public ScriptdotnetGrammar(bool expressionGrammar)
{
#region 1. Terminals
NumberLiteral n = TerminalFactory.CreateCSharpNumber("number");
IdentifierTerminal v = CreateScriptNetIdentifier("Identifier");
Terminal s = CreateScriptNetString("string");
Terminal @is = Symbol("is");
Terminal dot = Symbol(".", "dot");
Terminal less = Symbol("<");
Terminal greater = Symbol(">");
Terminal arrow = Symbol("->");
Terminal LSb = Symbol("[");
Terminal RSb = Symbol("]");
Terminal LCb = Symbol("(");
Terminal RCb = Symbol(")");
Terminal RFb = Symbol("}");
Terminal LFb = Symbol("{");
Terminal LMb = Symbol("<!");
Terminal RMb = Symbol("!>");
Terminal LGb = Symbol("<|");
Terminal RGb = Symbol("|>");
Terminal comma = Symbol(",");
Terminal semicolon = Symbol(";");
Terminal colon = Symbol(":");
#endregion
#region 2. Non-terminals
#region 2.1 Expressions
NonTerminal Expr = new NonTerminal("Expr", typeof(ScriptExpr));
NonTerminal ConstExpr = new NonTerminal("ConstExpr", typeof(ScriptConstExpr));
NonTerminal BinExpr = new NonTerminal("BinExpr", typeof(ScriptBinExpr));
NonTerminal UnaryExpr = new NonTerminal("UnaryExpr", typeof(ScriptUnaryExpr));
NonTerminal AssignExpr = new NonTerminal("AssignExpr", typeof(ScriptAssignExpr));
NonTerminal TypeConvertExpr = new NonTerminal("TypeConvertExpr", typeof(ScriptTypeConvertExpr));
NonTerminal IsExpr = new NonTerminal("IsExpr", typeof(ScriptIsExpr));
NonTerminal MetaExpr = new NonTerminal("MetaExpr", typeof(ScriptMetaExpr));
NonTerminal FuncDefExpr = new NonTerminal("FuncDefExpr", typeof(ScriptFunctionDefinition)); //typeof(ScriptFunctionDefExpression));
NonTerminal TypeExpr = new NonTerminal("TypeExpr", typeof(ScriptTypeExpr));
NonTerminal TypeConstructor = new NonTerminal("TypeConstructor", typeof(ScriptTypeConstructor));
NonTerminal FunctionCall = new NonTerminal("FunctionCall", typeof(ScriptFunctionCall));
NonTerminal ArrayResolution = new NonTerminal("ArrayResolution", typeof(ScriptArrayResolution));
NonTerminal BinOp = new NonTerminal("BinOp");
NonTerminal LUnOp = new NonTerminal("LUnOp");
NonTerminal RUnOp = new NonTerminal("RUnOp");
NonTerminal ArrayConstructor = new NonTerminal("ArrayConstructor", typeof(ScriptArrayConstructor));
NonTerminal MObjectConstructor = new NonTerminal("MObjectConstructor", typeof(ScriptMObject));
NonTerminal MObjectPart = new NonTerminal("MObjectPart", typeof(ScriptMObjectPart));
NonTerminal MObjectParts = new NonTerminal("MObjectPart", typeof(ScriptAst));
NonTerminal TypeList = new NonTerminal("TypeList", typeof(ScriptTypeExprList));
#endregion
#region 2.2 QualifiedName
//Expression List: expr1, expr2, expr3, ..
NonTerminal ExprList = new NonTerminal("ExprList", typeof(ScriptExprList));
//A name in form: a.b.c().d[1,2].e ....
NonTerminal NewStmt = new NonTerminal("NewStmt", typeof(ScriptNewStmt));
NonTerminal NewArrStmt = new NonTerminal("NewArrStmt", typeof(ScriptNewArrStmt));
NonTerminal QualifiedName = new NonTerminal("QualifiedName", typeof(ScriptQualifiedName));
NonTerminal GenericsPostfix = new NonTerminal("GenericsPostfix", typeof(ScriptGenericsPostfix));
NonTerminal GlobalList = new NonTerminal("GlobalList", typeof(ScriptGlobalList));
#endregion
#region 2.3 Statement
NonTerminal Condition = new NonTerminal("Condition", typeof(ScriptCondition));
NonTerminal Statement = new NonTerminal("Statement", typeof(ScriptStatement));
NonTerminal IfStatement = new NonTerminal("IfStatement", typeof(ScriptIfStatement));
NonTerminal WhileStatement = new NonTerminal("WhileStatement", typeof(ScriptWhileStatement));
NonTerminal ForStatement = new NonTerminal("ForStatement", typeof(ScriptForStatement));
NonTerminal ForEachStatement = new NonTerminal("ForEachStatement", typeof(ScriptForEachStatement));
NonTerminal OptionalExpression = new NonTerminal("OptionalExpression", typeof(ScriptExpr));
NonTerminal SwitchStatement = new NonTerminal("SwitchStatement", typeof(ScriptStatement));
NonTerminal SwitchStatements = new NonTerminal("SwitchStatements", typeof(ScriptSwitchStatement));
NonTerminal SwitchCaseStatement = new NonTerminal("SwitchCaseStatement", typeof(ScriptSwitchCaseStatement));
NonTerminal SwitchDefaultStatement = new NonTerminal("SwitchDefaultStatement", typeof(ScriptSwitchDefaultStatement));
NonTerminal UsingStatement = new NonTerminal("UsingStatement", typeof(ScriptUsingStatement));
NonTerminal TryCatchFinallyStatement = new NonTerminal("TryCatchFinallyStatement", typeof(ScriptTryCatchFinallyStatement));
NonTerminal FlowControlStatement = new NonTerminal("FlowControl", typeof(ScriptFlowControlStatement));
NonTerminal ExprStatement = new NonTerminal("ExprStatement", typeof(ScriptStatement));
//Block
NonTerminal BlockStatement = new NonTerminal("BlockStatement", typeof(ScriptStatement));
NonTerminal Statements = new NonTerminal("Statements(Compound)", typeof(ScriptCompoundStatement));
#endregion
#region 2.4 Program and Functions
NonTerminal Prog = new NonTerminal("Prog", typeof(ScriptProg));
NonTerminal Element = new NonTerminal("Element", typeof(ScriptAst));
NonTerminal Elements = new NonTerminal("Elements", typeof(ScriptElements));
NonTerminal FuncDef = new NonTerminal("FuncDef", typeof(ScriptFunctionDefinition));
NonTerminal FuncContract = new NonTerminal("FuncContract", typeof(ScriptFuncContract));
NonTerminal ParameterList = new NonTerminal("ParamaterList", typeof(ScriptFuncParameters));
NonTerminal FuncContractPre = new NonTerminal("Pre Conditions", typeof(ScriptFuncContractPre));
NonTerminal FuncContractPost = new NonTerminal("Post Conditions", typeof(ScriptFuncContractPost));
NonTerminal FuncContractInv = new NonTerminal("Invariant Conditions", typeof(ScriptFuncContractInv));
#endregion
#endregion
#region 3. BNF rules
#region 3.1 Expressions
ConstExpr.Rule = Symbol("true")
| "false"
| "null"
| s
| n;
BinExpr.Rule = Expr + BinOp + Expr
| IsExpr;
UnaryExpr.Rule = LUnOp + Expr;
IsExpr.Rule = Expr + @is + TypeExpr;
TypeConvertExpr.Rule = LCb + Expr + RCb + Expr.Q();
AssignExpr.Rule = QualifiedName + "=" + Expr
| QualifiedName + "++"
| QualifiedName + "--"
| QualifiedName + ":=" + Expr
| QualifiedName + "+=" + Expr
| QualifiedName + "-=" + Expr;
//TODO: MetaFeatures;
// <[ ] + > because of conflict a[1]>2
MetaExpr.Rule = LMb + Elements + RMb;
GlobalList.Rule = "global" + LCb + ParameterList + RCb;
FuncDefExpr.Rule = "function" + LCb + ParameterList + RCb
+ GlobalList.Q()
+ FuncContract.Q()
+ BlockStatement;
Expr.Rule = ConstExpr
| BinExpr
| UnaryExpr
| QualifiedName
| AssignExpr
| NewStmt
| FuncDefExpr
| NewArrStmt
| ArrayConstructor
| MObjectConstructor
| TypeConvertExpr
| MetaExpr
;
NewStmt.Rule = "new" + TypeConstructor;
NewArrStmt.Rule = "new" + TypeExpr + ArrayResolution;
BinOp.Rule = Symbol("+") | "-" | "*" | "/" | "%" | "^" | "&" | "|"
| "&&" | "||" | "==" | "!=" | greater | less
| ">=" | "<=";
LUnOp.Rule = Symbol("~") | "-" | "!" | "$";
ArrayConstructor.Rule = LSb + ExprList + RSb;
MObjectPart.Rule = v + arrow + Expr;
MObjectParts.Rule = MakePlusRule(MObjectParts, comma, MObjectPart);
MObjectConstructor.Rule = LSb + MObjectParts + RSb;
OptionalExpression.Rule = Expr.Q();
#endregion
#region 3.2 QualifiedName
TypeExpr.Rule = //MakePlusRule(TypeExpr, dot, v);
v + GenericsPostfix.Q()
| TypeExpr + dot + (v + GenericsPostfix.Q());
GenericsPostfix.Rule = LGb + TypeList + RGb;
FunctionCall.Rule = LCb + ExprList.Q() + RCb;
ArrayResolution.Rule = LSb + ExprList + RSb;
QualifiedName.Rule = v + (GenericsPostfix | ArrayResolution | FunctionCall).Star()
| QualifiedName + dot + v + (GenericsPostfix | ArrayResolution | FunctionCall).Star();
ExprList.Rule = MakePlusRule(ExprList, comma, Expr);
TypeList.Rule = MakePlusRule(TypeList, comma, TypeExpr);
TypeConstructor.Rule = TypeExpr + FunctionCall;
#endregion
#region 3.3 Statement
Condition.Rule = LCb + Expr + RCb;
IfStatement.Rule = "if" + Condition + Statement + ("else" + Statement).Q();
WhileStatement.Rule = "while" + Condition + Statement;
ForStatement.Rule = "for" + LCb + OptionalExpression + semicolon + OptionalExpression + semicolon + OptionalExpression + RCb + Statement;
ForEachStatement.Rule = "foreach" + LCb + v + "in" + Expr + RCb + Statement;
UsingStatement.Rule = "using" + LCb + Expr + RCb + BlockStatement;
TryCatchFinallyStatement.Rule = "try" + BlockStatement + "catch" + LCb + v + RCb + BlockStatement + "finally" + BlockStatement;
SwitchStatement.Rule = "switch" + LCb + Expr + RCb + LFb + SwitchStatements + RFb;
ExprStatement.Rule = Expr + semicolon;
FlowControlStatement.Rule = "break" + semicolon
| "continue" + semicolon
| "return" + Expr + semicolon
| "throw" + Expr + semicolon;
Statement.Rule = semicolon
| IfStatement //1. If
| WhileStatement //2. While
| ForStatement //3. For
| ForEachStatement //4. ForEach
| UsingStatement //5. Using
| SwitchStatement //6. Switch
| BlockStatement //7. Block
| TryCatchFinallyStatement //8. TryCatch
| ExprStatement //9. Expr
| FlowControlStatement; //10. FlowControl
Statements.SetOption(TermOptions.IsList);
Statements.Rule = Statements + Statement | Empty;
BlockStatement.Rule = LFb + Statements + RFb;
SwitchStatements.Rule = SwitchCaseStatement.Star() + SwitchDefaultStatement.Q();
SwitchCaseStatement.Rule = Symbol("case") + Expr + colon + Statements;
SwitchDefaultStatement.Rule = "default" + colon + Statements;
#endregion
#region 3.4 Prog
FuncContract.Rule = LSb +
FuncContractPre + semicolon +
FuncContractPost + semicolon +
FuncContractInv + semicolon +
RSb;
FuncContractPre.Rule = "pre" + LCb + ExprList.Q() + RCb;
FuncContractPost.Rule = "post" + LCb + ExprList.Q() + RCb;
FuncContractInv.Rule = "invariant" + LCb + ExprList.Q() + RCb;
ParameterList.Rule = MakeStarRule(ParameterList, comma, v);
FuncDef.Rule = "function" + v + LCb + ParameterList + RCb
+ GlobalList.Q()
+ FuncContract.Q()
+ BlockStatement;
Element.Rule = Statement | FuncDef;
Elements.SetOption(TermOptions.IsList);
Elements.Rule = Elements + Element | Empty;
Prog.Rule = Elements + Eof;
Terminal Comment = new CommentTerminal("Comment", "/*", "*/");
NonGrammarTerminals.Add(Comment);
Terminal LineComment = new CommentTerminal("LineComment", "//", "\n");
NonGrammarTerminals.Add(LineComment);
#endregion
#endregion
#region 4. Set starting symbol
if (!expressionGrammar)
Root = Prog; // Set grammar root
else
Root = Expr;
#endregion
#region 5. Operators precedence
RegisterOperators(1, "=", "+=", "-=", ":=");
RegisterOperators(2, "|", "||");
RegisterOperators(3, "&", "&&");
RegisterOperators(4, "==", "!=", ">", "<", ">=", "<=");
RegisterOperators(5, "is");
RegisterOperators(6, "+", "-");
RegisterOperators(7, "*", "/", "%");
RegisterOperators(8, Associativity.Right, "^");
RegisterOperators(9, "~", "!", "$", "++", "--");
RegisterOperators(10, ".");
//RegisterOperators(10, Associativity.Right, ".",",", ")", "(", "]", "[", "{", "}");
//RegisterOperators(11, Associativity.Right, "else");
#endregion
#region 6. Punctuation symbols
RegisterPunctuation( "(", ")", "[", "]", "{", "}", ",", ";" );
#endregion
}
public BasicGrammar()
{
#region Initialisation
// BASIC is not case sensitive...
this.CaseSensitive = false;
// By default, new-line characters are ignored. Because BASIC uses
// line breaks to delimit lines, we need to know where the line breaks
// are. The following line is required for this.
this.TokenFilters.Add(new CodeOutlineFilter(false));
// Define the Terminals
Terminal number = new NumberLiteral("NUMBER");
VariableIdentifierTerminal variable = new VariableIdentifierTerminal();
Terminal stringLiteral = new StringLiteral("STRING", "\"", ScanFlags.None);
//Important: do not add comment term to base.NonGrammarTerminals list - we do use this terminal in grammar rules
Terminal comment = new CommentTerminal("Comment", "REM", "\n");
Terminal comma = Symbol(",", "comma");
// Make sure reserved keywords of the BASIC language aren't mistaken
// for variables. Only the keywords ending with '$' could be mistaken
// for variables.
variable.AddKeywords(
"inkey$", "left$", "right$", "mid$", "chr$",
"space$", "str$", "string$"
);
// Define the non-terminals
NonTerminal PROGRAM = new NonTerminal("PROGRAM", typeof(ProgramNode));
NonTerminal LINE = new NonTerminal("LINE", typeof(LineNode));
NonTerminal STATEMENT_LIST = new NonTerminal("STATEMENT_LIST", typeof(StatementListNode));
NonTerminal STATEMENT = new NonTerminal("STATEMENT", typeof(StatementNode));
NonTerminal COMMAND = new NonTerminal("COMMAND", typeof(StatementNode)); //TODO: create command node
NonTerminal PRINT_STMT = new NonTerminal("PRINT_STMT", typeof(PrintStmtNode));
NonTerminal INPUT_STMT = new NonTerminal("INPUT_STMT", typeof(InputStmtNode));
NonTerminal IF_STMT = new NonTerminal("IF_STMT", typeof(IfElseStmtNode)); //TODO: join IfStmtNode and IfElseStmtNode in one
NonTerminal ELSE_CLAUSE_OPT = new NonTerminal("ELSE_CLAUSE_OPT", typeof(GenericJsBasicNode));
NonTerminal EXPR = new NonTerminal("EXPRESSION", typeof(ExpressionNode));
NonTerminal EXPR_LIST = new NonTerminal("EXPRESSION_LIST", typeof(ExprListNode));
NonTerminal BINARY_OP = new NonTerminal("BINARY_OP", typeof(BinaryOpNode));
NonTerminal BINARY_EXPR = new NonTerminal("BINARY_EXPR", typeof(GenericJsBasicNode)); //TODO: create Binary_expr node
NonTerminal BRANCH_STMT = new NonTerminal("BRANCH_STMT", typeof(BranchStmtNode));
NonTerminal ASSIGN_STMT = new NonTerminal("ASSIGN_STMT", typeof(AssignStmtNode));
NonTerminal FOR_STMT = new NonTerminal("FOR_STMT", typeof(ForStmtNode));
NonTerminal STEP_OPT = new NonTerminal("STEP_OPT", typeof(GenericJsBasicNode)); //TODO: create step specifier node
NonTerminal NEXT_STMT = new NonTerminal("NEXT_STMT", typeof(NextStmtNode));
NonTerminal LOCATE_STMT = new NonTerminal("LOCATE_STMT", typeof(LocateStmtNode));
NonTerminal WHILE_STMT = new NonTerminal("WHILE_STMT", typeof(WhileStmtNode));
NonTerminal WEND_STMT = new NonTerminal("WEND_STMT", typeof(WendStmtNode));
NonTerminal SWAP_STMT = new NonTerminal("SWAP_STMT", typeof(SwapStmtNode));
NonTerminal GLOBAL_FUNCTION_EXPR = new NonTerminal("GLOBAL_FUNCTION_EXPR", typeof(GlobalFunctionExpr));
NonTerminal ARG_LIST = new NonTerminal("ARG_LIST", typeof(GenericJsBasicNode));
NonTerminal FUNC_NAME = new NonTerminal("FUNC_NAME", typeof(GenericJsBasicNode));
NonTerminal COMMENT_STMT = new NonTerminal("COMMENT_STMT", typeof(RemStmtNode));
NonTerminal GLOBAL_VAR_EXPR = new NonTerminal("GLOBAL_VAR_EXPR", typeof(GenericJsBasicNode));
// Set the PROGRAM to be the root node of BASIC programs.
// A program is a bunch of lines
this.Root = PROGRAM;
#endregion
#region Grammar declaration
// A program is a collection of lines
PROGRAM.Rule = MakePlusRule(PROGRAM, null, LINE);
// A line can be an empty line, or it's a number followed by a statement list ended by a new-line.
LINE.Rule = NewLine | number + NewLine | number + STATEMENT_LIST + NewLine;
// A statement list is 1 or more statements separated by the ':' character
STATEMENT_LIST.Rule = MakePlusRule(STATEMENT_LIST, Symbol(":"), STATEMENT);
// A statement can be one of a number of types
STATEMENT.Rule = EXPR | ASSIGN_STMT | PRINT_STMT | INPUT_STMT | IF_STMT | COMMENT_STMT
| BRANCH_STMT | COMMAND | FOR_STMT | NEXT_STMT | LOCATE_STMT | SWAP_STMT
| WHILE_STMT | WEND_STMT;
// The different statements are defined here
PRINT_STMT.Rule = "print" + EXPR_LIST;
INPUT_STMT.Rule = "input" + EXPR_LIST + variable;
IF_STMT.Rule = "if" + EXPR + "then" + STATEMENT_LIST + ELSE_CLAUSE_OPT;
ELSE_CLAUSE_OPT.Rule = Empty | "else" + STATEMENT_LIST;
BRANCH_STMT.Rule = "goto" + number | "gosub" + number | "return";
ASSIGN_STMT.Rule = variable + "=" + EXPR;
LOCATE_STMT.Rule = "locate" + EXPR + comma + EXPR;
SWAP_STMT.Rule = "swap" + EXPR + comma + EXPR;
COMMAND.Rule = Symbol("end") | "cls";
COMMENT_STMT.Rule = comment;
// An expression is a number, or a variable, a string, or the result of a binary comparison.
EXPR.Rule = number | variable | stringLiteral | BINARY_EXPR
| GLOBAL_VAR_EXPR | GLOBAL_FUNCTION_EXPR | "(" + EXPR + ")";
BINARY_EXPR.Rule = EXPR + BINARY_OP + EXPR;
BINARY_OP.Rule = Symbol("+") | "-" | "*" | "/" | "=" | "<=" | ">=" | "<" | ">" | "<>" | "and" | "or";
//let's do operator precedence right here
RegisterOperators(50, "*", "/");
RegisterOperators(40, "+", "-");
RegisterOperators(30, "=", "<=", ">=", "<", ">", "<>");
RegisterOperators(20, "and", "or");
// Used by print and input to allow a bunch of expressions separated by whitespace,
// or be empty, for example:
// print
// print "Hi"
// print "Hi " a$
// All of these match "print" EXPR_LIST
EXPR_LIST.Rule = MakeStarRule(EXPR_LIST, null, EXPR);
FOR_STMT.Rule = "for" + ASSIGN_STMT + "to" + EXPR + STEP_OPT;
STEP_OPT.Rule = Empty | "step" + number;
NEXT_STMT.Rule = "next" + variable;
WHILE_STMT.Rule = "while" + EXPR;
WEND_STMT.Rule = "wend";
//TODO: check number of arguments for particular function in node constructor
GLOBAL_FUNCTION_EXPR.Rule = FUNC_NAME + "(" + ARG_LIST + ")";
FUNC_NAME.Rule = Symbol("len") | "left$" | "mid$" | "right$" | "abs" | "asc" | "chr$" | "csrlin$"
| "cvi" | "cvs" | "cvd" | "exp" | "fix" | "log" | "pos" | "sgn" | "sin" | "cos" | "tan"
| "instr" | "space$" | "spc" | "sqr" | "str$" | "string$" | "val" | "cint";
ARG_LIST.Rule = MakePlusRule(ARG_LIST, comma, EXPR);
GLOBAL_VAR_EXPR.Rule = Symbol("rnd") | "timer" | "inkey$" | "csrlin";
// By registering these strings as "punctuation", we exclude them from
// appearing in as nodes in the compiled node tree.
RegisterPunctuation("(", ")", ",");
#endregion
}
