:octocat: use Generator to list Kanji/Hanzi characters

This commit is contained in:
smiley
2023-03-08 12:27:00 +01:00
parent ccad9ca5e9
commit 021db2f0c9
2 changed files with 21 additions and 24 deletions
+10 -12
View File
@@ -11,10 +11,8 @@
namespace chillerlan\QRCodeTest\Data;
use chillerlan\QRCode\Data\Hanzi;
use Throwable;
use function bin2hex;
use function defined;
use function sprintf;
use Generator, Throwable;
use function bin2hex, chr, defined, sprintf;
/**
* Tests the Hanzi/GB2312 class
@@ -25,7 +23,7 @@ class HanziTest extends DataInterfaceTestAbstract{
protected string $testdata = '无可奈何燃花作香';
/**
* isGB2312() should pass on Hanzi/Hanzi characters and fail on everything else
* isGB2312() should pass on Hanzi/GB2312 characters and fail on everything else
*/
public function stringValidateProvider():array{
return [
@@ -41,22 +39,22 @@ class HanziTest extends DataInterfaceTestAbstract{
/**
* lists all characters in the valid GB2312 range
*/
public function hanziProvider():array{
$list = [];
public function hanziProvider():Generator{
for($byte1 = 0xa1; $byte1 < 0xf8; $byte1 += 0x1){
for($byte1 = 0xa1; $byte1 < 0xf8; $byte1++){
if($byte1 > 0xa9 && $byte1 < 0xb0){
continue;
}
for($byte2 = 0xa1; $byte2 < 0xff; $byte2++){
$list[] = [chr($byte1).chr($byte2)];
yield sprintf('0x%X', ($byte1 << 8 | $byte2)) => [
mb_convert_encoding(chr($byte1).chr($byte2), 'UTF-8', Hanzi::ENCODING),
];
}
}
return array_map(fn($chr) => mb_convert_encoding($chr, 'UTF-8', Hanzi::ENCODING), $list);
}
/**
@@ -75,8 +73,8 @@ class HanziTest extends DataInterfaceTestAbstract{
$this::markTestSkipped(sprintf(
'invalid glyph: %s => %s',
bin2hex($chr),
mb_convert_encoding($chr, Hanzi::ENCODING, mb_internal_encoding())
bin2hex(mb_convert_encoding($chr, Hanzi::ENCODING, 'UTF-8')),
$chr
));
}
}
+11 -12
View File
@@ -11,8 +11,8 @@
namespace chillerlan\QRCodeTest\Data;
use chillerlan\QRCode\Data\Kanji;
use Throwable;
use function array_map, bin2hex, chr, defined, mb_internal_encoding, sprintf;
use Generator, Throwable;
use function bin2hex, chr, defined, sprintf;
/**
* Tests the Kanji class
@@ -37,12 +37,13 @@ final class KanjiTest extends DataInterfaceTestAbstract{
}
/**
* lists the valid SJIS kanj
* lists the valid SJIS kanji
*/
public function kanjiProvider():array{
$list = [];
public function kanjiProvider():Generator{
$key = fn(int $byte1, int $byte2):string => sprintf('0x%X', ($byte1 << 8 | $byte2));
$val = fn(int $byte1, int $byte2):string => mb_convert_encoding(chr($byte1).chr($byte2), 'UTF-8', Kanji::ENCODING);
for($byte1 = 0x81; $byte1 < 0xeb; $byte1 += 0x1){
for($byte1 = 0x81; $byte1 < 0xeb; $byte1++){
// skip invalid/vendor ranges
if(($byte1 > 0x84 && $byte1 < 0x88) || ($byte1 > 0x9f && $byte1 < 0xe0)){
@@ -58,7 +59,7 @@ final class KanjiTest extends DataInterfaceTestAbstract{
continue;
}
$list[] = [chr($byte1).chr($byte2)];
yield $key($byte1, $byte2) => [$val($byte1, $byte2)];
}
}
@@ -66,15 +67,13 @@ final class KanjiTest extends DataInterfaceTestAbstract{
else{
for($byte2 = 0x9f; $byte2 < 0xfd; $byte2++){
$list[] = [chr($byte1).chr($byte2)];
yield $key($byte1, $byte2) => [$val($byte1, $byte2)];
}
}
}
// we need to put the joined byte sequence in a proper encoding
return array_map(fn($chr) => mb_convert_encoding($chr, Kanji::ENCODING, Kanji::ENCODING), $list);
}
/**
@@ -93,8 +92,8 @@ final class KanjiTest extends DataInterfaceTestAbstract{
$this::markTestSkipped(sprintf(
'invalid glyph: %s => %s',
bin2hex($chr),
mb_convert_encoding($chr, Kanji::ENCODING, mb_internal_encoding())
bin2hex(mb_convert_encoding($chr, Kanji::ENCODING, 'UTF-8')),
$chr
));
}
}