Split the parser from the code generator

This commit is contained in:
2025-06-11 22:36:35 +02:00
parent 00e557686b
commit e3f094c8a5
7 changed files with 409 additions and 230 deletions

View File

@ -35,9 +35,9 @@ begin
while token.kind <> lexerKindEnd do
INC(field_count);
INC(field_count);
REALLOCATE(field_declarations, TSIZE(AstFieldDeclaration) * field_count);
DEC(field_count);
INC(field_count);
REALLOCATE(field_declarations, TSIZE(AstFieldDeclaration) * field_count);
DEC(field_count);
current_field := field_declarations;
INC(current_field , TSIZE(AstFieldDeclaration) * (field_count - 1));
@ -109,7 +109,7 @@ begin
result^.length := token.integerKind;
token := transpiler_lex(lexer);
token := transpiler_lex(lexer)
end;
token := transpiler_lex(lexer);
result^.base := parse_type_expression(lexer);
@ -143,8 +143,8 @@ begin
REALLOCATE(result^.cases, TSIZE(Identifier) * case_count);
DEC(case_count);
current_case := result^.cases;
INC(current_case, TSIZE(Identifier) * (case_count - 1));
current_case^ := token.identifierKind;
INC(current_case, TSIZE(Identifier) * (case_count - 1));
current_case^ := token.identifierKind;
token := transpiler_lex(lexer)
end;
@ -190,7 +190,7 @@ begin
REALLOCATE(result^.parameters, TSIZE(PAstTypeExpression) * parameter_count);
DEC(parameter_count);
current_parameter := result^.parameters;
INC(current_parameter, TSIZE(PAstTypeExpression) * (parameter_count - 1));
INC(current_parameter, TSIZE(PAstTypeExpression) * (parameter_count - 1));
current_parameter^ := parse_type_expression(lexer);
@ -235,10 +235,10 @@ begin
return result
end;
proc parse_type_declaration(lexer: PLexer) -> PAstTypeDeclaration;
proc parse_type_declaration(lexer: PLexer) -> PAstTypedDeclaration;
var
token: LexerToken;
result: PAstTypeDeclaration;
result: PAstTypedDeclaration;
begin
token := lexer_current(lexer);
@ -254,16 +254,16 @@ begin
return result
end;
proc parse_type_part(lexer: PLexer) -> PPAstTypeDeclaration;
proc parse_type_part(lexer: PLexer) -> PPAstTypedDeclaration;
var
token: LexerToken;
result: PPAstTypeDeclaration;
current_declaration: PPAstTypeDeclaration;
result: PPAstTypedDeclaration;
current_declaration: PPAstTypedDeclaration;
declaration_count: CARDINAL;
begin
token := lexer_current(lexer);
ALLOCATE(result, TSIZE(PAstTypeDeclaration));
ALLOCATE(result, TSIZE(PAstTypedDeclaration));
current_declaration := result;
declaration_count := 0;
@ -273,16 +273,16 @@ begin
while token.kind = lexerKindIdentifier do
INC(declaration_count);
REALLOCATE(result, TSIZE(PAstTypeDeclaration) * (declaration_count + 1));
REALLOCATE(result, TSIZE(PAstTypedDeclaration) * (declaration_count + 1));
current_declaration := result;
INC(current_declaration, TSIZE(PAstTypeDeclaration) * (declaration_count - 1));
INC(current_declaration, TSIZE(PAstTypedDeclaration) * (declaration_count - 1));
current_declaration^ := parse_type_declaration(lexer);
token := transpiler_lex(lexer)
end
end;
if declaration_count <> 0 then
INC(current_declaration, TSIZE(PAstTypeDeclaration))
INC(current_declaration, TSIZE(PAstTypedDeclaration))
end;
current_declaration^ := nil;
@ -426,7 +426,7 @@ begin
REALLOCATE(result^.symbols, TSIZE(Identifier) * (symbol_count + 1));
current_symbol := result^.symbols;
INC(current_symbol, TSIZE(Identifier) * (symbol_count - 1));
INC(current_symbol, TSIZE(Identifier) * (symbol_count - 1));
current_symbol^ := token.identifierKind;
token := transpiler_lex(lexer)
@ -481,18 +481,18 @@ begin
NEW(literal);
literal^.kind := astLiteralKindInteger;
literal^.integer := token.integerKind;
literal^.integer := token.integerKind
end;
if (token.kind = lexerKindCharacter) or (token.kind = lexerKindString) then
NEW(literal);
literal^.kind := astLiteralKindString;
literal^.string := token.stringKind;
literal^.string := token.stringKind
end;
if token.kind = lexerKindNull then
NEW(literal);
literal^.kind := astLiteralKindNull;
literal^.kind := astLiteralKindNull
end;
if token.kind = lexerKindBoolean then
NEW(literal);
@ -522,7 +522,7 @@ begin
NEW(result);
result^.kind := astExpressionKindLiteral;
result^.literal := literal;
result^.literal := literal
end;
if (result = nil) & (next_token.kind = lexerKindMinus) then
NEW(result);
@ -549,10 +549,10 @@ begin
end;
if (result = nil) & (next_token.kind = lexerKindIdentifier) then
NEW(result);
result^.kind := astExpressionKindIdentifier;
result^.identifier := next_token.identifierKind;
next_token := transpiler_lex(lexer)
end;
@ -658,7 +658,7 @@ begin
result^.kind := astExpressionKindBinary;
result^.binary_operator := operator;
result^.lhs := left;
result^.rhs := right;
result^.rhs := right
end;
return result
@ -757,4 +757,243 @@ begin
return result
end;
proc parse_compound_statement(lexer: PLexer) -> AstCompoundStatement;
var
result: AstCompoundStatement;
token: LexerToken;
current_statement: PPAstStatement;
old_count: CARDINAL;
begin
result.count := 0;
result.statements := nil;
token := lexer_current(lexer);
while token.kind <> lexerKindEnd do
old_count := result.count;
INC(result.count);
REALLOCATE(result.statements, TSIZE(PAstStatement) * result.count);
current_statement := result.statements;
INC(current_statement, TSIZE(PAstStatement) * old_count);
current_statement^ := parse_statement(lexer);
token := lexer_current(lexer)
end;
return result
end;
proc parse_statement(lexer: PLexer) -> PAstStatement;
var
token: LexerToken;
statement: PAstStatement;
designator: PAstExpression;
begin
statement := nil;
token := transpiler_lex(lexer);
if token.kind = lexerKindIf then
statement := parse_if_statement(lexer)
end;
if token.kind = lexerKindWhile then
statement := parse_while_statement(lexer)
end;
if token.kind = lexerKindReturn then
statement := parse_return_statement(lexer)
end;
if token.kind = lexerKindIdentifier then
designator := parse_designator(lexer);
token := lexer_current(lexer);
if token.kind = lexerKindAssignment then
statement := parse_assignment_statement(lexer, designator)
end;
if token.kind <> lexerKindAssignment then
statement := parse_call_statement(lexer, designator)
end
end;
return statement
end;
proc parse_if_statement(lexer: PLexer) -> PAstStatement;
var
token: LexerToken;
result: PAstStatement;
begin
NEW(result);
result^.kind := astStatementKindIf;
token := transpiler_lex(lexer);
result^.if_condition := parse_expression(lexer);
result^.if_branch := parse_compound_statement(lexer);
token := transpiler_lex(lexer);
return result
end;
proc parse_while_statement(lexer: PLexer) -> PAstStatement;
var
token: LexerToken;
result: PAstStatement;
begin
NEW(result);
result^.kind := astStatementKindWhile;
token := transpiler_lex(lexer);
result^.while_condition := parse_expression(lexer);
result^.while_body := parse_compound_statement(lexer);
token := transpiler_lex(lexer);
return result
end;
proc parse_statement_part(lexer: PLexer) -> AstCompoundStatement;
var
token: LexerToken;
compound: AstCompoundStatement;
begin
compound.count := 0;
compound.statements := nil;
token := lexer_current(lexer);
if token.kind = lexerKindBegin then
compound := parse_compound_statement(lexer)
end;
return compound
end;
proc parse_procedure_heading(lexer: PLexer) -> PAstProcedureDeclaration;
var
token: LexerToken;
declaration: PAstProcedureDeclaration;
parameter_index: CARDINAL;
current_parameter: PAstTypedDeclaration;
begin
NEW(declaration);
token := transpiler_lex(lexer);
declaration^.name := token.identifierKind;
token := transpiler_lex(lexer);
declaration^.parameters := nil;
declaration^.parameter_count := 0;
token := transpiler_lex(lexer);
while token.kind <> lexerKindRightParen do
parameter_index := declaration^.parameter_count;
INC(declaration^.parameter_count);
REALLOCATE(declaration^.parameters, TSIZE(AstTypedDeclaration) * declaration^.parameter_count);
current_parameter := declaration^.parameters;
INC(current_parameter, TSIZE(AstTypedDeclaration) * parameter_index);
current_parameter^.identifier := token.identifierKind;
token := transpiler_lex(lexer);
token := transpiler_lex(lexer);
current_parameter^.type_expression := parse_type_expression(lexer);
token := transpiler_lex(lexer);
if token.kind = lexerKindComma then
token := transpiler_lex(lexer)
end
end;
token := transpiler_lex(lexer);
declaration^.return_type := nil;
(* Check for the return type and write it. *)
if token.kind = lexerKindArrow then
token := transpiler_lex(lexer);
declaration^.return_type := parse_type_expression(lexer);
token := transpiler_lex(lexer)
end;
token := transpiler_lex(lexer);
return declaration
end;
proc parse_procedure_declaration(lexer: PLexer) -> PAstProcedureDeclaration;
var
token: LexerToken;
declaration: PAstProcedureDeclaration;
begin
declaration := parse_procedure_heading(lexer);
declaration^.constants := parse_constant_part(lexer);
declaration^.variables := parse_variable_part(lexer);
declaration^.statements := parse_statement_part(lexer);
token := transpiler_lex(lexer);
token := transpiler_lex(lexer);
return declaration
end;
proc parse_procedure_part(lexer: PLexer) -> PPAstProcedureDeclaration;
var
token: LexerToken;
current_declaration: PPAstProcedureDeclaration;
result: PPAstProcedureDeclaration;
declaration_count: CARDINAL;
declaration_index: CARDINAL;
begin
token := lexer_current(lexer);
declaration_count := 0;
declaration_index := 0;
ALLOCATE(result, TSIZE(PAstProcedureDeclaration));
while token.kind = lexerKindProc do
INC(declaration_count);
REALLOCATE(result, TSIZE(PAstProcedureDeclaration) * (declaration_count + 1));
current_declaration := result;
INC(current_declaration, TSIZE(PAstProcedureDeclaration) * declaration_index);
current_declaration^ := parse_procedure_declaration(lexer);
token := lexer_current(lexer);
declaration_index := declaration_count
end;
current_declaration := result;
INC(current_declaration, TSIZE(PAstProcedureDeclaration) * declaration_index);
current_declaration^ := nil;
return result
end;
proc parse_module(lexer: PLexer) -> PAstModule;
var
token: LexerToken;
result: PAstModule;
begin
NEW(result);
token := transpiler_lex(lexer);
result^.main := true;
if token.kind = lexerKindModule then
result^.main := false
end;
token := transpiler_lex(lexer);
(* Write the module body. *)
token := transpiler_lex(lexer);
result^.imports := parse_import_part(lexer);
result^.constants := parse_constant_part(lexer);
result^.types := parse_type_part(lexer);
result^.variables := parse_variable_part(lexer);
result^.procedures := parse_procedure_part(lexer);
result^.statements := parse_statement_part(lexer);
token := transpiler_lex(lexer);
token := transpiler_lex(lexer);
return result
end;
end.