 1164 |
out => [undef, "\xFD", [q<unassigned-code-point-error>]], |
| 1165 |
}, |
| 1166 |
{ |
| 1167 |
id => q<l=1.fe>, |
| 1168 |
in => qq<\xFE>, |
| 1169 |
out => [undef, "\xFE", [q<unassigned-code-point-error>]], |
| 1170 |
}, |
| 1171 |
{ |
| 1172 |
id => q<l=1.ff>, |
| 1173 |
in => qq<\xFF>, |
| 1174 |
out => [undef, "\xFF", [q<unassigned-code-point-error>]], |
| 1175 |
}, |
| 1176 |
{ |
| 1177 |
id => q<l=2.0000>, |
| 1178 |
in => qq<\x00\x00>, |
| 1179 |
out => [undef, "\x00", undef, "\x00", undef], |
| 1180 |
}, |
| 1181 |
{ |
| 1182 |
id => q<l=2.0D0A>, |
| 1183 |
in => qq<\x0D\x0A>, |
| 1184 |
out => [undef, "\x0D", undef, "\x0A", undef], |
| 1185 |
}, |
| 1186 |
{ |
| 1187 |
id => q<l=2.1B28>, |
| 1188 |
in => qq<\x1B\x28>, |
| 1189 |
out => [undef, "\x1B", undef, "\x28", undef], |
| 1190 |
},# Error?? |
| 1191 |
{ |
| 1192 |
id => q<l=2.2020>, |
| 1193 |
in => qq<\x20\x20>, |
| 1194 |
out => [undef, "\x20", undef, "\x20", undef], |
| 1195 |
}, |
| 1196 |
{ |
| 1197 |
id => q<l=2.ab>, |
| 1198 |
in => qq<ab>, |
| 1199 |
out => [undef, "a", undef, "b", undef], |
| 1200 |
}, |
| 1201 |
{ |
| 1202 |
id => q<8040>, |
| 1203 |
in => qq<\x80\x40>, |
| 1204 |
out => [undef, "\x80", [q<unassigned-code-point-error>], |
| 1205 |
"\x40", undef], |
| 1206 |
}, |
| 1207 |
{ |
| 1208 |
id => q<8100>, |
| 1209 |
in => qq<\x81\x00>, |
| 1210 |
out => [undef, "\x81\x00", [q<unassigned-code-point-error>]], |
| 1211 |
}, |
| 1212 |
{ |
| 1213 |
id => q<8101>, |
| 1214 |
in => qq<\x81\x01>, |
| 1215 |
out => [undef, "\x81\x01", [q<unassigned-code-point-error>]], |
| 1216 |
}, |
| 1217 |
{ |
| 1218 |
id => q<813F>, |
| 1219 |
in => qq<\x81\x3F>, |
| 1220 |
out => [undef, "\x81\x3F", [q<unassigned-code-point-error>]], |
| 1221 |
}, |
| 1222 |
{ |
| 1223 |
id => q<8140>, |
| 1224 |
in => qq<\x81\x40>, |
| 1225 |
out => [undef, "\x{3000}", undef], |
| 1226 |
}, |
| 1227 |
{ |
| 1228 |
id => q<8141>, |
| 1229 |
in => qq<\x81\x41>, |
| 1230 |
out => [undef, "\x{3001}", undef], |
| 1231 |
}, |
| 1232 |
{ |
| 1233 |
id => q<8143>, |
| 1234 |
in => qq<\x81\x43>, |
| 1235 |
out => [undef, "\x{FF0C}", undef], # FULLWIDTH COMMA |
| 1236 |
}, |
| 1237 |
{ |
| 1238 |
id => q<8150>, |
| 1239 |
in => qq<\x81\x50>, |
| 1240 |
out => [undef, "\x{FFE3}", undef], # FULLWIDTH MACRON |
| 1241 |
}, |
| 1242 |
{ |
| 1243 |
id => q<815C>, |
| 1244 |
in => qq<\x81\x5C>, |
| 1245 |
out => [undef, "\x{2014}", undef], # EM DASH |
| 1246 |
}, |
| 1247 |
{ |
| 1248 |
id => q<815F>, |
| 1249 |
in => qq<\x81\x5F>, |
| 1250 |
out => [undef, "\x{005C}", undef], # REVERSE SOLIDUS |
| 1251 |
}, |
| 1252 |
{ |
| 1253 |
id => q<8160>, |
| 1254 |
in => qq<\x81\x60>, |
| 1255 |
out => [undef, "\x{301C}", undef], # WAVE DASH |
| 1256 |
}, |
| 1257 |
{ |
| 1258 |
id => q<8161>, |
| 1259 |
in => qq<\x81\x61>, |
| 1260 |
out => [undef, "\x{2016}", undef], # DOUBLE VERTICAL LINE |
| 1261 |
}, |
| 1262 |
{ |
| 1263 |
id => q<8163>, |
| 1264 |
in => qq<\x81\x63>, |
| 1265 |
out => [undef, "\x{2026}", undef], # HORIZONTAL ELLIPSIS |
| 1266 |
}, |
| 1267 |
{ |
| 1268 |
id => q<817C>, |
| 1269 |
in => qq<\x81\x7C>, |
| 1270 |
out => [undef, "\x{2212}", undef], # MINUS SIGN |
| 1271 |
}, |
| 1272 |
{ |
| 1273 |
id => q<817F>, |
| 1274 |
in => qq<\x81\x7F>, |
| 1275 |
out => [undef, "\x81\x7F", [q<unassigned-code-point-error>]], |
| 1276 |
}, |
| 1277 |
{ |
| 1278 |
id => q<818F>, |
| 1279 |
in => qq<\x81\x8F>, |
| 1280 |
out => [undef, "\x{FFE5}", undef], # FULLWIDTH YEN SIGN |
| 1281 |
}, |
| 1282 |
{ |
| 1283 |
id => q<8191>, |
| 1284 |
in => qq<\x81\x91>, |
| 1285 |
out => [undef, "\x{00A2}", undef], # CENT SIGN |
| 1286 |
}, |
| 1287 |
{ |
| 1288 |
id => q<8192>, |
| 1289 |
in => qq<\x81\x92>, |
| 1290 |
out => [undef, "\x{00A3}", undef], # POUND SIGN |
| 1291 |
}, |
| 1292 |
{ |
| 1293 |
id => q<81AC>, |
| 1294 |
in => qq<\x81\xAC>, |
| 1295 |
out => [undef, "\x{3013}", undef], # GETA MARK |
| 1296 |
}, |
| 1297 |
{ |
| 1298 |
id => q<81AD>, |
| 1299 |
in => qq<\x81\xAD>, |
| 1300 |
out => [undef, "\x81\xAD", [q<unassigned-code-point-error>]], |
| 1301 |
}, |
| 1302 |
{ |
| 1303 |
id => q<81B8>, |
| 1304 |
in => qq<\x81\xB8>, |
| 1305 |
out => [undef, "\x{2208}", undef], # ELEMENT OF |
| 1306 |
}, |
| 1307 |
{ |
| 1308 |
id => q<81CA>, |
| 1309 |
in => qq<\x81\xCA>, |
| 1310 |
out => [undef, "\x{00AC}", undef], # NOT SIGN |
| 1311 |
}, |
| 1312 |
{ |
| 1313 |
id => q<81FC>, |
| 1314 |
in => qq<\x81\xFC>, |
| 1315 |
out => [undef, "\x{25EF}", undef], # LARGE CIRCLE |
| 1316 |
}, |
| 1317 |
{ |
| 1318 |
id => q<81FD>, |
| 1319 |
in => qq<\x81\xFD>, |
| 1320 |
out => [undef, "\x81\xFD", [q<unassigned-code-point-error>]], |
| 1321 |
}, |
| 1322 |
{ |
| 1323 |
id => q<81FE>, |
| 1324 |
in => qq<\x81\xFE>, |
| 1325 |
out => [undef, "\x81\xFE", [q<unassigned-code-point-error>]], |
| 1326 |
}, |
| 1327 |
{ |
| 1328 |
id => q<81FF>, |
| 1329 |
in => qq<\x81\xFF>, |
| 1330 |
out => [undef, "\x81\xFF", [q<unassigned-code-point-error>]], |
| 1331 |
}, |
| 1332 |
{ |
| 1333 |
id => q<DDDE>, |
| 1334 |
in => qq<\xDD\xDE>, |
| 1335 |
out => [undef, "\x{FF9D}", undef, "\x{FF9E}", undef], |
| 1336 |
}, |
| 1337 |
{ |
| 1338 |
id => q<e040>, |
| 1339 |
in => qq<\xE0\x40>, |
| 1340 |
out => [undef, "\x{6F3E}", undef], |
| 1341 |
}, |
| 1342 |
{ |
| 1343 |
id => q<eaa4>, |
| 1344 |
in => qq<\xEA\xA4>, |
| 1345 |
out => [undef, "\x{7199}", undef], |
| 1346 |
}, |
| 1347 |
{ |
| 1348 |
id => q<eaa5>, |
| 1349 |
in => qq<\xEA\xA5>, |
| 1350 |
out => [undef, "\xEA\xA5", [q<unassigned-code-point-error>]], |
| 1351 |
}, |
| 1352 |
{ |
| 1353 |
id => q<eb40>, |
| 1354 |
in => qq<\xEB\x40>, |
| 1355 |
out => [undef, "\xEB\x40", [q<unassigned-code-point-error>]], |
| 1356 |
}, |
| 1357 |
{ |
| 1358 |
id => q<ed40>, |
| 1359 |
in => qq<\xED\x40>, |
| 1360 |
out => [undef, "\xED\x40", [q<unassigned-code-point-error>]], |
| 1361 |
}, |
| 1362 |
{ |
| 1363 |
id => q<effc>, |
| 1364 |
in => qq<\xEF\xFC>, |
| 1365 |
out => [undef, "\xEF\xFC", [q<unassigned-code-point-error>]], |
| 1366 |
}, |
| 1367 |
{ |
| 1368 |
id => q<f040>, |
| 1369 |
in => qq<\xF0\x40>, |
| 1370 |
out => [undef, "\xF0", [q<unassigned-code-point-error>], |
| 1371 |
"\x40", undef], |
| 1372 |
}, |
| 1373 |
{ |
| 1374 |
id => q<f140>, |
| 1375 |
in => qq<\xF1\x40>, |
| 1376 |
out => [undef, "\xF1", [q<unassigned-code-point-error>], |
| 1377 |
"\x40", undef], |
| 1378 |
}, |
| 1379 |
{ |
| 1380 |
id => q<fb40>, |
| 1381 |
in => qq<\xFB\x40>, |
| 1382 |
out => [undef, "\xFB", [q<unassigned-code-point-error>], |
| 1383 |
"\x40", undef], |
| 1384 |
}, |
| 1385 |
{ |
| 1386 |
id => q<fc40>, |
| 1387 |
in => qq<\xFc\x40>, |
| 1388 |
out => [undef, "\xFC", [q<unassigned-code-point-error>], |
| 1389 |
"\x40", undef], |
| 1390 |
}, |
| 1391 |
{ |
| 1392 |
id => q<fd40>, |
| 1393 |
in => qq<\xFD\x40>, |
| 1394 |
out => [undef, "\xFD", [q<unassigned-code-point-error>], |
| 1395 |
"\x40", undef], |
| 1396 |
}, |
| 1397 |
{ |
| 1398 |
id => q<fE40>, |
| 1399 |
in => qq<\xFE\x40>, |
| 1400 |
out => [undef, "\xFE", [q<unassigned-code-point-error>], |
| 1401 |
"\x40", undef], |
| 1402 |
}, |
| 1403 |
{ |
| 1404 |
id => q<ff40>, |
| 1405 |
in => qq<\xFF\x40>, |
| 1406 |
out => [undef, "\xFF", [q<unassigned-code-point-error>], |
| 1407 |
"\x40", undef], |
| 1408 |
}, |
| 1409 |
{ |
| 1410 |
id => q<81408142>, |
| 1411 |
in => qq<\x81\x40\x81\x42>, |
| 1412 |
out => [undef, "\x{3000}", undef, "\x{3002}", undef], |
| 1413 |
}, |
| 1414 |
); |
| 1415 |
|
| 1416 |
check_charset ('XML-Shift_JIS', $XML_CHARSET.'shift_jis', |
| 1417 |
[map {$_->{name} = 'Shift_JIS'; $_} @testdata]); |
| 1418 |
} |
| 1419 |
|
| 1420 |
## ISO-2022-JP |
| 1421 |
{ |
| 1422 |
|
| 1423 |
my @testdata = ( |
| 1424 |
{ |
| 1425 |
id => q<l=0>, |
| 1426 |
in => q<>, |
| 1427 |
out1 => [undef], |
| 1428 |
out2 => [undef], |
| 1429 |
}, |
| 1430 |
{ |
| 1431 |
id => q<l=1.00>, |
| 1432 |
in => qq<\x00>, |
| 1433 |
out1 => [undef, "\x00", undef], |
| 1434 |
out2 => [undef, "\x00", undef], |
| 1435 |
}, |
| 1436 |
{ |
| 1437 |
id => q<l=1.0d>, |
| 1438 |
in => qq<\x0D>, |
| 1439 |
out1 => [undef, "\x0D", undef], |
| 1440 |
out2 => [undef, "\x0D", undef], |
| 1441 |
}, # Error? |
| 1442 |
{ |
| 1443 |
id => q<0A>, |
| 1444 |
in => qq<\x0A>, |
| 1445 |
out1 => [undef, "\x0A", undef], |
| 1446 |
out2 => [undef, "\x0A", undef], |
| 1447 |
}, # Error? |
| 1448 |
{ |
| 1449 |
id => q<l=1.0e>, |
| 1450 |
in => qq<\x0E>, |
| 1451 |
out1 => [undef, "\x0E", [q<illegal-octets-error>]], |
| 1452 |
out2 => [undef, "\x0E", [q<illegal-octets-error>]], |
| 1453 |
}, |
| 1454 |
{ |
| 1455 |
id => q<l=1.0f>, |
| 1456 |
in => qq<\x0F>, |
| 1457 |
out1 => [undef, "\x0F", [q<illegal-octets-error>]], |
| 1458 |
out2 => [undef, "\x0F", [q<illegal-octets-error>]], |
| 1459 |
}, |
| 1460 |
{ |
| 1461 |
id => q<l=1.1b>, |
| 1462 |
in => qq<\x1B>, |
| 1463 |
out1 => [undef, "\x1B", [q<illegal-octets-error>]], |
| 1464 |
out2 => [undef, "\x1B", [q<illegal-octets-error>]], |
| 1465 |
}, |
| 1466 |
{ |
| 1467 |
id => q<l=1.a>, |
| 1468 |
in => q<a>, |
| 1469 |
out1 => [undef, "a", undef], |
| 1470 |
out2 => [undef, "a", undef], |
| 1471 |
}, |
| 1472 |
{ |
| 1473 |
id => q<l=1.20>, |
| 1474 |
in => qq<\x20>, |
| 1475 |
out1 => [undef, "\x20", undef], |
| 1476 |
out2 => [undef, "\x20", undef], |
| 1477 |
}, |
| 1478 |
{ |
| 1479 |
id => q<l=1.5C>, |
| 1480 |
in => qq<\x5C>, |
| 1481 |
out1 => [undef, "\x5C", undef], |
| 1482 |
out2 => [undef, "\x5C", undef], |
| 1483 |
}, |
| 1484 |
{ |
| 1485 |
id => q<l=1.7E>, |
| 1486 |
in => qq<\x7E>, |
| 1487 |
out1 => [undef, "\x7E", undef], |
| 1488 |
out2 => [undef, "\x7E", undef], |
| 1489 |
}, |
| 1490 |
{ |
| 1491 |
id => q<l=1.7F>, |
| 1492 |
in => qq<\x7F>, |
| 1493 |
out1 => [undef, "\x7F", undef], |
| 1494 |
out2 => [undef, "\x7F", undef], |
| 1495 |
}, |
| 1496 |
{ |
| 1497 |
id => q<l=1.80>, |
| 1498 |
in => qq<\x80>, |
| 1499 |
out1 => [undef, "\x80", [q<illegal-octets-error>]], |
| 1500 |
out2 => [undef, "\x80", [q<illegal-octets-error>]], |
| 1501 |
}, |
| 1502 |
{ |
| 1503 |
id => q<l=1.8c>, |
| 1504 |
in => qq<\x8C>, |
| 1505 |
out1 => [undef, "\x8C", [q<illegal-octets-error>]], |
| 1506 |
out2 => [undef, "\x8C", [q<illegal-octets-error>]], |
| 1507 |
}, |
| 1508 |
{ |
| 1509 |
id => q<l=1.8e>, |
| 1510 |
in => qq<\x8E>, |
| 1511 |
out1 => [undef, "\x8E", [q<illegal-octets-error>]], |
| 1512 |
out2 => [undef, "\x8E", [q<illegal-octets-error>]], |
| 1513 |
}, |
| 1514 |
{ |
| 1515 |
id => q<l=1.8f>, |
| 1516 |
in => qq<\x8F>, |
| 1517 |
out1 => [undef, "\x8F", [q<illegal-octets-error>]], |
| 1518 |
out2 => [undef, "\x8F", [q<illegal-octets-error>]], |
| 1519 |
}, |
| 1520 |
{ |
| 1521 |
id => q<l=1.a0>, |
| 1522 |
in => qq<\xA0>, |
| 1523 |
out1 => [undef, "\xA0", [q<illegal-octets-error>]], |
| 1524 |
out2 => [undef, "\xA0", [q<illegal-octets-error>]], |
| 1525 |
}, |
| 1526 |
{ |
| 1527 |
id => q<l=1.a1>, |
| 1528 |
in => qq<\xA1>, |
| 1529 |
out1 => [undef, "\xA1", [q<illegal-octets-error>]], |
| 1530 |
out2 => [undef, "\xA1", [q<illegal-octets-error>]], |
| 1531 |
}, |
| 1532 |
{ |
| 1533 |
id => q<l=1.a2>, |
| 1534 |
in => qq<\xA2>, |
| 1535 |
out1 => [undef, "\xA2", [q<illegal-octets-error>]], |
| 1536 |
out2 => [undef, "\xA2", [q<illegal-octets-error>]], |
| 1537 |
}, |
| 1538 |
{ |
| 1539 |
id => q<l=1.df>, |
| 1540 |
in => qq<\xdf>, |
| 1541 |
out1 => [undef, "\xDF", [q<illegal-octets-error>]], |
| 1542 |
out2 => [undef, "\xDF", [q<illegal-octets-error>]], |
| 1543 |
}, |
| 1544 |
{ |
| 1545 |
id => q<l=1.e0>, |
| 1546 |
in => qq<\xe0>, |
| 1547 |
out1 => [undef, "\xE0", [q<illegal-octets-error>]], |
| 1548 |
out2 => [undef, "\xE0", [q<illegal-octets-error>]], |
| 1549 |
}, |
| 1550 |
{ |
| 1551 |
id => q<l=1.ef>, |
| 1552 |
in => qq<\xEF>, |
| 1553 |
out1 => [undef, "\xEF", [q<illegal-octets-error>]], |
| 1554 |
out2 => [undef, "\xEF", [q<illegal-octets-error>]], |
| 1555 |
}, |
| 1556 |
{ |
| 1557 |
id => q<F0>, |
| 1558 |
in => qq<\xF0>, |
| 1559 |
out1 => [undef, "\xF0", [q<illegal-octets-error>]], |
| 1560 |
out2 => [undef, "\xF0", [q<illegal-octets-error>]], |
| 1561 |
}, |
| 1562 |
{ |
| 1563 |
id => q<l=1.fc>, |
| 1564 |
in => qq<\xFC>, |
| 1565 |
out1 => [undef, "\xFC", [q<illegal-octets-error>]], |
| 1566 |
out2 => [undef, "\xFC", [q<illegal-octets-error>]], |
| 1567 |
}, |
| 1568 |
{ |
| 1569 |
id => q<l=1.fd>, |
| 1570 |
in => qq<\xFD>, |
| 1571 |
out1 => [undef, "\xFD", [q<illegal-octets-error>]], |
| 1572 |
out2 => [undef, "\xFD", [q<illegal-octets-error>]], |
| 1573 |
}, |
| 1574 |
{ |
| 1575 |
id => q<l=1.fe>, |
| 1576 |
in => qq<\xFE>, |
| 1577 |
out1 => [undef, "\xFE", [q<illegal-octets-error>]], |
| 1578 |
out2 => [undef, "\xFE", [q<illegal-octets-error>]], |
| 1579 |
}, |
| 1580 |
{ |
| 1581 |
id => q<l=1.ff>, |
| 1582 |
in => qq<\xFF>, |
| 1583 |
out1 => [undef, "\xFF", [q<illegal-octets-error>]], |
| 1584 |
out2 => [undef, "\xFF", [q<illegal-octets-error>]], |
| 1585 |
}, |
| 1586 |
{ |
| 1587 |
id => q<l=2.0000>, |
| 1588 |
in => qq<\x00\x00>, |
| 1589 |
out1 => [undef, "\x00", undef, "\x00", undef], |
| 1590 |
out2 => [undef, "\x00", undef, "\x00", undef], |
| 1591 |
}, |
| 1592 |
{ |
| 1593 |
id => q<l=2.0D0A>, |
| 1594 |
in => qq<\x0D\x0A>, |
| 1595 |
out1 => [undef, "\x0D", undef, "\x0A", undef], |
| 1596 |
out2 => [undef, "\x0D", undef, "\x0A", undef], |
| 1597 |
}, |
| 1598 |
{ |
| 1599 |
id => q<l=2.1B1B>, |
| 1600 |
in => qq<\x1B\x1B>, |
| 1601 |
out1 => [undef, "\x1B", [q<illegal-octets-error>], |
| 1602 |
"\x1B", [q<illegal-octets-error>]], |
| 1603 |
out2 => [undef, "\x1B", [q<illegal-octets-error>], |
| 1604 |
"\x1B", [q<illegal-octets-error>]], |
| 1605 |
}, |
| 1606 |
{ |
| 1607 |
id => q<l=2.1B20>, |
| 1608 |
in => qq<\x1B\x20>, |
| 1609 |
out1 => [undef, "\x1B", [q<illegal-octets-error>], "\x20", undef], |
| 1610 |
out2 => [undef, "\x1B", [q<illegal-octets-error>], "\x20", undef], |
| 1611 |
}, |
| 1612 |
{ |
| 1613 |
id => q<l=2.1B24>, |
| 1614 |
in => qq<\x1B\x24>, |
| 1615 |
out1 => [undef, "\x1B", [q<illegal-octets-error>], "\x24", undef], |
| 1616 |
out2 => [undef, "\x1B", [q<illegal-octets-error>], "\x24", undef], |
| 1617 |
}, |
| 1618 |
{ |
| 1619 |
id => q<l=2.1B28>, |
| 1620 |
in => qq<\x1B\x28>, |
| 1621 |
out1 => [undef, "\x1B", [q<illegal-octets-error>], "\x28", undef], |
| 1622 |
out2 => [undef, "\x1B", [q<illegal-octets-error>], "\x28", undef], |
| 1623 |
}, |
| 1624 |
{ |
| 1625 |
id => q<l=2.2020>, |
| 1626 |
in => qq<\x20\x20>, |
| 1627 |
out1 => [undef, "\x20", undef, "\x20", undef], |
| 1628 |
out2 => [undef, "\x20", undef, "\x20", undef], |
| 1629 |
}, |
| 1630 |
{ |
| 1631 |
id => q<l=2.ab>, |
| 1632 |
in => qq<ab>, |
| 1633 |
out1 => [undef, "a", undef, "b", undef], |
| 1634 |
out2 => [undef, "a", undef, "b", undef], |
| 1635 |
}, |
| 1636 |
{ |
| 1637 |
id => q<8040>, |
| 1638 |
in => qq<\x80\x40>, |
| 1639 |
out1 => [undef, "\x80", [q<illegal-octets-error>], |
| 1640 |
"\x40", undef], |
| 1641 |
out2 => [undef, "\x80", [q<illegal-octets-error>], |
| 1642 |
"\x40", undef], |
| 1643 |
}, |
| 1644 |
{ |
| 1645 |
id => q<1B2440>, |
| 1646 |
in => qq<\x1B\x24\x40>, |
| 1647 |
out1 => [undef], |
| 1648 |
out2 => [undef], |
| 1649 |
eof_error => [q<invalid-state-error>], |
| 1650 |
}, |
| 1651 |
{ |
| 1652 |
id => q<1B2442>, |
| 1653 |
in => qq<\x1B\x24\x42>, |
| 1654 |
out1 => [undef], |
| 1655 |
out2 => [undef], |
| 1656 |
eof_error => [q<invalid-state-error>], |
| 1657 |
}, |
| 1658 |
{ |
| 1659 |
id => q<1B2840>, |
| 1660 |
in => qq<\x1B\x28\x40>, |
| 1661 |
out1 => [undef, "\x1B", [q<illegal-octets-error>], "(", undef, |
| 1662 |
"\x40", undef], |
| 1663 |
out2 => [undef, "\x1B", [q<illegal-octets-error>], "(", undef, |
| 1664 |
"\x40", undef], |
| 1665 |
}, |
| 1666 |
{ |
| 1667 |
id => q<1B2842>, |
| 1668 |
in => qq<\x1B\x28\x42>, |
| 1669 |
out1 => [undef], |
| 1670 |
out2 => [undef], |
| 1671 |
}, |
| 1672 |
{ |
| 1673 |
id => q<1B284A>, |
| 1674 |
in => qq<\x1B\x28\x4A>, |
| 1675 |
out1 => [undef], |
| 1676 |
out2 => [undef], |
| 1677 |
eof_error => [q<invalid-state-error>], |
| 1678 |
}, |
| 1679 |
{ |
| 1680 |
id => q<1B$B1B(B>, |
| 1681 |
in => qq<\x1B\x24\x42\x1B\x28\x42>, |
| 1682 |
out1 => [undef], |
| 1683 |
out2 => [undef], |
| 1684 |
}, |
| 1685 |
{ |
| 1686 |
id => q<1B(B1B(B>, |
| 1687 |
in => qq<\x1B\x28\x42\x1B\x28\x42>, |
| 1688 |
out1 => [undef], |
| 1689 |
out2 => [undef], |
| 1690 |
}, |
| 1691 |
{ |
| 1692 |
id => q<1B(Ba1B(B>, |
| 1693 |
in => qq<\x1B\x28\x42a\x1B\x28\x42>, |
| 1694 |
out1 => [undef, "a", undef], |
| 1695 |
out2 => [undef, "a", undef], |
| 1696 |
}, |
| 1697 |
{ |
| 1698 |
id => q<1B(Ba1B(B1B(B>, |
| 1699 |
in => qq<\x1B\x28\x42a\x1B\x28\x42\x1B\x28\x42>, |
| 1700 |
out1 => [undef, "a", undef], |
| 1701 |
out2 => [undef, "a", undef], |
| 1702 |
}, |
| 1703 |
{ |
| 1704 |
id => q<1B$42!!1B2842>, |
| 1705 |
in => qq<\x1B\x24\x42!!\x1B\x28\x42>, |
| 1706 |
out1 => [undef, "\x{3000}", undef], |
| 1707 |
out2 => [undef, "\x{3000}", undef], |
| 1708 |
}, |
| 1709 |
{ |
| 1710 |
id => q<1B$4221211B284A>, |
| 1711 |
in => qq<\x1B\x24\x42!!\x1B\x28\x4A>, |
| 1712 |
out1 => [undef, "\x{3000}", undef], |
| 1713 |
out2 => [undef, "\x{3000}", undef], |
| 1714 |
eof_error => [q<invalid-state-error>], |
| 1715 |
}, |
| 1716 |
{ |
| 1717 |
id => q<1B$4021211B2842>, |
| 1718 |
in => qq<\x1B\x24\x40!!\x1B\x28\x42>, |
| 1719 |
out1 => [undef, "\x{3000}", undef], |
| 1720 |
out2 => [undef, "\x{3000}", undef], |
| 1721 |
}, |
| 1722 |
{ |
| 1723 |
id => q<1B$402121211B2842>, |
| 1724 |
in => qq<\x1B\x24\x40!!!\x1B\x28\x42>, |
| 1725 |
out1 => [undef, "\x{3000}", undef, "!", [q<illegal-octets-error>]], |
| 1726 |
out2 => [undef, "\x{3000}", undef, "!", [q<illegal-octets-error>]], |
| 1727 |
}, |
| 1728 |
{ |
| 1729 |
id => q<1B$4021211B2442!!1B2842>, |
| 1730 |
in => qq<\x1B\x24\x40!!\x1B\x24\x42!!\x1B\x28\x42>, |
| 1731 |
out1 => [undef, "\x{3000}", undef, "\x{3000}", undef], |
| 1732 |
out2 => [undef, "\x{3000}", undef, "\x{3000}", undef], |
| 1733 |
}, |
| 1734 |
{ |
| 1735 |
id => q<1B$4021211B2440!!1B2842>, |
| 1736 |
in => qq<\x1B\x24\x40!!\x1B\x24\x40!!\x1B\x28\x42>, |
| 1737 |
out1 => [undef, "\x{3000}", undef, "\x{3000}", undef], |
| 1738 |
out2 => [undef, "\x{3000}", undef, "\x{3000}", undef], |
| 1739 |
}, |
| 1740 |
{ |
| 1741 |
id => q<1B$@!"1B(B\~|>, |
| 1742 |
in => qq<\x1B\x24\x40!"\x1B(B\\~|>, |
| 1743 |
out1 => [undef, "\x{3001}", undef, "\x5C", undef, |
| 1744 |
"\x7E", undef, "|", undef], |
| 1745 |
out2 => [undef, "\x{3001}", undef, "\x5C", undef, |
| 1746 |
"\x7E", undef, "|", undef], |
| 1747 |
}, |
| 1748 |
{ |
| 1749 |
id => q<1B$B!"1B(J\~|1B(B>, |
| 1750 |
in => qq<\x1B\x24\x42!"\x1B(J\\~|\x1B(B>, |
| 1751 |
out1 => [undef, "\x{3001}", undef, "\xA5", undef, |
| 1752 |
"\x{203E}", undef, "|", undef], |
| 1753 |
out2 => [undef, "\x{3001}", undef, "\xA5", undef, |
| 1754 |
"\x{203E}", undef, "|", undef], |
| 1755 |
}, |
| 1756 |
{ |
| 1757 |
id => q<78compat.3022(16-02)>, |
| 1758 |
in => qq<\x1B\$\@\x30\x22\x1B\$B\x30\x22\x1B(B>, |
| 1759 |
out1 => [undef, "\x{555E}", undef, "\x{5516}", undef], |
| 1760 |
out2 => [undef, "\x{5516}", undef, "\x{5516}", undef], |
| 1761 |
}, |
| 1762 |
{ |
| 1763 |
id => q<unassigned.2239>, |
| 1764 |
in => qq<\x1B\$\@\x22\x39\x1B\$B\x22\x39\x1B(B>, |
| 1765 |
out1 => [undef, "\x22\x39", [q<unassigned-code-point-error>], |
| 1766 |
"\x22\x39", [q<unassigned-code-point-error>]], |
| 1767 |
out2 => [undef, "\x22\x39", [q<unassigned-code-point-error>], |
| 1768 |
"\x22\x39", [q<unassigned-code-point-error>]], |
| 1769 |
}, |
| 1770 |
{ |
| 1771 |
id => q<83add.223A>, |
| 1772 |
in => qq<\x1B\$\@\x22\x3A\x1B\$B\x22\x3A\x1B(B>, |
| 1773 |
out1 => [undef, "\x22\x3A", [q<unassigned-code-point-error>], |
| 1774 |
"\x{2208}", undef], |
| 1775 |
out2 => [undef, "\x{2208}", undef, "\x{2208}", undef], |
| 1776 |
}, |
| 1777 |
{ |
| 1778 |
id => q<83add.2840>, |
| 1779 |
in => qq<\x1B\$\@\x28\x40\x1B\$B\x28\x40\x1B(B>, |
| 1780 |
out1 => [undef, "\x28\x40", [q<unassigned-code-point-error>], |
| 1781 |
"\x{2542}", undef], |
| 1782 |
out2 => [undef, "\x{2542}", undef, "\x{2542}", undef], |
| 1783 |
}, |
| 1784 |
{ |
| 1785 |
id => q<83add.7421>, |
| 1786 |
in => qq<\x1B\$\@\x74\x21\x1B\$B\x74\x21\x1B(B>, |
| 1787 |
out1 => [undef, "\x74\x21", [q<unassigned-code-point-error>], |
| 1788 |
"\x{582F}", undef], |
| 1789 |
out2 => [undef, "\x{5C2D}", undef, "\x{582F}", undef], |
| 1790 |
}, |
| 1791 |
{ |
| 1792 |
id => q<83swap.3033>, |
| 1793 |
in => qq<\x1B\$\@\x30\x33\x1B\$B\x30\x33\x1B(B>, |
| 1794 |
out1 => [undef, "\x{9C3A}", undef, "\x{9BF5}", undef], |
| 1795 |
out2 => [undef, "\x{9C3A}", undef, "\x{9BF5}", undef], |
| 1796 |
}, |
| 1797 |
{ |
| 1798 |
id => q<83swap.724D>, |
| 1799 |
in => qq<\x1B\$\@\x72\x4D\x1B\$B\x72\x4D\x1B(B>, |
| 1800 |
out1 => [undef, "\x{9BF5}", undef, "\x{9C3A}", undef], |
| 1801 |
out2 => [undef, "\x{9BF5}", undef, "\x{9C3A}", undef], |
| 1802 |
}, |
| 1803 |
{ |
| 1804 |
id => q<90add.7425>, |
| 1805 |
in => qq<\x1B\$\@\x74\x25\x1B\$B\x74\x25\x1B(B>, |
| 1806 |
out1 => [undef, "\x74\x25", [q<unassigned-code-point-error>], |
| 1807 |
"\x74\x25", [q<unassigned-code-point-error>]], |
| 1808 |
out2 => [undef, "\x{51DC}", undef, "\x{51DC}", undef], |
| 1809 |
}, |
| 1810 |
{ |
| 1811 |
id => q<90add.7426>, |
| 1812 |
in => qq<\x1B\$\@\x74\x26\x1B\$B\x74\x26\x1B(B>, |
| 1813 |
out1 => [undef, "\x74\x26", [q<unassigned-code-point-error>], |
| 1814 |
"\x74\x26", [q<unassigned-code-point-error>]], |
| 1815 |
out2 => [undef, "\x{7199}", undef, "\x{7199}", undef], |
| 1816 |
}, |
| 1817 |
); |
| 1818 |
|
| 1819 |
check_charset ('IETF-ISO-2022-JP', $IANA_CHARSET.'iso-2022-jp', |
| 1820 |
[map {$_->{out} = $_->{out1}; $_->{name} = 'iso-2022-jp'; $_} |
| 1821 |
@testdata]); |
| 1822 |
check_charset ('XML-ISO-2022-JP', $XML_CHARSET.'iso-2022-jp', |
| 1823 |
[map {$_->{out} = $_->{out2}; $_->{name} = 'ISO-2022-JP'; $_} |
| 1824 |
@testdata]); |
| 1825 |
} |
Parent Directory
| 
Revision Log