MY_COMPILER/frontend/lexer/Lexer.java

package frontend.lexer;

import java.util.ArrayList;

import error.Error;
import error.ErrorType;
import error.Errors;
import java.nio.file.Files;
import java.nio.file.Paths;
import java.io.IOException;

public class Lexer {
    private String input;
    private ArrayList<Token> tokens;
    private int position;
    private int line;
    private char currentChar;

    public Lexer(String input) {
        this.input = input;
        this.tokens = new ArrayList<Token>();
        this.position = 0;
        this.line = 1;
    }

    public void lex(Errors errors) {
        int sigComment = 0;
        while (this.position < this.input.length()) {
            currentChar = this.input.charAt(this.position);
            if (currentChar == ' ' || currentChar == '\t') {
                this.position++;
                continue;
            }
            if (currentChar == '\n' || currentChar == '\r') {
                if (currentChar == '\r') {
                    if (this.position + 1 < this.input.length() &&
                            this.input.charAt(this.position + 1) == '\n') {
                        this.position++;
                    }
                }
                this.line++;
                this.position++;
                if (sigComment == 1) {
                    sigComment = 0;
                }
                continue;
            }
            if (sigComment == 0) {
                if (currentChar == '/') {
                    if (this.position + 1 < this.input.length() &&
                            this.input.charAt(this.position + 1) == '/') {
                        sigComment = 1;
                        this.position += 2;
                        continue;
                    }
                    if (this.position + 1 < this.input.length() &&
                            this.input.charAt(this.position + 1) == '*') {
                        sigComment = 2;
                        this.position += 2;
                        continue;
                    }
                }
                if (this.isIntCons()) {
                    lexInt();
                    continue;
                }
                if (this.isStrCons()) {
                    lexStr();
                    continue;
                }
                if (this.isIdenfr()) {
                    lexIdenfr();
                    continue;
                }
                lexOp(errors);
            }
            if (sigComment == 2) {
                if (this.position + 1 < this.input.length() &&
                        this.currentChar == '*' && this.input.charAt(this.position + 1) == '/') {
                    sigComment = 0;
                    this.position += 2;
                    continue;
                }
            }
            if (sigComment != 0) {
                this.position++;
            }
        }
    }

    public boolean isIntCons() {
        return Character.isDigit(this.currentChar);
    }

    public boolean isStrCons() {
        return this.currentChar == '"';
    }

    public boolean isIdenfr() {
        return Character.isLowerCase(this.currentChar) || Character.isUpperCase(this.currentChar)
                || this.currentChar == '_';
    }

    public boolean isNotWp() {
        return !Character.isWhitespace(this.currentChar);
    }

    public boolean isOp() {
        return !this.isIntCons() && !this.isStrCons() && !this.isIdenfr();
    }

    public void lexInt() {
        StringBuilder sb = new StringBuilder();
        while (this.position < this.input.length() && this.isIntCons()) {
            sb.append(this.currentChar);
            this.position++;
            updateCurrentChar();
        }
        this.tokens.add(new Token(sb.toString(), this.line));
    }

    public void lexStr() {
        StringBuilder sb = new StringBuilder();
        sb.append(this.currentChar);
        this.position++;
        updateCurrentChar();
        while (this.position < this.input.length() && this.currentChar != '"') {
            sb.append(this.currentChar);
            this.position++;
            updateCurrentChar();
        }
        if (this.position < this.input.length() && this.currentChar == '"') {
            sb.append(this.currentChar);
            this.position++;
        }
        this.tokens.add(new Token(sb.toString(), this.line));
    }

    public void lexIdenfr() {
        StringBuilder sb = new StringBuilder();
        while (this.position < this.input.length() && (this.isIdenfr() || this.isIntCons())) {
            sb.append(this.currentChar);
            this.position++;
            updateCurrentChar();
        }
        this.tokens.add(new Token(sb.toString(), this.line));
    }

    public void lexOp(Errors errors) {
        StringBuilder sb = new StringBuilder();
        if (this.currentChar == '/' && this.position + 1 < this.input.length() &&
                (this.input.charAt(this.position + 1) == '/' ||
                        this.input.charAt(this.position + 1) == '*')) {
            return;
        }
        sb.append(this.currentChar);
        switch (this.currentChar) {
            case '&':
                readAnd(sb);
                break;
            case '|':
                readOr(sb);
                break;
            case '<':
                readEq(sb);
                break;
            case '>':
                readEq(sb);
                break;
            case '=':
                readEq(sb);
                break;
            case '!':
                readEq(sb);
                break;
            default:
                break;
        }
        this.position++;
        if (sb.toString().equals("&") || sb.toString().equals("|")) {
            errors.addError(new Error(this.line, ErrorType.a));
        }
        this.tokens.add(new Token(sb.toString(), this.line));
    }

    public void readAnd(StringBuilder sb) {
        if (this.position + 1 < this.input.length() &&
                this.input.charAt(this.position + 1) == '&') {
            this.position++;
            currentChar = this.input.charAt(this.position);
            sb.append(currentChar);
        }
    }

    public void readOr(StringBuilder sb) {
        if (this.position + 1 < this.input.length() &&
                this.input.charAt(this.position + 1) == '|') {
            this.position++;
            currentChar = this.input.charAt(this.position);
            sb.append(currentChar);
        }
    }

    public void readEq(StringBuilder sb) {
        if (this.position + 1 < this.input.length() &&
                this.input.charAt(this.position + 1) == '=') {
            this.position++;
            currentChar = this.input.charAt(this.position);
            sb.append(currentChar);
        }
    }

    public void updateCurrentChar() {
        if (this.position < this.input.length()) {
            currentChar = this.input.charAt(this.position);
        }
    }

    public void printTokens() {
        for (Token token : this.tokens) {
            token.adjustType();
            System.out.println(token.getType() + " " + token.getValue() + " " + token.getLine());
        }
    }

    public void writeToFile(String fileName) {
        try {
            StringBuilder sb = new StringBuilder();
            for (Token token : this.tokens) {
                token.adjustType();
                sb.append(token.toString());
            }
            Files.write(Paths.get(fileName), sb.toString().getBytes());
        } catch (IOException e) {
            e.printStackTrace();
        }
    }

    public ArrayList<Token> getTokens() {
        for (Token token : this.tokens) {
            token.adjustType();
        }
        return this.tokens;
    }
}