java-commons/muscript/src/main/java/io/gitlab/jfronny/muscript/compiler/Parser.java

package io.gitlab.jfronny.muscript.compiler;

import io.gitlab.jfronny.muscript.ast.*;
import io.gitlab.jfronny.muscript.ast.bool.*;
import io.gitlab.jfronny.muscript.ast.compare.Equal;
import io.gitlab.jfronny.muscript.ast.compare.Greater;
import io.gitlab.jfronny.muscript.ast.conditional.UnresolvedConditional;
import io.gitlab.jfronny.muscript.ast.dynamic.*;
import io.gitlab.jfronny.muscript.ast.dynamic.assign.DynamicAssign;
import io.gitlab.jfronny.muscript.ast.literal.DynamicLiteral;
import io.gitlab.jfronny.muscript.ast.math.*;
import io.gitlab.jfronny.muscript.ast.string.Concatenate;
import io.gitlab.jfronny.muscript.compiler.lexer.LegacyLexer;
import io.gitlab.jfronny.muscript.compiler.lexer.Lexer;
import io.gitlab.jfronny.muscript.data.Script;
import io.gitlab.jfronny.muscript.data.dynamic.additional.DFinal;
import io.gitlab.jfronny.muscript.error.*;
import org.jetbrains.annotations.Nullable;

import java.util.*;

public class Parser extends VersionedComponent {
    private final Lexer lexer;

    private Lexer.Token previous = null;

    public static Expr<?> parse(MuScriptVersion version, String source) {
        return parse(version, source, null);
    }

    public static Expr<?> parse(MuScriptVersion version, String source, String file) {
        return new Parser(new LegacyLexer(version, source, file)).parse().optimize();
    }

    public static Script parseScript(MuScriptVersion version, String source) {
        return parseScript(version, source, null);
    }

    public static Script parseScript(MuScriptVersion version, String source, String file) {
        return new Parser(new LegacyLexer(version, source, file)).parseScript().optimize();
    }

    public static Script parseMultiScript(MuScriptVersion version, String startFile, SourceFS filesystem) {
        return new Script(parseMultiScript(version, startFile, filesystem, new HashSet<>()).stream().flatMap(Script::stream).toList());
    }

    private static List<Script> parseMultiScript(MuScriptVersion version, String startFile, SourceFS filesystem, Set<String> alreadyIncluded) {
        alreadyIncluded.add(startFile);
        boolean isIncludes = true;
        StringBuilder src = new StringBuilder();
        List<Script> includes = new LinkedList<>();
        int row = 0;
        final String includePrefix = "#include ";
        for (String s : filesystem.read(startFile).split("\n")) {
            row++;
            if (s.isBlank()) {
                src.append("\n");
            } else if (s.startsWith(includePrefix)) {
                if (isIncludes) {
                    String file = s.substring(includePrefix.length());
                    src.append("// include ").append(file).append("\n");
                    if (!alreadyIncluded.contains(file)) {
                        includes.addAll(parseMultiScript(version, file, filesystem, alreadyIncluded));
                    }
                } else {
                    throw new ParseException(PrettyPrintError.builder()
                            .setLocation(new PrettyPrintError.Location(s, 0, row), new PrettyPrintError.Location(s, s.length() - 1, row))
                            .setMessage("Includes MUST be located at the top of the file")
                            .build());
                }
            } else {
                isIncludes = false;
                src.append(s).append("\n");
            }
        }
        includes.add(parseScript(version, src.toString(), startFile));
        return includes;
    }

    public Parser(io.gitlab.jfronny.muscript.compiler.Lexer lexer) {
        this(new LegacyLexer(lexer));
    }

    public Parser(Lexer lexer) {
        super(lexer.version());
        this.lexer = lexer;
    }

    /**
     * Generate a single expression.
     * Must exhaust the source!
     *
     * @return the resulting expression
     */
    public Expr<?> parse() {
        advance();
        Expr<?> expr = expression();
        if (!isAtEnd() && version.contains(MuScriptVersion.V2))
            throw new ParseException(PrettyPrintError.builder(lexer.location()).setMessage("Unexpected element after end of expression").build());
        return expr;
    }

    /**
     * Generate a script instance.
     * Multiple instructions will be executed in sequence and the result of the last one will be returned.
     *
     * @return the resulting expression
     */
    public Script parseScript() {
        advance();
        List<Expr<?>> expressions = new LinkedList<>();
        while (!isAtEnd()) {
            expressions.add(expression());
            // Consume semicolon if present
            if (!lexer.wasNewlinePassed() & !match(Token.Semicolon) & !isAtEnd() & version.contains(MuScriptVersion.V3)) {
                throw error("Either a semicolon or a new line must separate expressions in scripts");
            }
        }
        if (expressions.isEmpty()) throw new ParseException(PrettyPrintError.builder(lexer.location()).setMessage("Missing any elements in closure").build());
        return new Script(expressions);
    }

    // Expressions
    private Expr<?> expression() {
        try {
            return conditional();
        } catch (RuntimeException e) {
            if (e instanceof ParseException) throw e;
            else if (e instanceof LocationalException le) {
                throw new ParseException(le.asPrintable(), le.getCause());
            } else throw error(e.getMessage());
        }
    }

    private Expr<?> conditional() {
        Expr<?> expr = and();

        if (match(Token.QuestionMark)) {
            CodeLocation location = previous.location();
            Expr<?> trueExpr = expression();
            consume(Token.Colon, "Expected ':' after first part of condition");
            Expr<?> falseExpr = expression();
            expr = new UnresolvedConditional(location, asBool(expr), trueExpr, falseExpr);
        }

        return expr;
    }

    private Expr<?> and() {
        Expr<?> expr = or();

        while (match(Token.And)) {
            CodeLocation location = previous.location();
            Expr<?> right = or();
            expr = new And(location, asBool(expr), asBool(right));
        }

        return expr;
    }

    private Expr<?> or() {
        Expr<?> expr = equality();

        while (match(Token.Or)) {
            CodeLocation location = previous.location();
            Expr<?> right = equality();
            expr = new Or(location, asBool(expr), asBool(right));
        }

        return expr;
    }

    private Expr<?> equality() {
        Expr<?> expr = concat();

        while (match(Token.EqualEqual, Token.BangEqual)) {
            Token op = previous.token();
            CodeLocation location = previous.location();
            Expr<?> right = concat();
            BoolExpr e = new Equal(location, expr, right);
            if (op == Token.BangEqual) e = new Not(location, e);
            expr = e;
        }

        return expr;
    }

    private Expr<?> concat() {
        Expr<?> expr = comparison();

        while (match(Token.Concat)) {
            CodeLocation location = previous.location();
            Expr<?> right = comparison();
            expr = new Concatenate(location, asString(expr), asString(right));
        }

        return expr;
    }

    private Expr<?> comparison() {
        Expr<?> expr = term();

        while (match(Token.Greater, Token.GreaterEqual, Token.Less, Token.LessEqual)) {
            Token op = previous.token();
            CodeLocation location = previous.location();
            NumberExpr right = asNumber(term());
            expr = switch (op) {
                case Greater -> new Greater(location, asNumber(expr), right);
                case GreaterEqual -> new Not(location, new Greater(location, right, asNumber(expr)));
                case Less -> new Greater(location, right, asNumber(expr));
                case LessEqual -> new Not(location, new Greater(location, asNumber(expr), right));
                default -> throw new IllegalStateException();
            };
        }

        return expr;
    }

    private Expr<?> term() {
        Expr<?> expr = factor();

        while (match(Token.Plus, Token.Minus)) {
            Token op = previous.token();
            CodeLocation location = previous.location();
            NumberExpr right = asNumber(factor());
            expr = switch (op) {
                case Plus -> new Plus(location, asNumber(expr), right);
                case Minus -> new Minus(location, asNumber(expr), right);
                default -> throw new IllegalStateException();
            };
        }

        return expr;
    }

    private Expr<?> factor() {
        Expr<?> expr = exp();

        while (match(Token.Star, Token.Slash, Token.Percentage)) {
            Token op = previous.token();
            CodeLocation location = previous.location();
            NumberExpr right = asNumber(exp());
            expr = switch (op) {
                case Star -> new Multiply(location, asNumber(expr), right);
                case Slash -> new Divide(location, asNumber(expr), right);
                case Percentage -> new Modulo(location, asNumber(expr), right);
                default -> throw new IllegalStateException();
            };
        }

        return expr;
    }

    private Expr<?> exp() {
        Expr<?> expr = unary();

        while (match(Token.UpArrow)) {
            CodeLocation location = previous.location();
            NumberExpr right = asNumber(unary());
            expr = new Power(location, asNumber(expr), right);
        }

        return expr;
    }

    private Expr<?> unary() {
        if (match(Token.Bang, Token.Minus)) {
            Token op = previous.token();
            CodeLocation location = previous.location();
            Expr<?> right = unary();
            return switch (op) {
                case Bang -> new Not(location, asBool(right));
                case Minus -> new Invert(location, asNumber(right));
                default -> throw new IllegalStateException();
            };
        }

        return call();
    }

    private Expr<?> call() {
        Expr<?> expr = primary();

        while (match(Token.LeftParen, Token.Dot, Token.LeftBracket, Token.DoubleColon)) {
            CodeLocation location = previous.location();
            expr = switch (previous.token()) {
                case LeftParen -> finishCall(location, expr);
                case Dot -> {
                    Lexer.Token name = consume(Token.Identifier, "Expected field name after '.'");
                    yield new Get(location, asDynamic(expr), Expr.literal(name.location(), name.lexeme()));
                }
                case DoubleColon -> {
                    DynamicExpr callable;
                    if (match(Token.Identifier)) {
                        callable = new Variable(previous.location(), previous.lexeme());
                    } else if (match(Token.LeftParen)) {
                        callable = expression().asDynamicExpr();
                        consume(Token.RightParen, "Expected ')' after expression");
                    } else throw error("Bind operator requires right side to be a literal identifier or to be wrapped in parentheses.");
                    yield new Bind(location, callable, expr.asDynamicExpr());
                }
                case LeftBracket -> {
                    expr = new Get(location, asDynamic(expr), expression());
                    consume(Token.RightBracket, "Expected closing bracket");
                    yield expr;
                }
                default -> throw new IllegalStateException();
            };
        }

        return expr;
    }

    private Expr<?> finishCall(CodeLocation location, Expr<?> callee) {
        List<Call.Arg> args = new ArrayList<>(2);

        if (!check(Token.RightParen)) {
            do {
                args.add(new Call.Arg(asDynamic(expression()), match(Token.Ellipsis)));
            } while (match(Token.Comma));
        }

        consume(Token.RightParen, "Expected ')' after function arguments");
        return new Call(location, asDynamic(callee), args);
    }

    private Expr<?> primary() {
        if (match(Token.Null)) return Expr.literalNull(previous.location());
        if (match(Token.String)) return Expr.literal(previous.location(), previous.lexeme());
        if (match(Token.True, Token.False)) return Expr.literal(previous.location(), previous.lexeme().equals("true"));
        if (match(Token.Number)) return Expr.literal(previous.location(), Double.parseDouble(previous.lexeme()));
        if (match(Token.Identifier)) {
            CodeLocation location = previous.location();
            String name = previous.lexeme();
            if (match(Token.Assign)) return new DynamicAssign(location, name, expression().asDynamicExpr());
            else return new Variable(location, name);
        }

        if (match(Token.LeftParen)) {
            Expr<?> expr = expression();
            consume(Token.RightParen, "Expected ')' after expression");
            return expr;
        }

        if (match(Token.LeftBrace)) {
            int start = previous.start();
            if (match(Token.Arrow)) return finishClosure(start, null, false);
            if (match(Token.RightBrace)) return new DynamicLiteral<>(location(start, previous.start()), DFinal.of(Map.of()));
            consume(Token.Identifier, "Expected arrow or identifier as first element in closure or object");
            String first = previous.lexeme();
            if (check(Token.Arrow)) return finishClosure(start, first, false);
            if (match(Token.Ellipsis)) return finishClosure(start, first, true);
            if (check(Token.Comma)) return finishClosure(start, first, false);
            if (match(Token.Assign)) {
                return finishObject(start, first, expression().asDynamicExpr());
            }
            throw error("Unexpected");
        }

        throw error("Expected expression.");
    }

    private Expr<?> finishClosure(int start, @Nullable String firstArg, boolean firstVariadic) {
        List<String> boundArgs = new LinkedList<>();
        boolean variadic = false;
        if (firstArg != null) {
            boundArgs.add(firstArg);
            if (firstVariadic) {
                consume(Token.Arrow, "Variadic argument MUST be the last argument");
                variadic = true;
            } else {
                while (!match(Token.Arrow)) {
                    consume(Token.Comma, "Closure parameters MUST be comma-seperated");
                    consume(Token.Identifier, "Closure arguments MUST be identifiers");
                    boundArgs.add(previous.lexeme());
                    if (match(Token.Ellipsis)) {
                        variadic = true;
                        consume(Token.Arrow, "Variadic argument MUST be the last argument");
                        break;
                    }
                }
            }
        }
        List<Expr<?>> expressions = new LinkedList<>();
        while (!match(Token.RightBrace)) {
            expressions.add(expression());
            // Consume semicolon if present
            if (!lexer.wasNewlinePassed() & !match(Token.Semicolon) & version.contains(MuScriptVersion.V3)) {
                if (match(Token.RightBrace)) break;
                throw error("Either a semicolon or a new line must separate expressions in closures");
            }
        }
        int end = previous.start();
        return new Closure(location(start, end), boundArgs, expressions, variadic);
    }

    private Expr<?> finishObject(int start, @Nullable String firstArg, @Nullable DynamicExpr firstValue) {
        Map<String, DynamicExpr> content = new LinkedHashMap<>();
        content.put(firstArg, firstValue);
        while (match(Token.Comma)) {
            consume(Token.Identifier, "Object element MUST start with an identifier");
            String name = previous.lexeme();
            consume(Token.Assign, "Object element name and value MUST be seperated with '='");
            content.put(name, expression().asDynamicExpr());
        }
        consume(Token.RightBrace, "Expected end of object");
        return new ObjectLiteral(location(start, previous.start()), content);
    }

    // Type conversion
    private BoolExpr asBool(Expr<?> expression) {
        try {
            return expression.asBoolExpr();
        } catch (TypeMismatchException e) {
            throw error(e.getMessage(), expression);
        }
    }

    private NumberExpr asNumber(Expr<?> expression) {
        try {
            return expression.asNumberExpr();
        } catch (TypeMismatchException e) {
            throw error(e.getMessage(), expression);
        }
    }

    private StringExpr asString(Expr<?> expression) {
        try {
            return expression.asStringExpr();
        } catch (TypeMismatchException e) {
            throw error(e.getMessage(), expression);
        }
    }

    private DynamicExpr asDynamic(Expr<?> expression) {
        try {
            return expression.asDynamicExpr();
        } catch (TypeMismatchException e) {
            throw error(e.getMessage(), expression);
        }
    }

    // Helpers
    private CodeLocation location(int chStart, int chEnd) {
        return new CodeLocation(chStart, chEnd, lexer.getSource(), lexer.getFile());
    }

    private ParseException error(String message) {
        int loc = lexer.getPrevious().current() - 1;
        return new ParseException(PrettyPrintError.builder(location(loc, loc)).setMessage(message).build());
    }

    private ParseException error(String message, Expr<?> expr) {
        return new ParseException(PrettyPrintError.builder(expr.location).setMessage(message).build());
    }

    private Lexer.Token consume(Token token, String message) {
        if (check(token)) return advance();
        throw error(message + " but got " + lexer.getPrevious().token());
    }

    private boolean match(Token... tokens) {
        for (Token token : tokens) {
            if (check(token)) {
                advance();
                return true;
            }
        }

        return false;
    }

    private boolean check(Token token) {
        if (isAtEnd()) return false;
        return lexer.getPrevious().token() == token;
    }

    private Lexer.Token advance() {
        previous = lexer.getPrevious();

        lexer.advance();

        if (lexer.getPrevious().token() == Token.Error) {
            throw error(lexer.getPrevious().lexeme());
        }

        return previous;
    }

    private boolean isAtEnd() {
        return lexer.getPrevious().token() == Token.EOF;
    }

    // Parse Exception
    public static class ParseException extends RuntimeException {
        public final PrettyPrintError error;

        public ParseException(PrettyPrintError error) {
            super(error.toString());
            this.error = error;
        }

        public ParseException(PrettyPrintError error, Throwable cause) {
            super(error.toString(), cause);
            this.error = error;
        }
    }
}