private MiniPL()
{
// Create NFA-type things for tokens using regular operations
Regex reBlockCommentStart = Regex.Concat("/*");
Regex reBlockCommentEnd = Regex.Concat("*/");
Regex reLineComment = Regex.Concat("//").Concat(Regex.Not('\n').Star());
Regex reWhitespace = Regex.Union(" \t\r\n").Star().Union(reLineComment);
Regex reString = Regex.Char('"').Concat(Regex.Char('\\').Concat(Regex.Any()).Union(Regex.Not('"', '\\')).Star()).Concat(Regex.Char('"'));
Regex reBinaryOperator = Regex.Union("+-*/<=&");
Regex reUnaryOperator = Regex.Char('!');
Regex reKeyword = Regex.Union(Regex.Concat("var"), Regex.Concat("for"), Regex.Concat("end"), Regex.Concat("in"),
Regex.Concat("do"), Regex.Concat("read"), Regex.Concat("print"), Regex.Concat("assert"));
Regex reType = Regex.Union(Regex.Concat("bool"), Regex.Concat("int"), Regex.Concat("string"));
Regex reParenRight = Regex.Char(')'),
reParenLeft = Regex.Char('('),
reColon = Regex.Char(':'),
reSemicolon = Regex.Char(';'),
reAssignment = Regex.Concat(":="),
reDots = Regex.Concat("..");
Regex reIdentifier = Regex.Union(Regex.Range('A', 'Z'), Regex.Range('a', 'z'))
.Concat(Regex.Union(Regex.Range('A', 'Z'), Regex.Range('a', 'z'), Regex.Range('0', '9'), Regex.Char('_')).Star());
Regex reInteger = Regex.Range('0', '9').Plus();
// Define token types
tokenTypes["block_comment_start"] = new TokenType("block_comment_start", reBlockCommentStart);
tokenTypes["block_comment_end"] = new TokenType("block_comment_end", reBlockCommentEnd);
tokenTypes["int"] = new TokenType("int", reInteger);
tokenTypes["whitespace"] = new TokenType("whitespace", reWhitespace, priority: TokenType.Priority.Whitespace);
tokenTypes["string"] = new TokenType("string", reString);
tokenTypes["binary_op"] = new TokenType("binary op", reBinaryOperator);
tokenTypes["unary_op"] = new TokenType("unary op", reUnaryOperator);
tokenTypes["keyword"] = new TokenType("keyword", reKeyword, priority: TokenType.Priority.Keyword);
tokenTypes["type"] = new TokenType("type", reType, priority: TokenType.Priority.Keyword);
tokenTypes["left_paren"] = new TokenType("left paren", reParenLeft);
tokenTypes["right_paren"] = new TokenType("right paren", reParenRight);
tokenTypes["colon"] = new TokenType("colon", reColon);
tokenTypes["semicolon"] = new TokenType("semicolon", reSemicolon);
tokenTypes["assignment"] = new TokenType("assignment", reAssignment);
tokenTypes["dots"] = new TokenType("dots", reDots);
tokenTypes["identifier"] = new TokenType("identifier", reIdentifier);
// create combined automaton and scanner object
TokenAutomaton automaton = TokenType.CombinedAutomaton(tokenTypes.Values.ToArray());
scanner = new Scanner(automaton, tokenTypes["block_comment_start"], tokenTypes["block_comment_end"]);
// Define nonterminal variables of CFG
nonterminals["program"] = new Nonterminal("PROG");
nonterminals["statements"] = new Nonterminal("STMTS");
nonterminals["statements_head"] = new Nonterminal("STMTS_HEAD");
nonterminals["statements_tail"] = new Nonterminal("STMTS_TAIL");
nonterminals["statement"] = new Nonterminal("STMT");
nonterminals["declaration"] = new Nonterminal("DECL");
nonterminals["declaration_assignment"] = new Nonterminal("DECL_ASSIGN");
nonterminals["expression"] = new Nonterminal("EXPR");
nonterminals["unary_operation"] = new Nonterminal("UNARY_OP");
nonterminals["binary_operation"] = new Nonterminal("BINARY_OP");
nonterminals["operand"] = new Nonterminal("OPND");
// Define terminal variables of CFG
terminals["identifier"] = new Terminal(tokenTypes["identifier"]);
terminals["assert"] = new Terminal("assert");
terminals["print"] = new Terminal("print");
terminals["read"] = new Terminal("read");
terminals["for"] = new Terminal("for");
terminals["in"] = new Terminal("in");
terminals["end"] = new Terminal("end");
terminals["do"] = new Terminal("do");
terminals["var"] = new Terminal("var");
terminals["type"] = new Terminal(tokenTypes["type"]);
terminals["string"] = new Terminal(tokenTypes["string"]);
terminals["int"] = new Terminal(tokenTypes["int"]);
terminals[")"] = new Terminal(")");
terminals["("] = new Terminal("(");
terminals[".."] = new Terminal("..");
terminals[":="] = new Terminal(":=");
terminals[":"] = new Terminal(":");
terminals[";"] = new Terminal(";");
terminals["binary_operator"] = new Terminal(tokenTypes["binary_op"]);
terminals["unary_operator"] = new Terminal(tokenTypes["unary_op"]);
// Create the Mini-PL grammar
grammar = new CFG(nonterminals["program"], terminals.Values, nonterminals.Values);
// define production rules for the grammar
grammar.AddProductionRule(nonterminals["program"], new ISymbol[] { nonterminals["statements"] });
grammar.AddProductionRule(nonterminals["statements"], new ISymbol[] { nonterminals["statements_head"], nonterminals["statements_tail"] });
grammar.AddProductionRule(nonterminals["statements_head"], new ISymbol[] { nonterminals["statement"], terminals[";"] });
grammar.AddProductionRule(nonterminals["statements_tail"], new ISymbol[] { nonterminals["statements_head"], nonterminals["statements_tail"] });
grammar.AddProductionRule(nonterminals["statements_tail"], new ISymbol[] { Terminal.EPSILON });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { nonterminals["declaration"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["identifier"], terminals[":="], nonterminals["expression"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["for"], terminals["identifier"], terminals["in"], nonterminals["expression"], terminals[".."], nonterminals["expression"], terminals["do"],
nonterminals["statements"], terminals["end"], terminals["for"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["read"], terminals["identifier"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["print"], nonterminals["expression"] });
grammar.AddProductionRule(nonterminals["statement"], new ISymbol[] { terminals["assert"], terminals["("], nonterminals["expression"], terminals[")"] });
grammar.AddProductionRule(nonterminals["declaration"], new ISymbol[] { terminals["var"], terminals["identifier"], terminals[":"], terminals["type"], nonterminals["declaration_assignment"] });
grammar.AddProductionRule(nonterminals["declaration_assignment"], new ISymbol[] { terminals[":="], nonterminals["expression"] });
grammar.AddProductionRule(nonterminals["declaration_assignment"], new ISymbol[] { Terminal.EPSILON });
grammar.AddProductionRule(nonterminals["expression"], new ISymbol[] { nonterminals["unary_operation"] });
grammar.AddProductionRule(nonterminals["expression"], new ISymbol[] { nonterminals["operand"], nonterminals["binary_operation"] });
grammar.AddProductionRule(nonterminals["unary_operation"], new ISymbol[] { terminals["unary_operator"], nonterminals["operand"] });
grammar.AddProductionRule(nonterminals["binary_operation"], new ISymbol[] { terminals["binary_operator"], nonterminals["operand"] });
grammar.AddProductionRule(nonterminals["binary_operation"], new ISymbol[] { Terminal.EPSILON });
grammar.AddProductionRule(nonterminals["operand"], new ISymbol[] { terminals["int"] });
grammar.AddProductionRule(nonterminals["operand"], new ISymbol[] { terminals["string"] });
grammar.AddProductionRule(nonterminals["operand"], new ISymbol[] { terminals["identifier"] });
grammar.AddProductionRule(nonterminals["operand"], new ISymbol[] { terminals["("], nonterminals["expression"], terminals[")"] });
// use ; as synchronizing token for Mini-PL
parser = new Parser(grammar, terminals[";"]);
}