/**
 * Created by natalia on 05.04.14.
 */

import java.util.regex.Matcher;
import java.util.regex.Pattern;

import java.io.*;
import java.util.ArrayList;

class Reader {
    private File openedFile_;
    private BufferedReader reader_;
    private ArrayList<String> readedText_;

    public Reader(String path) {
        openedFile_ = new File(path);
        try {
            reader_ = new BufferedReader(new InputStreamReader(new FileInputStream(openedFile_), "UTF-8"));
        } catch (UnsupportedEncodingException exc) {
            System.out.println(" UTF-8 exception. " + exc.getMessage());
        } catch (FileNotFoundException exc) {
            System.out.println("File Not Found. " + exc.getMessage());
        }
        readedText_ = new ArrayList<String>();
    }

    public ArrayList<String> Read() {
        String currentString;
        try {
            while((currentString = reader_.readLine()) != null) {
                readedText_.add(currentString);
            }
        } catch (IOException exc) {
            System.out.println("IOException while reading file. " + exc.getMessage());
        }
        return new ArrayList<String>(readedText_);
    }
}

public class lex {
    static private int strNum = 0;

    private void print(String s, int pos, int mend) {
        System.out.println("IDENT ("+strNum+","+pos+")-("+strNum+","+(pos+mend)+"): "+s);
    }
    private static void findLex2(String s, Pattern p) {
        int pos = 0;
        Matcher m;
        while(s.length() != 0){
            m = p.matcher(s);
            if(m.find()) {
                if (m.group(1) != null) {
                    System.out.println("RULE II ("+strNum+","+pos+")-("+strNum+","+
                            (pos+m.end())+"): "+m.group(1));

                } else {
                    System.out.println("??? ("+strNum+","+pos+")-("+strNum+","+
                            (pos+m.end())+"): "+m.group(7));
                }
                pos += m.end();
                s = s.substring(m.end());
                //System.out.println(s);
                //for(int i = 1; i< 7;i++)
                //    System.out.println("g: " +m.group(i));
            } else {
                pos++;
                if(s.charAt(0) != ' ') {
                    System.out.println("ERROR at (" + strNum + "," + pos + ")");
                }
                s = s.substring(1);
            }
        }
        strNum++;
    }

    private static void findLex(String s, Pattern p) {
        int pos = 0;
        Matcher m;
        while(s.length() != 0){
            m = p.matcher(s);
            if(m.find()) {
                if (m.group(1) != null) {
                    System.out.println("AXIOM ("+strNum+","+pos+")-("+strNum+","+
                            (pos+m.end())+"): "+m.group(1));
                } else if(m.group(5) != null) {
                    System.out.println("TERM ("+strNum+","+pos+")-("+strNum+","+
                            (pos+m.end())+"): "+m.group(5));
                } else if(m.group(3) != null) {
                    System.out.println("NTERM ("+strNum+","+pos+")-("+strNum+","+
                            (pos+m.end())+"): "+m.group(3));
                } else {
                    System.out.println("??? ("+strNum+","+pos+")-("+strNum+","+
                            (pos+m.end())+"): "+m.group(7));
                    findLex2(m.group(7), Pattern.compile("\\$RULE ([A-Z]\\'{0,1} =([( [A-Z]\\'{0,1})( \"[a-z]\"| \"\\*\"| \"\\+\"| \"\\(\"| \"\\)\")]+\\n| \\$EPS\\n)+)"));
                }
                pos += m.end();
                s = s.substring(m.end());
                //System.out.println(s);
                //for(int i = 1; i< 7;i++)
                //    System.out.println("g: " +m.group(i));
            } else {
                pos++;
                if(s.charAt(0) != ' ') {
                    System.out.println("ERROR at (" + strNum + "," + pos + ")");
                }
                s = s.substring(1);
            }
        }
        strNum++;
    }

    public static void splitLexems(String rules) {

    }

    public static void main(String args[])
    {
        // Текст для сопоставления
        String text = "$AXIOM E\n" +
                "$TERM";

        // Регулярные выражения
        String ident = "^\\p{L}.{0,8}\\p{L}";//"^[A-Za-z]+|^[(][0-9]+[)]";+";//
        //String number = //"^0|^[1-9][0-9]*";
        //String oper = //"^[(][)]|^[:][=]|^[:]";
        String strlit = "^\'([^\\\\]|\\\\n|\\\\\'|\\\\[0-9A-Fa-f]{4,4}|\\p{L})*\'";//|[[\\][n]]*|[[\\][0-9A-Fa-f]{,4}]*|[.]*][\']\'";
        String keyword = "^z|^forward|^for";
        String pattern = "("+keyword+")|("+strlit+")|("+ident+")";

        String nt = "[A-Z]\\'{0,1}";
        String t = "\"[a-z]\"| \"\\*\"| \"\\+\"| \"\\(\"| \"\\)\"";

        String axiom = "\\$AXIOM ([A-Z]\\\'{0,1})\\n";
        String nterm = "\\$NTERM( [A-Z]\\'{0,1})+\\n";
        String term = "\\$TERM( \"[a-z]\"| \"\\*\"| \"\\+\"| \"\\(\"| \"\\)\")+\\n";
        String rule = "\\$RULE ([A-Z]\\'{0,1} =([( [A-Z]\\'{0,1})( \"[a-z]\"| \"\\*\"| \"\\+\"| \"\\(\"| \"\\)\")]+\\n| \\$EPS\\n)+)";

        pattern = "("+axiom+")|("+nterm+")|("+term+")|(("+rule+")+)";
        //String

        // Компиляция регулярного выражения
        Pattern p = Pattern.compile(pattern);

        Reader reader = new Reader("test.txt");
        ArrayList<String> mas = reader.Read();
        //System.out.println(reader1.Read());
        // Сопоставление текста с регулярным выражением
        String program = "";
        for(String s: mas)
            program += (s+'\n');
        findLex(program, p);

        for (String s : mas) {
            //System.out.println();
            //findLex(s+'\n', p);
        }
    }
}