368 lines
7.6 KiB
C++
368 lines
7.6 KiB
C++
module;
|
|
|
|
#include <stdexcept>
|
|
#include <format>
|
|
#include <iterator>
|
|
#include <vector>
|
|
#include <memory>
|
|
|
|
export module jlx:ast;
|
|
|
|
import :tokenizer;
|
|
|
|
namespace jlx {
|
|
export enum ast_type {
|
|
Root,
|
|
Expression,
|
|
Block,
|
|
FunctionDeclaration,
|
|
VariableDeclaration,
|
|
LiteralValue,
|
|
IfStatement,
|
|
};
|
|
|
|
export enum class literal_value_type {
|
|
Boolean,
|
|
Numeric,
|
|
String,
|
|
Character,
|
|
};
|
|
|
|
export template<class T>
|
|
concept token_iterator = requires() {
|
|
requires std::same_as<typename T::value_type, jlx::token>;
|
|
std::bidirectional_iterator<T>;
|
|
};
|
|
|
|
struct statement {
|
|
ast_type type;
|
|
|
|
statement(ast_type type) : type(type) {
|
|
|
|
}
|
|
|
|
virtual ~statement() = default;
|
|
};
|
|
|
|
struct root_statement : public statement {
|
|
root_statement() : statement(Root) {
|
|
|
|
}
|
|
|
|
root_statement(std::vector<std::unique_ptr<statement>> statements) : statement(Root), statements(std::move(statements)) {
|
|
|
|
}
|
|
|
|
std::vector<std::unique_ptr<statement>> statements;
|
|
|
|
~root_statement() override = default;
|
|
};
|
|
|
|
struct expression : public statement {
|
|
expression() : statement(Expression) {
|
|
|
|
}
|
|
|
|
~expression() override = default;
|
|
};
|
|
|
|
struct block : public statement {
|
|
block() : statement(Block) {
|
|
|
|
}
|
|
|
|
block(std::vector<std::unique_ptr<statement>> statements) : statement(Block), statements(std::move(statements)) {
|
|
|
|
}
|
|
|
|
std::vector<std::unique_ptr<statement>> statements;
|
|
|
|
~block() override = default;
|
|
};
|
|
|
|
struct function_parameter {
|
|
std::string name;
|
|
std::string type;
|
|
};
|
|
|
|
struct function_declaration : public statement {
|
|
function_declaration() : statement(FunctionDeclaration) {
|
|
|
|
}
|
|
|
|
function_declaration(std::string name, std::vector<function_parameter> parameters, std::optional<std::string> return_type, std::unique_ptr<block> body) :
|
|
statement(FunctionDeclaration), name(std::move(name)), parameters(std::move(parameters)), return_type(std::move(return_type)), body(std::move(body)) {
|
|
|
|
}
|
|
|
|
std::string name;
|
|
std::vector<function_parameter> parameters;
|
|
std::optional<std::string> return_type;
|
|
std::unique_ptr<block> body;
|
|
|
|
|
|
~function_declaration() override = default;
|
|
};
|
|
|
|
struct variable_declaration : public statement {
|
|
variable_declaration() : statement(VariableDeclaration) {
|
|
|
|
}
|
|
|
|
bool constant;
|
|
std::string name;
|
|
std::optional<std::string> type;
|
|
std::unique_ptr<expression> initial_expression;
|
|
|
|
~variable_declaration() override = default;
|
|
};
|
|
|
|
struct if_statement : public statement {
|
|
if_statement() : statement(IfStatement) {
|
|
|
|
}
|
|
|
|
std::unique_ptr<expression> condition;
|
|
std::unique_ptr<block> block;
|
|
};
|
|
|
|
export template<token_iterator T, std::sentinel_for<T> E>
|
|
class parser {
|
|
T current;
|
|
E last;
|
|
|
|
inline void fail_invalid_token(const token& t) {
|
|
throw std::runtime_error(std::format("Invalid token {} at {}:{}", t.content, t.line, t.col).c_str());
|
|
}
|
|
|
|
inline void fail_invalid_eof() {
|
|
throw std::runtime_error("Unexpected end-of-file");
|
|
}
|
|
|
|
void next() {
|
|
current++;
|
|
|
|
if (current == last) {
|
|
fail_invalid_eof();
|
|
}
|
|
}
|
|
|
|
std::unique_ptr<block> parse_block() {
|
|
if (current->type != Punctuation || current->content != "{") {
|
|
fail_invalid_token(*current);
|
|
}
|
|
next();
|
|
|
|
while(current->type != Punctuation && current->content != "}") {
|
|
|
|
}
|
|
|
|
return nullptr;
|
|
}
|
|
|
|
std::unique_ptr<variable_declaration> parse_variable_declaration() {
|
|
if (current->type != Keyword) {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
std::string name;
|
|
std::optional<std::string> type = std::nullopt;
|
|
std::unique_ptr<expression> starting_value = nullptr;
|
|
bool constant;
|
|
|
|
if (current->content == "let") {
|
|
constant = true;
|
|
} else if (current->content == "var") {
|
|
constant = false;
|
|
} else {
|
|
fail_invalid_token(*current);
|
|
return nullptr;
|
|
}
|
|
|
|
next();
|
|
|
|
if (current->type != Identifier) {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
name = current->content;
|
|
|
|
next();
|
|
|
|
if (current->type == Punctuation && current->content == ":") {
|
|
next();
|
|
|
|
if (current->type != Identifier) {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
type = current->content;
|
|
|
|
next();
|
|
}
|
|
|
|
if (current->type == Operator && current->content == "=") {
|
|
next();
|
|
|
|
starting_value = parse_expression();
|
|
}
|
|
|
|
auto var = std::make_unique<variable_declaration>();
|
|
var->constant = constant;
|
|
var->name = std::move(name);
|
|
var->type = std::move(type);
|
|
var->initial_expression = std::move(starting_value);
|
|
|
|
return var;
|
|
}
|
|
|
|
std::unique_ptr<if_statement> parse_if_statement() {
|
|
if (current->type != Keyword || current->content != "if") {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
next();
|
|
|
|
if (current->type != Punctuation || current->content != "(") {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
next();
|
|
|
|
auto expr = parse_expression();
|
|
|
|
if (current->type != Punctuation || current->content != ")") {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
next();
|
|
|
|
auto block = parse_block();
|
|
|
|
auto statement = std::make_unique<if_statement>();
|
|
statement->block = std::move(block);
|
|
statement->condition = std::move(expr);
|
|
|
|
return statement;
|
|
}
|
|
|
|
std::unique_ptr<expression> parse_expression() {
|
|
return nullptr;
|
|
}
|
|
|
|
std::unique_ptr<function_declaration> parse_function() {
|
|
if (current->type != Keyword || current->content != "fun") {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
next();
|
|
|
|
if (current->type != Identifier) {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
std::string function_name = current->content;
|
|
|
|
next();
|
|
|
|
if (current->type != Punctuation || current->content != "(") {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
next();
|
|
|
|
std::vector<function_parameter> params;
|
|
std::optional<std::string> return_type;
|
|
|
|
bool first = true;
|
|
while (current->type != Punctuation && current->content != ")") {
|
|
if (!first) {
|
|
if(current->type != Punctuation || current->content != ",") {
|
|
fail_invalid_token(*current);
|
|
} else {
|
|
next();
|
|
}
|
|
}
|
|
|
|
std::string name;
|
|
|
|
if (current->type != Identifier) {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
name = current->content;
|
|
|
|
next();
|
|
|
|
if (current->type != Punctuation || current->content != ":") {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
next();
|
|
|
|
auto param_type = parse_type();
|
|
params.emplace_back(std::move(name), std::move(param_type));
|
|
|
|
next();
|
|
first = false;
|
|
}
|
|
next();
|
|
|
|
if (current->type == Punctuation && current->content == ":") {
|
|
next();
|
|
|
|
return_type = parse_type();
|
|
next();
|
|
}
|
|
|
|
auto block = parse_block();
|
|
|
|
return std::make_unique<function_declaration>(std::move(function_name), std::move(params), std::move(return_type), std::move(block));
|
|
}
|
|
|
|
std::string parse_type(){
|
|
if (current->type != Identifier) {
|
|
fail_invalid_token(*current);
|
|
}
|
|
|
|
return current->content;
|
|
}
|
|
|
|
std::unique_ptr<statement> parse_top_level_statement() {
|
|
if (current == last) {
|
|
return nullptr;
|
|
}
|
|
|
|
if (current->type == token_type::Keyword) {
|
|
if (current->content == "let" || current->content == "var") {
|
|
return parse_variable_declaration();
|
|
} else if (current->content == "if") {
|
|
return parse_if_statement();
|
|
} else if (current->content == "fun") {
|
|
return parse_function();
|
|
}
|
|
}
|
|
|
|
return nullptr;
|
|
}
|
|
|
|
public:
|
|
parser(T current, E last) : current(current), last(last) {
|
|
|
|
}
|
|
|
|
std::unique_ptr<statement> parse() {
|
|
|
|
std::vector<std::unique_ptr<statement>> top_level_statements;
|
|
while(current != last) {
|
|
auto s = parse_top_level_statement();
|
|
if (s == nullptr) {
|
|
throw std::runtime_error("No statement parsed...");
|
|
}
|
|
top_level_statements.push_back(std::move(s));
|
|
}
|
|
|
|
return std::make_unique<root_statement>(std::move(top_level_statements));
|
|
}
|
|
};
|
|
}
|