view Implab/Automaton/RegularExpressions/Grammar.cs @ 172:92d5278d1b10 ref20160224

Working on text scanner
author cin
date Mon, 14 Mar 2016 01:19:38 +0300
parents e227e78d72e4
children 0c3c69fe225b
line wrap: on
line source

using Implab;
using System;
using System.Collections.Generic;
using System.Linq;

namespace Implab.Automaton.RegularExpressions {
    /// <summary>
    /// Базовый абстрактный класс. Грамматика, позволяет формулировать выражения над алфавитом типа <c>char</c>.
    /// </summary>
    public abstract class Grammar<TSymbol, TTag> {
        
        protected abstract IAlphabetBuilder<TSymbol> AlphabetBuilder {
            get;
        }

        protected SymbolToken<TTag> UnclassifiedToken() {
            return new SymbolToken<TTag>(DFAConst.UNCLASSIFIED_INPUT);
        }

        protected void DefineAlphabet(IEnumerable<TSymbol> alphabet) {
            Safe.ArgumentNotNull(alphabet, "alphabet");

            foreach (var ch in alphabet)
                AlphabetBuilder.DefineSymbol(ch);
        }

        protected Token<TTag> SymbolToken(TSymbol symbol) {
            return Token<TTag>.New(TranslateOrAdd(symbol));
        }

        protected Token<TTag> SymbolToken(IEnumerable<TSymbol> symbols) {
            Safe.ArgumentNotNull(symbols, "symbols");

            return Token<TTag>.New(TranslateOrAdd(symbols).ToArray());
        }

        protected Token<TTag> SymbolSetToken(params TSymbol[] set) {
            return SymbolToken(set);
        }

        int TranslateOrAdd(TSymbol ch) {
            var t = AlphabetBuilder.Translate(ch);
            if (t == DFAConst.UNCLASSIFIED_INPUT)
                t = AlphabetBuilder.DefineSymbol(ch);
            return t;
        }

        IEnumerable<int> TranslateOrAdd(IEnumerable<TSymbol> symbols) {
            return symbols.Distinct().Select(TranslateOrAdd);
        }

        int TranslateOrDie(TSymbol ch) {
            var t = AlphabetBuilder.Translate(ch);
            if (t == DFAConst.UNCLASSIFIED_INPUT)
                    throw new ApplicationException(String.Format("Symbol '{0}' is UNCLASSIFIED", ch));
            return t;
        }

        IEnumerable<int> TranslateOrDie(IEnumerable<TSymbol> symbols) {
            return symbols.Distinct().Select(TranslateOrDie);
        }

        protected Token<TTag> SymbolTokenExcept(IEnumerable<TSymbol> symbols) {
            Safe.ArgumentNotNull(symbols, "symbols");

            return Token<TTag>.New( Enumerable.Range(0, AlphabetBuilder.Count).Except(TranslateOrDie(symbols)).ToArray() );
        }

        protected abstract IAlphabetBuilder<TSymbol> CreateAlphabet();

        protected RegularDFA<TSymbol, TTag> BuildDFA(Token<TTag> regexp) {
            
            var dfa = new RegularDFA<TSymbol, TTag>(AlphabetBuilder);

            var visitor = new RegularExpressionVisitor<TTag>();
            regexp.Accept( visitor );

            visitor.BuildDFA(dfa);

            if (dfa.IsFinalState(dfa.InitialState))
                throw new ApplicationException("The specified language contains empty token");

            return dfa.Optimize(CreateAlphabet());
        }

    }


}