Skip to content

Instantly share code, notes, and snippets.

@theraot

theraot/Parser.cs

Last active December 18, 2021 23:39
Show Gist options
  • Save theraot/e07af2f8d3acae3f33eb47f0698865c4 to your computer and use it in GitHub Desktop.
Save theraot/e07af2f8d3acae3f33eb47f0698865c4 to your computer and use it in GitHub Desktop.
Download ZIP
Raw
Parser.cs
#r "nuget:Theraot.Core/3.0.3"
// This file is for RoslynPad
using System;
using System.Collections;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using Theraot.Core;
public class ParseException : Exception
{
public ParseException(string message, string @string, int position, Pattern pattern)
: base(message)
{
Pattern = pattern;
String = @string;
Position = position;
}
public Pattern Pattern { get; }
public int Position { get; }
public string String { get; }
}
public class Symbol : IEnumerable<Symbol>
{
public static readonly Symbol[] EmptySymbols = new Symbol[0];
public Symbol(Pattern pattern)
{
Pattern = pattern;
}
public Pattern Pattern { get; }
public virtual IEnumerator<Symbol> GetEnumerator()
{
yield return this;
}
IEnumerator IEnumerable.GetEnumerator()
{
return GetEnumerator();
}
}
public class TerminalSymbol : Symbol
{
public TerminalSymbol(Pattern pattern, string value)
: base(pattern)
{
Value = value;
}
public string Value { get; }
public override string ToString()
{
return Value;
}
}
public class CompositeSymbol : Symbol
{
public CompositeSymbol(Pattern pattern, IEnumerable<Symbol> symbols)
: base(pattern)
{
Symbols = symbols;
}
private IEnumerable<Symbol> Symbols { get; }
public override string ToString()
{
var builder = new StringBuilder();
foreach (var symbol in Symbols)
{
builder.Append(symbol);
}
return builder.ToString();
}
public override IEnumerator<Symbol> GetEnumerator()
{
return GraphHelper.ExploreBreadthFirstGraph(this, symbol => symbol is CompositeSymbol compositeSymbol ? compositeSymbol.Symbols : EmptySymbols, EqualityComparer<Symbol>.Default).GetEnumerator();
}
}
public sealed class Pattern
{
private Func<StringProcessor, Symbol> _parse;
private Pattern(string name)
{
Name = name;
_parse = _ => new CompositeSymbol(this, Symbol.EmptySymbols);
}
public string Name { get; }
public static Pattern Conjunction(params Pattern[] patterns)
{
return Conjunction("(" + string.Join(" + ", from subPattern in patterns select subPattern.Name) + ")", patterns);
}
public static Pattern Conjunction(string name, params Pattern[] patterns)
{
var pattern = new Pattern(name);
pattern._parse = processor => new CompositeSymbol(pattern, (from subPattern in patterns select subPattern.Parse(processor)).ToArray());
return pattern;
}
public static Pattern Custom(string name, Func<StringProcessor, string> callback)
{
if (callback == null)
{
throw new ArgumentNullException(nameof(callback));
}
var pattern = new Pattern(name);
pattern._parse = processor =>
{
var position = processor.Position;
var greedy = processor.Greedy;
string? result = null;
try
{
result = callback(processor);
}
finally
{
if (result == null)
{
processor.Position = position;
processor.Greedy = greedy;
throw new ParseException($"Expected {name}", processor.String, processor.Position, pattern);
}
}
return new TerminalSymbol(pattern, result);
};
return pattern;
}
public static Pattern Disjunction(params Pattern[] patterns)
{
return Disjunction("(" + string.Join(" | ", from subPattern in patterns select subPattern.Name) + ")", patterns);
}
public static Pattern Disjunction(string name, params Pattern[] patterns)
{
var pattern = new Pattern(name);
pattern._parse = processor =>
{
var position = processor.Position;
var greedy = processor.Greedy;
foreach (var subPattern in patterns)
{
Symbol? result = null;
try
{
result = subPattern.Parse(processor);
}
catch (ParseException exception)
{
// swallow
Theraot.No.Op(exception);
}
finally
{
if (result == null)
{
processor.Position = position;
processor.Greedy = greedy;
}
}
if (result != null)
{
return new CompositeSymbol(pattern, new[] { result });
}
}
throw new ParseException($"Expected {name}", processor.String, processor.Position, pattern);
};
return pattern;
}
public static Pattern Empty()
{
return Empty(string.Empty);
}
public static Pattern Empty(string name)
{
return new Pattern(name);
}
public static Pattern Literal(string literal)
{
return Literal("\"" + literal.Replace("\"", "\\\"") + "\"", literal);
}
public static Pattern Literal(string name, string literal)
{
var pattern = new Pattern(name);
pattern._parse = processor =>
{
if (processor.Read(literal))
{
return new TerminalSymbol(pattern, literal);
}
throw new ParseException($"Expected {name} ({literal})", processor.String, processor.Position, pattern);
};
return pattern;
}
public static Pattern Literal(char literal)
{
return Literal("\"" + (literal == '"' ? "\\\"" : literal.ToString()) + "\"", literal);
}
public static Pattern Literal(string name, char literal)
{
var pattern = new Pattern(name);
pattern._parse = processor =>
{
if (processor.Read(literal))
{
return new TerminalSymbol(pattern, literal.ToString());
}
throw new ParseException($"Expected {name}", processor.String, processor.Position, pattern);
};
return pattern;
}
public static Pattern Optional(Pattern subPattern)
{
return Optional("[" + subPattern.Name + "]", subPattern);
}
public static Pattern Optional(string name, Pattern subPattern)
{
var pattern = new Pattern(name);
pattern._parse = processor =>
{
var position = processor.Position;
var greedy = processor.Greedy;
try
{
return new CompositeSymbol(pattern, new[] { subPattern.Parse(processor) });
}
catch (ParseException)
{
processor.Position = position;
processor.Greedy = greedy;
return new CompositeSymbol(pattern, Symbol.EmptySymbols);
}
};
return pattern;
}
public static Pattern OptionalRepetition(Pattern subPattern)
{
return OptionalRepetition(subPattern.Name + "*", subPattern);
}
public static Pattern OptionalRepetition(string name, Pattern subPattern)
{
var pattern = new Pattern(name);
pattern._parse = processor =>
{
var symbols = new List<Symbol>();
while (!processor.EndOfString)
{
var position = processor.Position;
var greedy = processor.Greedy;
Symbol? result = null;
try
{
result = subPattern.Parse(processor);
}
catch (ParseException exception)
{
// swallow
Theraot.No.Op(exception);
}
finally
{
if (result == null)
{
processor.Position = position;
processor.Greedy = greedy;
}
}
if (result == null)
{
break;
}
symbols.Add(result);
}
return new CompositeSymbol(pattern, symbols);
};
return pattern;
}
public static Pattern Repetition(Pattern subPattern)
{
return Repetition(subPattern.Name + "+", subPattern);
}
public static Pattern Repetition(string name, Pattern subPattern)
{
var pattern = new Pattern(name);
pattern._parse = processor =>
{
var symbols = new List<Symbol>();
while (!processor.EndOfString)
{
var position = processor.Position;
var greedy = processor.Greedy;
Symbol? result = null;
try
{
result = subPattern.Parse(processor);
}
catch (ParseException exception)
{
// swallow
Theraot.No.Op(exception);
}
finally
{
if (result == null)
{
processor.Position = position;
processor.Greedy = greedy;
}
}
if (result == null)
{
break;
}
symbols.Add(result);
}
if (symbols.Count == 0)
{
throw new ParseException($"Expected {pattern.Name}", processor.String, processor.Position, pattern);
}
return new CompositeSymbol(pattern, symbols);
};
return pattern;
}
public Symbol Parse(string str)
{
var processor = new StringProcessor(str);
return Parse(processor);
}
private Symbol Parse(StringProcessor processor)
{
return _parse?.Invoke(processor);
}
}
const string input = "\"hello \" + \"world \" + \" + \" + \"hello\"";
var quoteSymbol = Pattern.Literal("QuoteSymbol", '"');
var nonQuoteSymbol = Pattern.Custom("NonQuoteSymbol", s => s.ReadUntil('"'));
var String = Pattern.Conjunction("String", quoteSymbol, nonQuoteSymbol, quoteSymbol);
var whiteSpace = Pattern.Custom("WhiteSpace", s => s.ReadWhile(char.IsWhiteSpace));
var plusSymbol = Pattern.Literal("PlusSymbol", '+');
var document = Pattern.Repetition(Pattern.Conjunction(whiteSpace, String, whiteSpace, plusSymbol));
var results = from symbol in document.Parse(input) where symbol.Pattern == String select symbol.ToString();
foreach (var v in results.ToArray())
{
Console.WriteLine(v);
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment