public StringExpression(ASTNode parent, Scope scope, IList <LToken> tokens) : base(parent, scope, tokens)
{
if (this.Tokens.Count != 1)
{
throw new Exception("A StringExpression should only be composed of a single LToken of type String.");
}
LToken strtok = this.Tokens[0];
if (strtok.ID == "literalstring")
{
this.UnescapedValue = ( String )strtok.Value;
this.EscapedValue = this.UnescapedValue;
this.StartDelimiter = strtok.Raw.Substring(0, strtok.Raw.IndexOf('[', 1) + 1);
this.EndDelimiter = this.StartDelimiter.Replace('[', ']');
this.IsLiteralString = true;
}
else
{
this.StartDelimiter = strtok.Raw[0].ToString( );
this.EndDelimiter = this.StartDelimiter;
this.UnescapedValue = ( String )strtok.Value;
var unescaped = new StringBuilder( );
foreach (var c in ( String )strtok.Value)
{
if (c == this.StartDelimiter[0] || c == '\n')
{
unescaped
.Append('\\')
.Append(c);
}
else if (!LJUtils.IsGraph(c) && !LJUtils.IsSpace(c))
{
unescaped.AppendFormat("\\x{0:X2}", ( Int32 )c);
}
else
{
unescaped.Append(c);
}
}
this.EscapedValue = unescaped.ToString( );
}
}
/// <summary>
/// </summary>
/// <param name="input">Should be in Extended ASCII</param>
/// <param name="isGlua"></param>
public GLuaLexer(String input) : base(CInterop.ConvertStringToASCII(input))
{
input = CInterop.ConvertStringToASCII(input);
if (input.Any(ch => ch > 255))
{
throw new LexException("Input data is not in Extended ASCII.", this.Location);
}
// Silently eat up BOM
this.Consume("\xEF\xBB\xBF");
this.consumeNewlinesAutomatically = true;
this.whitespaceID = "whitespace";
// This won't be really used tbh since the base lexer
// class isn't ready to handle floating point numbers
// just yet
this.numberSettings = new IntegerLexSettings
{
DefaultType = IntegerLexSettings.NumberType.Decimal,
HexadecimalPrefix = "0x",
};
this.charSettings = new CharLexSettings
{
// \000 -> \255
DecimalEscapePrefix = "\\",
DecimalEscapeMaxLengh = 3,
DecimalEscapeMaxValue = 255,
// \x00 -> \xFF
HexadecimalEscapePrefix = "\\x",
HexadecimalEscapeMaxLengh = 2
};
// String settings. Not many pitfalls other than \
// making a string multiline
this.stringSettings = new StringLexSettings
{
CharSettings = this.charSettings,
NewlineEscape = "\\"
};
this.stringSettings
.CharSettings
.RegisterEscapeConstant(@"\0", '\0')
.RegisterEscapeConstant(@"\a", '\a')
.RegisterEscapeConstant(@"\b", '\b')
.RegisterEscapeConstant(@"\f", '\f')
.RegisterEscapeConstant(@"\n", '\n')
.RegisterEscapeConstant(@"\r", '\r')
.RegisterEscapeConstant(@"\t", '\t')
.RegisterEscapeConstant(@"\v", '\v')
.RegisterEscapeConstant(@"\'", '\'')
.RegisterEscapeConstant("\\\"", '"')
.RegisterEscapeConstant(@"\\", '\\');
var notalnum = new Func <Char, Boolean> (ch => !LJUtils.IsAlNum(ch));
#region Binary Operators
this.tokenManager
.AddToken("=", "=", TokenType.Operator)
// Arithmetic
.AddToken("+", "+", TokenType.Operator)
.AddToken("-", "-", TokenType.Operator, next => next != '-')
.AddToken("%", "%", TokenType.Operator)
.AddToken("/", "/", TokenType.Operator)
.AddToken("*", "*", TokenType.Operator)
.AddToken("^", "^", TokenType.Operator)
// Binary
.AddToken("<<", "<<", TokenType.Operator)
.AddToken(">>", ">>", TokenType.Operator)
.AddToken("|", "|", TokenType.Operator)
.AddToken("&", "&", TokenType.Operator)
.AddToken("~", "~", TokenType.Operator)
// String
.AddToken("..", "..", TokenType.Operator)
// Boolean
.AddToken("==", "==", TokenType.Operator)
.AddToken("<", "<", TokenType.Operator)
.AddToken("<=", "<=", TokenType.Operator)
.AddToken("~=", "~=", TokenType.Operator)
.AddToken("!=", "!=", TokenType.Operator)
.AddToken(">", ">", TokenType.Operator)
.AddToken(">=", ">=", TokenType.Operator)
.AddToken("&&", "&&", TokenType.Operator)
.AddToken("||", "||", TokenType.Operator)
.AddToken("and", "and", TokenType.Operator, notalnum)
.AddToken("or", "or", TokenType.Operator, notalnum);
#endregion Binary Operators
#region Unary Operators
this.tokenManager
.AddToken("not", "not", TokenType.Operator, notalnum)
.AddToken("#", "#", TokenType.Operator)
.AddToken("!", "!", TokenType.Operator);
// - and ~ are in binary ops since they can be either.
#endregion Unary Operators
#region Symbols
this.tokenManager
.AddToken(",", ",", TokenType.Punctuation)
.AddToken(".", ".", TokenType.Punctuation, ch => !LJUtils.IsDigit(ch))
.AddToken(":", ":", TokenType.Punctuation)
.AddToken("::", "::", TokenType.Punctuation)
.AddToken(";", ";", TokenType.Punctuation)
.AddToken("...", "...", TokenType.Identifier)
.AddToken("(", "(", TokenType.LParen)
.AddToken(")", ")", TokenType.RParen)
.AddToken("[", "[", TokenType.LBracket, next => next != '[' && next != '=')
.AddToken("]", "]", TokenType.RBracket)
.AddToken("{", "{", TokenType.LCurly)
.AddToken("}", "}", TokenType.RCurly);
#endregion Symbols
#region Keywords
this.tokenManager
.AddToken("break", "break", TokenType.Keyword, notalnum)
.AddToken("continue", "continue", TokenType.Keyword, notalnum)
.AddToken("goto", "goto", TokenType.Keyword, notalnum)
.AddToken("do", "do", TokenType.Keyword, notalnum)
.AddToken("end", "end", TokenType.Keyword, notalnum)
.AddToken("while", "while", TokenType.Keyword, notalnum)
.AddToken("repeat", "repeat", TokenType.Keyword, notalnum)
.AddToken("until", "until", TokenType.Keyword, notalnum)
.AddToken("if", "if", TokenType.Keyword, notalnum)
.AddToken("then", "then", TokenType.Keyword, notalnum)
.AddToken("elseif", "elseif", TokenType.Keyword, notalnum)
.AddToken("else", "else", TokenType.Keyword, notalnum)
.AddToken("for", "for", TokenType.Keyword, notalnum)
.AddToken("in", "in", TokenType.Keyword, notalnum)
.AddToken("function", "function", TokenType.Keyword, notalnum)
.AddToken("local", "local", TokenType.Keyword, notalnum)
.AddToken("return", "return", TokenType.Keyword, notalnum)
.AddToken("nil", "nil", TokenType.Keyword, notalnum)
.AddToken("true", "true", TokenType.Keyword, notalnum)
.AddToken("false", "false", TokenType.Keyword, notalnum);
#endregion Keywords
}
protected override Boolean TryReadToken(out Token tok)
{
var peekedChar = ( Char )this.reader.Peek( );
SourceLocation start = this.reader.Location;
// Token starts null and might be set by any of the following
tok = null;
// Shebang (I'll consider this a leading flair because
// it's technnically a comment)
if (this.Location.Line == 0 && this.Consume("#!"))
{
var exec = this.reader.ReadLine( );
tok = this.CreateToken("shebang", $"#!{exec}", exec, TokenType.Comment, start.To(this.Location));
}
// Numbers
else if (LJUtils.IsDigit(peekedChar) || (peekedChar == '.' && LJUtils.IsDigit(( Char )this.reader.Peek(1))))
{
try
{
(var Raw, var Value) = this.ReadLuaNumber( );
tok = this.CreateToken("number", Raw, Value, TokenType.Number, start.To(this.reader.Location));
}
catch (LexException)
{
tok = null;
}
}
// Identifiers (keywords are read by the tokenmanager)
else if (LJUtils.IsIdent(peekedChar) || peekedChar == '_')
{
var ident = this.reader.ReadStringWhile(LJUtils.IsIdent);
tok = this.CreateToken("ident", ident, ident, TokenType.Identifier, start.To(this.reader.Location));
}
// Normal strings
else if (this.Consume('\'', '"'))
{
// LexerBase string reader doesn't adds
// delimiters to the raw
try
{
(var Raw, var Value) = this.ReadString(peekedChar.ToString( ), false, this.stringSettings);
Raw = $"{peekedChar}{Raw}{peekedChar}";
tok = this.CreateToken("string", Raw, Value, TokenType.String, start.To(this.Location));
}
catch (LexException)
{
tok = null;
}
}
// Stuff that has/is :
else if (this.Consume(':'))
{
// Goto labels
if (this.Consume(':'))
{
var name = this.reader.ReadStringWhile(LJUtils.IsIdent);
this.Expect("::");
tok = this.CreateToken("gotoLabel", $"::{name}::", name, TokenType.Identifier, start.To(this.Location));
}
// Normal self-functions
else
{
tok = this.CreateToken(":", ":", ":", TokenType.Punctuation, start.To(this.Location));
}
}
// comments
else if (this.reader.IsNext("//") || this.reader.IsNext("--") || this.reader.IsNext("/*"))
{
(var Raw, var Value) = this.ReadComment( );
tok = CreateToken("comment", Raw, Value, TokenType.Comment, start.To(this.Location));
}
else if (this.Consume('['))
{
(var Raw, var Value) = this.ReadLiteralString( );
tok = CreateToken("literalstring", Raw, Value, TokenType.String, start.To(this.Location));
}
return(tok != null);
}
protected override Boolean CharIsWhitepace(Char ch) => LJUtils.IsSpace(ch);
protected override Boolean CharIsHex(Char ch) => LJUtils.IsXDigit(ch);
protected override Boolean CharIsDec(Char ch) => LJUtils.IsDigit(ch);
