/**
  TinyScanner.java
  scanner that gets tokens for TINY
  a simple language that is not very complex
  for a few examples of TINY programs see
  http://www.cs.stedwards.edu/~lbaker/cosc4342/CcompilerBookcode/sample.tny
  and s1.tny, s2.tny, sample1.tny in same directory
  @author COSC 4342 
  @version 1.0
*/
import java.util.*;
import java.io.*;

public class TinyScanner
{
  // various states in the scanner.
  public final char EOF = (char) -1;
  public final int START=1;
  public final int INASSIGN=2;
  public final int INCOMMENT=3;
  public final int INNUM=4;
  public final int INID=5;
  public final int DONE=6;
  // various token definitions

 public static final int // op codes for the individual tokens
   EOFtok = -1, 
   IFtok = 0,  
   THENtok = 1,  
   ELSEtok = 2,   
   ENDtok=3, 
   REPEATtok = 4,    
   UNTILtok = 5, 
   READtok = 6,   
   WRITEtok = 7,
   NUMtok = 9,   
   IDtok = 10,  
   LPARENtok = 11, /* ( */
   RPARENtok = 12, /* ) */
   SEMItok = 13,   /*  ;        */
   EQtok = 14,    /*  =  */
   LTtok = 15,     /*  <   */
   ASSIGNtok = 16,    /*  :=       */
   PLUStok = 17,   
   MINUStok = 18,         /*  + -      */
   DIVtok = 19,    
   MULtok = 20,           /*  /  *      */
   ERRORtok = 21;


  public final String reservedWords[]= {
    "if","then", "else", "end", "repeat", "until",
    "read", "write" };
  public final int numReservedWords = reservedWords.length;
 
  String tokenString;
  String lineBuf="";    // stores each input line of scanned file
  int lineno=0;         // line number of scanned file 
  int linepos=0;        // position in the line

  BufferedReader bufRdr;    // input file buffered reader 
  File inFile;   // source file to scan
  
  public static void main(String args[])
  {
      if(args.length == 0)
        {
          System.out.println(
            "You must give a filename to scan on the" +
            " command line such as java TinyScanner file1.tny " +
            ", now exiting program. ");
          return;
        }
      TinyScanner myScanner = new TinyScanner (args[0]);
      Token tok = myScanner.getToken();
      while( tok.getNumber()!= TinyScanner.EOFtok ){
          System.out.println("token: " + tok.toString());
          tok = myScanner.getToken();
      }
  
    System.out.println("***  Scanning is complete, End-of-File reached.  ***");
  }


/**
 * Returns the next character from the input buffer
 
 * @param  none
 *         
 * @return     the next char from the line Buffer 
 * @see         
 * 
 */
 char getNextChar()
    {
      char ch=EOF;     
      if( lineBuf == "" || lineBuf==null || (linepos >= lineBuf.length()))
        {
          linepos=0;
          try
            {
              lineBuf=bufRdr.readLine();
              lineno++;               // we must try to get another line

              // check if we found end of file, set char to EOF
              if(lineBuf==null)
                {
                  ch = EOF;
                }
              else if (lineBuf.equals(""))  // we found a blank line
                {
                  System.out.println(lineno+" : " + lineBuf);
                  lineBuf=lineBuf+"\n";        // put newline onto line
                  ch = getNextChar();
                }
              else
                {
                  System.out.println(lineno + " : " + lineBuf);
                  lineBuf = lineBuf + "\n";  // add newline to lineBuf
                  linepos = 0;
                  ch = lineBuf.charAt(linepos++);
                } 
              
            } catch(IOException e) 
              { 
                System.out.println("IOException in getchar: " + e );
              }
        }
      else                   // we have a character on current line to read
        ch = (char) lineBuf.charAt(linepos++);
      
      return ch;
    }
/** 
  * this function simulates putting a char back into the input buffer
  * it does this by decrementing the line position index
  * @param none
  * @returns  nothing
  */

  void ungetNextChar()
    {
      linepos--;
    }

  /**
     constructs this with fname as file name to use as
     file to open tries to open file
     @param  fname - name of file to read (scan) from
    */

  public TinyScanner(String fname)
    {
      inFile=new File(fname);
      try{
          bufRdr = new BufferedReader(new FileReader(inFile));
      }catch(FileNotFoundException e)
        {  
          System.out.println("error opening file: " + fname + " for input. " );
        }
                                              
    }
/**
 * Returns the location of String s in the reserved words table
 * @param  s - the String to look for as a reserved word
 *         
 * @return  location (index) in table if found, -1 if not found
 *         
 * 
 */
  int reservedLookup(String s)
    {
      int i;

      for(i=0; i< numReservedWords; i++)
        { 
          if(reservedWords[i].equals(s))
            return i;
        }
      return -1;
    }

/**
 * returns the next Token from the input buffer
 *  repeatedly calls getNextChar() until a token is recognized
 * @param none 
 * @return next Token which could be the  EOFtok
*/
  Token getToken()
    {
      boolean save=true;
      Token currentToken=null;
      int state=START;
      int loc =0;
      char c= EOF;
      tokenString = "";
      while(state != DONE)
        {
          c = getNextChar();
          save=true;
          switch(state)
            {
              case START:
                if(Character.isLetter(c))  // is it alpha?
                  state=INID;
                else if(Character.isDigit(c)) // is it a digit?
                  state=INNUM;
                else if(c==':')
                  state=INASSIGN;
                else if (c=='{')
                  {
                    save=false;
                    state=INCOMMENT;
                  }
                else if (Character.isWhitespace(c))
                  save=false;
                else
                  {
                    state = DONE;
                    save=false;
                    switch(c)
                      {
                        case '=':
                          currentToken=new Token(EQtok);
                          break;
                        case '<':
                          currentToken= new Token (LTtok);
                          break;
                        case '+':
                          currentToken= new Token (PLUStok);
                          break;
                        case '-': 
                          currentToken = new Token (MINUStok);
                          break;             
                        case '*':
                          currentToken = new Token (MULtok);
                          break;
                        case '/':
                          currentToken = new Token (DIVtok);
                          break;
                        case '(':
                          currentToken = new Token  (LPARENtok);
                          break;
                        case ')':
                          currentToken = new Token (RPARENtok);
                          break;
                        case ';':
                          currentToken = new Token (SEMItok);
                          break;
                        case EOF: 
                          currentToken = new Token (EOFtok);
                          break;
                        default:
                          currentToken = new Token  (ERRORtok);
                          break;
                      }
                  }
                break;
              case INCOMMENT:
                save = false;
                if (c == '}') state = START;
                break;
              case INASSIGN:
                state = DONE;
                if (c == '=')
                  currentToken = new Token(ASSIGNtok);
                else
                  { /* backup in the input */
                    ungetNextChar();
                    save = false;
                    currentToken = new Token (ERRORtok);
                  }
                break;
              case INNUM:
                if (!(c>='0' && c <='9'))
                  { /* backup in the input */
                    ungetNextChar();
                    save = false;
                    state = DONE;
                    currentToken = new Token(NUMtok);
                    currentToken.setValue(tokenString);
                  }
                break;
              case INID:
                if (!((c>='A' && c <='Z')|| (c >='a' && c<='z')))
                  { /* backup in the input */
                    ungetNextChar();
                    save = false;
                    state = DONE;
                    currentToken = new Token(IDtok);
                    currentToken.setValue(tokenString);
                  }
                break;
       
              default: /* should never happen */
                System.out.println("Scanner Bug: state= " + state);
                state = DONE;
                currentToken = new Token(ERRORtok);
                break;
            } // end of switch
          if (save)
            tokenString=tokenString + c;
        } // endwhile
      currentToken.setValue(tokenString);
      // check to see if we have an identifier, if so check if reserved word
      if (currentToken.getNumber() == (IDtok))
        {   
          loc = this.reservedLookup(tokenString);
          // set token number according to location in reserved word table
          if(loc>=0&&loc<=7)
            currentToken.setNumber(loc);
        }
        
      return currentToken;
    }

/** class Token
 * stores the token number and the string value of the token
 *
 */
  class Token
    {
      private int number;   // number of the token (used by parser really )
      private String value; // characters of the token

/**
 * creates a default token with number 0 and empty string of chars.
 * @param  none
 *         
 * @return  Token built with default values
 * @see         
 * 
 */
      public Token()
        {
          value="";
          number=0;
        }
/**
 * creates a Token with given number  and empty string of chars.
 * @param  num - the token number (assumed valid)
 *         
 * @return  Token built with given number
 * @see         
 * 
 */
      public Token(int num)
        {
          number = num;
          value = "";
        }
/**
 * creates a Token with given number and String 
 * @param  num - the token number (assumed valid)
 * @param  s - string value to store corresponds to name or value of token        
 * @return  Token built with given number
 * @see         
 * 
 */
      public Token(int num, String s)
        {
          value = "";
          number=num;
          if (num == NUMtok)
            value = s;
          else if(num == IDtok)
            value = s;
        }

      int getNumber()
        {
          return number;
        }
      String getValue()
        {
          return value;
        }
 
      void setValue(String v)
        { 
          value = v;
        }
      void setNumber(int n)
        {
          number = n;
        }

      /** Returns a printable version of Token
       * @param : none
       *  @returns: printable version of Token with name
       */
      public String toString()
        {
          switch(number)
            {
              case -1: value="EOFtok";  break;
              case 0: value="IFtok";  break;
              case 1: value="THENtok";  break;
              case 2: value="ELSEtok";  break;
              case 3: value="ENDtok";  break;
              case 4: value="REPEATtok";  break;              
              case 5: value="UNTILtok";  break;
              case 6: value="READtok";  break;
              case 7: value="WRITEtok";  break;
              case 9:  break;  // number is set already in getToken
              case 10: break;  // id value is set already in getToken
              case 11: value="LPAREN: (";  break;
              case 12: value="RPAREN: )";  break;
              case 13: value="SEMI: ;";  break;
              case 14: value="EQ: =";  break;
              case 15: value="LT: <";  break;
              case 16: value="ASSIGN: :=";  break;
              case 17: value="PLUS: +";  break;
              case 18: value="MINUS: -";  break;
              case 19: value="MUL: *";  break;
              case 20: value="DIV: /";  break;
              case 21: value="ERRORtok: " + value;  break;
            }
          String temp="";
          if ( number== 9)
            temp = number + " : NUMtok value:  " + value ;
          else if (number==10)
            temp = number + " : IDtok name: " + value;
          else
            temp = number + " : " + value;
          return temp;
        }

  } // end of Token class
} // end of Scanner class