IndexedAlphabetBase.cs
50 lines
| 1.8 KiB
| text/x-csharp
|
CSharpLexer
cin
|
r289 | using Implab; | ||
| using System; | ||||
| using System.Collections.Generic; | ||||
| using System.Diagnostics; | ||||
| using System.Linq; | ||||
| namespace Implab.Automaton { | ||||
| /// <summary> | ||||
| /// Indexed alphabet is the finite set of symbols where each symbol has a zero-based unique index. | ||||
| /// </summary> | ||||
| /// <remarks> | ||||
| /// Indexed alphabets are usefull in bulting efficient translations from source alphabet | ||||
| /// to the input alphabet of the automaton. It's assumed that the index to the symbol match | ||||
| /// is well known and documented. | ||||
| /// </remarks> | ||||
| public abstract class IndexedAlphabetBase<T> : MapAlphabet<T> { | ||||
| protected IndexedAlphabetBase() :base(true, null) { | ||||
| } | ||||
| public abstract int GetSymbolIndex(T symbol); | ||||
| /// <summary> | ||||
| /// Gets the translation map from the index of the symbol to it's class this is usefull for the optimized input symbols transtaion. | ||||
| /// </summary> | ||||
| /// <remarks> | ||||
| /// The map is continous and start from the symbol with zero code. The last symbol | ||||
| /// in the map is the last classified symbol in the alphabet, i.e. the map can be | ||||
| /// shorter then the whole alphabet. | ||||
| /// </remarks> | ||||
| /// <returns>The translation map.</returns> | ||||
| public int[] GetTranslationMap() { | ||||
| var map = new Dictionary<int, int>(); | ||||
| int max = 0; | ||||
| foreach (var p in Mappings) { | ||||
| var index = GetSymbolIndex(p.Key); | ||||
| max = Math.Max(max, index); | ||||
| map[index] = p.Value; | ||||
| } | ||||
| var result = new int[max + 1]; | ||||
| for (int i = 0; i < result.Length; i++) | ||||
| map.TryGetValue(i, out result[i]); | ||||
| return result; | ||||
| } | ||||
| } | ||||
| } | ||||