MensBeam
/
Intl
3
0
Fork 0
Code Pull Requests Releases 10 Activity
Browse Source

Style fixes

multi-byte
J. King 4 years ago
parent
4299bf0100
commit
a57dde6dbd
22 changed files with 142 additions and 121 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 10
      lib/Encoding/Big5.php
  2. 6
      lib/Encoding/EUCJP.php
  3. 109
      lib/Encoding/Encoder.php
  4. 16
      lib/Encoding/ISO2022JP.php
  5. 2
      lib/Encoding/ShiftJIS.php
  6. 4
      lib/Encoding/UTF16BE.php
  7. 2
      lib/Encoding/UTF16LE.php
  8. 10
      lib/Encoding/UTF8.php
  9. 1
      tests/cases/Encoding/TestBig5.php
  10. 25
      tests/cases/Encoding/TestEUCJP.php
  11. 1
      tests/cases/Encoding/TestEUCKR.php
  12. 1
      tests/cases/Encoding/TestGB18030.php
  13. 31
      tests/cases/Encoding/TestISO2022JP.php
  14. 29
      tests/cases/Encoding/TestShiftJIS.php
  15. 1
      tests/cases/Encoding/TestSingleByte.php
  16. 1
      tests/cases/Encoding/TestUTF16LE.php
  17. 1
      tests/cases/Encoding/TestUTF8.php
  18. 1
      tests/cases/Encoding/TestXUserDefined.php
  19. 4
      tests/cases/TestEncoding.php
  20. 2
      tests/lib/DecoderTest.php
  21. 2
      tools/mkindex.php
  22. 4
      tools/mklabels.php

10
lib/Encoding/Big5.php
View File

@ -9,11 +9,11 @@ namespace MensBeam\Intl\Encoding;
class Big5 extends AbstractEncoding implements StatelessEncoding {
const NAME = "Big5";
const LABELS = [
"big5",
"big5-hkscs",
"cn-big5",
"csbig5",
"x-x-big5"
"big5",
"big5-hkscs",
"cn-big5",
"csbig5",
"x-x-big5",
];
const TABLE_DOUBLES = [
1133 => [0x00CA, 0x0304],

6
lib/Encoding/EUCJP.php
View File

File diff suppressed because one or more lines are too long

109
lib/Encoding/Encoder.php
View File

@ -22,7 +22,7 @@ class Encoder {
if (!$l || !$l['encoder']) {
throw new EncoderException("Label '$label' does not have an encoder", Encoder::E_UNAVAILABLE_ENCODER);
} else {
$this->name = $['name'];
$this->name = $s['name'];
$this->fatal = $fatal;
}
}
@ -32,41 +32,76 @@ class Encoder {
throw new EncoderException("Encountered code point outside Unicode range ($codePoint)", self::E_INVALID_CODE_POINT);
}
switch ($this->name) {
case "UTF-8": return return UTF8::encode($codePoint, $this->fatal);
case "Big5": return Big5::encode($codePoint, $this->fatal);
case "EUC-JP": return EUCJP::encode($codePoint, $this->fatal);
case "EUC-KR": return EUCKR::encode($codePoint, $this->fatal);
case "gb18030": return GB18030::encode($codePoint, $this->fatal);
case "GBK": return GBK::encode($codePoint, $this->fatal);
case "IBM866": return IBM866::encode($codePoint, $this->fatal);
case "ISO-8859-2": return ISO88592::encode($codePoint, $this->fatal);
case "ISO-8859-3": return ISO88593::encode($codePoint, $this->fatal);
case "ISO-8859-4": return ISO88594::encode($codePoint, $this->fatal);
case "ISO-8859-5": return ISO88595::encode($codePoint, $this->fatal);
case "ISO-8859-6": return ISO88596::encode($codePoint, $this->fatal);
case "ISO-8859-7": return ISO88597::encode($codePoint, $this->fatal);
case "ISO-8859-8": return ISO88598::encode($codePoint, $this->fatal);
case "ISO-8859-8-I": return ISO88598I::encode($codePoint, $this->fatal);
case "ISO-8859-10": return ISO885910::encode($codePoint, $this->fatal);
case "ISO-8859-13": return ISO885913::encode($codePoint, $this->fatal);
case "ISO-8859-14": return ISO885914::encode($codePoint, $this->fatal);
case "ISO-8859-15": return ISO885915::encode($codePoint, $this->fatal);
case "ISO-8859-16": return ISO885916::encode($codePoint, $this->fatal);
case "KOI8-R": return KOI8R::encode($codePoint, $this->fatal);
case "KOI8-U": return KOI8U::encode($codePoint, $this->fatal);
case "macintosh": return Macintosh::encode($codePoint, $this->fatal);
case "Shift_JIS": return ShiftJIS::encode($codePoint, $this->fatal);
case "windows1250": return Windows1250::encode($codePoint, $this->fatal);
case "windows1251": return Windows1251::encode($codePoint, $this->fatal);
case "windows1252": return Windows1252::encode($codePoint, $this->fatal);
case "windows1253": return Windows1253::encode($codePoint, $this->fatal);
case "windows1254": return Windows1254::encode($codePoint, $this->fatal);
case "windows1255": return Windows1255::encode($codePoint, $this->fatal);
case "windows1256": return Windows1256::encode($codePoint, $this->fatal);
case "windows1257": return Windows1257::encode($codePoint, $this->fatal);
case "windows874": return Windows874::encode($codePoint, $this->fatal);
case "x-mac-cyrillic": return XMacCyrillic::encode($codePoint, $this->fatal);
case "x-user-defined": return XUserDefined::encode($codePoint, $this->fatal);
case "UTF-8":
return UTF8::encode($codePoint, $this->fatal);
case "Big5":
return Big5::encode($codePoint, $this->fatal);
case "EUC-JP":
return EUCJP::encode($codePoint, $this->fatal);
case "EUC-KR":
return EUCKR::encode($codePoint, $this->fatal);
case "gb18030":
return GB18030::encode($codePoint, $this->fatal);
case "GBK":
return GBK::encode($codePoint, $this->fatal);
case "IBM866":
return IBM866::encode($codePoint, $this->fatal);
case "ISO-8859-2":
return ISO88592::encode($codePoint, $this->fatal);
case "ISO-8859-3":
return ISO88593::encode($codePoint, $this->fatal);
case "ISO-8859-4":
return ISO88594::encode($codePoint, $this->fatal);
case "ISO-8859-5":
return ISO88595::encode($codePoint, $this->fatal);
case "ISO-8859-6":
return ISO88596::encode($codePoint, $this->fatal);
case "ISO-8859-7":
return ISO88597::encode($codePoint, $this->fatal);
case "ISO-8859-8":
return ISO88598::encode($codePoint, $this->fatal);
case "ISO-8859-8-I":
return ISO88598I::encode($codePoint, $this->fatal);
case "ISO-8859-10":
return ISO885910::encode($codePoint, $this->fatal);
case "ISO-8859-13":
return ISO885913::encode($codePoint, $this->fatal);
case "ISO-8859-14":
return ISO885914::encode($codePoint, $this->fatal);
case "ISO-8859-15":
return ISO885915::encode($codePoint, $this->fatal);
case "ISO-8859-16":
return ISO885916::encode($codePoint, $this->fatal);
case "KOI8-R":
return KOI8R::encode($codePoint, $this->fatal);
case "KOI8-U":
return KOI8U::encode($codePoint, $this->fatal);
case "macintosh":
return Macintosh::encode($codePoint, $this->fatal);
case "Shift_JIS":
return ShiftJIS::encode($codePoint, $this->fatal);
case "windows1250":
return Windows1250::encode($codePoint, $this->fatal);
case "windows1251":
return Windows1251::encode($codePoint, $this->fatal);
case "windows1252":
return Windows1252::encode($codePoint, $this->fatal);
case "windows1253":
return Windows1253::encode($codePoint, $this->fatal);
case "windows1254":
return Windows1254::encode($codePoint, $this->fatal);
case "windows1255":
return Windows1255::encode($codePoint, $this->fatal);
case "windows1256":
return Windows1256::encode($codePoint, $this->fatal);
case "windows1257":
return Windows1257::encode($codePoint, $this->fatal);
case "windows874":
return Windows874::encode($codePoint, $this->fatal);
case "x-mac-cyrillic":
return XMacCyrillic::encode($codePoint, $this->fatal);
case "x-user-defined":
return XUserDefined::encode($codePoint, $this->fatal);
case "ISO-2022-JP":
if ($codePoint === 0xE || $codePoint === 0xF || $codePoint === 0x1B) {
return $this->err($codePoint, 0xFFFD);
@ -102,7 +137,7 @@ class Encoder {
return $lead.$trail;
}
return $this->err($codePoint);
}
}
}
}

16
lib/Encoding/ISO2022JP.php
View File

@ -27,7 +27,7 @@ class ISO2022JP extends AbstractEncoding implements StatefulEncoding {
protected $modeMark = \PHP_INT_MIN;
protected $modeStack = [];
protected $dirtyEOF = 0;
public function __construct(string $string, bool $fatal = false, bool $allowSurrogates = false) {
parent::__construct($string, $fatal, $allowSurrogates);
$this->stateProps[] = "dirtyEOF";
@ -44,7 +44,7 @@ class ISO2022JP extends AbstractEncoding implements StatefulEncoding {
public function nextCode() {
$this->posChar++;
$state = $this->mode;
while (true) {
while (true) {
$b = @$this->string[$this->posByte++];
$eof = ($b === "");
$b = ord($b);
@ -142,7 +142,7 @@ class ISO2022JP extends AbstractEncoding implements StatefulEncoding {
$this->modeMark = $this->posByte;
return $mode;
}
public static function encode(array $codePoints, bool $fatal = true): string {
return "";
}
@ -167,7 +167,7 @@ class ISO2022JP extends AbstractEncoding implements StatefulEncoding {
} else {
$this->posByte -= ($this->mode === self::LEAD_BYTE_STATE ? 2 : 1);
}
// check for a mode change that is not also an error character
// check for a mode change that is not also an error character
if ($this->posByte === $this->modeMark && $this->posByte !== $this->errMark) {
$this->posByte -= 3;
list($this->modeMark, $this->mode) = array_pop($this->modeStack);
@ -175,13 +175,13 @@ class ISO2022JP extends AbstractEncoding implements StatefulEncoding {
}
return $distance;
}
protected function stateSave(): array {
$out = parent::stateSave();
$out['modeCount'] = sizeof($this->modeStack);
return $out;
}
protected function stateApply(array $state) {
while (sizeof($this->modeStack) > $state['modeCount']) {
list($this->modeMark, $this->mode) = array_pop($this->modeStack);
@ -189,7 +189,7 @@ class ISO2022JP extends AbstractEncoding implements StatefulEncoding {
unset($state['modeCount']);
parent::stateApply($state);
}
public function rewind() {
$this->modeStack = [];
$this->modeMark = \PHP_INT_MIN;
@ -197,7 +197,7 @@ class ISO2022JP extends AbstractEncoding implements StatefulEncoding {
$this->dirtyEOF = 0;
parent::rewind();
}
public function eof(): bool {
return $this->posByte === $this->lenByte || ($this->posByte === ($this->lenByte - 3) && $this->peekCode() === false);
}

2
lib/Encoding/ShiftJIS.php
View File

@ -50,7 +50,7 @@ class ShiftJIS extends AbstractEncoding implements StatelessEncoding {
if ($pointer >= 8836 && $pointer <= 10715) {
return 0xE000 - 8836 + $pointer;
} else {
$codePoint = self::TABLE_CODES[$pointer] ?? self::TABLE_CODES_EXTRA[$pointer] ?? null;
$codePoint = self::TABLE_CODES[$pointer] ?? self::TABLE_CODES_EXTRA[$pointer] ?? null;
}
}
if ($codePoint === null) {

4
lib/Encoding/UTF16BE.php
View File

@ -10,7 +10,7 @@ class UTF16BE extends UTF16 {
const BE = true;
const NAME = "UTF-16BE";
const LABELS = [
"unicodefffe",
"utf-16be"
"unicodefffe",
"utf-16be",
];
}

2
lib/Encoding/UTF16LE.php
View File

@ -16,6 +16,6 @@ class UTF16LE extends UTF16 {
"unicode",
"unicodefeff",
"utf-16",
"utf-16le"
"utf-16le",
];
}

10
lib/Encoding/UTF8.php
View File

@ -9,12 +9,12 @@ namespace MensBeam\Intl\Encoding;
class UTF8 extends AbstractEncoding implements StatelessEncoding {
const NAME = "UTF-8";
const LABELS = [
"unicode-1-1-utf-8",
"unicode11utf8",
"unicode-1-1-utf-8",
"unicode11utf8",
"unicode20utf8",
"utf-8",
"utf8",
"x-unicode20utf8"
"utf-8",
"utf8",
"x-unicode20utf8",
];
protected $selfSynchronizing = true;

1
tests/cases/Encoding/TestBig5.php
View File

@ -136,7 +136,6 @@ class TestBig5 extends \MensBeam\Intl\Test\CoderDecoderTest {
return parent::testIterateThroughAStringAllowingSurrogates($input, $strictExp, $relaxedExp);
}
/**
* @covers MensBeam\Intl\Encoding\Big5::seekBack
*/

25
tests/cases/Encoding/TestEUCJP.php
View File

File diff suppressed because one or more lines are too long

1
tests/cases/Encoding/TestEUCKR.php
View File

@ -136,7 +136,6 @@ class TestEUCKR extends \MensBeam\Intl\Test\CoderDecoderTest {
return parent::testIterateThroughAStringAllowingSurrogates($input, $strictExp, $relaxedExp);
}
/**
* @covers MensBeam\Intl\Encoding\EUCKR::seekBack
*/

1
tests/cases/Encoding/TestGB18030.php
View File

@ -144,7 +144,6 @@ class TestGB18030 extends \MensBeam\Intl\Test\CoderDecoderTest {
return parent::testIterateThroughAStringAllowingSurrogates($input, $strictExp, $relaxedExp);
}
/**
* @covers MensBeam\Intl\Encoding\GB18030::seekBack
*/

31
tests/cases/Encoding/TestISO2022JP.php
View File

@ -7,8 +7,6 @@ declare(strict_types=1);
namespace MensBeam\Intl\TestCase\Encoding;
use MensBeam\Intl\Encoding\ISO2022JP;
use MensBeam\Intl\Encoding\Encoding;
use MensBeam\Intl\Encoding\EncoderException;
class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
protected $testedClass = ISO2022JP::class;
@ -64,7 +62,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
* @dataProvider provideCodePoints
* @covers MensBeam\Intl\Encoding\ISO2022JP::encode
* @covers MensBeam\Intl\Encoding\ISO2022JP::errEnc
*/
*/
public function testEncodeCodePoints(bool $fatal, $input, $exp) {
return parent::testEncodeCodePoints($fatal, $input, $exp);
}
@ -74,7 +72,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
* @covers MensBeam\Intl\Encoding\ISO2022JP::__construct
* @covers MensBeam\Intl\Encoding\ISO2022JP::nextCode
* @covers MensBeam\Intl\Encoding\ISO2022JP::modeSet
*/
*/
public function testDecodeMultipleCharactersAsCodePoints(string $input, array $exp) {
return parent::testDecodeMultipleCharactersAsCodePoints($input, $exp);
}
@ -84,7 +82,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
* @covers MensBeam\Intl\Encoding\ISO2022JP::__construct
* @covers MensBeam\Intl\Encoding\ISO2022JP::nextChar
* @covers MensBeam\Intl\Encoding\ISO2022JP::modeSet
*/
*/
public function testDecodeMultipleCharactersAsStrings(string $input, array $exp) {
return parent::testDecodeMultipleCharactersAsStrings($input, $exp);
}
@ -92,7 +90,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
/**
* @dataProvider provideStrings
* @covers MensBeam\Intl\Encoding\ISO2022JP::seekBack
*/
*/
public function testSTepBackThroughAString(string $input, array $exp) {
return parent::testSTepBackThroughAString($input, $exp);
}
@ -102,7 +100,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
* @covers MensBeam\Intl\Encoding\ISO2022JP::posChar
* @covers MensBeam\Intl\Encoding\ISO2022JP::posByte
* @covers MensBeam\Intl\Encoding\ISO2022JP::rewind
*/
*/
public function testSeekThroughAString() {
return parent::testSeekThroughAString();
}
@ -111,7 +109,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
* @covers MensBeam\Intl\Encoding\ISO2022JP::posChar
* @covers MensBeam\Intl\Encoding\ISO2022JP::posByte
* @covers MensBeam\Intl\Encoding\ISO2022JP::eof
*/
*/
public function testTraversePastTheEndOfAString() {
return parent::testTraversePastTheEndOfAString();
}
@ -120,7 +118,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
* @covers MensBeam\Intl\Encoding\ISO2022JP::peekChar
* @covers MensBeam\Intl\Encoding\ISO2022JP::stateSave
* @covers MensBeam\Intl\Encoding\ISO2022JP::stateApply
*/
*/
public function testPeekAtCharacters() {
return parent::testPeekAtCharacters();
}
@ -129,7 +127,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
* @covers MensBeam\Intl\Encoding\ISO2022JP::peekCode
* @covers MensBeam\Intl\Encoding\ISO2022JP::stateSave
* @covers MensBeam\Intl\Encoding\ISO2022JP::stateApply
*/
*/
public function testPeekAtCodePoints() {
return parent::testPeekAtCodePoints();
}
@ -140,14 +138,14 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
* @covers MensBeam\Intl\Encoding\ISO2022JP::lenByte
* @covers MensBeam\Intl\Encoding\ISO2022JP::stateSave
* @covers MensBeam\Intl\Encoding\ISO2022JP::stateApply
*/
*/
public function testGetStringLength(string $input, array $points) {
return parent::testGetStringLength($input, $points);
}
/**
* @covers MensBeam\Intl\Encoding\ISO2022JP::errDec
*/
*/
public function testReplacementModes() {
return parent::testReplacementModes();
}
@ -157,7 +155,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
* @covers MensBeam\Intl\Encoding\ISO2022JP::rewind
* @covers MensBeam\Intl\Encoding\ISO2022JP::chars
* @covers MensBeam\Intl\Encoding\ISO2022JP::codes
*/
*/
public function testIterateThroughAString(string $input, array $exp) {
return parent::testIterateThroughAString($input, $exp);
}
@ -165,12 +163,11 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
/**
* @dataProvider provideStrings
* @coversNothing
*/
*/
public function testIterateThroughAStringAllowingSurrogates(string $input, array $strictExp, array $relaxedExp = null) {
return parent::testIterateThroughAStringAllowingSurrogates($input, $strictExp, $relaxedExp);
}
/**
* @covers MensBeam\Intl\Encoding\ISO2022JP::seekBack
*/
@ -180,7 +177,7 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
/**
* @group optional
*/
*/
public function testPedanticallyDecodeSingleCharactersAsCodePoint() {
$series = [
];
@ -196,5 +193,3 @@ class TestISO2022JP extends \MensBeam\Intl\Test\CoderDecoderTest {
}
}
}

29
tests/cases/Encoding/TestShiftJIS.php
View File

File diff suppressed because one or more lines are too long

1
tests/cases/Encoding/TestSingleByte.php
View File

@ -214,7 +214,6 @@ class TestSingleByte extends \MensBeam\Intl\Test\CoderDecoderTest {
return parent::testIterateThroughAStringAllowingSurrogates($input, $exp, $exp);
}
/**
* @dataProvider provideClasses
* @coversNothing

1
tests/cases/Encoding/TestUTF16LE.php
View File

@ -126,7 +126,6 @@ class TestUTF16LE extends \MensBeam\Intl\Test\DecoderTest {
return parent::testIterateThroughAStringAllowingSurrogates($input, $strictExp, $relaxedExp);
}
/**
* @covers MensBeam\Intl\Encoding\UTF16::seekBack
*/

1
tests/cases/Encoding/TestUTF8.php
View File

@ -136,7 +136,6 @@ class TestUTF8 extends \MensBeam\Intl\Test\CoderDecoderTest {
return parent::testIterateThroughAStringAllowingSurrogates($input, $strictExp, $relaxedExp);
}
/**
* @covers MensBeam\Intl\Encoding\UTF8::seekBack
*/

1
tests/cases/Encoding/TestXUserDefined.php
View File

@ -117,7 +117,6 @@ class TestXUserDefined extends \MensBeam\Intl\Test\DecoderTest {
return parent::testIterateThroughAStringAllowingSurrogates($input, $strictExp, $relaxedExp);
}
/**
* @coversNothing
*/

4
tests/cases/TestEncoding.php
View File

@ -49,7 +49,9 @@ class TestEncoding extends \PHPUnit\Framework\TestCase {
}
$out = [];
foreach ($labels as $label => $name) {
$out[] = [(string) $label, ['label' => (string) $label, 'name' => $name, 'class' => $names[$name]]];
$class = $names[$name];
$encoder = !in_array($name, ["UTF-16LE", "UTF-16BE", "replacement"]);
$out[] = [(string) $label, ['label' => (string) $label, 'name' => $name, 'encoder' => $encoder, 'class' => $class]];
}
return $out;
}

2
tests/lib/DecoderTest.php
View File

@ -94,7 +94,7 @@ abstract class DecoderTest extends \PHPUnit\Framework\TestCase {
$this->assertSame(0, $s->seek(4));
$this->assertSame(7, $s->posChar());
$this->assertSame($off[7], $s->posByte());
$this->assertSame(1, $s->seek(1));
$this->assertSame(7, $s->posChar());
if ($this->testedClass !== ISO2022JP::class) {

2
tools/mkindex.php
View File

@ -111,7 +111,7 @@ ARRAY_LITERAL;
// search the Big5 rump for duplicates
$dupes = make_override_array($nhk);
// remove those duplicates which should use the last code point
foreach([0x2550, 0x255E, 0x2561, 0x256A, 0x5341, 0x5345] as $code) {
foreach ([0x2550, 0x255E, 0x2561, 0x256A, 0x5341, 0x5345] as $code) {
unset($dupes[$code]);
}
// serialize and print; Hong Kong characters are kept separate as they are not used in encoding

4
tools/mklabels.php
View File

@ -1,6 +1,6 @@
<?php
// this script read and names and labels from each concrete
// class in the Encoding set and generates tables mapping labels
// this script read and names and labels from each concrete
// class in the Encoding set and generates tables mapping labels
// to names and names to classes
use MensBeam\Intl\Encoding\Encoding;

Write Preview
Loading…
Cancel
Save