// scanner.cpp -- scanner class implementation
#include "scanner.h"
#include "chset.h"
#include "Token.h"
#include "FileManager.h"
#include "TokFiler.h"
#include "rwtable.h"
#include "idsymtab.h"
#include <string.h>
#include <iostream.h>
#include <stdlib.h>

scanner::scanner() {  // scanner class constructor
   chset t1("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz");
   chset t2("0123456789");
   chset t3("+-*/=<>:(),;.^[]{}#'\"");  // 20 delimiters
   chset t4("`~!@$%&_|'\\?");
   chset t5(".");
   letters = t1; digits = t2; delimiters = t3; illegalChars = t4;
   decimal = t5;
   aRwtablePtr = new rwtable;
   anIdSymTabPtr = new idSymbolTable;
   lineNumber = 1;  // line number starts with 1.
   word[79] = '\0'; // a word length will be 80, end with null char.
   strcatHelper[1] = '\0';  // helper for string concatenation.
}  // end of scanner constructor.

void scanner::error(unsigned int n, unsigned int line) {
   cout << "\n\nError on line " << line << ", #" << n << ": ";
   switch (n) {
      case 1: cout << "Numeric constant contains a letter!"; break;
      case 2: cout << "Illegal symbol find in program!"; break;
      case 3: cout << "Numeric value too large!"; break;
      case 4: cout << "Sum of all identifier characters exceeds limit!"; break;
      case 5: cout << "Lang2 compiler not capable handling floats!"; break;
      case 6: cout << "Nested comments not allowed!"; break;
      case 7: cout << "Comments should end with right }"; break;
   }  // end of switch.
   cout << "\n\n";
   aFileManager.closeProgram();
   aFileManager.seeErrorLines(line);
   aTokFiler.removeTokenFile();
   exit(1);
}  // end of method error().

void scanner::scanIdentifier() {
   const NUMOFRWORDS = 35;

   while ( (ch != ' ') && (ch != '\\') && (ch != '\t') && (ch != '\n') &&
           !delimiters.in(ch) && !illegalChars.in(ch) &&
           !aFileManager.ENDOFPROGRAM ) {
      strcatHelper[0] = ch;
      strcat(word, strcatHelper);
      ch = aFileManager.getNextChar();

      if (illegalChars.in(ch)) {
         strcatHelper[0] = ch;
         strcat(word, strcatHelper);
         error(2, lineNumber);
      }  // end of if
   }  // end of while loop.

   int rwHashValue = -1;
   if ( (aRwtablePtr != NULL) && (strlen(word) != 0) ) {
      rwHashValue = aRwtablePtr->rwhash(word);
   }
   if ( (rwHashValue >= 0) && (rwHashValue <= NUMOFRWORDS) ) {
      if (strcmp(word, aRwtablePtr->resWord(rwHashValue)) == 0) {
         tok.setClass(RESERVED);
         tok.setValue(rwHashValue);
         aTokFiler.emitToken(tok);
      } else {
         tok.setClass(IDENTIFIER);
      }      
   } else {
      tok.setClass(IDENTIFIER);
   }  // end of outer if-else.

   if (tok.getClass() == IDENTIFIER) {
      int ididx;
      bool insertOK;
      anIdSymTabPtr->searchAndMaybeInsert(word, ididx, insertOK);
      if (!insertOK) {
         error(4, lineNumber);
      }  // end of inner if.
      tok.setValue(ididx);
      aTokFiler.emitToken(tok);
   }  // end of outer if.
}  // end of method scanIdentifier().

void scanner::strToInt(unsigned int& n, bool& ok) {
   const maxVal = 32767;
   n = 0; ok = true;
   int i = 0, len = strlen(word);
   int digit;

   while (ok && (i < len)) {
      digit = word[i] - '0';
      if (n < ( (maxVal - digit) / 10) ) {
         n = 10* n + digit;
         i++;
      } else {
         ok = false;
      }
   }  // end of while
}  // end of method strToInt().

void scanner::scanNumeric() {
   while ( (ch != ' ') &&  (ch != '\\') && (ch != '\t') && 
           (ch != '\n') && !(delimiters.in(ch)) && 
           !(illegalChars.in(ch)) && !aFileManager.ENDOFPROGRAM ) {
      strcatHelper[0] = ch;
      strcat(word, strcatHelper);
      ch = aFileManager.getNextChar();
      if (letters.in(ch)) {
         strcatHelper[0] = ch;
         strcat(word, strcatHelper);
         error(1, lineNumber);
      } else if (illegalChars.in(ch) ) {
         strcatHelper[0] = ch;
         strcat(word, strcatHelper);
         error(2, lineNumber);
      } else if (decimal.in(ch) ) {
         strcatHelper[0] = ch;
         strcat(word, strcatHelper);
         error(5, lineNumber);
      }  // end of if-else-if.
   }  // end of while.

   unsigned int value;
   bool ok;
   strToInt(value, ok);  // w need not be passed since data member.
   if (!ok) 
      error(3, lineNumber);
   tok.setClass(NUMBER);
   tok.setValue(value);
   aTokFiler.emitToken(tok);
}  // end of method scanNumerical().
      
void scanner::scanSpecial() {
   unsigned int special;
   strcatHelper[0] = ch;
   strcat(word, strcatHelper);
   switch (ch) {
      case ';': case '.': case ',': case '(': case ')':
      case '+': case '-': case '*': case '/': case '=':
      case '[': case ']': case '{': case '}': case '#':
      case '\'': case '"': case '^':
         switch (ch) {
            case ';': special = SEMICOLON; break;
            case '.': special = PERIOD; break; 
            case ',': special = COMMA; break;
            case '(': special = LPAREN; break;
            case ')': special = RPAREN; break;
            case '+': special = PLUS; break;
            case '-': special = MINUS; break;
            case '*': special = MULTIPLY; break;
            case '/': special = DIVIDE; break;
            case '=': special = EQUAL; break;
            case '[': special = LSQUARE; break;
            case ']': special = RSQUARE; break;
            case '{': special = LCURLY; break;
            case '}': special = RCURLY; break;
            case '"': special = DQUOTE; break;
            case '\'': special = SQUOTE; break;
            case '#': special = NOTEQUAL; break;
            case '^': special = CARAT; break;
         }  // end of inner switch.
         ch = aFileManager.getNextChar();
         break;
      case ':':
         special = COLON;
         ch = aFileManager.getNextChar();
         if (ch == '=') {
           strcatHelper[0] = ch;
           strcat(word, strcatHelper);
           special = COLONEQUAL;
           ch = aFileManager.getNextChar();
         }
         break; 
      case '>':
         special = GREATER;
         ch = aFileManager.getNextChar();
         if (ch == '=') {
           strcatHelper[0] = ch;
           strcat(word, strcatHelper);
           special = GREATEREQUAL;
           ch = aFileManager.getNextChar();
         }
         break; 
      case '<':
         special = LESS;
         ch = aFileManager.getNextChar();
         if (ch == '=') {
           strcatHelper[0] = ch;
           strcat(word, strcatHelper);
           special = LESSEQUAL;
           ch = aFileManager.getNextChar();
         }
         break; 
   }  // end of switch.
   if (special == LCURLY) 
      scanComments(special);
   tok.setClass(SYMBOL);
   tok.setValue(special);
   aTokFiler.emitToken(tok);
}  // end of method scanSpecial().

void scanner::scanComments(unsigned int special) {
   int lCurlyCnt = 0, rCurlyCnt = 0;
   lCurlyCnt++;
   if (ch == '\n') {  // deal with NEWLINE
      lineNumber++;
      tok.setClass(NEWLINE);
      tok.setValue(lineNumber);
      aTokFiler.emitToken(tok);
   }
   // discard all comment blocks.
   while ((ch != '}') && !aFileManager.ENDOFPROGRAM) { 
      ch = aFileManager.getNextChar();
      if (ch == '\n') {
         lineNumber++;
         tok.setClass(NEWLINE);
         tok.setValue(lineNumber);
         aTokFiler.emitToken(tok);
      }
      if (ch == '{') {
         lCurlyCnt++;
         error(6, lineNumber);
      } 
   }  // end of while loop.
   if (ch == '}') {
      rCurlyCnt++;
   } 
   // cout <<"In scanComments, lCurlyCnt == "<<lCurlyCnt<<" rCurlyCnt == " 
   //     << rCurlyCnt << endl;
   if (lCurlyCnt > rCurlyCnt++) {
      error(7, lineNumber);
   } 
}

char* scanner::getNextWord() {
   word[0] = '\0';
   while ( (ch == ' ') ||  (ch == '\t') ) { // throw away white spaces.
      ch = aFileManager.getNextChar();
   }

   if ( (ch=='\\') && (aFileManager.peekNextChar()=='\\') ) {
      //comments: from "\\" to end of current line
      while ( (ch != '\n') && !(aFileManager.ENDOFPROGRAM) ) {
         ch = aFileManager.getNextChar();
      }  // end of while.
   }

   if (letters.in(ch)) {
      scanIdentifier();
   } else if (digits.in(ch)) {
      scanNumeric();
   } else if (delimiters.in(ch)) {
      scanSpecial();
   } else if (ch == '\n') {
      lineNumber++;
      tok.setClass(NEWLINE);
      tok.setValue(lineNumber);
      aTokFiler.emitToken(tok);
      ch = aFileManager.getNextChar();
   } else if (aFileManager.ENDOFPROGRAM) {
      ; //do nothing -- eof reached
   } else {
      strcatHelper[0] = ch;
      strcat(word, strcatHelper);
      error(2, lineNumber);
      cout << "In getNextWord() 1, word == " << word << endl;
   }

   return word;
}  // end of method getNextWord().
 
void scanner::program() {
   aTokFiler.createTokenFile();
   tok.setClass(NEWLINE);
   tok.setValue(lineNumber);
   aTokFiler.emitToken(tok);
   ch = aFileManager.getNextChar();
   int wordCount = 0;
   aFileManager.ENDOFPROGRAM = false;

   while (!aFileManager.ENDOFPROGRAM) {
      char* aStr = getNextWord();
      wordCount++;
      if (wordCount > 100) {
         aFileManager.ENDOFPROGRAM = true;
      }
   }
   tok.setClass(ENDOFTEXT);
   tok.setValue(0);
   aTokFiler.emitToken(tok);
   aTokFiler.closeTokenFile();
}  // end of method program().