private static void Main(string[] args)
{
//TestCompiler();
if (args.Length == 0)
{
Console.WriteLine("No llex or alx file specified");
return;
}
else if (!File.Exists(args[0]))
{
Console.WriteLine("Invalid llex file {0}", args[0]);
return;
}
Console.WriteLine("Generating lexer...");
string outputFile = args.Length > 1 ? args[1] : new FileInfo(args[0]).FullName + ".cs";
var lexer = args[0].ToLower().EndsWith(".alx") ?
AlxFile.Interpret(args[0]) :
LLexFile.Interpret(args[0]);
File.WriteAllText(outputFile, lexer);
Console.WriteLine("Lexer written to {0}", outputFile);
}
public static string Interpret(string file)
{
var interpreter = new AphidInterpreter();
interpreter.InterpretFile(file);
var retVal = interpreter.GetReturnValue();
var llexFile = new LLexFile();
retVal.Bind(llexFile);
var tokenTable = new TokenTable();
tokenTable.Ignore = llexFile.Ignore;
var nameInfo = LLexNameInfo.Parse(llexFile.Name);
int z = 0;
var modeTable = llexFile.Modes.ToDictionary(x => x.Mode, x => z++);
var tokenTypes = new List<string>();
foreach (var mode in llexFile.Modes)
{
tokenTable.SetMode(modeTable[mode.Mode]);
foreach (var token in mode.Tokens)
{
if (token.Regex != null)
{
var regexLexer = new RegexLexer(token.Regex);
var tokens = regexLexer.GetTokens();
var parser = new RegexParser(tokens.ToArray());
var ast = parser.Parse();
var compiler = new RegexCompiler(ast);
var strings = compiler.ExpandRegex();
foreach (var l in strings)
{
if (token.Code != null)
{
tokenTable.AddLexemeCode(l, token.Code);
continue;
}
if (!tokenTypes.Contains(token.TokenType))
tokenTypes.Add(token.TokenType);
if (!string.IsNullOrEmpty(token.NewMode))
{
tokenTable.Add(l, token.TokenType, modeTable[token.NewMode]);
}
else
{
tokenTable.Add(l, token.TokenType);
}
}
}
else if (token.Code != null)
{
tokenTable.AddCode(token.Code);
}
}
foreach (var keyword in mode.Keywords)
{
tokenTable.AddKeyword(keyword);
var t = keyword + "Keyword";
if (mode.KeywordTail != null)
{
tokenTable.AddLexemeCode(keyword, mode.KeywordTail.Replace("{Keyword}", t));
}
else
{
tokenTable.Add(keyword, t);
}
}
if (!string.IsNullOrEmpty(mode.KeywordDefault))
{
var k = mode.Keywords
.SelectMany(x => Enumerable
.Range(1, x.Length - 1)
.Select(y => x.Remove(y))
.ToArray())
.Distinct()
.ToArray();
foreach (var i in k)
{
if (tokenTable.Lists.Any(x => x.Value.Any(y => y.Lexeme == i)))
{
continue;
}
tokenTable.AddLexemeCode(i, mode.KeywordDefault);
}
}
}
var generator = new LexerGenerator(tokenTable) { IgnoreCase = llexFile.IgnoreCase };
var lexer = generator.Generate();
return lexer
.Replace("{Lexer}", nameInfo.LexerName)
.Replace("{Token}", nameInfo.TokenName)
.Replace("{TokenType}", nameInfo.TokenTypeName)
.Replace("{LexerNamespace}", nameInfo.Namespace);
}
public static string Interpret(string file)
{
var interpreter = new AphidInterpreter();
interpreter.InterpretFile(file);
var retVal = interpreter.GetReturnValue();
var llexFile = new LLexFile();
retVal.Bind(llexFile);
var tokenTable = new TokenTable();
tokenTable.Ignore = llexFile.Ignore;
var nameInfo = LLexNameInfo.Parse(llexFile.Name);
int z = 0;
var modeTable = llexFile.Modes.ToDictionary(x => x.Mode, x => z++);
var tokenTypes = new List <string>();
foreach (var mode in llexFile.Modes)
{
tokenTable.SetMode(modeTable[mode.Mode]);
foreach (var token in mode.Tokens)
{
if (token.Regex != null)
{
var regexLexer = new RegexLexer(token.Regex);
var tokens = regexLexer.GetTokens();
var parser = new RegexParser(tokens.ToArray());
var ast = parser.Parse();
var compiler = new RegexCompiler(ast);
var strings = compiler.ExpandRegex();
foreach (var l in strings)
{
if (token.Code != null)
{
tokenTable.AddLexemeCode(l, token.Code);
continue;
}
if (!tokenTypes.Contains(token.TokenType))
{
tokenTypes.Add(token.TokenType);
}
if (!string.IsNullOrEmpty(token.NewMode))
{
tokenTable.Add(l, token.TokenType, modeTable[token.NewMode]);
}
else
{
tokenTable.Add(l, token.TokenType);
}
}
}
else if (token.Code != null)
{
tokenTable.AddCode(token.Code);
}
}
foreach (var keyword in mode.Keywords)
{
tokenTable.AddKeyword(keyword);
var t = keyword + "Keyword";
if (mode.KeywordTail != null)
{
tokenTable.AddLexemeCode(keyword, mode.KeywordTail.Replace("{Keyword}", t));
}
else
{
tokenTable.Add(keyword, t);
}
}
if (!string.IsNullOrEmpty(mode.KeywordDefault))
{
var k = mode.Keywords
.SelectMany(x => Enumerable
.Range(1, x.Length - 1)
.Select(y => x.Remove(y))
.ToArray())
.Distinct()
.ToArray();
foreach (var i in k)
{
if (tokenTable.Lists.Any(x => x.Value.Any(y => y.Lexeme == i)))
{
continue;
}
tokenTable.AddLexemeCode(i, mode.KeywordDefault);
}
}
}
var generator = new LexerGenerator(tokenTable);
var lexer = generator.Generate();
return(lexer
.Replace("{Lexer}", nameInfo.LexerName)
.Replace("{Token}", nameInfo.TokenName)
.Replace("{TokenType}", nameInfo.TokenTypeName)
.Replace("{LexerNamespace}", nameInfo.Namespace));
}
public static string From(AphidObject retVal)
{
var llexFile = new LLexFile();
retVal.Bind(llexFile);
var tokenTable = new TokenTable {
Ignore = llexFile.Ignore
};
var nameInfo = LLexNameInfo.Parse(llexFile.Name);
var z = 0;
var modeTable = llexFile.Modes.ToDictionary(x => x.Mode, x => z++);
var tokenTypes = new List <string>();
foreach (var mode in llexFile.Modes)
{
tokenTable.SetMode(modeTable[mode.Mode]);
foreach (var token in mode.Tokens)
{
if (token.Regex != null)
{
var regexLexer = new RegexLexer(token.Regex);
var tokens = regexLexer.GetTokens();
var parser = new RegexParser(tokens.ToArray());
var ast = parser.Parse();
var compiler = new RegexCompiler(ast);
var strings = compiler.ExpandRegex();
foreach (var l in strings)
{
if (token.Code != null)
{
tokenTable.AddLexemeCode(l, token.Code);
continue;
}
if (!tokenTypes.Contains(token.TokenType))
{
tokenTypes.Add(token.TokenType);
}
if (!string.IsNullOrEmpty(token.NewMode))
{
tokenTable.Add(l, token.TokenType, modeTable[token.NewMode]);
}
else
{
tokenTable.Add(l, token.TokenType);
}
}
}
else if (token.Code != null)
{
tokenTable.AddCode(token.Code);
}
else if (token.TokenType != null)
{
tokenTable.AddCode("return {TokenType}." + token.TokenType + ";\r\n");
}
else
{
throw new NotImplementedException(
"Token with no regex, code, or type not supported.");
}
}
foreach (var keyword in mode.Keywords ?? Array.Empty <string>())
{
tokenTable.AddKeyword(keyword);
var t = keyword + "Keyword";
if (mode.KeywordTail != null)
{
tokenTable.AddLexemeCode(keyword, mode.KeywordTail.Replace("{Keyword}", t));
}
else
{
tokenTable.Add(keyword, t);
}
}
if (!string.IsNullOrEmpty(mode.KeywordDefault))
{
var k = mode.Keywords
.SelectMany(x => Enumerable
.Range(1, x.Length - 1)
.Select(y => x.Remove(y))
.ToArray())
.Distinct()
.ToArray();
foreach (var i in k)
{
if (tokenTable.Lists.Any(x => x.Value.Any(y => y.Lexeme == i)))
{
continue;
}
tokenTable.AddLexemeCode(i, mode.KeywordDefault);
}
}
}
var generator = new LexerGenerator(tokenTable)
{
IgnoreCase = llexFile.IgnoreCase
};
var lexer = generator.Generate();
return(lexer
.Replace("{Lexer}", nameInfo.LexerName)
.Replace("{Token}", nameInfo.TokenName)
.Replace("{TokenType}", nameInfo.TokenTypeName)
.Replace("{LexerNamespace}", nameInfo.Namespace));
}