// file: $isip/class/pr/LanguageModelJSGF/lmjsgf_09.cc
//

// system include files
//
#include <typeinfo>

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

// method: convertJSGFtoABNF
//
// arguments:
//   
// return: 
//
ProductionRuleSet LanguageModelJSGF::convertJSGFtoABNF(Vector<JSGFToken> grammar_a) {

  ProductionRuleSet abnf_grammar;
  String start_rule_name;
  int32 token_index = 0;
  
  // iterate over all tokens and construct rules
  //
  while(token_index < grammar_a.length()) {
    
    ProductionRule current_rule;
    
    // find the next rule_name token
    //
    while(token_index < grammar_a.length() &&
	  grammar_a(token_index).getTokenType() != JSGFToken::RULE_NAME) {
      token_index++;
    }
    
    if (token_index < grammar_a.length()) {
      current_rule.clear();
      
      String rule_name(grammar_a(token_index).rulename_d);
      
      current_rule.setRuleName(rule_name);
      token_index++;
    }
    
    float32 weight = 0;
    
    // loop over tokens until a ';' operator is encountered
    //
    while(token_index < grammar_a.length() &&
	  grammar_a(token_index).operator_d(0).ne(L";") &&
	  current_rule.getRuleName().length() > 0) {
      
      if (grammar_a(token_index).getTokenType() == JSGFToken::WEIGHT) {
	
	weight = grammar_a(token_index).weight_d;
	token_index++;
      }
      else if (grammar_a(token_index).getTokenType() == JSGFToken::OPERATOR) {
	
	if (grammar_a(token_index).operator_d(0).eq(L"(")) {
	  current_rule.append(ProductionRuleTokenType::OPEN_PAREN);
	  token_index++;
	}
	else if (grammar_a(token_index).operator_d(0).eq(L")")) {
	  current_rule.append(ProductionRuleTokenType::CLOSE_PAREN);
	  token_index++;
	}
	else if (grammar_a(token_index).operator_d(0).eq(L"|")) {
	  current_rule.append(ProductionRuleTokenType::ALTERNATION);
	  token_index++;
	}
	else {
	  token_index++;
	}		    
      }
      else if (grammar_a(token_index).getTokenType() == JSGFToken::TERMINAL) {
	if (grammar_a(token_index).terminal_d.eq(START_SYMBOL)) {
	  current_rule.setRuleType(ProductionRule::START);
	}
	else if (grammar_a(token_index).terminal_d.eq(TERM_SYMBOL)) {
	  
	  current_rule.append(ProductionRuleTokenType::OPEN_PAREN, L"", weight);
	  current_rule.append(ProductionRuleTokenType::NON_TERMINAL,
			      grammar_a(token_index).terminal_d);
	  current_rule.append(ProductionRuleTokenType::CLOSE_PAREN);

	  ProductionRule epsilon_rule;

	  epsilon_rule.setRuleName(grammar_a(token_index).terminal_d);
	  epsilon_rule.append(ProductionRuleTokenType::EPSILON);

	  // check to see if we already added an epsilon rule.  if not
	  // add it to the productionruleset
	  //
	  if (!abnf_grammar.contains(&epsilon_rule)) {
	    abnf_grammar.concat(epsilon_rule);
	  }	  
	}
	else {		
	  current_rule.append(ProductionRuleTokenType::TERMINAL,
			      grammar_a(token_index).terminal_d);
	  current_rule.append(ProductionRuleTokenType::CONCATENATION);
	}
	token_index++;	      
      }
      else if (grammar_a(token_index).getTokenType() == JSGFToken::RULE_NAME) {
	
	// don't do anything if this is the start rule reference.  We've already
	// handled this by making this rule a START rule
	//
	if (grammar_a(token_index).rulename_d.ne(START_SYMBOL_REFERENCE)) {
	  current_rule.append(ProductionRuleTokenType::OPEN_PAREN, L"", weight);
		current_rule.append(ProductionRuleTokenType::NON_TERMINAL,
				    grammar_a(token_index).rulename_d);	  
		current_rule.append(ProductionRuleTokenType::CLOSE_PAREN);
		token_index++;
	}
      }
      else {
	token_index++;
      }
    }
    abnf_grammar.concat(current_rule);
  }
  
  
  return abnf_grammar;
}

// method: convertBNFtoJSGF
//
// arguments:
//   
// return: 
//
Vector<JSGFToken> LanguageModelJSGF::convertBNFtoJSGF(ProductionRuleSet rules_a) {

  Vector<JSGFToken> grammar;
  JSGFToken current_token;
  Vector<SearchSymbol> graph_symbols;
  String term_rulename;
  
  // we first need to determine the unique nodes in this graph.  this is accomplished
  // by finding unique pairs of rule_names and terminal tokens
  //
  Vector< Pair<String, SearchSymbol> > nodes;

  // we also need to keep up with the nodes that have already been visited.
  //
  Vector< Pair<String, SearchSymbol> > nodes_visited;  
  
  // loop over all rules in this graph
  //
  for (int32 i = 0; i < rules_a.length(); i++) {
    
    rules_a(i).gotoFirst();
    
    do {
      
      if (rules_a(i).length() > 0) {

	SearchSymbol symbol = rules_a(i).getValue();
	
	// try to identify unique nodes by finding unique pairs of rule
	// names and terminal tokens. this is an attempt to minimize
	// the graph as much as possible.
	//
	Pair<String, SearchSymbol> node(rules_a(i).getRuleName(),
					symbol);
	
	if (rules_a(i).getType() == ProductionRuleTokenType::TERMINAL
	    && !nodes.contains(&node)) {
	  nodes.concat(node);
	}
      }
    } while (rules_a(i).gotoNext());
  }
  
  // we need to find which rule is the epsilon rule
  //
  for (int32 i = 0; i < rules_a.length(); i++) {
    rules_a(i).gotoFirst();
    if (rules_a(i).length() == 1 && rules_a(i).getType() == ProductionRuleTokenType::EPSILON) {
      term_rulename.assign(rules_a(i).getRuleName());
    }    
  }

  // all rules will be converted to JSGFTokens and added to a single
  // vector of  JSGFTokens.
  //

  // add the appropriate header information
  //  - find start rule and set the grammar name to the name of the start rule
  //  - add the appropriate keywords, and JSGF headers
  //

  // first token is the header token
  //
  current_token.setTokenType(JSGFToken::HEADER);
  current_token.setHeader(PARAM_JSGF_VERSION);
  grammar.concat(current_token);
  current_token.clear();
  
  // next token is the grammar keyword and declaration
  //
  current_token.setTokenType(JSGFToken::GRAMMAR_NAME);
  current_token.setKeyword(KEYWORD_GRAMMAR);
  String grammar_definition(GRAMMAR_DEF_BASE);
  String grammar_name;
  
  // look for start rule and set grammar definition
  //
  int32 index = -1;

  do {
    index++;
    if (rules_a(index).getRuleType() == ProductionRule::START || rules_a(index).length() == 0) {
      grammar_name.assign(rules_a(index).getRuleName());
    }
  } while(rules_a(index).getRuleType() != ProductionRule::START && index < rules_a.length()-1);

  grammar_definition.concat(grammar_name);
  current_token.setGrammarName(grammar_definition);
  grammar.concat(current_token);
  current_token.clear();
  
  // close grammar definition with semicolon operator
  //
  current_token.setTokenType(JSGFToken::OPERATOR);
  current_token.operator_d(0).assign(OPERATOR_SEMICOLON);
  grammar.concat(current_token);
  current_token.clear();

  // add the public keyword
  //
  current_token.setTokenType(JSGFToken::KEYWORD);
  current_token.setKeyword(KEYWORD_PUBLIC);
  grammar.concat(current_token);
  current_token.clear();

  // set the rulename token
  //
  current_token.setTokenType(JSGFToken::RULE_NAME);
  current_token.setRuleName(grammar_name);
  current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_OPENANGLEBRACKET;
  current_token.operator_d(OPERATOR_CLOSE_POS) = OPERATOR_CLOSEANGLEBRACKET;
  grammar.concat(current_token);
  current_token.clear();  
  
  // add the equal operator
  //
  current_token.setTokenType(JSGFToken::OPERATOR);
  current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_EQUALS;  
  grammar.concat(current_token);
  current_token.clear();

  // we will first define the JSGF start rule if it's not a dummy  
  //
  if (rules_a.length() > 1) {
    current_token.setTokenType(JSGFToken::TERMINAL);
    current_token.setTerminal(START_SYMBOL);  
    grammar.concat(current_token);
    current_token.clear();
  }

  Vector<ProductionRule> temp_rules;
  
  // the first JSGF rule following the public keyword needs to be the start rule
  //
  for (int32 i = 0; i < rules_a.length(); i++) {
    if (rules_a(i).getRuleType() == ProductionRule::START && rules_a(i).getRuleName().eq(grammar_name)) {
      temp_rules.concat(rules_a(i));
      rules_a.deleteRange(i--,1);
    } 
  }

  if (temp_rules.length() > 1) {
    current_token.setTokenType(JSGFToken::OPERATOR);
    current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_OPENPARENS;  
    grammar.concat(current_token);
    current_token.clear();
  }

  for (int32 i = 0; i < temp_rules.length(); i++) {

    temp_rules(i).gotoFirst();
    ProductionRule ref_rule;
    String rulename;
    float32 weight = 0;
    Pair<String, SearchSymbol> node;
    
    current_token.setTokenType(JSGFToken::OPERATOR);
    current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_OPENPARENS;  
    grammar.concat(current_token);
    current_token.clear();
    
    if (temp_rules(i).length() == 1 && temp_rules(i).getType() == ProductionRuleTokenType::NON_TERMINAL) {
      ref_rule = findRule(temp_rules(i).getValue(), rules_a);
      weight = temp_rules(i).getWeight();
    }
    else {
      Error::handle(L"LanguageModelJSGF",
		    L"convertJSGFtoABNF - invalid start rule format",
		    Error::ERROR, __FILE__,
		    __LINE__);
    }

    current_token.setTokenType(JSGFToken::WEIGHT);
    current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_FWDSLASH;  
    current_token.operator_d(OPERATOR_CLOSE_POS) = OPERATOR_FWDSLASH;  
    current_token.setWeight(weight);
    grammar.concat(current_token);
    current_token.clear();
    
    ref_rule.gotoFirst();

    node.assign(ref_rule.getRuleName(), ref_rule.getValue());

    rulename.assign(RULE_NAME_BASE);
    rulename.concat(nodes.first(node));

    current_token.setTokenType(JSGFToken::RULE_NAME);
    current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_OPENANGLEBRACKET;
    current_token.operator_d(OPERATOR_CLOSE_POS) = OPERATOR_CLOSEANGLEBRACKET;
    current_token.setRuleName(rulename);
    grammar.concat(current_token);
    current_token.clear();
          
    current_token.setTokenType(JSGFToken::OPERATOR);
    current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_CLOSEPARENS;  
    grammar.concat(current_token);
    current_token.clear();    

    if (i < temp_rules.length()-1) {
      current_token.setTokenType(JSGFToken::OPERATOR);
      current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_ALTERNATION;
      grammar.concat(current_token);
      current_token.clear();
    }    
  }

  if (temp_rules.length() > 1) {
    current_token.setTokenType(JSGFToken::OPERATOR);
    current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_CLOSEPARENS;  
    grammar.concat(current_token);
    current_token.clear();
  }

  current_token.setTokenType(JSGFToken::OPERATOR);
  current_token.operator_d(0).assign(OPERATOR_SEMICOLON);
  grammar.concat(current_token);
  current_token.clear();
  
  // iterate over rules.  for each rule, check the remaining rules to find rules
  // with the same rulename and nonterminal.  when these are found, and the
  // appropriate JSGF tokens have been added to the token vector, remove them
  // from the rules_a list.  this will be implemented with an outerloop to loop
  // over rules, and an inner loop to loop over the remaining rules and find
  // matches.
  //
  for (int32 i = 0; i < rules_a.length(); i++) {

    temp_rules.clear();
    ProductionRule current_rule = rules_a(i);
    Pair<String, SearchSymbol> node;
    
    node.assign(current_rule.getRuleName(), current_rule.getValue());

    if (current_rule.length() == 3 && nodes_visited.first(node) < 0) {
      
      nodes_visited.concat(node);

      String rulename(RULE_NAME_BASE);
      rulename.concat(nodes.first(node));

      current_token.setTokenType(JSGFToken::RULE_NAME);
      current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_OPENANGLEBRACKET;
      current_token.operator_d(OPERATOR_CLOSE_POS) = OPERATOR_CLOSEANGLEBRACKET;      current_token.setRuleName(rulename);
      grammar.concat(current_token);
      current_token.clear();

      current_token.setTokenType(JSGFToken::OPERATOR);
      current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_EQUALS;  
      grammar.concat(current_token);
      current_token.clear();

      current_token.setTokenType(JSGFToken::TERMINAL);
      current_token.setTerminal(current_rule.getValue());  
      grammar.concat(current_token);
      current_token.clear();
      
      // the first JSGF rule following the public keyword needs to be the start rule
      //
      for (int32 j = i; j < rules_a.length(); j++) {
	if (rules_a(j).getRuleName().eq(current_rule.getRuleName())) {
	  temp_rules.concat(rules_a(j));
	} 
      }

      if (temp_rules.length() > 1) {
	current_token.setTokenType(JSGFToken::OPERATOR);
	current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_OPENPARENS;  
	grammar.concat(current_token);
	current_token.clear();
      }

      for (int32 i = 0; i < temp_rules.length(); i++) {
	
	temp_rules(i).gotoFirst();
	ProductionRule ref_rule;
	String rulename;
	float32 weight;
	Pair<String, SearchSymbol> node;
	
	current_token.setTokenType(JSGFToken::OPERATOR);
	current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_OPENPARENS;  
	grammar.concat(current_token);
	current_token.clear();
	
	if (temp_rules(i).length() == 3) {

	  temp_rules(i).gotoNext();
	  weight = temp_rules(i).getWeight();	     
	  temp_rules(i).gotoNext();
	  
	  if(temp_rules(i).getType() == ProductionRuleTokenType::NON_TERMINAL) {
	    ref_rule = findRule(temp_rules(i).getValue(), rules_a);
	  }
	  else {
	    Error::handle(L"LanguageModelJSGF",
			  L"convertJSGFtoABNF - invalid rule format",
			  Error::ERROR, __FILE__,
			  __LINE__);
	  }
	  
	  
	  current_token.setTokenType(JSGFToken::WEIGHT);
	  current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_FWDSLASH;  
	  current_token.operator_d(OPERATOR_CLOSE_POS) = OPERATOR_FWDSLASH;  
	  current_token.setWeight(weight);
	  grammar.concat(current_token);
	  current_token.clear();

	  if (ref_rule.getRuleName().eq(term_rulename)) {
	    
	    rulename.assign(PARAM_JSGF_TERM_SYMBOL);
	  }
	  else {
	    ref_rule.gotoFirst();
	    
	    node.assign(ref_rule.getRuleName(), ref_rule.getValue());

	    rulename.assign(RULE_NAME_BASE);
	    rulename.concat(nodes.first(node));
	  }
	  
	  current_token.setTokenType(JSGFToken::RULE_NAME);
	  current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_OPENANGLEBRACKET;
	  current_token.operator_d(OPERATOR_CLOSE_POS) = OPERATOR_CLOSEANGLEBRACKET;
	  current_token.setRuleName(rulename);
	  current_token.setRuleName(rulename);
	  grammar.concat(current_token);
	  current_token.clear();
	  
	  current_token.setTokenType(JSGFToken::OPERATOR);
	  current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_CLOSEPARENS;  
	  grammar.concat(current_token);
	  current_token.clear();    
	  
	  if (i < temp_rules.length()-1) {
	    current_token.setTokenType(JSGFToken::OPERATOR);
	    current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_ALTERNATION;
	    grammar.concat(current_token);
	    current_token.clear();
	  }
	  else if (temp_rules.length() > 1) {
	    current_token.setTokenType(JSGFToken::OPERATOR);
	    current_token.operator_d(OPERATOR_OPEN_POS) = OPERATOR_CLOSEPARENS;
	    grammar.concat(current_token);
	    current_token.clear();

	  }	  	  
	}
      }
      current_token.setTokenType(JSGFToken::OPERATOR);
      current_token.operator_d(0).assign(OPERATOR_SEMICOLON);
      grammar.concat(current_token);
      current_token.clear();	          
    }
  }
  return grammar;

}

// method: findRule
//
// arguments:
//   
// return: 
//
ProductionRule LanguageModelJSGF::findRule(String ruleref_a, ProductionRuleSet rules_a) {

  ProductionRule def;
  
  for (int32 i = 0; i < rules_a.length(); i++) {

    if (rules_a(i).getRuleName().eq(ruleref_a)) {
      rules_a(i).gotoFirst();
      if (rules_a(i).getType() == ProductionRuleTokenType::TERMINAL ||
	  rules_a(i).getType() == ProductionRuleTokenType::EPSILON) {
	return rules_a(i);
      }
      else if (rules_a(i).getType() != ProductionRuleTokenType::EPSILON) {
	return findRule(rules_a(i).getValue(), rules_a);
      }
      
    }
  }

  return def;
}