283 lines
7.5 KiB
C#
283 lines
7.5 KiB
C#
using Common;
|
|
using Syntax.Diagnostics;
|
|
|
|
namespace Syntax.Tokenization;
|
|
|
|
public static class Tokenizer
|
|
{
|
|
private static readonly Dictionary<string, Symbol> Keywords = new()
|
|
{
|
|
["namespace"] = Symbol.Namespace,
|
|
["func"] = Symbol.Func,
|
|
["if"] = Symbol.If,
|
|
["else"] = Symbol.Else,
|
|
["while"] = Symbol.While,
|
|
["break"] = Symbol.Break,
|
|
["continue"] = Symbol.Continue,
|
|
["return"] = Symbol.Return,
|
|
["alloc"] = Symbol.Alloc,
|
|
["struct"] = Symbol.Struct,
|
|
["let"] = Symbol.Let,
|
|
["calls"] = Symbol.Calls,
|
|
};
|
|
|
|
private static readonly Dictionary<string, Modifier> Modifiers = new()
|
|
{
|
|
["export"] = Modifier.Export,
|
|
["extern"] = Modifier.Extern,
|
|
};
|
|
|
|
private static readonly Dictionary<char[], Symbol> Chians = new()
|
|
{
|
|
[['=', '=']] = Symbol.Equal,
|
|
[['!', '=']] = Symbol.NotEqual,
|
|
[['<', '=']] = Symbol.LessThanOrEqual,
|
|
[['>', '=']] = Symbol.GreaterThanOrEqual,
|
|
[[':', ':']] = Symbol.DoubleColon,
|
|
};
|
|
|
|
private static readonly Dictionary<char, Symbol> Chars = new()
|
|
{
|
|
[';'] = Symbol.Semicolon,
|
|
[':'] = Symbol.Colon,
|
|
['('] = Symbol.OpenParen,
|
|
[')'] = Symbol.CloseParen,
|
|
['{'] = Symbol.OpenBrace,
|
|
['}'] = Symbol.CloseBrace,
|
|
['['] = Symbol.OpenBracket,
|
|
[']'] = Symbol.CloseBracket,
|
|
[','] = Symbol.Comma,
|
|
['.'] = Symbol.Period,
|
|
['='] = Symbol.Assign,
|
|
['<'] = Symbol.LessThan,
|
|
['>'] = Symbol.GreaterThan,
|
|
['+'] = Symbol.Plus,
|
|
['-'] = Symbol.Minus,
|
|
['*'] = Symbol.Star,
|
|
['/'] = Symbol.ForwardSlash,
|
|
['!'] = Symbol.Bang,
|
|
['^'] = Symbol.Caret,
|
|
['&'] = Symbol.Ampersand,
|
|
};
|
|
|
|
private static SourceText _sourceText;
|
|
private static int _index;
|
|
|
|
public static IEnumerable<Token> Tokenize(SourceText sourceText, out IEnumerable<Diagnostic> diagnostics)
|
|
{
|
|
_sourceText = sourceText;
|
|
_index = 0;
|
|
|
|
List<Token> tokens = [];
|
|
while (ParseToken().TryGetValue(out var token))
|
|
{
|
|
tokens.Add(token);
|
|
}
|
|
|
|
// TODO: Implement diagnostics
|
|
diagnostics = [];
|
|
return tokens;
|
|
}
|
|
|
|
private static void ConsumeWhitespace()
|
|
{
|
|
while (Peek().TryGetValue(out var character) && char.IsWhiteSpace(character))
|
|
{
|
|
Next();
|
|
}
|
|
}
|
|
|
|
private static Optional<Token> ParseToken()
|
|
{
|
|
ConsumeWhitespace();
|
|
var startIndex = _index;
|
|
|
|
if (!Peek().TryGetValue(out var current))
|
|
{
|
|
return Optional<Token>.Empty();
|
|
}
|
|
|
|
if (current == '/' && Peek(1).TryGetValue(out var nextChar) && nextChar == '/')
|
|
{
|
|
Next();
|
|
Next();
|
|
|
|
if (Peek().TryGetValue(out var thirdChar) && thirdChar == '/')
|
|
{
|
|
Next();
|
|
var buffer = string.Empty;
|
|
while (Peek().TryGetValue(out var character) && character != '\n')
|
|
{
|
|
buffer += character;
|
|
Next();
|
|
}
|
|
|
|
Next();
|
|
return new DocumentationToken(CreateSpan(startIndex), buffer);
|
|
}
|
|
|
|
while (Peek().TryGetValue(out var character) && character != '\n')
|
|
{
|
|
Next();
|
|
}
|
|
|
|
Next();
|
|
return ParseToken();
|
|
}
|
|
|
|
if (char.IsLetter(current) || current == '_')
|
|
{
|
|
var buffer = string.Empty;
|
|
|
|
while (Peek().TryGetValue(out var next) && (char.IsLetterOrDigit(next) || next == '_'))
|
|
{
|
|
buffer += next;
|
|
Next();
|
|
}
|
|
|
|
if (Keywords.TryGetValue(buffer, out var keywordSymbol))
|
|
{
|
|
return new SymbolToken(CreateSpan(startIndex), keywordSymbol);
|
|
}
|
|
|
|
if (Modifiers.TryGetValue(buffer, out var modifer))
|
|
{
|
|
return new ModifierToken(CreateSpan(startIndex), modifer);
|
|
}
|
|
|
|
if (buffer is "true" or "false")
|
|
{
|
|
return new LiteralToken(CreateSpan(startIndex), LiteralKind.Bool, buffer);
|
|
}
|
|
|
|
return new IdentifierToken(CreateSpan(startIndex), buffer);
|
|
}
|
|
|
|
if (char.IsDigit(current))
|
|
{
|
|
var isFloat = false;
|
|
var buffer = string.Empty;
|
|
|
|
while (Peek().TryGetValue(out var next))
|
|
{
|
|
if (next == '.')
|
|
{
|
|
if (isFloat)
|
|
{
|
|
throw new Exception("More than one period found in float literal");
|
|
}
|
|
|
|
isFloat = true;
|
|
buffer += next;
|
|
Next();
|
|
}
|
|
else if (char.IsDigit(next))
|
|
{
|
|
buffer += next;
|
|
Next();
|
|
}
|
|
else
|
|
{
|
|
break;
|
|
}
|
|
}
|
|
|
|
return new LiteralToken(CreateSpan(startIndex), isFloat ? LiteralKind.Float : LiteralKind.Integer, buffer);
|
|
}
|
|
|
|
// TODO: Revisit this
|
|
foreach (var chain in Chians)
|
|
{
|
|
if (current != chain.Key[0]) continue;
|
|
|
|
for (var i = 1; i < chain.Key.Length; i++)
|
|
{
|
|
var c = Peek(i);
|
|
if (!c.HasValue || c.Value != chain.Key[i]) break;
|
|
|
|
if (i == chain.Key.Length - 1)
|
|
{
|
|
for (var j = 0; j <= i; j++)
|
|
{
|
|
Next();
|
|
}
|
|
|
|
return new SymbolToken(CreateSpan(startIndex), chain.Value);
|
|
}
|
|
}
|
|
}
|
|
|
|
if (Chars.TryGetValue(current, out var charSymbol))
|
|
{
|
|
Next();
|
|
return new SymbolToken(CreateSpan(startIndex), charSymbol);
|
|
}
|
|
|
|
if (current == '"')
|
|
{
|
|
Next();
|
|
var buffer = string.Empty;
|
|
|
|
while (true)
|
|
{
|
|
if (!Peek().TryGetValue(out var next))
|
|
{
|
|
throw new Exception("Unclosed string literal");
|
|
}
|
|
|
|
if (next == '"')
|
|
{
|
|
Next();
|
|
break;
|
|
}
|
|
|
|
buffer += next;
|
|
Next();
|
|
}
|
|
|
|
return new LiteralToken(CreateSpan(startIndex), LiteralKind.String, buffer);
|
|
}
|
|
|
|
throw new Exception($"Unknown character {current}");
|
|
}
|
|
|
|
private static SourceLocation CreateLocation(int index)
|
|
{
|
|
var line = 1;
|
|
var column = 1;
|
|
for (var i = 0; i < Math.Min(index, _sourceText.Content.Length - 1); i++)
|
|
{
|
|
if (_sourceText.Content[i] == '\n')
|
|
{
|
|
column = 1;
|
|
line += 1;
|
|
}
|
|
else
|
|
{
|
|
column += 1;
|
|
}
|
|
}
|
|
|
|
return new SourceLocation(line, column);
|
|
}
|
|
|
|
private static SourceSpan CreateSpan(int startIndex)
|
|
{
|
|
return new SourceSpan(_sourceText, CreateLocation(startIndex), CreateLocation(_index));
|
|
}
|
|
|
|
private static Optional<char> Peek(int offset = 0)
|
|
{
|
|
if (_index + offset < _sourceText.Content.Length)
|
|
{
|
|
return _sourceText.Content[_index + offset];
|
|
}
|
|
|
|
return Optional<char>.Empty();
|
|
}
|
|
|
|
private static void Next()
|
|
{
|
|
_index++;
|
|
}
|
|
} |