// file: $isip/class/asr/JSGFParser/jp_06.cc
// version: $Id: jp_06.cc 10499 2006-03-15 21:43:17Z may $
//

// isip include files
//
#include "JSGFParser.h"

// method: parseGrammar
//
// arguments: none
//
// return: a bool8 value indicating status
//
// This method checks through the class-protected token vector to
// validate the input JSGF grammar while saving some usefule data
//
bool8 JSGFParser::parseGrammar() {
  
  // declare a current token
  //
  JSGFToken curr_tok;
  
  // get the number of tokens in the token vector
  //
  int32 num_tok = token_vect_d.length();

  // make sure the class-protected token index is zero
  //
  token_index_d = 0;

  // set the first token as the current token
  //
  if (token_index_d < num_tok) {
    curr_tok = token_vect_d(token_index_d);
  }
  else {
    return Error::handle(name(), L"missing grammar header",
			 Error::TEST, __FILE__, __LINE__);
  }

  // the first token in the grammar must be a header token
  //
  if (curr_tok.getTokenType() == JSGFToken::HEADER) {

    // move to the next token
    //
    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(), L"missing grammar name declaration",
			   Error::TEST, __FILE__, __LINE__);
    }
  }
  else {
    Console::put(L"***unexpected token:");
    curr_tok.printToken();
    return Error::handle(name(),
	   L"missing self-identifying header at the beginning of the grammar",
			 Error::TEST, __FILE__, __LINE__);
  }
  
  // The current token must be a grammar declaration
  //
  if (curr_tok.getTokenType() == JSGFToken::GRAMMAR_NAME) {
    
    // store the grammar name as a class-protected data
    //
    grammar_name_d.assign(curr_tok.grammar_name_d);

    // move to the next
    //
    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(), L"missing grammar body",
			   Error::TEST, __FILE__, __LINE__);
    }
  }
  else {
    Console::put(L"***unexpected token:");
    curr_tok.printToken();
    return Error::handle(name(),
			 L"missing grammar declaration after the header",
			 Error::TEST, __FILE__, __LINE__);
  }

  // the current token must be a semi-colon
  //
  if (curr_tok.operator_d(0).eq(L";")) {
    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(), L"missing grammar body",
			   Error::TEST, __FILE__, __LINE__);
    }
  }
  else {
    return Error::handle(name(),
			 L"missing semi-colon in grammar declaration",
			 Error::TEST, __FILE__, __LINE__);
  }
  
  // The current token can be an import grammar declaration
  //
  while (curr_tok.getTokenType() == JSGFToken::IMPORT_GRAMMAR_NAME) {

    // get import rules
    //
    //getImport(curr_tok);
      
    // move to the next token
    //
    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(), L"missing grammar body",
			   Error::TEST, __FILE__, __LINE__);
    }

    // the current token must be a semi-colon
    //
    if (curr_tok.operator_d(0).eq(L";")) {
      token_index_d++;
      if (token_index_d < num_tok) {
	curr_tok = token_vect_d(token_index_d);
      }
      else {
	return Error::handle(name(), L"missing grammar body",
			     Error::TEST, __FILE__, __LINE__);
      }
    }
    else {
      return Error::handle(name(),
			   L"missing semi-colon in import grammar declaration",
			   Error::TEST, __FILE__, __LINE__);
    }
  } // end: while (curr_tok.getTokenType() == 3)

  // Then starting from the current token, the rule definitions come
  // so make sure each rule definition is valid
  //
  while (token_index_d < num_tok) {
    parseRuleDefinition();
    token_index_d++;
  }

  // gracefully exit
  //
  return true;
}


// method: parseRuleDefinition
//
// arguments: none
//
// return: a bool8 value indicating status
//
// This method checks if a rule definition is valid
//
bool8 JSGFParser::parseRuleDefinition() {

  // set grouping and optional grouping match flag as zero
  //
  grouping_match_d = 0;
  op_grouping_match_d = 0;
  
  // get the total number of tokens
  //
  int32 num_tok = token_vect_d.length();

  // initalize current token
  //
  JSGFToken curr_tok = token_vect_d(token_index_d);
  
  // The first token can be a public keyword
  //
  if (curr_tok.getTokenType() == JSGFToken::KEYWORD) {

    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(), L"missing rule definition",
			 Error::TEST, __FILE__, __LINE__);
    }
  }

  // The current token can be a rule name
  //
  if (curr_tok.getTokenType() == JSGFToken::RULE_NAME) {

    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(), L"missing rule definition",
			 Error::TEST, __FILE__, __LINE__);
    }
  }
  // error if the current token is neither of the above
  //
  else {
    Console::put(L"***unexpected token:");
    curr_tok.printToken();
    return Error::handle(name(), L"invalid rule definition",
			 Error::TEST, __FILE__, __LINE__);
  }

  // The current token must be operator=
  //
  if (curr_tok.operator_d(0).eq(L"=")) {
    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(),
			   L"missing rule exapnsion for rule definition",
			   Error::TEST, __FILE__, __LINE__);
    }
  }
  else {
    Console::put(L"***unexpected token:");
    curr_tok.printToken();
    return Error::handle(name(),
			 L"missing operator= in the rule definition",
			 Error::TEST, __FILE__, __LINE__);
  }

  // starting from the current token, check if a valid rule expansion exists
  // if there isn't an expansion, this is a dummy symbol
  //
  if (curr_tok.operator_d(0).ne(L";")) {
    parseRuleExpansion();
  }

  // check grouping flags
  //
  if (grouping_match_d != 0) {
    return Error::handle(name(), L"missing left brace ( in grouping",
			 Error::TEST, __FILE__, __LINE__);
  }

  if (op_grouping_match_d != 0) {
    return Error::handle(name(),
			 L"missing left bracket [ in optional grouping",
			 Error::TEST, __FILE__, __LINE__);
  }
  
  // gracefully exit
  //
  return true;
}


// method: parseRuleExpansion
//
// arguments: none
//
// return: a bool8 value indicating status
//
// This method checks if a rule expansion is valid
//
bool8 JSGFParser::parseRuleExpansion() {

  // get the total number of tokens
  //
  int32 num_tok = token_vect_d.length();

  // initalize current token
  //
  JSGFToken curr_tok = token_vect_d(token_index_d);
  
  while (curr_tok.operator_d(0).ne(L";") &&
	 curr_tok.operator_d(0).ne(L")") &&
	 curr_tok.operator_d(0).ne(L"]")) {

    // bool8 variable to check if alternative is weighted or not
    //
    bool8 weighted = false;
    
    // the current token can be weight
    //
    if (curr_tok.getTokenType() == JSGFToken::WEIGHT) {

      // update the weighted flag
      //
      weighted = true;

      // move to the next
      //
      token_index_d++;
      if (token_index_d < num_tok) {
	curr_tok = token_vect_d(token_index_d);
      }
      else {
	return Error::handle(name(),
			     L"missing semi-colon for rule definition",
			     Error::TEST, __FILE__, __LINE__);
      }
    }

    // the current token must lead to a unit
    //
    parseUnit();
    curr_tok = token_vect_d(token_index_d);

    // current token can be a vertical bar | indicating alternative
    //
    if (curr_tok.operator_d(0).eq(L"|")) {
      token_index_d++;
      if (token_index_d < num_tok) {
	curr_tok = token_vect_d(token_index_d);
      }
      else {
	return Error::handle(name(),
			     L"missing semi-colon for rule definition",
			     Error::TEST, __FILE__, __LINE__);
      }
    
      // semi-colon can not follow vertical bar
      //
      if (curr_tok.operator_d(0).eq(L";") ||
	  curr_tok.operator_d(0).eq(L")") ||
	  curr_tok.operator_d(0).eq(L"]")) {

	return Error::handle(name(),
               	      L"invalid rule component after the vertical bar |",
			     Error::TEST, __FILE__, __LINE__);
      }

      // if a weight is specified for one item in a set of alternatives
      // then a weight must be specified for every item
      // if a weight token is set for the unit before the vertical bar |
      // the current token is required to be weight
      //
      if ((weighted) && (curr_tok.getTokenType() != JSGFToken::WEIGHT)) {
	Console::put(L"***unexpected token:");
	curr_tok.printToken();
	return Error::handle(name(),
		      L"no matching weight token after the vertical bar |",
			     Error::TEST, __FILE__, __LINE__);
      }


      // error if a weight token is not set for the unit before
      // the vertical bar |  while the current token is weight token
      //
      if ((!weighted) && (curr_tok.getTokenType() == JSGFToken::WEIGHT)) {
	Console::put(L"***unexpected token:");
	curr_tok.printToken();
	return Error::handle(name(),
		      L"no matching weight token before the vertical bar |",
			     Error::TEST, __FILE__, __LINE__);
      }
    }
  } // end: while (curr_tok.operator_d(0).ne(L";") && ...)
  
  // update grouping match flags
  //
  if (curr_tok.operator_d(0).eq(L")")) {
    grouping_match_d--;
  }
  else if (curr_tok.operator_d(0).eq(L"]")) {
    op_grouping_match_d--;
  }  
    
  // gracefully exit
  //
  return true;
}


// method: parseUnit
//
// arguments: none
//
// return: a bool8 value indicating status
//
// This method checks if a rule expansion is valid
//
bool8 JSGFParser::parseUnit() {

  // get the total number of tokens
  //
  int32 num_tok = token_vect_d.length();

  // initalize current token
  //
  JSGFToken curr_tok = token_vect_d(token_index_d);

  // the current can be terminal symbol, quoted token or rule reference
  //
  if ((curr_tok.getTokenType() == JSGFToken::RULE_NAME)
      || (curr_tok.getTokenType() == JSGFToken::TERMINAL) ||
      (curr_tok.getTokenType() == JSGFToken::QUOTED_TOKEN)) {
  
    // move to the next
    //
    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(),
			   L"missing semi-colon for rule definition",
			   Error::TEST, __FILE__, __LINE__);
    }
  }
  // the current token can lead to a grouping with () parentheses
  //
  else if (curr_tok.operator_d(0).eq(L"(")){
    grouping_match_d++;
    parseGrouping();
    curr_tok = token_vect_d(token_index_d);
  }
  // the current token can lead to an optional grouping with [] brackets
  //
  else if (curr_tok.operator_d(0).eq(L"[")){
    op_grouping_match_d++;
    parseOptionalGrouping();
    curr_tok = token_vect_d(token_index_d);
  }
  // otherwise invalid
  //
  else {
    Console::put(L"***unexpected token:");
    curr_tok.printToken();
    return Error::handle(name(),
			 L"invalid use of rule component at this position",
			 Error::TEST, __FILE__, __LINE__);
  }

  // the current token can be a unary operator * or +
  //
  if (curr_tok.operator_d(0).eq(L"*") || curr_tok.operator_d(0).eq(L"+")) {
     token_index_d++;
     if (!(token_index_d < num_tok)) {
       return Error::handle(name(),
			    L"missing semi-colon for rule definition",
			    Error::TEST, __FILE__, __LINE__);
     }
  }
  // or the current token can also be a tag
  //
  else if (curr_tok.getTokenType() == JSGFToken::TAG) {

    // multiple tags followed are allowed
    //
    while (curr_tok.getTokenType() == JSGFToken::TAG) {
      token_index_d++;
      if (token_index_d < num_tok) {
	curr_tok = token_vect_d(token_index_d);
      }
      else {
	return Error::handle(name(),
			     L"missing semi-colon for rule definition",
			     Error::TEST, __FILE__, __LINE__);
      }
    }
  }

  // gracefully exit
  //
  return true;
}


// method: parseGrouping
//
// arguments: none
//
// return: a bool8 value indicating status
//
// This method checks if a rule expansion is valid
//
bool8 JSGFParser::parseGrouping() {

  // get the total number of tokens
  //
  int32 num_tok = token_vect_d.length();

  // declare current token
  //
  JSGFToken curr_tok;

  // move to the next
  //
  token_index_d++;
  curr_tok = token_vect_d(token_index_d);
  
  // make sure the current token is not right brace )
  // empty grouping () is illegal 
  //
  if (curr_tok.operator_d(0).eq(L")")) {
    return Error::handle(name(), L"empty grouping () is illegal",
			 Error::TEST, __FILE__, __LINE__);
  }
  
  // followed can be any valid rule expansion
  //
  if (token_index_d < num_tok) {
    parseRuleExpansion();
  }
  else {
    return Error::handle(name(), L"invalid using ( left brace",
			 Error::TEST, __FILE__, __LINE__);
  }
  
  // get the current token
  //
  curr_tok = token_vect_d(token_index_d);

  // the current token must be ) right parentheses
  //
  if (curr_tok.operator_d(0).eq(L")")) {

    // move to the next
    //
    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(),
			   L"missing semi-colon for rule definition",
			   Error::TEST, __FILE__, __LINE__);
    } 
  }
  else {
    Console::put(L"***unexpected token:");
    curr_tok.printToken();
    return Error::handle(name(), L"missing ) right brace in grouping",
			 Error::TEST, __FILE__, __LINE__);
  }

  // gracefully exit
  //
  return true;
}


// method: parseOptionalGrouping
//
// arguments: none
//
// return: a bool8 value indicating status
//
// This method checks if a rule expansion is valid
//
bool8 JSGFParser::parseOptionalGrouping() {

  // get the total number of tokens
  //
  int32 num_tok = token_vect_d.length();

  // declare current token
  //
  JSGFToken curr_tok;

  // move to the next
  //
  token_index_d++;
  curr_tok = token_vect_d(token_index_d);

  // make sure the current token is not right brace )
  // empty optional grouping [] is illegal 
  //
  if (curr_tok.operator_d(0).eq(L"]")) {
    return Error::handle(name(), L"empty optional grouping [] is illegal",
			 Error::TEST, __FILE__, __LINE__);
  }
  
  // followed can be any valid rule expansion
  //
  if (token_index_d < num_tok) {
    parseRuleExpansion();
  }
  else {
    return Error::handle(name(), L"invalid using [ left brace",
			 Error::TEST, __FILE__, __LINE__);
  }
  
  // get the current token
  //
  curr_tok = token_vect_d(token_index_d);

  // the current token must be ] right bracket
  //
  if (curr_tok.operator_d(0).eq(L"]")) {

    // move to the next
    //
    token_index_d++;
    if (token_index_d < num_tok) {
      curr_tok = token_vect_d(token_index_d);
    }
    else {
      return Error::handle(name(),
			   L"missing semi-colon for rule definition",
			   Error::TEST, __FILE__, __LINE__);
    } 
  }
  else {
    Console::put(L"***unexpected token:");
    curr_tok.printToken();
    return Error::handle(name(),
			 L"missing ] right bracket in optional grouping",
			 Error::TEST, __FILE__, __LINE__);
  }
  
  // gracefully exit
  //
  return true;
}