Grammar.cs
99 lines
| 3.3 KiB
| text/x-csharp
|
CSharpLexer
cin
|
r177 | using Implab; | ||
| using System; | ||||
| using System.Collections.Generic; | ||||
| using System.Linq; | ||||
| using Implab.Automaton; | ||||
| using Implab.Automaton.RegularExpressions; | ||||
| namespace Implab.Formats { | ||||
| /// <summary> | ||||
| /// Базовый абстрактный класс. Грамматика, позволяет формулировать выражения над алфавитом типа <c>char</c>. | ||||
| /// </summary> | ||||
|
cin
|
r178 | public abstract class Grammar<TSymbol> { | ||
|
cin
|
r177 | |||
| protected abstract IAlphabetBuilder<TSymbol> AlphabetBuilder { | ||||
| get; | ||||
| } | ||||
|
cin
|
r178 | protected SymbolToken UnclassifiedToken() { | ||
| return new SymbolToken(AutomatonConst.UNCLASSIFIED_INPUT); | ||||
|
cin
|
r177 | } | ||
| protected void DefineAlphabet(IEnumerable<TSymbol> alphabet) { | ||||
| Safe.ArgumentNotNull(alphabet, "alphabet"); | ||||
| foreach (var ch in alphabet) | ||||
| AlphabetBuilder.DefineSymbol(ch); | ||||
| } | ||||
|
cin
|
r178 | protected Token SymbolToken(TSymbol symbol) { | ||
| return Token.New(TranslateOrAdd(symbol)); | ||||
|
cin
|
r177 | } | ||
|
cin
|
r178 | protected Token SymbolToken(IEnumerable<TSymbol> symbols) { | ||
|
cin
|
r177 | Safe.ArgumentNotNull(symbols, "symbols"); | ||
|
cin
|
r178 | return Token.New(TranslateOrAdd(symbols).ToArray()); | ||
|
cin
|
r177 | } | ||
|
cin
|
r178 | protected Token SymbolSetToken(params TSymbol[] set) { | ||
|
cin
|
r177 | return SymbolToken(set); | ||
| } | ||||
| int TranslateOrAdd(TSymbol ch) { | ||||
| var t = AlphabetBuilder.Translate(ch); | ||||
|
cin
|
r178 | if (t == AutomatonConst.UNCLASSIFIED_INPUT) | ||
|
cin
|
r177 | 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); | ||||
|
cin
|
r178 | if (t == AutomatonConst.UNCLASSIFIED_INPUT) | ||
|
cin
|
r177 | throw new ApplicationException(String.Format("Symbol '{0}' is UNCLASSIFIED", ch)); | ||
| return t; | ||||
| } | ||||
| IEnumerable<int> TranslateOrDie(IEnumerable<TSymbol> symbols) { | ||||
| return symbols.Distinct().Select(TranslateOrDie); | ||||
| } | ||||
|
cin
|
r178 | protected Token SymbolTokenExcept(IEnumerable<TSymbol> symbols) { | ||
|
cin
|
r177 | Safe.ArgumentNotNull(symbols, "symbols"); | ||
|
cin
|
r178 | return Token.New( Enumerable.Range(0, AlphabetBuilder.Count).Except(TranslateOrDie(symbols)).ToArray() ); | ||
|
cin
|
r177 | } | ||
| protected abstract IndexedAlphabetBase<TSymbol> CreateAlphabet(); | ||||
|
cin
|
r178 | protected ScannerContext<TTag> BuildScannerContext<TTag>(Token regexp) { | ||
|
cin
|
r177 | |||
| var dfa = new RegularDFA<TSymbol, TTag>(AlphabetBuilder); | ||||
|
cin
|
r178 | var visitor = new RegularExpressionVisitor<TTag>(dfa); | ||
| regexp.Accept(visitor); | ||||
| visitor.BuildDFA(); | ||||
|
cin
|
r177 | |||
| if (dfa.IsFinalState(dfa.InitialState)) | ||||
| throw new ApplicationException("The specified language contains empty token"); | ||||
| var ab = CreateAlphabet(); | ||||
| var optimal = dfa.Optimize(ab); | ||||
| return new ScannerContext<TTag>( | ||||
| optimal.CreateTransitionTable(), | ||||
| optimal.CreateFinalStateTable(), | ||||
| optimal.CreateTagTable(), | ||||
| optimal.InitialState, | ||||
| ab.GetTranslationMap() | ||||
| ); | ||||
| } | ||||
| } | ||||
| } | ||||