TokenFactory和Tokenizer基础

This commit is contained in:
Jerry Yan 2021-01-22 14:21:11 +08:00
parent 10bf8e6481
commit 074f1805eb
7 changed files with 240 additions and 42 deletions

View File

@ -0,0 +1,32 @@
<?php
/**
* @filename DefaultFactory.php
* @author Jerry Yan <792602257@qq.com>
* @date 2020/12/17 14:48
*/
namespace JerryYan\DSL\Token\Factory;
use JerryYan\DSL\Token\Token;
use JerryYan\DSL\Token\TokenAnd;
use JerryYan\DSL\Token\TokenOr;
use JerryYan\DSL\Token\TokenVar;
class DefaultFactory extends FactoryInterface
{
protected $tokenMap = [
Token::AND => TokenAnd::class,
Token::OR => TokenOr::class,
Token::VAR => TokenVar::class,
];
protected $tokenNameMap = [
"" => Token::AND,
"或者" => Token::OR,
"" => Token::OR,
];
protected $undefinedTokenClass = TokenVar::class;
}

View File

@ -0,0 +1,38 @@
<?php
/**
* @filename FactoryInterface.php
* @author Jerry Yan <792602257@qq.com>
* @date 2021/1/22 13:42
*/
namespace JerryYan\DSL\Token\Factory;
use JerryYan\DSL\Token\TokenInterface;
use JerryYan\DSL\Token\TokenUndefined;
abstract class FactoryInterface
{
/** @var array<string, \JerryYan\DSL\Token\TokenInterface> Token类型及映射类 */
protected $tokenMap = [];
/** @var array<string, string> Token别名映射 */
protected $tokenNameMap = [];
/** @var \JerryYan\DSL\Token\TokenInterface 默认Token类 */
protected $undefinedTokenClass = TokenUndefined::class;
public function getTokenByName(string $name): TokenInterface
{
$originalName = $name;
if (isset($this->tokenNameMap[$name])) {
$name = $this->tokenNameMap[$name];
}
if (!isset($this->tokenMap[$name])) {
return new $this->undefinedTokenClass($originalName);
} else {
return new $this->tokenMap[$name]($originalName);
}
}
}

View File

@ -1,37 +0,0 @@
<?php
/**
* @filename TokenFactory.php
* @author Jerry Yan <792602257@qq.com>
* @date 2020/12/17 14:48
*/
namespace JerryYan\DSL\Token;
class TokenFactory
{
/** @var array<string, string> Token类型及映射类 */
private $tokenMap = [
Token::AND => TokenAnd::class,
Token::OR => TokenOr::class,
Token::VAR => TokenVar::class,
];
protected $tokenNameMap = [
];
public function getTokenByName(string $name): TokenInterface
{
$originalName = $name;
if (isset($this->tokenNameMap[$name])) {
$name = $this->tokenNameMap[$name];
}
if (!isset($this->tokenMap[$name])) {
return new TokenUndefined($originalName);
} else {
return new $this->tokenMap[$name]($originalName);
}
}
}

View File

@ -40,12 +40,12 @@ abstract class TokenInterface
*/
public function getFirstToken(): ?TokenInterface
{
if ($this->hasPrevToken()) return $this->prevToken->getFirstToken();
if ($this->hasPrevToken()) return $this->getPrevToken()->getFirstToken();
else return $this;
}
public function getLastToken(): ?TokenInterface
{
if ($this->hasNextToken()) return $this->nextToken->getLastToken();
if ($this->hasNextToken()) return $this->getNextToken()->getLastToken();
else return $this;
}
public function hasPrevToken(): bool
@ -62,4 +62,22 @@ abstract class TokenInterface
public function getNextToken(): ?TokenInterface{
return $this->nextToken;
}
public function getPrevTokenLength(): int
{
if ($this->hasPrevToken()) return $this->getPrevToken()->getPrevTokenLength() + 1;
else return 0;
}
public function getNextTokenLength(): int
{
if ($this->hasNextToken()) return $this->getNextToken()->getNextTokenLength() + 1;
else return 0;
}
public function getLength(): int
{
return $this->getPrevTokenLength() + $this->getNextTokenLength() + 1;
}
public function count(): int
{
return $this->getLength();
}
}

View File

@ -10,13 +10,14 @@ namespace JerryYan\DSL\Tokenizer;
use JerryYan\DSL\Reader\ReaderInterface;
use JerryYan\DSL\Token\TokenFactory;
use JerryYan\DSL\Token\Factory\FactoryInterface;
use JerryYan\DSL\Token\TokenInterface;
class Tokenizer extends TokenizerInterface
{
public function __construct(TokenFactory $tokenFactory)
/** @var FactoryInterface token工厂 */
protected $tokenFactory;
public function __construct(FactoryInterface $tokenFactory)
{
$this->tokenFactory = $tokenFactory;
}

View File

@ -0,0 +1,96 @@
<?php
/**
* @filename TokenInterfaceTest.php
* @author Jerry Yan <792602257@qq.com>
* @date 2021/1/22 13:52
*/
namespace JerryYan\DSL\Test\Token;
use JerryYan\DSL\Token\TokenAnd;
use JerryYan\DSL\Token\TokenDefine;
use JerryYan\DSL\Token\TokenInterface;
use JerryYan\DSL\Token\TokenOr;
use JerryYan\DSL\Token\TokenUndefined;
use JerryYan\DSL\Token\TokenVar;
use PHPUnit\Framework\TestCase;
class TokenInterfaceTest extends TestCase
{
/** @var TokenInterface[] TokenClass */
private $tokenTypes = [
TokenAnd::class,
TokenDefine::class,
TokenOr::class,
TokenUndefined::class,
TokenVar::class,
];
protected $chainFirst;
protected $chainLast;
protected function setUp(): void
{
/** @var ?TokenInterface $next */
$last = NULL;
foreach ($this->tokenTypes as $cls) {
/** @var TokenInterface $current */
$current = new $cls('');
$current->setPrevToken($last);
if ($last !== NULL) $last->setNextToken($current);
$last = $current;
}
$this->chainLast = $last;
$this->chainFirst = $last->getFirstToken();
}
public function testGetFirstToken()
{
$this->assertInstanceOf($this->tokenTypes[0], $this->chainLast->getFirstToken(), '链表头类型非预期');
}
public function testGetPrevToken()
{
$this->assertInstanceOf($this->tokenTypes[count($this->tokenTypes)-2], $this->chainLast->getPrevToken(), '链表尾前一类型非预期');
$this->assertNull($this->chainFirst->getPrevToken(), '链表头前一类型非空');
}
public function testCount()
{
$this->assertEquals($this->chainFirst->count(), $this->chainLast->count(), '链表头尾计数不同');
$this->assertEquals(count($this->tokenTypes), $this->chainLast->count(), '链表尾的长度不同');
}
public function testGetNextToken()
{
$this->assertInstanceOf($this->tokenTypes[1], $this->chainFirst->getNextToken(), '链表头后一类型非预期');
$this->assertNull($this->chainLast->getNextToken(), '链表尾前一类型非空');
}
public function testGetLastToken()
{
$this->assertInstanceOf($this->tokenTypes[count($this->tokenTypes)-1], $this->chainFirst->getLastToken(), '链表尾类型非预期');
}
public function testHasPrevToken()
{
$this->assertTrue($this->chainLast->hasPrevToken());
$this->assertFalse($this->chainFirst->hasPrevToken());
}
public function testHasNextToken()
{
$this->assertTrue($this->chainFirst->hasNextToken());
$this->assertFalse($this->chainLast->hasNextToken());
}
public function testGetNextTokenLength()
{
$this->assertEquals(count($this->tokenTypes) - 2, $this->chainFirst->getNextToken()->getNextTokenLength(), '链表头后一后长度不对');
$this->assertEquals(0, $this->chainLast->getNextTokenLength(), '链表尾后长度不对');
}
public function testGetPrevTokenLength()
{
$this->assertEquals(count($this->tokenTypes) - 2, $this->chainLast->getPrevToken()->getPrevTokenLength(), '链表尾前一前长度不对');
$this->assertEquals(0, $this->chainFirst->getPrevTokenLength(), '链表头前长度不对');
}
}

View File

@ -0,0 +1,50 @@
<?php
/**
* @filename TokenizerTest.php
* @author Jerry Yan <792602257@qq.com>
* @date 2021/1/22 13:33
*/
namespace JerryYan\DSL\Test\Tokenizer;
use JerryYan\DSL\Reader\StringReader;
use JerryYan\DSL\Token\Factory\DefaultFactory;
use JerryYan\DSL\Token\TokenAnd;
use JerryYan\DSL\Token\TokenInterface;
use JerryYan\DSL\Token\TokenOr;
use JerryYan\DSL\Token\TokenVar;
use JerryYan\DSL\Tokenizer\Tokenizer;
use PHPUnit\Framework\TestCase;
class TokenizerTest extends TestCase
{
protected $tokenizer;
protected $reader;
private $text = "这个 和 那个 或者 那个 和 这个";
private $textTokenType = [
TokenVar::class,
TokenAnd::class,
TokenVar::class,
TokenOr::class,
TokenVar::class,
TokenAnd::class,
TokenVar::class,
];
protected function setUp(): void
{
$this->tokenizer = new Tokenizer(new DefaultFactory());
$this->reader = new StringReader($this->text);
}
public function testTokenize()
{
$tokens = $this->tokenizer->tokenize($this->reader);
$this->assertInstanceOf(TokenInterface::class, $tokens);
$index = 0;
do {
$this->assertInstanceOf($this->textTokenType[$index], $tokens);
$tokens = $tokens->getNextToken();
$index ++;
} while ($tokens->hasNextToken());
}
}