Parser - Type checking

opast.hh

@@ -1,6 +1,7 @@
 #pragma once
 #include "odbg.hh"
 #include <ebcl/SRDData.hh>
+#include <ebcl/Sets.hh>
 
 
 namespace opast {
@@ -1152,11 +1153,21 @@ class T_BinaryOperatorNode : public A_ExpressionNode
 
 /*= PARSER ===================================================================*/
 
+// Parser output. Consists in a root node as well as other details (table of
+// constants, data types, etc...)
+struct T_ParserOutput
+{
+	T_RootNode root;
+	T_KeyValueTable< T_String , E_DataType > types;
+	T_Set< float > constants{ UseTag< IndexBacked< > >( ) };
+};
+
+// The actual parser
 class T_Parser : public A_PrivateImplementation
 {
     private:
 	T_Array< T_SRDError > errors_;
-	T_OwnPtr< T_RootNode > rootNode_;
+	T_OwnPtr< T_ParserOutput > output_;
 
     public:
 	T_Parser( ) noexcept;
@@ -1165,8 +1176,8 @@ class T_Parser : public A_PrivateImplementation
 
 	T_Array< T_SRDError > const& errors( ) const noexcept
 		{ return errors_; }
-	T_OwnPtr< T_RootNode > result( ) noexcept
+	T_OwnPtr< T_ParserOutput > result( ) noexcept
-		{ return std::move( rootNode_ ); }
+		{ return std::move( output_ ); }
 };
 
 

opparser.cc

@@ -147,7 +147,7 @@ struct T_ParserImpl_
 
 	// ---------------------------------------------------------------------
 
-	T_OwnPtr< T_RootNode >& root;
+	T_OwnPtr< T_ParserOutput >& output;
 	T_Array< T_SRDError >& errors;
 	T_Visitor< A_Node > visitor{ opast::ASTVisitorBrowser };
 	T_MultiArray< uint32_t > calls;
@@ -161,18 +161,26 @@ struct T_ParserImpl_
 			} };
 
 	T_ParserImpl_( T_Array< T_SRDError >* errors ,
-			T_OwnPtr< T_RootNode >* root ) noexcept;
+			T_OwnPtr< T_ParserOutput >* root ) noexcept;
 
 	// ---------------------------------------------------------------------
 
-	void parseTopLevel( T_SRDList const& list ) noexcept;
+	void main( T_SRDList const& list ) noexcept;
+
+	bool checkRequiredBlocks( T_SRDList const& list ) noexcept;
+	bool checkCalls( ) noexcept;
+	bool checkInstructionRestrictions( ) noexcept;
+	bool collectGlobalTypes( ) noexcept;
+
+	// ---------------------------------------------------------------------
+
+	bool parseTopLevel( T_SRDList const& list ) noexcept;
+
 	void parseFunction( T_SRDList const& funcList ) noexcept;
 	void parseFunctionArguments(
 			T_FuncNode& function ,
 			T_SRDToken const& argsToken ) noexcept;
 
-	// ---------------------------------------------------------------------
-
 	void parseInstructions(
 			T_InstrListNode& instructions ,
 			T_SRDList const& input ,
@@ -263,48 +271,47 @@ struct T_ParserImpl_
 
 inline T_ParserImpl_::T_ParserImpl_(
 		T_Array< T_SRDError >* const errors ,
-		T_OwnPtr< T_RootNode >* const root ) noexcept
+		T_OwnPtr< T_ParserOutput >* const output ) noexcept
-	: root( *root ) , errors( *errors )
+	: output( *output ) , errors( *errors )
 { }
 
+void T_ParserImpl_::main(
+		T_SRDList const& input ) noexcept
+{
+	parseTopLevel( input )
+		&& checkRequiredBlocks( input )
+		&& checkCalls( )
+		&& checkInstructionRestrictions( )
+		&& collectGlobalTypes( );
+}
+
 /*----------------------------------------------------------------------------*/
 
-void T_ParserImpl_::parseTopLevel(
+bool T_ParserImpl_::checkRequiredBlocks(
 		T_SRDList const& input ) noexcept
 {
-	for ( auto const& t : input ) {
-		if ( t.type( ) == E_SRDTokenType::LIST && t.list( ).size( ) > 0 ) {
-			parseFunction( t.list( ) );
-		} else {
-			errors.addNew( "function, init or frame list expected" ,
-					t.location( ) );
-		}
-	}
-	if ( !errors.empty( ) ) {
-		return;
-	}
-
 	const T_SRDLocation missingErrLoc( ([&input]() {
 		if ( input.size( ) != 0 ) {
 			return T_SRDLocation( input[ 0 ].location( ).source( ) , 1 , 1 );
 		}
 		return T_SRDLocation{};
 	})( ));
-	if ( !root->hasInit( ) ) {
+	if ( !output->root.hasInit( ) ) {
 		errors.addNew( "no initialisation block" , missingErrLoc );
 	}
-	if ( !root->hasFrame( ) ) {
+	if ( !output->root.hasFrame( ) ) {
-		errors.addNew( "no initialisation block" , missingErrLoc );
+		errors.addNew( "no frame rendering block" , missingErrLoc );
 	}
-	if ( !errors.empty( ) ) {
+	return errors.empty( );
-		return;
 }
 
+bool T_ParserImpl_::checkCalls( ) noexcept
+{
 	calls.clear( );
 	uint32_t cfi;
-	for ( cfi = 0 ; cfi < root->nFunctions( ) ; cfi ++ ) {
+	for ( cfi = 0 ; cfi < output->root.nFunctions( ) ; cfi ++ ) {
 		calls.next( );
-		visitor.visit( root->function( cfi ) , [&]( A_Node& node , bool exit ) -> bool {
+		visitor.visit( output->root.function( cfi ) , [&]( A_Node& node , bool exit ) -> bool {
 			if ( exit || dynamic_cast< A_ExpressionNode* >( &node ) ) {
 				return false;
 			}
@@ -313,14 +320,14 @@ void T_ParserImpl_::parseTopLevel(
 			}
 
 			auto& call( (T_CallInstrNode&) node );
-			const auto callee( root->functionIndex( call.id( ) ) );
+			const auto callee( output->root.functionIndex( call.id( ) ) );
 			if ( callee >= 0 ) {
 				if ( !calls.contains( cfi , callee ) ) {
 					calls.add( (uint32_t) callee );
 				}
 
 				// Check argument count while we're at it
-				auto& fn( (T_FuncNode&) root->function( callee ) );
+				auto& fn( (T_FuncNode&) output->root.function( callee ) );
 				if ( fn.arguments( ) != call.arguments( ) ) {
 					T_StringBuilder sb;
 					sb << "function expects " << fn.arguments( )
@@ -336,12 +343,13 @@ void T_ParserImpl_::parseTopLevel(
 			return false;
 		} );
 	}
-	if ( !errors.empty( ) ) {
+	return errors.empty( );
-		return;
 }
 
+bool T_ParserImpl_::checkInstructionRestrictions( ) noexcept
+{
 	T_InstrRestriction callInfo[ calls.size( ) ];
-	callGraphVisitor.visit( root->functionIndex( "*init*" ) ,
+	callGraphVisitor.visit( output->root.functionIndex( "*init*" ) ,
 		[&]( uint32_t id , const bool exit ) -> bool {
 			if ( exit || callInfo[ id ] & E_InstrRestriction::INIT ) {
 				return false;
@@ -349,7 +357,7 @@ void T_ParserImpl_::parseTopLevel(
 			callInfo[ id ] |= E_InstrRestriction::INIT;
 			return true;
 		} );
-	callGraphVisitor.visit( root->functionIndex( "*frame*" ) ,
+	callGraphVisitor.visit( output->root.functionIndex( "*frame*" ) ,
 		[&]( uint32_t id , const bool exit ) -> bool {
 			if ( exit || callInfo[ id ] & E_InstrRestriction::FRAME ) {
 				return false;
@@ -357,8 +365,8 @@ void T_ParserImpl_::parseTopLevel(
 			callInfo[ id ] |= E_InstrRestriction::FRAME;
 			return true;
 		} );
-	for ( auto i = 0u ; i < root->nFunctions( ) ; i ++ ) {
+	for ( auto i = 0u ; i < output->root.nFunctions( ) ; i ++ ) {
-		visitor.visit( root->function( i ) ,
+		visitor.visit( output->root.function( i ) ,
 			[&]( A_Node& node , bool exit ) {
 				if ( exit ) {
 					return false;
@@ -374,6 +382,96 @@ void T_ParserImpl_::parseTopLevel(
 				return true;
 			} );
 	}
+	return errors.empty( );
+}
+
+bool T_ParserImpl_::collectGlobalTypes( ) noexcept
+{
+	struct T_Decl_ {
+		E_DataType type;
+		T_SRDLocation location;
+
+		T_Decl_( const E_DataType dt ,
+				T_SRDLocation const& loc ) noexcept
+			: type( dt ) , location( loc )
+		{ }
+
+		T_Decl_( const E_DataType dt ) noexcept
+			: type( dt ) , location{}
+		{ }
+	};
+	T_KeyValueTable< T_String , T_Decl_ > type;
+	type.add( T_String::Pooled( "time" ) , T_Decl_{ E_DataType::BUILTIN } );
+	type.add( T_String::Pooled( "width" ) , T_Decl_{ E_DataType::BUILTIN } );
+	type.add( T_String::Pooled( "height" ) , T_Decl_{ E_DataType::BUILTIN } );
+
+	visitor.visit( output->root , [&]( A_Node& node , bool exit ) -> bool {
+		if ( exit ) {
+			return false;
+		}
+
+		E_DataType dt{ E_DataType::UNKNOWN };
+		T_String id;
+		T_SRDLocation location;
+		if ( node.type( ) == A_Node::OP_SET ) {
+			dt = E_DataType::VARIABLE;
+			id = dynamic_cast< T_SetInstrNode& >( node ).id( );
+			location = dynamic_cast< T_SetInstrNode& >( node ).idLocation( );
+		} else {
+			auto* np{ dynamic_cast< A_ResourceDefInstrNode* >( &node ) };
+			if ( !np ) {
+				return !dynamic_cast< A_ExpressionNode* >( &node );
+			}
+			dt = np->dataType( );
+			id = np->id( );
+			location = np->idLocation( );
+		}
+		assert( dt != E_DataType::UNKNOWN );
+
+		T_Decl_ const* const existing( type.get( id ) );
+		if ( !existing ) {
+			type.add( id , T_Decl_{ dt , location } );
+			return false;
+		}
+		if ( existing->type == dt ) {
+			return false;
+		}
+
+		T_StringBuilder sb;
+		if ( existing->type == E_DataType::BUILTIN ) {
+			sb << "trying to redefine built-in variable " << id;
+		} else {
+			sb << "'" << id << "' redeclared as " << dt
+				<< "; previous declaration as "
+				<< existing->type << " at "
+				<< existing->location;
+		}
+		errors.addNew( std::move( sb ) , location );
+
+		return false;
+	} );
+
+	for ( auto const& k : type.keys( ) ) {
+		output->types.add( k , type.get( k )->type );
+	}
+
+	return errors.empty( );
+}
+
+/*------------------------------------------------------------------------------*/
+
+bool T_ParserImpl_::parseTopLevel(
+		T_SRDList const& input ) noexcept
+{
+	for ( auto const& t : input ) {
+		if ( t.type( ) == E_SRDTokenType::LIST && t.list( ).size( ) > 0 ) {
+			parseFunction( t.list( ) );
+		} else {
+			errors.addNew( "function, init or frame list expected" ,
+					t.location( ) );
+		}
+	}
+	return errors.empty( );
 }
 
 void T_ParserImpl_::parseFunction(
@@ -402,15 +500,15 @@ void T_ParserImpl_::parseFunction(
 			return;
 		}
 
-		fn = NewOwned< T_FuncNode >( funcList[ 1 ].stringValue( ) , *root );
+		fn = NewOwned< T_FuncNode >( funcList[ 1 ].stringValue( ) , output->root );
 		parseFunctionArguments( dynamic_cast< T_FuncNode& >( *fn ) ,
 				funcList[ 2 ] );
 	} else {
-		fn = NewOwned< T_SpecialFuncNode >( ftw == "init" , *root );
+		fn = NewOwned< T_SpecialFuncNode >( ftw == "init" , output->root );
 	}
 	fn->location( ) = fw.location( );
 
-	const auto af( root->addFunction( fn ) );
+	const auto af( output->root.addFunction( fn ) );
 	if ( af.dupLocation ) {
 		T_StringBuilder esb( "duplicate " );
 		switch ( fn->type( ) ) {
@@ -1388,15 +1486,15 @@ P_ExpressionNode T_ParserImpl_::parseUnaryOp(
 /*----------------------------------------------------------------------------*/
 
 T_Parser::T_Parser( ) noexcept
-	: A_PrivateImplementation( new T_ParserImpl_( &errors_ , &rootNode_ ) ) ,
+	: A_PrivateImplementation( new T_ParserImpl_( &errors_ , &output_ ) ) ,
-		errors_( 64 ) , rootNode_{}
+		errors_( 64 ) , output_{}
 {}
 
 bool T_Parser::parse(
 		T_SRDList const& input ) noexcept
 {
 	errors_.clear( );
-	rootNode_ = NewOwned< T_RootNode >( );
+	output_ = NewOwned< T_ParserOutput >( );
-	p< T_ParserImpl_ >( ).parseTopLevel( input );
+	p< T_ParserImpl_ >( ).main( input );
 	return errors_.empty( );
 }

parsercheck.cc

@@ -37,75 +37,10 @@ void WriteSRDError(
 /*============================================================================*/
 
 T_Array< T_SRDError > parserPrototypes(
-		T_OwnPtr< T_RootNode > const& root )
+		T_OwnPtr< T_ParserOutput > const& output )
 {
 	T_Visitor< opast::A_Node > visitor{ ASTVisitorBrowser };
 	T_Array< T_SRDError > errors;
-
-	struct T_Decl_ {
-		E_DataType type;
-		T_SRDLocation location;
-
-		T_Decl_( const E_DataType dt ,
-				T_SRDLocation const& loc ) noexcept
-			: type( dt ) , location( loc )
-		{ }
-
-		T_Decl_( const E_DataType dt ) noexcept
-			: type( dt ) , location{}
-		{ }
-	};
-	T_KeyValueTable< T_String , T_Decl_ > type;
-	type.add( T_String::Pooled( "time" ) , T_Decl_{ E_DataType::BUILTIN } );
-	type.add( T_String::Pooled( "width" ) , T_Decl_{ E_DataType::BUILTIN } );
-	type.add( T_String::Pooled( "height" ) , T_Decl_{ E_DataType::BUILTIN } );
-
-	visitor.visit( *root , [&]( A_Node& node , bool exit ) -> bool {
-		if ( exit ) {
-			return false;
-		}
-
-		E_DataType dt{ E_DataType::UNKNOWN };
-		T_String id;
-		T_SRDLocation location;
-		if ( node.type( ) == A_Node::OP_SET ) {
-			dt = E_DataType::VARIABLE;
-			id = dynamic_cast< T_SetInstrNode& >( node ).id( );
-			location = dynamic_cast< T_SetInstrNode& >( node ).idLocation( );
-		} else {
-			auto* np{ dynamic_cast< A_ResourceDefInstrNode* >( &node ) };
-			if ( !np ) {
-				return !dynamic_cast< A_ExpressionNode* >( &node );
-			}
-			dt = np->dataType( );
-			id = np->id( );
-			location = np->idLocation( );
-		}
-		assert( dt != E_DataType::UNKNOWN );
-
-		T_Decl_ const* const existing( type.get( id ) );
-		if ( !existing ) {
-			type.add( id , T_Decl_{ dt , location } );
-			return false;
-		}
-		if ( existing->type == dt ) {
-			return false;
-		}
-
-		T_StringBuilder sb;
-		if ( existing->type == E_DataType::BUILTIN ) {
-			sb << "trying to redefine built-in variable " << id;
-		} else {
-			sb << "'" << id << "' redeclared as " << dt
-				<< "; previous declaration as "
-				<< existing->type << " at "
-				<< existing->location;
-		}
-		errors.addNew( std::move( sb ) , location );
-
-		return false;
-	} );
-
 	return errors;
 }