Extract simple expression parser

2025-10-08 12:06:07 +02:00
parent af9ad5b712
commit d25c2d41d1
1 changed files with 259 additions and 61 deletions
--- a/boot/stage15.elna
+++ b/boot/stage15.elna
@@ -128,10 +128,13 @@ type
 		_goto,
 		eof
 	);
-	NodeKind := (
+	NodeKind = (
 		integer_literal,
 		string_literal,
-		character_literal
+		character_literal,
+		variable_expression,
+		field_access_expression,
+		dereference_expression
 	);

 const
@@ -623,21 +626,68 @@ begin
 	_write_c('\n');
 end;

-proc _compile_variable_expression();
+proc _variable_expression_size();
+	return 12
+end;
+
+proc _variable_expression_get_name(this: Word);
+begin
+	this := this + 4;
+	return this^
+end;
+
+proc _variable_expression_set_name(this: Word, value: Word);
+begin
+	this := this + 4;
+	this^ := value
+end;
+
+proc _variable_expression_get_length(this: Word);
+begin
+	this := this + 8;
+	return this^
+end;
+
+proc _variable_expression_set_length(this: Word, value: Word);
+begin
+	this := this + 8;
+	this^ := value
+end;
+
+proc _parse_variable_expression();
 var
 	name: Word;
-	lookup_result: Word;
 	name_token: Word;
+	result: Word;
 begin
 	name := _lexer_global_get_start();
 	name_token := _lexer_global_get_end() - name;
-	lookup_result := _symbol_table_lookup(@symbol_table_global, name, name_token);
+	_lexer_skip_token();

+	result := memory_free_pointer;
+	memory_free_pointer := memory_free_pointer + _variable_expression_size();
+
+	_node_set_kind(result, NodeKind.variable_expression);
+	_variable_expression_set_name(result, name);
+	_variable_expression_set_length(result, name_token);
+
+	return result
+end;
+
+proc _compile_variable_expression(variable_expression: Word);
+var
+	name: Word;
+	name_token: Word;
+	lookup_result: Word;
+begin
+	name := _variable_expression_get_name(variable_expression);
+	name_token := _variable_expression_get_length(variable_expression);
+
+	lookup_result := _symbol_table_lookup(@symbol_table_local, name, name_token);
 	if lookup_result <> 0 then
-		_compile_enumeration_value(lookup_result)
+		_compile_local_designator(lookup_result)
 	else
-		_compile_designator();
-		_write_z("\tlw t0, (t0)\n\0")
+		_compile_global_designator(variable_expression)
 	end
 end;

@@ -645,9 +695,12 @@ end;
 * Compiled take address expression, starting with an "@" sign.
 *)
 proc _compile_address_expression();
+var
+	designator: Word;
 begin
 	_lexer_skip_token();
-	_compile_designator()
+	designator := _parse_designator();
+	_compile_designator(designator)
 end;

 (**
@@ -706,6 +759,7 @@ begin
 	length := _string_length(token_start);
 	_lexer_skip_token();

+	_node_set_kind(result, NodeKind.string_literal);
 	_string_literal_node_set_value(result, token_start);
 	_string_literal_node_set_length(result, length);

@@ -731,6 +785,99 @@ begin
 	_write_z("\tadd t0, t0, t1\n\0")
 end;

+proc _parse_simple_expression();
+var
+	current_character: Word;
+	parser_node: Word;
+	token_kind: Word;
+begin
+	parser_node := 0;
+	_lexer_read_token(@token_kind);
+
+	if token_kind = LexerTokenKind.character then
+		parser_node := _parse_character_literal()
+	elsif token_kind = LexerTokenKind.integer then
+		parser_node := _parse_integer_literal()
+	elsif token_kind = LexerTokenKind.string then
+		parser_node := _parse_string_literal()
+	elsif token_kind = LexerTokenKind.identifier then
+		current_character := _lexer_global_get_start();
+		current_character := _load_byte(current_character);
+
+		(* This is a call if the statement starts with an underscore. *)
+		if current_character <> '_' then
+			parser_node := _parse_variable_expression()
+		end
+	end;
+	return parser_node
+end;
+
+proc _dereference_expression_size();
+	return 8
+end;
+
+proc _dereference_expression_get_pointer(this: Word);
+begin
+	this := this + 4;
+	return this^
+end;
+
+proc _dereference_expression_set_pointer(this: Word, value: Word);
+begin
+	this := this + 4;
+	this^ := value
+end;
+
+proc _parse_dereference_expression(simple_expression: Word);
+var
+	result: Word;
+begin
+	result := memory_free_pointer;
+	memory_free_pointer := memory_free_pointer + _dereference_expression_size();
+
+	_node_set_kind(result, NodeKind.dereference_expression);
+	_dereference_expression_set_pointer(result, simple_expression);
+	_lexer_skip_token();
+
+	return result
+end;
+
+proc _parse_designator();
+var
+	simple_expression: Word;
+	token_kind: Word;
+begin
+	simple_expression := _parse_simple_expression();
+
+	_lexer_read_token(@token_kind);
+
+	if token_kind = LexerTokenKind.hat then
+		simple_expression := _parse_dereference_expression(simple_expression)
+	elsif token_kind = LexerTokenKind.dot then
+		simple_expression := _parse_field_access_expression(simple_expression)
+	end;
+	return simple_expression
+end;
+
+proc _compile_simple_expression(parser_node: Word);
+var
+	is_address: Word;
+begin
+	is_address := 0;
+
+	if _node_get_kind(parser_node) = NodeKind.character_literal then
+		_compile_character_literal(parser_node)
+	elsif _node_get_kind(parser_node) = NodeKind.string_literal then
+		_compile_string_literal(parser_node)
+	elsif _node_get_kind(parser_node) = NodeKind.integer_literal then
+		_compile_integer_literal(parser_node)
+	else
+		_compile_variable_expression(parser_node);
+		is_address := 1
+	end;
+	return is_address
+end;
+
 proc _compile_term();
 var
 	current_character: Word;
@@ -739,15 +886,11 @@ var
 begin
 	_lexer_read_token(@token_kind);

-	if token_kind = LexerTokenKind.character then
-		parser_node := _parse_character_literal();
-		_compile_character_literal(parser_node)
-	elsif token_kind = LexerTokenKind.string then
-		parser_node := _parse_string_literal();
-		_compile_string_literal(parser_node)
-	elsif token_kind = LexerTokenKind.integer then
-		parser_node := _parse_integer_literal();
-		_compile_integer_literal(parser_node)
+	parser_node := _parse_designator();
+	if parser_node <> 0 then
+		if _compile_designator(parser_node) = 1 then
+			_write_z("\tlw t0, (t0) # Designator is an address.\n\0")
+		end
 	elsif token_kind = LexerTokenKind.at then
 		_compile_address_expression()
 	elsif token_kind = LexerTokenKind.minus then
@@ -755,16 +898,8 @@ begin
 	elsif token_kind = LexerTokenKind.not then
 		_compile_not_expression()
 	elsif token_kind = LexerTokenKind.identifier then
-		current_character := _lexer_global_get_start();
-		current_character := _load_byte(current_character);
-
-		(* This is a call if the statement starts with an underscore. *)
-		if current_character = '_' then
-			_compile_call();
-			_write_z("\tmv t0, a0\n\0")
-		else
-			_compile_variable_expression()
-		end
+		_compile_call();
+		_write_z("\tmv t0, a0\n\0")
 	end
 end;

@@ -979,28 +1114,64 @@ begin
 	_write_z("\taddi t0, sp, \0");
 	variable_offset := _parameter_info_get_offset(symbol);
 	_write_i(variable_offset);
-	_write_c('\n');
-	_lexer_skip_token()
+	_write_c('\n')
 end;

-proc _compile_global_designator();
+proc _compile_global_designator(variable_expression: Word);
 var
 	name: Word;
-	token_kind: Word;
 	token_length: Word;
 begin
 	_write_z("\tla t0, \0");

-	_lexer_read_token(@token_kind);
-	name := _lexer_global_get_start();
-	token_length := _lexer_global_get_end() - name;
-	_write_s(name, token_length);
-	_lexer_skip_token();
+	name := _variable_expression_get_name(variable_expression);
+	token_length := _variable_expression_get_length(variable_expression);

+	_write_s(name, token_length);
 	_write_c('\n')
 end;

-proc _compile_enumeration_value(symbol: Word);
+proc _field_access_expression_size();
+	return 16
+end;
+
+proc _field_access_expression_get_aggregate(this: Word);
+begin
+	this := this + 4;
+	return this^
+end;
+
+proc _field_access_expression_set_aggregate(this: Word, value: Word);
+begin
+	this := this + 4;
+	this^ := value
+end;
+
+proc _field_access_expression_get_field(this: Word);
+begin
+	this := this + 8;
+	return this^
+end;
+
+proc _field_access_expression_set_field(this: Word, value: Word);
+begin
+	this := this + 8;
+	this^ := value
+end;
+
+proc _field_access_expression_get_length(this: Word);
+begin
+	this := this + 12;
+	return this^
+end;
+
+proc _field_access_expression_set_length(this: Word, value: Word);
+begin
+	this := this + 12;
+	this^ := value
+end;
+
+proc _compile_enumeration_value(field_access_expression: Word);
 var
 	enumeration_type: Word;
 	members: Word;
@@ -1011,20 +1182,22 @@ var
 	member_name: Word;
 	member_length: Word;
 	counter: Word;
+	symbol: Word;
 begin
+	symbol := _field_access_expression_get_aggregate(field_access_expression);
+	value_name := _variable_expression_get_name(symbol);
+	name_length := _variable_expression_get_length(symbol);
+
+	symbol := _symbol_table_lookup(@symbol_table_global, value_name, name_length);
+
 	enumeration_type := _type_info_get_type(symbol);
 	members := _enumeration_type_get_members(enumeration_type);
 	members_length := _enumeration_type_get_length(enumeration_type);

-	(* Skip enumeration type name and dot. Read the enumeration value. *)
-	_lexer_skip_token();
-	_lexer_read_token(@token_type);
-	_lexer_skip_token();
 	_lexer_read_token(@token_type);

-	value_name := _lexer_global_get_start();
-	name_length := _lexer_global_get_end() - value_name;
-	_lexer_skip_token();
+	value_name := _field_access_expression_get_field(field_access_expression);
+	name_length := _field_access_expression_get_length(field_access_expression);
 	counter := 1;

 	.compile_enumeration_value_members;
@@ -1044,35 +1217,60 @@ begin
 	end
 end;

-proc _compile_designator();
+proc _parse_field_access_expression(aggregate: Word);
+var
+	token_kind: Word;
+	name: Word;
+	name_token: Word;
+	result: Word;
+begin
+	(* Skip dot. Read the enumeration value. *)
+	_lexer_skip_token();
+	_lexer_read_token(@token_kind);
+
+	name := _lexer_global_get_start();
+	name_token := _lexer_global_get_end() - name;
+	result := memory_free_pointer;
+	_lexer_skip_token();
+	memory_free_pointer := memory_free_pointer + _field_access_expression_size();
+
+	_node_set_kind(result, NodeKind.field_access_expression);
+	_field_access_expression_set_aggregate(result, aggregate);
+	_field_access_expression_set_field(result, name);
+	_field_access_expression_set_length(result, name_token);
+
+	return result
+end;
+
+proc _compile_designator(parser_node: Word);
 var
 	name_token: Word;
 	lookup_result: Word;
 	token_kind: Word;
-	name: Word;
+	parser_node: Word;
+	is_address: Word;
 begin
-	_lexer_read_token(@token_kind);
-	name := _lexer_global_get_start();
-	name_token := _lexer_global_get_end() - name;
-	lookup_result := _symbol_table_lookup(@symbol_table_local, name, name_token);
+	is_address := 1;

-	if lookup_result <> 0 then
-		_compile_local_designator(lookup_result)
-	else
-		_compile_global_designator()
-	end;
-	_lexer_read_token(@token_kind);
-	if token_kind = LexerTokenKind.hat then
-		_lexer_skip_token();
+	if _node_get_kind(parser_node) = NodeKind.dereference_expression then
+		parser_node := _dereference_expression_get_pointer(parser_node);
+		_compile_simple_expression(parser_node);
 		_write_z("\tlw t0, (t0)\n\0")
-	end
+	elsif _node_get_kind(parser_node) = NodeKind.field_access_expression then
+		_compile_enumeration_value(parser_node);
+		is_address := 0
+	else
+		is_address := _compile_simple_expression(parser_node)
+	end;
+	return is_address
 end;

 proc _compile_assignment();
 var
 	token_kind: Word;
 begin
-	_compile_designator();
+	token_kind := _parse_designator();
+	_compile_designator(token_kind);

 	(* Save the assignee address on the stack. *)
 	_write_z("\tsw t0, 60(sp)\n\0");