Add conditional types support and parser features

Kirill Nesmeyanov · Kirill Nesmeyanov · commit 8aea70f85287 · 2023-11-23T22:55:20.000+03:00
diff --git a/bin/build b/bin/build
@@ -37,6 +37,7 @@ $grammar = (new Compiler())
         ->withClassUsage('TypeLang\\Parser\\Node')
         ->withClassUsage('TypeLang\\Parser\\Exception')
         ->withClassUsage('TypeLang\\Parser\\Exception\\SemanticException')
+        ->withClassUsage('TypeLang\\Parser\\Exception\\FeatureNotAllowedException')
 ;
 
 file_put_contents(__DIR__ . '/../resources/grammar.php', $grammar->generate());
diff --git a/composer.json b/composer.json
@@ -13,8 +13,8 @@
         "psr/cache": "^1.0|^2.0|^3.0",
         "psr/simple-cache": "^1.0|^2.0|^3.0",
         "psr/log": "^1.0|^2.0|^3.0",
-        "phplrt/lexer": "^3.4",
-        "phplrt/parser": "^3.4"
+        "phplrt/lexer": "^3.5",
+        "phplrt/parser": "^3.5"
     },
     "autoload": {
         "psr-4": {
@@ -25,7 +25,7 @@
         "nikic/php-parser": "^4.17",
         "phpdocumentor/reflection-docblock": "^5.3",
         "friendsofphp/php-cs-fixer": "^3.35",
-        "phplrt/compiler": "^3.4",
+        "phplrt/compiler": "^3.5",
         "phpunit/phpunit": "^10.4",
         "rector/rector": "^0.18",
         "vimeo/psalm": "^5.15"
diff --git a/resources/grammar.php b/resources/grammar.php
diff --git a/resources/grammar.pp2 b/resources/grammar.pp2
@@ -12,6 +12,7 @@
 %include grammar/callable               // callable(mixed): void
 %include grammar/shape-fields           // array { key: int, ...<TKey, TValue> }
 %include grammar/named-type             // Map<int<0, max>, non-empty-string>
+%include grammar/ternary                // T is A ? B : C
 
 %pragma root Type
 
@@ -26,9 +27,25 @@
  */
 
 #Type
-  : LogicalType()
+  : Expression()
   ;
 
+/**
+ * -----------------------------------------------------------------------------
+ *  Ternary Expression
+ * -----------------------------------------------------------------------------
+ *
+ *  Ternary conditional expressions, like:
+ *  - T is A ? B : C - for equality type conditions.
+ *  - T is not A ? B ? C - for non-equality type conditions.
+ *
+ */
+
+Expression
+  : TernaryExpressionOrLogicalType()
+  ;
+
+
 /**
  * -----------------------------------------------------------------------------
  *  Logical Statements
@@ -89,7 +106,7 @@ PrefixedNullableType -> {
 
     return $children;
 }
-  : <T_NULLABLE> TypesList()
+  : <T_QMARK> TypesList()
   | TypesList()
   ;
 
diff --git a/resources/grammar/callable.pp2 b/resources/grammar/callable.pp2
@@ -2,6 +2,10 @@
 CallableType -> {
     $name = \array_shift($children);
 
+    if ($this->callables === false) {
+        throw FeatureNotAllowedException::fromFeature('callable types', $offset);
+    }
+
     $parameters = isset($children[0]) && $children[0] instanceof Node\Stmt\Callable\ParametersListNode
         ? \array_shift($children)
         : new Node\Stmt\Callable\ParametersListNode();
diff --git a/resources/grammar/common.pp2 b/resources/grammar/common.pp2
@@ -20,6 +20,7 @@ Identifier -> {
     return new Node\Identifier($children->getValue());
 }
   : <T_NAME>
+  | <T_EQ>
   | <T_BOOL_LITERAL>
   | <T_NULL_LITERAL>
   ;
diff --git a/resources/grammar/lexemes.pp2 b/resources/grammar/lexemes.pp2
@@ -15,7 +15,9 @@
 
 // Identifier
 
-%token  T_THIS                  \$this
+%token  T_THIS                  \$this(?![a-zA-Z0-9\-_\x80-\xff])
+%token  T_EQ                    (?i)is(?![a-zA-Z0-9\-_\x80-\xff])
+%token  T_NEQ                   (?i)is\h+not(?![a-zA-Z0-9\-_\x80-\xff])
 %token  T_VARIABLE              \$[a-zA-Z_\x80-\xff][a-zA-Z0-9\-_\x80-\xff]*
 %token  T_NAME                  [a-zA-Z_\x80-\xff][a-zA-Z0-9\-_\x80-\xff]*
 
@@ -36,7 +38,7 @@
 %token  T_COLON                 :
 %token  T_ASSIGN                =
 %token  T_NS_DELIMITER          \\
-%token  T_NULLABLE              \?
+%token  T_QMARK                 \?
 %token  T_NOT                   \!
 %token  T_OR                    \|
 %token  T_AMP                   &
diff --git a/resources/grammar/literals.pp2 b/resources/grammar/literals.pp2
@@ -1,12 +1,21 @@
 
 Literal
+  : RawLiteral()
+  | ConstMaskLiteral()
+  | ClassConstLiteral()
+  ;
+
+RawLiteral -> {
+    if ($this->literals === false) {
+        throw FeatureNotAllowedException::fromFeature('literal values', $offset);
+    }
+    return $children;
+}
   : StringLiteral()
   | FloatLiteral()
   | IntLiteral()
   | BoolLiteral()
   | NullLiteral()
-  | ConstMaskLiteral()
-  | ClassConstLiteral()
   ;
 
 VariableLiteral -> {
diff --git a/resources/grammar/shape-fields.pp2 b/resources/grammar/shape-fields.pp2
@@ -4,6 +4,10 @@ ShapeFields -> {
         return new Node\Stmt\Shape\FieldsListNode();
     }
 
+    if ($this->shapes === false) {
+        throw FeatureNotAllowedException::fromFeature('shape fields', $offset);
+    }
+
     $parameters = null;
 
     if (\end($children) instanceof Node\Stmt\Template\ArgumentsListNode) {
@@ -90,7 +94,7 @@ ExplicitField -> {
         default => new Node\Stmt\Shape\NamedFieldNode($name, $value, $optional),
     };
 }
-  : ShapeKey() (<T_NULLABLE>)? ::T_COLON:: ShapeValue()
+  : ShapeKey() (<T_QMARK>)? ::T_COLON:: ShapeValue()
   ;
 
 ImplicitField -> {
diff --git a/resources/grammar/ternary.pp2 b/resources/grammar/ternary.pp2
@@ -0,0 +1,47 @@
+
+TernaryExpressionOrLogicalType -> {
+    $count = \count($children);
+
+    if ($count === 1) {
+        return $children[0];
+    }
+
+    if ($this->conditional === false) {
+        throw FeatureNotAllowedException::fromFeature('conditional expressions', $offset);
+    }
+
+    $condition = match ($children[1]->getName()) {
+        'T_EQ' => new Node\Stmt\Condition\EqualConditionNode(
+            $children[0],
+            $children[2],
+        ),
+        'T_NEQ' => new Node\Stmt\Condition\NotEqualConditionNode(
+            $children[0],
+            $children[2],
+        ),
+        default => throw new SemanticException(
+            \sprintf('Invalid conditional operator "%s"', $children[1]->getValue()),
+            $offset,
+            SemanticException::CODE_INVALID_OPERATOR,
+        ),
+    };
+
+    return new Node\Stmt\TernaryConditionNode(
+        $condition,
+        $children[3],
+        $children[4],
+    );
+}
+  : LogicalType() OptionalTernaryExpressionSuffix()
+  | VariableLiteral() TernaryExpressionSuffix()
+  ;
+
+OptionalTernaryExpressionSuffix
+  : TernaryExpressionSuffix()?
+  ;
+
+TernaryExpressionSuffix
+  : (<T_EQ>|<T_NEQ>) (Type()|VariableLiteral())
+    ::T_QMARK:: Type()
+    ::T_COLON:: Type()
+  ;
diff --git a/src/Builder.php b/src/Builder.php
@@ -31,11 +31,7 @@ public function build(Context $context, mixed $result): mixed
         $result = ($this->reducers[$context->state])($context, $result);
 
         if ($result instanceof Node && $result->offset === 0) {
-            $processed = $context->lastProcessedToken;
-            $ordinal = $context->lastOrdinalToken;
-
-            $result->offset = $processed->getOffset();
-            $result->offsetTo = $ordinal?->getOffset() ?? $result->offset;
+            $result->offset = $context->lastProcessedToken->getOffset();
         }
 
         return $result;
diff --git a/src/Exception/FeatureNotAllowedException.php b/src/Exception/FeatureNotAllowedException.php
@@ -0,0 +1,15 @@
+<?php
+
+declare(strict_types=1);
+
+namespace TypeLang\Parser\Exception;
+
+final class FeatureNotAllowedException extends SemanticException
+{
+    public static function fromFeature(string $name, int $offset = 0): self
+    {
+        $message = \sprintf('%s not allowed by parser configuration', $name);
+
+        return new self($message, $offset);
+    }
+}
diff --git a/src/Exception/SemanticException.php b/src/Exception/SemanticException.php
@@ -12,7 +12,9 @@ class SemanticException extends \LogicException implements ParserExceptionInterf
 
     final public const CODE_VARIADIC_WITH_DEFAULT = 0x03;
 
-    public const CODE_LAST = self::CODE_VARIADIC_WITH_DEFAULT;
+    final public const CODE_INVALID_OPERATOR = 0x04;
+
+    public const CODE_LAST = self::CODE_INVALID_OPERATOR;
 
     /**
      * @param int<0, max> $offset
diff --git a/src/Node/Node.php b/src/Node/Node.php
@@ -11,11 +11,6 @@ abstract class Node implements NodeInterface
      */
     public int $offset = 0;
 
-    /**
-     * @var int<0, max>
-     */
-    public int $offsetTo = 0;
-
     public function getOffset(): int
     {
         return $this->offset;
diff --git a/src/Node/Stmt/Condition/Condition.php b/src/Node/Stmt/Condition/Condition.php
@@ -0,0 +1,16 @@
+<?php
+
+declare(strict_types=1);
+
+namespace TypeLang\Parser\Node\Stmt\Condition;
+
+use TypeLang\Parser\Node\Statement;
+use TypeLang\Parser\Node\Stmt\TypeStatement;
+
+abstract class Condition extends Statement
+{
+    public function __construct(
+        public readonly TypeStatement $subject,
+        public readonly TypeStatement $target,
+    ) {}
+}
diff --git a/src/Node/Stmt/Condition/EqualConditionNode.php b/src/Node/Stmt/Condition/EqualConditionNode.php
@@ -0,0 +1,7 @@
+<?php
+
+declare(strict_types=1);
+
+namespace TypeLang\Parser\Node\Stmt\Condition;
+
+final class EqualConditionNode extends Condition {}
diff --git a/src/Node/Stmt/Condition/NotEqualConditionNode.php b/src/Node/Stmt/Condition/NotEqualConditionNode.php
@@ -0,0 +1,7 @@
+<?php
+
+declare(strict_types=1);
+
+namespace TypeLang\Parser\Node\Stmt\Condition;
+
+final class NotEqualConditionNode extends Condition {}
diff --git a/src/Node/Stmt/TernaryConditionNode.php b/src/Node/Stmt/TernaryConditionNode.php
@@ -0,0 +1,16 @@
+<?php
+
+declare(strict_types=1);
+
+namespace TypeLang\Parser\Node\Stmt;
+
+use TypeLang\Parser\Node\Stmt\Condition\Condition;
+
+final class TernaryConditionNode extends TypeStatement
+{
+    public function __construct(
+        public readonly Condition $condition,
+        public readonly TypeStatement $then,
+        public readonly TypeStatement $else,
+    ) {}
+}
diff --git a/src/Parser.php b/src/Parser.php
@@ -56,10 +56,19 @@ final class Parser implements ParserInterface
      */
     private readonly \WeakMap $integerPool;
 
-    private readonly BuilderInterface $builer;
+    private readonly BuilderInterface $builder;
+
+    /**
+     * @var int<0, max>
+     */
+    public int $lastProcessedTokenOffset = 0;
 
     public function __construct(
-        private readonly bool $tolerant = false,
+        public readonly bool $tolerant = false,
+        public readonly bool $conditional = true,
+        public readonly bool $shapes = true,
+        public readonly bool $callables = true,
+        public readonly bool $literals = true,
     ) {
         /** @psalm-var GrammarConfigArray $grammar */
         $grammar = require __DIR__ . '/../resources/grammar.php';
@@ -70,7 +79,7 @@ public function __construct(
         /** @var \WeakMap<TokenInterface, IntLiteralNode> */
         $this->integerPool = new \WeakMap();
 
-        $this->builer = new Builder($grammar['reducers']);
+        $this->builder = new Builder($grammar['reducers']);
         $this->lexer = $this->createLexer($grammar);
         $this->parser = $this->createParser($this->lexer, $grammar);
     }
@@ -85,7 +94,7 @@ private function createParser(LexerInterface $lexer, array $grammar): ParserComb
             grammar: $grammar['grammar'],
             options: [
                 ParserConfigsInterface::CONFIG_INITIAL_RULE => $grammar['initial'],
-                ParserConfigsInterface::CONFIG_AST_BUILDER => $this->builer,
+                ParserConfigsInterface::CONFIG_AST_BUILDER => $this->builder,
                 ParserConfigsInterface::CONFIG_ALLOW_TRAILING_TOKENS => $this->tolerant,
             ],
         );
@@ -122,6 +131,11 @@ public function parse(#[Language('PHP')] mixed $source): ?TypeStatement
             try {
                 foreach ($this->parser->parse($source) as $stmt) {
                     if ($stmt instanceof TypeStatement) {
+                        $context = $this->parser->getLastExecutionContext();
+                        $token = $context->buffer->current();
+
+                        $this->lastProcessedTokenOffset = $token->getOffset();
+
                         return $stmt;
                     }
                 }
diff --git a/tests/Functional/LinterStubs/LinterStubsTestCase.php b/tests/Functional/LinterStubs/LinterStubsTestCase.php
@@ -120,13 +120,13 @@ protected function onFail(\Throwable $e, string $expr, string $location): void
         //
         // Known issues in phpdoc
         //
-        if (
+        if (true
             // Non-const expression in typedef (will not support)
-            \str_contains($expr, 'func_num_args() > ') ||
+            || \str_contains($expr, 'func_num_args() > ')
             // Conditional types with variables not supported
-            \str_starts_with($e->getMessage(), 'Syntax error, unexpected "$') ||
+            // || \str_starts_with($e->getMessage(), 'Syntax error, unexpected "$')
             // Conditional types with template params not supported
-            \str_starts_with($e->getMessage(), 'Syntax error, unexpected "is"')
+            // || \str_starts_with($e->getMessage(), 'Syntax error, unexpected "is"')
         ) {
             $this->markTestIncomplete("Test is flagged as a known issue:\n" . $message);
         }

Original file line number	Diff line number	Diff line change
`@@ -37,6 +37,7 @@ $grammar = (new Compiler())`
`37`	`37`	`->withClassUsage('TypeLang\\Parser\\Node')`
`38`	`38`	`->withClassUsage('TypeLang\\Parser\\Exception')`
`39`	`39`	`->withClassUsage('TypeLang\\Parser\\Exception\\SemanticException')`
	`40`	`+ ->withClassUsage('TypeLang\\Parser\\Exception\\FeatureNotAllowedException')`
`40`	`41`	`;`
`41`	`42`
`42`	`43`	`file_put_contents(__DIR__ . '/../resources/grammar.php', $grammar->generate());`
Original file line number	Diff line number	Diff line change
`@@ -20,6 +20,7 @@ Identifier -> {`
`20`	`20`	`return new Node\Identifier($children->getValue());`
`21`	`21`	`}`
`22`	`22`	`: <T_NAME>`
	`23`	`+ \| <T_EQ>`
`23`	`24`	`\| <T_BOOL_LITERAL>`
`24`	`25`	`\| <T_NULL_LITERAL>`
`25`	`26`	`;`
Original file line number	Diff line number	Diff line change
`@@ -4,6 +4,10 @@ ShapeFields -> {`
`4`	`4`	`return new Node\Stmt\Shape\FieldsListNode();`
`5`	`5`	`}`
`6`	`6`
	`7`	`+ if ($this->shapes === false) {`
	`8`	`+ throw FeatureNotAllowedException::fromFeature('shape fields', $offset);`
	`9`	`+ }`
	`10`	`+`
`7`	`11`	`$parameters = null;`
`8`	`12`
`9`	`13`	`if (\end($children) instanceof Node\Stmt\Template\ArgumentsListNode) {`
`@@ -90,7 +94,7 @@ ExplicitField -> {`
`90`	`94`	`default => new Node\Stmt\Shape\NamedFieldNode($name, $value, $optional),`
`91`	`95`	`};`
`92`	`96`	`}`
`93`		`- : ShapeKey() (<T_NULLABLE>)? ::T_COLON:: ShapeValue()`
	`97`	`+ : ShapeKey() (<T_QMARK>)? ::T_COLON:: ShapeValue()`
`94`	`98`	`;`
`95`	`99`
`96`	`100`	`ImplicitField -> {`