|
55
|
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 public abstract class AlphabetBase<T> : IAlphabet<T> {
|
|
|
11 public const int UNCLASSIFIED = 0;
|
|
|
12
|
|
|
13 int m_nextId = 1;
|
|
|
14 int[] m_map;
|
|
|
15
|
|
|
16 public int Count {
|
|
|
17 get { return m_nextId; }
|
|
|
18 }
|
|
|
19
|
|
59
|
20 protected AlphabetBase(int mapSize) {
|
|
|
21 m_map = new int[mapSize];
|
|
55
|
22 }
|
|
|
23
|
|
|
24 protected AlphabetBase(int[] map) {
|
|
|
25 Debug.Assert(map != null);
|
|
|
26
|
|
|
27 m_map = map;
|
|
|
28 m_nextId = map.Max() + 1;
|
|
|
29 }
|
|
|
30
|
|
|
31 public int DefineSymbol(T symbol) {
|
|
|
32 var index = GetSymbolIndex(symbol);
|
|
|
33 if (m_map[index] == UNCLASSIFIED)
|
|
|
34 m_map[index] = m_nextId++;
|
|
|
35 return m_map[index];
|
|
|
36 }
|
|
|
37
|
|
|
38 public int DefineClass(IEnumerable<T> symbols) {
|
|
|
39 Safe.ArgumentNotNull(symbols, "symbols");
|
|
|
40 symbols = symbols.Distinct();
|
|
|
41
|
|
|
42 foreach (var symbol in symbols) {
|
|
|
43 var index = GetSymbolIndex(symbol);
|
|
|
44 if (m_map[index] == UNCLASSIFIED)
|
|
|
45 m_map[GetSymbolIndex(symbol)] = m_nextId;
|
|
|
46 else
|
|
|
47 throw new InvalidOperationException(String.Format("Symbol '{0}' already in use", symbol));
|
|
|
48 }
|
|
|
49 return m_nextId++;
|
|
|
50 }
|
|
|
51
|
|
|
52 public List<T>[] CreateReverseMap() {
|
|
|
53 return
|
|
|
54 Enumerable.Range(UNCLASSIFIED, Count)
|
|
|
55 .Select(
|
|
|
56 i => InputSymbols
|
|
|
57 .Where(x => i != UNCLASSIFIED && m_map[GetSymbolIndex(x)] == i)
|
|
|
58 .ToList()
|
|
|
59 )
|
|
|
60 .ToArray();
|
|
|
61 }
|
|
|
62
|
|
|
63 public int[] Reclassify(IAlphabet<T> newAlphabet, IEnumerable<ICollection<int>> classes) {
|
|
|
64 Safe.ArgumentNotNull(newAlphabet, "newAlphabet");
|
|
|
65 Safe.ArgumentNotNull(classes, "classes");
|
|
|
66 var reverseMap = CreateReverseMap();
|
|
|
67
|
|
|
68 int[] translationMap = new int[Count];
|
|
|
69
|
|
|
70 foreach (var scl in classes) {
|
|
|
71 // skip if the supper class contains the unclassified element
|
|
|
72 if (scl.Contains(UNCLASSIFIED))
|
|
|
73 continue;
|
|
|
74 var range = new List<T>();
|
|
|
75 foreach (var cl in scl) {
|
|
|
76 if (cl < 0 || cl >= reverseMap.Length)
|
|
|
77 throw new ArgumentOutOfRangeException(String.Format("Class {0} is not valid for the current alphabet", cl));
|
|
|
78 range.AddRange(reverseMap[cl]);
|
|
|
79 }
|
|
|
80 var newClass = newAlphabet.DefineClass(range);
|
|
|
81 foreach (var cl in scl)
|
|
|
82 translationMap[cl] = newClass;
|
|
|
83 }
|
|
|
84
|
|
|
85 return translationMap;
|
|
|
86 }
|
|
|
87
|
|
|
88 public int Translate(T symbol) {
|
|
|
89 return m_map[GetSymbolIndex(symbol)];
|
|
|
90 }
|
|
|
91
|
|
|
92 public abstract int GetSymbolIndex(T symbol);
|
|
|
93
|
|
|
94 public abstract IEnumerable<T> InputSymbols { get; }
|
|
|
95
|
|
|
96 public int[] GetTranslationMap() {
|
|
|
97 return m_map;
|
|
|
98 }
|
|
|
99 }
|
|
|
100 }
|
