diff --git a/README.md b/README.md
index 07d5b7f..163bf83 100644
--- a/README.md
+++ b/README.md
@@ -11,10 +11,11 @@
## src目录解析
-- `Grammar` 语法解析器，验证语法及拆分语法
+- `AST` 语法树
+- `Grammar` 语法器，提供Token识别的规则及种类
- `Output` 输出器，将解析的内容输出（并不执行）
-- `Lexer` 词法解析器，将语义转换成正常的语法供`Grammar`使用
-- `Reader` 读取器，供`Tokenizer`读取使用
+- `Lexer` 词法分析器，二次解析`Token`供`Tokenizer`读取使用
+- `Parser` 语法解析器，生成语法树
+- `Reader` 读取器，供`Lexer`读取使用
- `Token` 所有的Token
-- `Token/Factory` Token工厂，生成Token用的
- `Tokenizer` 转换成Token用的
diff --git a/src/Token/Factory/DefaultFactory.php b/src/Grammar/DefaultGrammar.php
similarity index 73%
rename from src/Token/Factory/DefaultFactory.php
rename to src/Grammar/DefaultGrammar.php
index 51a6764..b28be9f 100644
--- a/src/Token/Factory/DefaultFactory.php
+++ b/src/Grammar/DefaultGrammar.php
@@ -1,12 +1,12 @@
<?php
/**
- * @filename DefaultFactory.php
+ * @filename DefaultGrammar.php
* @author Jerry Yan <792602257@qq.com>
- * @date 2020/12/17 14:48
+ * @date 2020/12/17 14:13
*/
-namespace JerryYan\DSL\Token\Factory;
+namespace JerryYan\DSL\Grammar;
use JerryYan\DSL\Token\Token;
@@ -25,9 +25,9 @@ use JerryYan\DSL\Token\TokenNumber;
use JerryYan\DSL\Token\TokenUseVariable;
use JerryYan\DSL\Token\TokenVariable;
-class DefaultFactory extends FactoryInterface
+class DefaultGrammar implements GrammarInterface
{
- protected $tokenMap = [
+ protected $tokens = [
Token::FAKE => TokenFake::class,
Token::CURRY => TokenCurry::class,
Token::LOGIC_AND => TokenLogicAnd::class,
@@ -44,5 +44,21 @@ class DefaultFactory extends FactoryInterface
Token::USE_VARIABLE => TokenUseVariable::class,
];
- protected $undefinedTokenClass = TokenVariable::class;
+ protected $undefinedTokenType = TokenVariable::class;
+
+ /**
+ * @inheritDoc
+ */
+ public function getTokenTypes(): array
+ {
+ return $this->tokens;
+ }
+
+ /**
+ * @inheritDoc
+ */
+ public function getUndefinedTokenType(): string
+ {
+ return $this->undefinedTokenType;
+ }
}
\ No newline at end of file
diff --git a/src/Grammar/Grammar.php b/src/Grammar/Grammar.php
deleted file mode 100644
index fcb279e..0000000
--- a/src/Grammar/Grammar.php
+++ /dev/null
@@ -1,15 +0,0 @@
-<?php
-/**
- * @filename Grammar.php
- * @author Jerry Yan <792602257@qq.com>
- * @date 2020/12/17 14:13
- */
-
-
-namespace JerryYan\DSL\Grammar;
-
-
-class Grammar extends GrammarInterface
-{
-
-}
\ No newline at end of file
diff --git a/src/Grammar/GrammarInterface.php b/src/Grammar/GrammarInterface.php
index 79ff66a..43ab48a 100644
--- a/src/Grammar/GrammarInterface.php
+++ b/src/Grammar/GrammarInterface.php
@@ -9,7 +9,19 @@
namespace JerryYan\DSL\Grammar;
-abstract class GrammarInterface
+interface GrammarInterface
{
+ /**
+ * @return array<string, class-string>
+ * @author Jerry Yan <792602257@qq.com>
+ * @date 2021/1/26 16:27
+ */
+ public function getTokenTypes(): array;
+ /**
+ * @return class-string
+ * @author Jerry Yan <792602257@qq.com>
+ * @date 2021/1/26 16:27
+ */
+ public function getUndefinedTokenType(): string;
}
\ No newline at end of file
diff --git a/src/Token/Factory/FactoryInterface.php b/src/Token/Factory/FactoryInterface.php
deleted file mode 100644
index dd36590..0000000
--- a/src/Token/Factory/FactoryInterface.php
+++ /dev/null
@@ -1,62 +0,0 @@
-<?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, class-string<TokenInterface>> Token类型及映射类 */
- protected $tokenMap = [];
- /** @var array<string, class-string<TokenInterface>> Token别名映射 */
- protected $tokenNameMap = [];
- /** @var array<string, class-string<TokenInterface>> Token别名映射 */
- protected $regexNameMap = [];
- /** @var class-string<TokenInterface> 默认Token类 */
- protected $undefinedTokenClass = TokenUndefined::class;
-
- public function __construct()
- {
- /**
- * @var string $key
- * @var TokenInterface $token
- */
- foreach ($this->tokenMap as $key=>$token) {
- foreach ($token::$alias as $name) {
- $this->tokenNameMap[$name] = $key;
- }
- foreach ($token::$regexAlias as $name) {
- $this->regexNameMap[$name] = $key;
- }
- }
- }
-
- public function getTokenByName(string $name): TokenInterface
- {
- $originalName = $name;
- if (isset($this->tokenNameMap[$name])) {
- $name = $this->tokenNameMap[$name];
- } else {
- foreach ($this->regexNameMap as $regex => $newName) {
- if (preg_match($regex, $name) === 1) {
- $name = $newName; break;
- }
- }
- }
- if (!isset($this->tokenMap[$name])) {
- return new $this->undefinedTokenClass($originalName);
- } else {
- return new $this->tokenMap[$name]($originalName);
- }
- }
-
-}
\ No newline at end of file
diff --git a/src/Tokenizer/Tokenizer.php b/src/Tokenizer/Tokenizer.php
index 9430c1e..2b4257b 100644
--- a/src/Tokenizer/Tokenizer.php
+++ b/src/Tokenizer/Tokenizer.php
@@ -9,17 +9,37 @@
namespace JerryYan\DSL\Tokenizer;
+use JerryYan\DSL\Grammar\GrammarInterface;
use JerryYan\DSL\Reader\ReaderInterface;
-use JerryYan\DSL\Token\Factory\FactoryInterface;
use JerryYan\DSL\Token\TokenInterface;
class Tokenizer extends TokenizerInterface
{
- /** @var FactoryInterface token工厂 */
- protected $tokenFactory;
- public function __construct(FactoryInterface $tokenFactory)
+ /** @var array<string, class-string<TokenInterface>> Token类型及映射类 */
+ protected $tokenMapping = [];
+ /** @var array<string, class-string<TokenInterface>> Token别名映射 */
+ protected $tokenNameMapping = [];
+ /** @var array<string, class-string<TokenInterface>> Token别名映射 */
+ protected $regexNameMapping = [];
+ /** @var GrammarInterface 语法器 */
+ protected $grammar;
+
+ public function __construct(GrammarInterface $grammar)
{
- $this->tokenFactory = $tokenFactory;
+ $this->grammar = $grammar;
+ /**
+ * @var string $key
+ * @var TokenInterface $token
+ */
+ foreach ($grammar->getTokenTypes() as $key=>$token) {
+ $this->tokenMapping[$key] = $token;
+ foreach ($token::$alias as $name) {
+ $this->tokenNameMapping[$name] = $key;
+ }
+ foreach ($token::$regexAlias as $name) {
+ $this->regexNameMapping[$name] = $key;
+ }
+ }
}
/**
@@ -33,11 +53,38 @@ class Tokenizer extends TokenizerInterface
do
{
$currentTokenName = $reader->getCurrentToken();
- $currentToken = $this->tokenFactory->getTokenByName($currentTokenName);
+ $currentToken = $this->getTokenByName($currentTokenName);
$currentToken->setPrevToken($lastToken);
if ($lastToken !== NULL) $lastToken->setNextToken($currentToken);
$lastToken = $currentToken;
}while($reader->moveToNextToken());
return $lastToken->getFirstToken();
}
+
+ /**
+ * 根据文字获取Token
+ * @param string $name
+ * @return TokenInterface
+ * @author Jerry Yan <792602257@qq.com>
+ * @date 2021/1/26 16:41
+ */
+ protected function getTokenByName(string $name): TokenInterface
+ {
+ $originalName = $name;
+ if (isset($this->tokenNameMapping[$name])) {
+ $name = $this->tokenNameMapping[$name];
+ } else {
+ foreach ($this->regexNameMapping as $regex => $newName) {
+ if (preg_match($regex, $name) === 1) {
+ $name = $newName; break;
+ }
+ }
+ }
+ if (!isset($this->tokenMapping[$name])) {
+ $undefinedType = $this->grammar->getUndefinedTokenType();
+ return new $undefinedType($originalName);
+ } else {
+ return new $this->tokenMapping[$name]($originalName);
+ }
+ }
}
\ No newline at end of file
diff --git a/tests/Tokenizer/TokenizerTest.php b/tests/Tokenizer/TokenizerTest.php
index 8a18a2b..3dc7154 100644
--- a/tests/Tokenizer/TokenizerTest.php
+++ b/tests/Tokenizer/TokenizerTest.php
@@ -7,8 +7,8 @@
namespace JerryYan\DSL\Test\Tokenizer;
+use JerryYan\DSL\Grammar\DefaultGrammar;
use JerryYan\DSL\Reader\StringReader;
-use JerryYan\DSL\Token\Factory\DefaultFactory;
use JerryYan\DSL\Token\TokenCurry;
use JerryYan\DSL\Token\TokenLogicAnd;
use JerryYan\DSL\Token\TokenInterface;
@@ -43,7 +43,7 @@ class TokenizerTest extends TestCase
];
protected function setUp(): void
{
- $this->tokenizer = new Tokenizer(new DefaultFactory());
+ $this->tokenizer = new Tokenizer(new DefaultGrammar());
$this->reader = new StringReader($this->text);
}