Initial commit

This commit is contained in:
2025-11-10 20:44:10 +11:00
commit 18fadc18ed
8 changed files with 3500 additions and 0 deletions

790
src/main.cpp Normal file
View File

@@ -0,0 +1,790 @@
#include <iostream>
#include <utility>
#include <vector>
#include <optional>
#include <variant>
#include <cctype>
#include <complex>
#include <sstream>
#include <map>
#include "../libs/linenoise.hpp"
enum class ValueTypes {
Identifier, Int, Double, String, List, Nil
};
enum class ParserErrorType {
UnexpectedEndOfInput,
UnexpectedToken,
MissingOpenParen,
MissingCloseParen,
UnexpectedTopLevelToken
};
enum class InterpreterErrorType {
UnknownInstruction,
UnexpectedToken,
IncorrectTokenType,
MathError,
UnknownEnvironment
};
class InterpretingError : public std::exception {
std::string message;
InterpreterErrorType errorType;
public:
explicit InterpretingError(InterpreterErrorType errorType) : errorType(errorType) {
std::stringstream oss;
oss << "Interpreting error: ";
switch (errorType) {
case InterpreterErrorType::UnexpectedToken:
oss << "Unexpected token";
break;
case InterpreterErrorType::UnknownInstruction:
oss << "Unknown instruction";
break;
case InterpreterErrorType::IncorrectTokenType:
oss << "Incorrect token type";
break;
case InterpreterErrorType::MathError:
oss << "Math error";
break;
case InterpreterErrorType::UnknownEnvironment:
oss << "Unknown environment";
break;
}
message = oss.str();
}
[[nodiscard]] const char* what() const noexcept override {
return message.c_str();
}
};
class ParsingError : public std::exception {
std::string message;
ParserErrorType errorType;
int position;
std::string sourceCode;
public:
ParsingError(ParserErrorType type, int pos, std::string source = "")
: errorType(type), position(pos), sourceCode(std::move(source)) {
std::ostringstream oss;
oss << "Parsing error at position " << pos << ": ";
switch (type) {
case ParserErrorType::UnexpectedEndOfInput:
oss << "Unexpected end of input";
break;
case ParserErrorType::UnexpectedToken:
oss << "Unexpected token";
break;
case ParserErrorType::MissingOpenParen:
oss << "Expected '(' at start of expression";
break;
case ParserErrorType::MissingCloseParen:
oss << "Missing closing ')'";
break;
case ParserErrorType::UnexpectedTopLevelToken:
oss << "Unexpected token at top level (expected '(')";
break;
}
message = oss.str();
}
[[nodiscard]] const char* what() const noexcept override {
return message.c_str();
}
[[nodiscard]] ParserErrorType getErrorType() const { return errorType; }
[[nodiscard]] int getPosition() const { return position; }
[[nodiscard]] const std::string& getSourceCode() const { return sourceCode; }
};
class Value {
std::variant<int, double, std::string, void*, std::vector<Value>> value;
public:
ValueTypes type;
std::optional<int> getInt() {
if (std::holds_alternative<int>(value)) {
return std::get<int>(value);
}
return {};
}
std::optional<double> getDouble() {
if (std::holds_alternative<double>(value)) {
return std::get<double>(value);
}
return {};
}
std::optional<std::string> getString() {
if (std::holds_alternative<std::string>(value)) {
return std::get<std::string>(value);
}
return {};
}
std::optional<std::vector<Value>> getList() {
if (std::holds_alternative<std::vector<Value>>(value)) {
return std::get<std::vector<Value>>(value);
}
return {};
}
void print() {
switch (type) {
case ValueTypes::String: {
std::cout << getString().value();
break;
}
case ValueTypes::Int: {
std::cout << getInt().value();
break;
}
case ValueTypes::Double: {
std::cout << getDouble().value();
break;
}
case ValueTypes::List: {
auto list = getList().value();
bool first = true;
for (auto& listElement : list) {
if (!first) {
std::cout << ", ";
} else {
first = false;
}
listElement.print();
}
}
case ValueTypes::Nil: {
std::cout << "\033[2;3;96m" << "nil" << "\033[0m";
}
}
}
bool operator==(Value &otherValue) {
if (type != otherValue.type) {
return false;
}
switch (type) {
case ValueTypes::String: {
if (getString() == otherValue.getString()) {
return true;
}
return false;
}
case ValueTypes::Int: {
if (getInt() == otherValue.getInt()) {
return true;
}
return false;
}
case ValueTypes::Double: {
if (getDouble() == otherValue.getDouble()) {
return true;
}
return false;
}
case ValueTypes::List: {
auto list = getList().value();
auto compareList = otherValue.getList().value();
if (list.size() != compareList.size()) {
return false;
}
for (int i = 0; i < list.size(); i++) {
if (list[i] == compareList[i]) {
return true;
}
return false;
}
return true;
}
default: return false;
}
}
explicit Value() : value(nullptr), type(ValueTypes::Nil) {}
explicit Value(int in) : value(in), type(ValueTypes::Int) {}
explicit Value(double in) : value(in), type(ValueTypes::Double) {}
explicit Value(std::string in) : value(in), type(ValueTypes::String) {}
explicit Value(std::vector<Value> in) : value(in), type(ValueTypes::List) {}
Value(std::string in, bool isIdentifier) : value(in), type(isIdentifier ? ValueTypes::Identifier : ValueTypes::String) {}
};
class Instruction {
public:
std::string instruction;
std::vector<std::variant<Instruction, Value>> args;
Instruction(std::string instruction, std::vector<std::variant<Instruction, Value>> args) : instruction(std::move(instruction)), args(std::move(args)) {}
Instruction() = default;
};
class Parser {
std::string input;
std::vector<std::string> split;
int currentChar = -1;
int current = -1;
std::optional<char> consumeChar() {
currentChar++;
if (currentChar >= input.size()) {
return {};
}
return input[currentChar];
}
std::optional<std::string> consume() {
current++;
if (current >= split.size()) {
return {};
}
return split[current];
}
[[nodiscard]] std::optional<std::string> peek() const {
if (current + 1 >= split.size()) {
return {};
}
return split[current + 1];
}
// Helper to detect value type
static Value parseValue(const std::string& token) {
// If it's nil, just return nil
if (token == "nil") {
return Value();
}
// Check if it's a string literal
if (token.front() == '"' && token.back() == '"') {
return Value(token.substr(1, token.length() - 2));
}
// Try to parse as integer
bool isInt = true;
bool isDouble = false;
for (size_t i = 0; i < token.length(); i++) {
char c = token[i];
if (i == 0 && c == '-') continue; // Allow negative sign
if (c == '.') {
if (isDouble) { // Second dot
isInt = false;
break;
}
isDouble = true;
isInt = false;
} else if (!std::isdigit(c)) {
isInt = false;
isDouble = false;
break;
}
}
if (isInt) {
return Value(std::stoi(token));
}
if (isDouble) {
return Value(std::stod(token));
}
// If it isn't any value, return an identifier
return {token, true};
}
// Recursive parser for expressions
Instruction parseExpression() {
auto topt = consume();
if (!topt) {
throw ParsingError(ParserErrorType::UnexpectedEndOfInput, current, input);
}
std::string token = topt.value();
if (token != "(") {
throw ParsingError(ParserErrorType::MissingOpenParen, current, input);
}
topt = consume();
if (!topt) {
throw ParsingError(ParserErrorType::UnexpectedEndOfInput, current, input);
}
std::string instruction = topt.value();
std::vector<std::variant<Instruction, Value>> args;
while (true) {
topt = peek();
if (!topt) {
throw ParsingError(ParserErrorType::MissingCloseParen, current, input);
}
token = topt.value();
if (token == ")") {
consume();
break;
} else if (token == "(") {
args.emplace_back(parseExpression());
} else {
consume();
args.emplace_back(parseValue(token));
}
}
return {instruction, args};
}
public:
std::vector<Instruction> instructions;
explicit Parser(std::string in) : input(std::move(in)) {
// Lexer logic
std::string buf;
bool inString = false;
while (auto copt = consumeChar()) {
char c = copt.value();
if (c != '"' && inString) {
buf.push_back(c);
continue;
}
switch (c) {
case '(': case ')': {
if (!buf.empty()) {
split.push_back(buf);
buf.clear();
}
split.emplace_back(1, c);
break;
}
case ' ': case '\n':
if (!buf.empty()) {
split.push_back(buf);
buf.clear();
}
break;
case '"': {
inString = !inString;
buf.push_back('"');
if (!inString) {
split.push_back(buf);
buf.clear();
}
break;
}
default: {
buf.push_back(c);
}
}
}
// Parser logic - parse all top-level expressions
while (peek()) {
auto token = peek().value();
if (token == "(") {
instructions.push_back(parseExpression());
} else {
throw ParsingError(ParserErrorType::UnexpectedTopLevelToken, current + 1, input);
}
}
}
Parser() = default;
};
class Interpreter {
public:
std::vector<Instruction> instructions = {};
std::map<std::string, Value> environment = {{"pippleVersion", Value("0.0.1")}};
Value interpretInstruction(Instruction& instruction) {
// Preprocess identifiers
for (auto& arg : instruction.args) {
if (std::holds_alternative<Value>(arg) && instruction.instruction != "let" && instruction.instruction != "set") {
if (std::get<Value>(arg).type == ValueTypes::Identifier) {
std::string id = std::get<Value>(arg).getString().value();
if (!environment.contains(id)) {
throw InterpretingError(InterpreterErrorType::UnknownEnvironment);
}
arg = environment[id];
}
}
}
// Instructions at the top of this function require input to not be preprocessed
if (instruction.instruction == "if") {
if (instruction.args.size() < 2) {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
bool toContinue = true;
if (std::holds_alternative<Instruction>(instruction.args[0])) {
toContinue = interpretInstruction(std::get<Instruction>(instruction.args[0])).type != ValueTypes::Nil;
}
if (std::holds_alternative<Value>(instruction.args[0])) {
toContinue = std::get<Value>(instruction.args[0]).type != ValueTypes::Nil;
}
if (toContinue) {
bool first = true;
Value returnValue;
for (auto& arg : instruction.args) {
if (first) {
first = false;
} else {
if (std::holds_alternative<Instruction>(arg)) {
returnValue = interpretInstruction(std::get<Instruction>(arg));
}
}
}
return returnValue;
}
return Value();
}
if (instruction.instruction == "while") {
if (instruction.args.size() < 2) {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
while (true) {
Value conditionResult;
if (std::holds_alternative<Instruction>(instruction.args[0])) {
Instruction condCopy = std::get<Instruction>(instruction.args[0]);
conditionResult = interpretInstruction(condCopy);
} else if (std::holds_alternative<Value>(instruction.args[0])) {
conditionResult = std::get<Value>(instruction.args[0]);
}
if (conditionResult.type == ValueTypes::Nil) {
break;
}
for (size_t i = 1; i < instruction.args.size(); i++) {
if (std::holds_alternative<Instruction>(instruction.args[i])) {
Instruction bodyCopy = std::get<Instruction>(instruction.args[i]);
if (bodyCopy.instruction == "break") {
return Value();
}
interpretInstruction(bodyCopy);
}
}
}
return Value();
}
// Preprocess instructions inside instructions
for (auto &arg : instruction.args) {
if (std::holds_alternative<Instruction>(arg)) {
arg = interpretInstruction(std::get<Instruction>(arg));
}
}
// Instructions that don't require preprocessing go below this line
// It is safe to use std::get<Value>(...) on all arguments
if (instruction.instruction == "+" || instruction.instruction == "add") {
double result = 0;
for (const auto& arg : instruction.args) {
auto value = std::get<Value>(arg);
switch (value.type) {
case ValueTypes::Int: {
result += value.getInt().value();
break;
}
case ValueTypes::Double: {
result += value.getDouble().value();
break;
}
default: {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
}
}
if (result == static_cast<int>(result)) {
return Value(static_cast<int>(result));
}
return Value(result);
}
if (instruction.instruction == "-" || instruction.instruction == "subtract") {
if (instruction.args.empty()) return Value(0);
auto firstValue = std::get<Value>(instruction.args[0]);
double result = 0;
if (firstValue.type == ValueTypes::Int) {
result = firstValue.getInt().value();
} else if (firstValue.type == ValueTypes::Double) {
result = firstValue.getDouble().value();
} else {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
for (size_t i = 1; i < instruction.args.size(); i++) {
auto value = std::get<Value>(instruction.args[i]);
if (value.type == ValueTypes::Int) {
result -= value.getInt().value();
} else if (value.type == ValueTypes::Double) {
result -= value.getDouble().value();
} else {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
}
if (result == static_cast<int>(result)) {
return Value(static_cast<int>(result));
}
return Value(result);
}
if (instruction.instruction == "*" || instruction.instruction == "multiply") {
if (instruction.args.empty()) return Value(1);
auto firstValue = std::get<Value>(instruction.args[0]);
double result = 0;
if (firstValue.type == ValueTypes::Int) {
result = firstValue.getInt().value();
} else if (firstValue.type == ValueTypes::Double) {
result = firstValue.getDouble().value();
} else {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
for (size_t i = 1; i < instruction.args.size(); i++) {
auto value = std::get<Value>(instruction.args[i]);
if (value.type == ValueTypes::Int) {
result *= value.getInt().value();
} else if (value.type == ValueTypes::Double) {
result *= value.getDouble().value();
} else {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
}
if (result == static_cast<int>(result)) {
return Value(static_cast<int>(result));
}
return Value(result);
}
if (instruction.instruction == "/" || instruction.instruction == "divide") {
if (instruction.args.empty()) return Value(1);
auto firstValue = std::get<Value>(instruction.args[0]);
double result = 0;
if (firstValue.type == ValueTypes::Int) {
result = firstValue.getInt().value();
} else if (firstValue.type == ValueTypes::Double) {
result = firstValue.getDouble().value();
} else {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
for (size_t i = 1; i < instruction.args.size(); i++) {
auto value = std::get<Value>(instruction.args[i]);
if (value.type == ValueTypes::Int) {
if (value.getInt().value() == 0) {
throw InterpretingError(InterpreterErrorType::MathError);
}
result /= value.getInt().value();
} else if (value.type == ValueTypes::Double) {
if (value.getDouble().value() == 0) {
throw InterpretingError(InterpreterErrorType::MathError);
}
result /= value.getDouble().value();
} else {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
}
if (result == static_cast<int>(result)) {
return Value(static_cast<int>(result));
}
return Value(result);
}
if (instruction.instruction == "==" || instruction.instruction == "equal") {
bool first = true;
Value firstVal;
for (const auto& arg : instruction.args) {
auto value = std::get<Value>(arg);
if (first) {
first = false;
firstVal = value;
} else {
if (value != firstVal) {
return Value();
}
}
}
return Value(1);
}
if (instruction.instruction == "!=" || instruction.instruction == "inequal") {
bool first = true;
Value firstVal;
for (const auto& arg : instruction.args) {
auto value = std::get<Value>(arg);
if (first) {
first = false;
firstVal = value;
} else {
if (value == firstVal) {
return Value();
}
}
}
return Value(1);
}
if (instruction.instruction == "input") {
std::string input;
linenoise::Readline("", input);
return Value(input);
}
if (instruction.instruction == "let") {
if (instruction.args.size() < 2) {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
auto idval = std::get<Value>(instruction.args[0]);
std::string id;
if (idval.type == ValueTypes::Identifier) {
id = idval.getString().value();
} else {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
environment[id] = std::get<Value>(instruction.args[1]);
return {environment[id]};
}
if (instruction.instruction == "set") {
if (instruction.args.size() < 2) {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
auto idval = std::get<Value>(instruction.args[0]);
std::string id;
if (idval.type == ValueTypes::Identifier) {
id = idval.getString().value();
} else {
throw InterpretingError(InterpreterErrorType::IncorrectTokenType);
}
if (!environment.contains(id)) {
throw InterpretingError(InterpreterErrorType::UnknownEnvironment);
}
environment[id] = std::get<Value>(instruction.args[1]);
return {environment[id]};
}
if (instruction.instruction == "print") {
for (const auto& arg : instruction.args) {
auto value = std::get<Value>(arg);
value.print();
std::cout << " ";
}
std::cout << "\n";
return Value();
}
if (instruction.instruction == "log") {
for (const auto& arg : instruction.args) {
std::cout << "\033[2;3;93m";
auto value = std::get<Value>(arg);
value.print();
std::cout << " ";
}
std::cout << "\033[0m\n";
return Value();
}
if (instruction.instruction == "exit") {
if (!instruction.args.empty()) {
auto value = std::get<Value>(instruction.args[0]);
if (value.type == ValueTypes::Int) {
exit(value.getInt().value());
}
}
exit(0);
}
throw InterpretingError(InterpreterErrorType::UnknownInstruction);
}
explicit Interpreter(std::vector<Instruction> in) : instructions(std::move(in)) {
for (Instruction &instruction : instructions) {
interpretInstruction(instruction);
}
}
Interpreter() = default;
};
int main(int argc, char** argv) {
bool isInteractive = true;
std::string program;
if (argc > 1) {
isInteractive = false;
std::ifstream ifs(argv[1]);
std::stringstream buffer;
if (ifs.is_open()) {
buffer << ifs.rdbuf();
}
program = buffer.str();
}
linenoise::SetMultiLine(true);
linenoise::SetHistoryMaxLen(50);
Parser parser;
Interpreter interpreter;
while (true) {
if (isInteractive) {
if (linenoise::Readline("pipple> ", program)) {
return 0;
}
}
try {
parser = Parser(program);
} catch (const ParsingError& e) {
std::cerr << "Parse error: " << e.what() << std::endl;
// Show code context
const std::string& source = e.getSourceCode();
if (!source.empty()) {
int pos = e.getPosition();
// Calculate character position in source from token position
int charPos = 0;
int tokenCount = 0;
bool inString = false;
for (size_t i = 0; i < source.length() && tokenCount < pos; i++) {
char c = source[i];
if (c == '"') inString = !inString;
if (!inString && (c == '(' || c == ')' || c == ' ' || c == '\n')) {
if (i > 0 && source[i-1] != ' ' && source[i-1] != '\n' && source[i-1] != '(' && source[i-1] != ')') {
tokenCount++;
}
if (c == '(' || c == ')') tokenCount++;
}
charPos = static_cast<int>(i);
}
// Show context (30 chars before and after)
int start = std::max(0, charPos - 30);
int end = std::min(static_cast<int>(source.length()), charPos + 30);
std::cerr << "\nCode context:\n";
if (start > 0) std::cerr << "...";
std::cerr << source.substr(start, end - start);
if (end < static_cast<int>(source.length())) std::cerr << "...";
std::cerr << "\n";
// Show pointer to error position
int pointerPos = (start > 0 ? 3 : 0) + (charPos - start);
std::cerr << std::string(pointerPos, ' ') << "^\n";
}
if (!isInteractive) return 1;
continue;
} catch (const std::exception& e) {
std::cerr << "Unexpected error: " << e.what() << std::endl;
if (!isInteractive) return 1;
}
try {
for (auto instruction : parser.instructions) {
interpreter.interpretInstruction(instruction);
}
} catch (const InterpretingError& e) {
std::cerr << e.what() << std::endl;
if (!isInteractive) return 1;
}
if (!isInteractive) {
break;
}
}
return 0;
}