|
158
|
1 using Implab;
|
|
|
2 using System;
|
|
|
3 using System.Collections.Generic;
|
|
|
4 using System.Diagnostics;
|
|
|
5 using System.Linq;
|
|
|
6 using System.Text;
|
|
|
7 using System.Threading.Tasks;
|
|
|
8
|
|
|
9 namespace Implab.Parsing {
|
|
|
10 /// <summary>
|
|
|
11 /// Indexed alphabet is the finite set of symbols where each symbol has a zero-based unique index.
|
|
|
12 /// </summary>
|
|
|
13 public abstract class IndexedAlphabetBase<T> : IAlphabet<T> {
|
|
|
14 public const int UNCLASSIFIED = 0;
|
|
|
15
|
|
|
16 int m_nextId = 1;
|
|
|
17 readonly int[] m_map;
|
|
|
18
|
|
|
19 public int Count {
|
|
|
20 get { return m_nextId; }
|
|
|
21 }
|
|
|
22
|
|
|
23 protected IndexedAlphabetBase(int mapSize) {
|
|
|
24 m_map = new int[mapSize];
|
|
|
25 }
|
|
|
26
|
|
|
27 protected IndexedAlphabetBase(int[] map) {
|
|
|
28 Debug.Assert(map != null);
|
|
|
29
|
|
|
30 m_map = map;
|
|
|
31 m_nextId = map.Max() + 1;
|
|
|
32 }
|
|
|
33
|
|
|
34 public int DefineSymbol(T symbol) {
|
|
|
35 var index = GetSymbolIndex(symbol);
|
|
|
36 if (m_map[index] == UNCLASSIFIED)
|
|
|
37 m_map[index] = m_nextId++;
|
|
|
38 return m_map[index];
|
|
|
39 }
|
|
|
40
|
|
|
41 public int DefineClass(IEnumerable<T> symbols) {
|
|
|
42 Safe.ArgumentNotNull(symbols, "symbols");
|
|
|
43 symbols = symbols.Distinct();
|
|
|
44
|
|
|
45 foreach (var symbol in symbols) {
|
|
|
46 var index = GetSymbolIndex(symbol);
|
|
|
47 if (m_map[index] == UNCLASSIFIED)
|
|
|
48 m_map[GetSymbolIndex(symbol)] = m_nextId;
|
|
|
49 else
|
|
|
50 throw new InvalidOperationException(String.Format("Symbol '{0}' already in use", symbol));
|
|
|
51 }
|
|
|
52 return m_nextId++;
|
|
|
53 }
|
|
|
54
|
|
|
55 public List<T>[] CreateReverseMap() {
|
|
|
56 return
|
|
|
57 Enumerable.Range(UNCLASSIFIED, Count)
|
|
|
58 .Select(
|
|
|
59 i => InputSymbols
|
|
|
60 .Where(x => i != UNCLASSIFIED && m_map[GetSymbolIndex(x)] == i)
|
|
|
61 .ToList()
|
|
|
62 )
|
|
|
63 .ToArray();
|
|
|
64 }
|
|
|
65
|
|
|
66 public int[] Reclassify(IAlphabet<T> newAlphabet, IEnumerable<ICollection<int>> classes) {
|
|
|
67 Safe.ArgumentNotNull(newAlphabet, "newAlphabet");
|
|
|
68 Safe.ArgumentNotNull(classes, "classes");
|
|
|
69 var reverseMap = CreateReverseMap();
|
|
|
70
|
|
|
71 int[] translationMap = new int[Count];
|
|
|
72
|
|
|
73 foreach (var scl in classes) {
|
|
|
74 // skip if the supper class contains the unclassified element
|
|
|
75 if (scl.Contains(UNCLASSIFIED))
|
|
|
76 continue;
|
|
|
77 var range = new List<T>();
|
|
|
78 foreach (var cl in scl) {
|
|
|
79 if (cl < 0 || cl >= reverseMap.Length)
|
|
|
80 throw new ArgumentOutOfRangeException(String.Format("Class {0} is not valid for the current alphabet", cl));
|
|
|
81 range.AddRange(reverseMap[cl]);
|
|
|
82 }
|
|
|
83 var newClass = newAlphabet.DefineClass(range);
|
|
|
84 foreach (var cl in scl)
|
|
|
85 translationMap[cl] = newClass;
|
|
|
86 }
|
|
|
87
|
|
|
88 return translationMap;
|
|
|
89 }
|
|
|
90
|
|
|
91 public virtual int Translate(T symbol) {
|
|
|
92 return m_map[GetSymbolIndex(symbol)];
|
|
|
93 }
|
|
|
94
|
|
|
95 public abstract int GetSymbolIndex(T symbol);
|
|
|
96
|
|
|
97 public abstract IEnumerable<T> InputSymbols { get; }
|
|
|
98
|
|
|
99 /// <summary>
|
|
|
100 /// Gets the translation map from the index of the symbol to it's class this is usefull for the optimized input symbols transtaion.
|
|
|
101 /// </summary>
|
|
|
102 /// <returns>The translation map.</returns>
|
|
|
103 public int[] GetTranslationMap() {
|
|
|
104 return m_map;
|
|
|
105 }
|
|
|
106 }
|
|
|
107 }
|
